See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/326669076

Science, Mind and the Universe: An Introduction to Natural Philosophy

Book · July 2018

CITATIONS READS

0 1,622

1 author:

Gerald Ulrich
Charité Universitätsmedizin Berlin
221 PUBLICATIONS   1,330 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

The epistemologic foundations of neurosciences View project

The two domains of electroencephalography View project

All content following this page was uploaded by Gerald Ulrich on 30 July 2018.

The user has requested enhancement of the downloaded file.

To my colleagues from

IAG
IUGG
ICSU
ÖAW

in gratitude
Preface
Schuster, bleib bei deinem Leisten!
German proverb

“The rabbi spoke three times. The first talk was brilliant: clear and
simple. I understood every word. The second was even better: deep
and subtle. I didn’t understand much, but the rabbi understood all of
it. The third was by far the finest: a great and unforgettable experience.
I understood nothing, and the rabbi didn’t understand much either.”
This was one of Niels Bohr’s favorite anecdotes (Folse, 1985, p. 258).
All books on philosophy belong to one of the three types of the
rabbi’s talks. In an introductory text like the present book, the reader
is entitled to expect that it belongs to “Rabbi Type 1”. This is at least
what we have tried to achieve.
Our higher polytechnical schools have frequently become “Technical
Universities” or “Universities of Technology”. This implies that they
intend not only to give a profound scientific or professional education,
but also to offer a touch of “universality”. Now, the common interdisci-
plinary background of all scientific, engineering and medical disciplines
is becoming increasingly “philosophical”.
On all aspects of philosophy of mathematics and natural science
there are excellent and even brilliant monographs. The intention of the
present book is much more modest: to provide an absolutely introduc-
tory yet rather systematic and comprehensive textbook taylored to the
interests of students of science, technology, and medicine.
This book is written not by a professional philosopher, but by a
practicing scientist. The lack of philosophical depth is thus perhaps
partly made up by a knowledge of the mentality and interests of stu-
dents of scientific disciplines and by active research experience in sci-
ence, in which interdisciplinary and philosophical questions are arising
naturally and to an ever increasing extent.

vii
viii

The author’s philosophical interests date back to the age of 17

years, when he happened to come across the brilliant book “An Out-
line of Philosophy” by Bertrand Russell (Russell 1927). Later his inter-
ests changed from Russell to his more profound and more many–sided
colleague Alfred North Whitehead. To me there is only one modern
philosopher of comparable depth and universality: Carl Friedrich von
Weizsäcker.
In 1974 and 1975 I gave a course on the philosophy of mathematics
and natural sciences at the Graz University of Technology. Later my
time was absorbed by other duties but my interest remained, and finally
I decided to resume this course again and to write the present book.
The scheme of the book may be illustrated by the following diagram
showing the “sign posts” of it in a somewhat provocative way:
Plato

?
Kant

?
Hegel
@
@
@
Kurt Gödel Niels Bohr
To avoid misunderstandings: the names stand for certain directions
of thinking and have no direct relation to “greatness”. (Otherwise one
might ask: why Plato and not Aristotle, why Bohr and not Einstein or
Heisenberg?)
Only the first three are “professional” philosophers. Gödel is the
greatest of modern logicians, and Bohr, through his “principle of com-
plementarity”, is the physicist who has made dialectics respectable in
the philosophy of natural science. When he was awarded a title of nobil-
ity for his scientific merits, he chose for his coat of arms the inscription
“Contraria sunt complementa ”.
Dialectics has frequently fallen into discredit by its dogmatic use
and misuse, and also Hegel has suffered from this circumstance. Nev-
ertheless dialectic thinking does have its use in natural philosophy, and
 ix

the justly famous bestseller “Gödel, Escher, Bach” by D.R. Hofstadter

is thoroughly dialectical although here dialectics is discreetly disguised
under the term of “paradoxes of self–reference”.
It is also remarkable that it is rare indeed that all three names form-
ing the bottom triangle of the figure are quoted simultaneously in the
literature: philosophers speak of Hegel, mathematical logicians quote
Gödel, and physicists refer to Bohr. Practically the only exception are
recent books by C.F. von Weizsäcker and, on a similarly high level, the
work “Filozofija znanosti i humanizam ” by Ivan Supek (SNL, Zagreb,
1991).
Emphasis is also on a basic element of uncertainty, which seems
to pervade nature and our knowledge of it. It is expressed by Gödel’s
incompleteness theorem, by fuzzy logic, by Heisenberg’s uncertainty re-
lation, by other random fluctuations and random measuring errors, etc.,
which have fascinated the imagination of mathematicians, physicists,
astronomers, and geodesists since C.F. Gauss.
The main intention of the book is didactic simplicity. Another pur-
pose is a more or less systematic unification of trends of philosophy of
science in Anglo–Saxon countries and in Central Europe. I believe that
not only Kant, but also some other German philosophers (among them
Fichte, Hegel, and Weizsäcker) have made contributions to our field
which deserve being known also in English–speaking countries. This is
one of the several reasons for writing this text in English.
Standard books on the philosophy of science (in German: Wis-
senschaftstheorie) such as (Carnap 1966), (Hempel 1966) or (Popper
1977) chiefly deal with the logical and methodological basis common to
all natural sciences, such as axiomatics and induction. The present text
treats such problems also, but more briefly. Emphasis is on general re-
sults of modern science (e.g. quantum theory), their interpretation and
their implications for standard philosophical problems, such as the re-
lation between matter and mind, the meaning of a law of nature, free
will etc.
The term “natural philosophy” has a proud English tradition:
physicists from Isaac Newton to W. Thomson (Lord Kelvin) used it
for what we would today call “theoretical physics with a philosoph-
ical touch”. The word “Naturphilosophie” was used in Germany,
by Goethe, Schelling, and Hegel, in quite a different, philosophical–
speculative sense. Again, the present work seeks to reflect something
of both connotations. Here, “natural philosophy” is used in very much
the same sense as in Whitrow’s well–known book “The Natural Philos-
ophy of Time” (Whitrow 1980).
x

The book should be comprehensible not only to undergraduate

students of mathematics, physics, natural science, technology, and
medicine, but also to a broader group of readers interested in philo-
sophical implications of modern science. Mathematical formulas are
kept to a minimum: while possibly of help to the mathematically versed
reader, they can be bypassed without serious problems. Similarly, the
references to selected additional works may provide further informa-
tion; they should, however, not be necessary to a basic understanding
of the present book, which is essentially self–contained.
For didactic purposes even a rigorous logical structure has some-
times been sacrificed: important concepts such as materialism and ide-
alism have been discussed, to increasing detail and on increasing level,
in several places of the book. Generally, redundancies and repetitions
are not avoided if they improve the readability. In this way, the book
resembles an informal university lecture.
Since this is an introductory book, it cannot be very profound. It
probably also contains quite a few errors. I have, however, tried to make
it interesting reading: it would be a crime to treat such a fascinating
subject, one of the greatest adventures of the human spirit, in a dull
manner.
Acknowledgments. Writing such a general book requires, but also
finds, much more interest and encouragement than a technical mono-
graph. This book owes its existence to an initial impulse provided by
Christian Poitevin (Brussels) and Georges Balmino (Toulouse), and to
the sponanteous enthusiasm of Heinz Draheim (Karlsruhe).
The publisher of my previous technical books, Dr. Christoph
Müller–Wirth of Wichmann Verlag (Karlsruhe), eagerly took up the
idea of this book and followed the different stages of its realiza-
tion with his interest and advice. Several colleagues from vari-
ous fields of science and philosophy kindly agreed to critically read
the manuscript: L. Jonathan Cohen (Oxford), Franz Moser (Graz),
Hellmuth Petsche (Wien), Wilfried Schröder (Bremen), Hans Georg
Schwarzacher (Wien), Ivan Supek (Zagreb), and Hans–Jürgen Treder
(Potsdam–Babelsberg). Their remarks helped remove some major er-
rors and many minor mistakes and even misprints, besides providing
additional information and inspiration. Their criticism and encourage-
ment were absolutely essential. Of course I bear the full responsibility
for the contents of the book and for any remaining errors.
Dr. Konrad Rautz (Graz) was of invaluable help in editing: he drew
the figures, composed the index, helped in proofreading, and provided
advice in difficult questions of word–processing. My secretary Mrs.
 xi

Ruth Hödl did all the word–processing in her usual efficient, dedicated,
and painstaking way. Last but not least, my wife Gerlinde read various
versions of the manuscript and was my adviser in questions of biology
and theology, besides confirming that the book can be read also without
mathematics. All this help is gratefully acknowledged.

Graz, February 1995

Helmut Moritz
xii
Contents

Preface vii

A Human Perception and Thinking 1

1 The human brain 3
1.1 Brain and nervous system . . . . . . . . . . . . . . . . . 3
1.2 Brain and mind . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Human perception . . . . . . . . . . . . . . . . . . . . . 14
1.4 Evolutionary theory of knowledge . . . . . . . . . . . . . 17

2 Logic and mathematics 23

2.1 Elements of symbolic logic . . . . . . . . . . . . . . . . . 23
2.2 The axiomatic method . . . . . . . . . . . . . . . . . . . 32
2.3 Logical paradoxes and Gödel’s theorem . . . . . . . . . . 34
2.4 Inexact concepts, “fuzzy logic” . . . . . . . . . . . . . . . 38
2.5 Dialectic thinking . . . . . . . . . . . . . . . . . . . . . . 44
2.6 Geometry: dimensions two to infinity . . . . . . . . . . . 59

B Natural Science 71
3 Physics 73
3.1 Classical mechanics and determinism . . . . . . . . . . . 73
3.2 Deterministic chaos . . . . . . . . . . . . . . . . . . . . . 80
3.3 Probability . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.4 The theory of relativity . . . . . . . . . . . . . . . . . . . 90
3.5 Quantum theory . . . . . . . . . . . . . . . . . . . . . . 99
3.6 Elementary particles . . . . . . . . . . . . . . . . . . . . 111
3.7 Space and time; cosmology . . . . . . . . . . . . . . . . . 120

xiii
xiv CONTENTS

3.8 Inverse problems . . . . . . . . . . . . . . . . . . . . . . 134

3.9 Induction, verification, falsification . . . . . . . . . . . . 145
3.10 The structure of scientific revolutions . . . . . . . . . . . 155

4 Systems, information, evolution 159

4.1 Feedback, regulation, downward causation . . . . . . . . 159
4.2 Self–organization . . . . . . . . . . . . . . . . . . . . . . 166
4.3 Entropy, information, evolution . . . . . . . . . . . . . . 173
4.4 Data and errors . . . . . . . . . . . . . . . . . . . . . . . 183
4.5 Complexity and reductionism . . . . . . . . . . . . . . . 186

C Philosophy 195
5 Philosophy for scientists 197
5.1 Realism, idealism and dualism . . . . . . . . . . . . . . . 197
5.2 The three–world model . . . . . . . . . . . . . . . . . . . 207
5.3 Subject and object . . . . . . . . . . . . . . . . . . . . . 214
5.4 Historical landmarks . . . . . . . . . . . . . . . . . . . . 225

6 Implications of science 237

6.1 Matter and mind . . . . . . . . . . . . . . . . . . . . . . 237
6.2 Materialism, idealism and the outer world . . . . . . . . 239
6.3 Time, creativity, block universe . . . . . . . . . . . . . . 240
6.4 Freedom of the will . . . . . . . . . . . . . . . . . . . . . 245
6.5 Laws of nature . . . . . . . . . . . . . . . . . . . . . . . 251
6.6 Theories of everything . . . . . . . . . . . . . . . . . . . 263
6.7 The Absolute . . . . . . . . . . . . . . . . . . . . . . . . 269
6.8 Pluralism . . . . . . . . . . . . . . . . . . . . . . . . . . 274

Selected additional reading 277

Index 285
 Part A
Human Perception and
Thinking
Chapter 1
The human brain

1.1 The brain and the nervous system

Philosophers, use your brain!
J.Z. Young

If there is an agreement about some fact in natural philosophy, it

is about the fact that human thinking is inseparably related to our
brain. It seems therefore appropriate to start lectures about natural
philosophy with a review of some basic knowledge about the structure
of our brain.
Very much is known in brain research, but in view of the huge com-
plexity of the topic, it is still relatively little. Our present treatment
further simplifies our knowledge to a few basic items which one should
know in order to speak meaningfully about some topics in natural phi-
losophy.
That the physiological or anatomical structure of our brain is at all
relevant to philosophy, is not generally accepted in philosophy. The
laws of logic and mathematics are so much more rigorous than our
mainly empirical knowledge about brain structure, and these “laws of
thinking” do not seem to depend very much on the architecture of
our brain. Thus books on philosophy do not generally start with the
physiology of the brain. It seems, however, that modern trends in the
theory of knowledge, known by the name of evolutionary epistemology,
do depend on mechanisms of perception (sight, hearing, etc.) and on
the processing of the data of perception by the brain. The same holds
for artificial intelligence and for other advanced aspects of automatic

3
4 CHAPTER 1. THE HUMAN BRAIN

computation which, though not directly the subject of the book, do

have some points of contact, such as algorithmic and axiomatic methods
and their limitation by Gödel’s theorem.
A brief introduction will be sufficient for our present purpose, how-
ever. Additional literature is abundant and may be consulted in case
of interest. On the other hand, the reader should not be bothered by
details which may confuse someone who is not used to think in terms
of neuroscience. For a first reading it is suggested that the reader does
not stop at some unusual details but skips them and continues to read
in a cursory fashion.
This is generally recommended for reading the present book (and
philosophical works in general). Don’t get stuck at the first difficulty
but continue reading. When you are through with the book, start again
from the beginning: many difficulties will be clear now. If necessary,
iterate the procedure, concentrating on the passages which you have
found difficult. Never try to conquer a difficulty by force, but rather
by patience.

The architecture of the brain

Main parts of the brain are
– the brainstem (“reptilian brain”),
– the limbic system (“mammalian brain”) and
– the cerebrum (“conscious brain”).

Figure 1.1: Main parts of the brain

The brainstem, from the point of view of biological evolution, is the

oldest part of the brain (some 500 million years). With characteristic
1.1. BRAIN AND NERVOUS SYSTEM 5

oversimplification, it is called reptilian brain because it looks somewhat

like the entire brain of a reptile. It is mainly concerned with the most
primitive functions of life support: control of breathing and heartbeat.
The cerebellum (Fig. 1.1) is an intermediate structure concerned
with coordinating the activity of the muscles and maintaining bodily
equilibrium (cf. sec. 4.1). It seems to be responsible for largely uncon-
scious skills such as automobile driving or piano playing.
The limbic system is a ringlike structure which was the highest
component in the brain of a reptile but is most highly developed in
mammals. It regulates body temperature, blood pressure, etc. (“home-
ostasis”). It controls emotional reactions necessary for survival: ap-
petite, agressiveness, flight reactions, and sexual feelings. It also seems
to be basic for learning and memory (the so–called hippocampus).
The most important structure of the limbic system is the hypotha-
lamus (located under the thalamus, Fig. 1.1). It is the regulator for
body functions and states: eating, drinking, sleeping, waking, chemical
balances, hormones, etc. It is the body’s thermostat for blood temper-
ature and acts by feedback, as also technological regulators do (see sec.
4.1).
The hypothalamus is intimately connected (by chemical and electri-
cal messages) to the adjacent pituitary gland. This is the master gland
of the body which regulates it by hormones, directly or by stimulating
other glands to emit hormones.
The thalamus (Fig. 1.1) serves as a central relay station for external
sensory, especially visual, information. This information is then relayed
to certain areas of the cortex (see below) for final processing.
The cerebrum is the part of the brain which is specifically related
to human intelligence. It consists of two hemispheres, the left and the
right hemisphere, which are connected by the corpus callosum. There
is a curious cross–connection: the left hemisphere receives information
from the right half of the body (e.g., the right eye), and also controls
the right half of the body (e.g., the right hand). Similarly, the right
hemisphere is related to the left side of the body.
Generally speaking and oversimplifying, the left hemisphere mainly
takes care of logical and analytic thinking and language, whereas the
right hemisphere is chiefly responsible for intuitive, “synthetic”, activi-
ties such as geometric intuition and music. Both hemispheres, however,
closely cooperate and form “one system”.
Most activities of the cerebrum go on at its surface, the cortex. The
cortex is about 3 mm thick and is intricately wrinkled and folded, so
that its huge surface fits into a relatively small skull.
6 CHAPTER 1. THE HUMAN BRAIN

Certain areas of the cortex are related to certain activities: the

primary visual area, the auditory cortex, the body sensory areas, the
motor cortex, and the speech areas (Broca’s and Wernicke’s areas).
These areas, however, are by no means to be taken in a strict and
exclusive sense; in a way, all activities affect the whole brain.
Corresponding to its function, the cortex is densely packed with
nerve cells (neurons) whose axes (axons, see below) are orthogonal to
the surface of the cortex , so that we have a structure reminiscent of
level surfaces and plumb lines in geodesy. Rational thinking goes on
mainly in the cortex; the limbic system (hypothalamus etc.) seems to
contribute the emotional background.

Neurons
Each brain contains about 1011 nerve cells or neurons. Fig. 1.2
shows, very schematically, a typical nerve cell. The synaptic buttons
(some 1014 !) attach to dendrites or somas of neighboring cells. There is
a tiny “synaptic cleft” between a button and the next dendrite, which is
bridged by chemicals (neurotransmitters ) sent from button to dendrite.

Figure 1.2: A typical neuron

The average length of neuronic fibers in the cortex is 4.1 km/mm 3 !

How is the nerve signal transmitted along the axon? It by no
means acts like a copper wire, say; rather, the transmission resembles
a transversal seismic wave! A transversal wave oscillates in a direc-
tion orthogonal to the direction of propagation. Similarly, the nerve
signal is a progressing impulse of diffusion of certain ions (K+ , Na+ ,
Cl− ) across the tube–like membrane forming the axon. The rather
complicated details of the process are without relevance to the present
context.
1.1. BRAIN AND NERVOUS SYSTEM 7

The normal speed is only around 5 meters per second; the speed
can be increased by a myelin sheath of the axon up to 100 m/s. Again,
the reader is referred to the literature for details.
Important is the following fact: the “firing” of an impulse is all or
none. The shape and size of the impulse is irrelevant; important is
only whether the neuron fires or not. Synapses (consisting of synaptic
buttons and synaptic clefts) may be excitatory or inhibitory. Their
combined effect on the neuron determines whether it fires or not. The
system: all (1) or nothing (0) is essentially digital, using a dual number
system (consisting of zero’s and one’s) which is not unlike the dual
system used by a digital computer. Very simple examples may be found
in (Penrose 1989, pp. 393–394).
Each historical period compares the human body to its favorite ma-
chines: formerly it was a mechanical or heat engine, with a pump for
the heart. Now it is fashionable to compare the brain to a huge parallel–
processing digital computer. (There seems to be some “analogue com-
putation” too, but on a minor scale, by “non–spiking neurons”.) An
important difference to present computers is a high redundancy: if part
of the “brain computer” fails, other parts may take over.
Assumptions about the brain computation have lead to the con-
struction of artificial neural networks, which, together with more con-
ventional computing devices, play a considerable role in artificial intel-
ligence (AI).
H. Petsche (Vienna) pointed out to me that the “digital firing” (all
or none) mentioned above does not imply that the brain works exactly
like a digital computer. From (Cohen and Stewart 1994, p. 454) we
quote:
Nerve cells don’t “compute” digitally, but they do use discrete pulses to
communicate over large distances.
That “brain computation” is not really “digital”. For a detailed
treatment of “neural computing” cf. (Kohonen 1988, Chapter 9: a
very readable non–mathematical chapter in a otherwise rather techni-
cal book). Also the discussion in (Penrose 1989, pp. 392–399) is highly
relevant.
Let us also not forget that “computation” or logical thinking is
only a tiny fraction of our intellectual activity which also includes joys,
desires, fears, poetry, music, etc., cf. end of sec. 2.1.
Suggested additional reading. The interesting reader may start with
the didactic masterpiece (Ornstein and Thompson 1984) and continue
with more technical books such as (Eccles 1973) or (Thompson 1985).
8 CHAPTER 1. THE HUMAN BRAIN

More advanced is (Oeser and Seitelberger 1988). The brain as a com-

puter is treated very clearly in the introductory book (Churchland 1988)
where also the elements of AI and artificial neural networks may be
found. A profound study of AI (and many other things) can be found
in the classics (Hofstadter 1979) and (Penrose 1989).

1.2 Brain and mind

The ghost in the machine.
Gilbert Ryle

The relation between brain and mind is one of the most controversial
problems in philosophy. It is closely related to the problem of matter
and mind.
At a first glance, there seems to be no problem at all. Our thoughts,
feelings, and emotions are clearly mental , belonging to the mind. A
tree, a house, or a stone are clearly material , consisting of matter: we
can kick a stone, live in a house, or fall from a tree, breaking our leg.
What could be more material?
Modern science gives a rather different picture of matter. It consists
of extremely small molecules, separated by large distances. Molecules
consist of atoms, and the atoms have a tiny nucleus with electrons or-
biting around them like planets orbiting around the sun, and separated
by similarly huge (in a relative sense) distances. So matter essentially
consists of empty space. If my foot kicks a stone, an empty space kicks
another empty space (to be honest, what matters are the forces that
act in these “empty spaces”).
If modern science makes matter less “material”, it also makes mind
less “mental”, as the philosopher Bertrand Russell said. The study of
our brain has shown us that, e.g., emotions are related to the activity
of the (material!) limbic system, and even the most sublime manifes-
tations of human thinking seem to correspond to the firing of certain
neurons in the cortex.
A fair and didactically useful comparison between brain and mind
seems to be possible by modern computer terminology: the brain is
the hardware, the mind is the software of human thinking. The same
software can be run on different computers, and the ideas contained in
the present book (supposing that it contains any) can be understood
by different readers, that is, by different brains.
1.2. BRAIN AND MIND 9

This analogy: brain = ˆ hardware, mind = ˆ software, seems to be

reasonably appropriate if we bear in mind that it is a crude picture
only which should not be taken too seriously. Difficulties start when
we ask for the “ontological status” of mind versus matter: is the men-
tal software “essentially different” from the “brain hardware”? Some
philosophers think that such questions make no sense, others think dif-
ferently. Is the mind a “substance”? Dear reader, don’t ask me what
a substance is. Webster’s New Collegiate Dictionary gives various def-
initions, e.g., “that which constitutes anything what it is”. Do you
understand this? I don’t.
More seriously, a substance may be regarded as something that
can exist independently of everything else, or at least independently of
other “substances”. In this sense, if mind, or “soul”, is a substance,
then it can exist independently of matter, and a consequence could be
the immortality of the soul !
Dear reader, you will see how these questions are burdened with
emotions. Far from being meaningless or purely theoretical phantoms,
they have inspired noble actions of unselfish humanity as well as caused
the most abominable wars.
But let us return to science. In a way, mind seems to emerge from
the activity of the brain. On a lower level, life may be an emergent
property of matter. “Emergent” means something new, qualitatively
different. Here we might have an agreement.
If lifeless matter becomes sufficiently highly organized (containing
complex organic molecules), then life may emerge, and if living brain
tissues are particularly highly organized, the phenomenon of mind may
emerge.
This seems to be acceptable to agnostics, dialectic materialists and
Christians alike. Differences are again to be expected (and found) in
ontological questions: in which way does mind, emerging from matter
(the brain tissue), differ from matter? Is it only an “outgrowth” of
matter or does it have an independent reality? (“Emergence” only
implies that mind is “something new”.)
Let us summarize the points which could be agreed upon by a ma-
jority of scientists, philosophers, and theologians:

(1) The mind is related to the brain somewhat as software is related

to hardware; see above.

(2) Mind is a new aspect of brain activity that emerged when biolog-
ical evolution led to the genus “man”.
10 CHAPTER 1. THE HUMAN BRAIN

These formulations are deliberately vague so as to be widely accept-

able (of course, we take biological evolution for granted).
So far, we have used the adjective “ontological” twice without defin-
ing ontology. Here Webster’s definition is excellent:
“the science of being or reality; the branch of knowledge that investigates
the nature, essential properties, and relations of being.”
If you don’t understand this, learn its use in the same way as a
child learns to use new words: by (largely unconscious) practice. Just
go ahead reading!

Theories of mind and brain

– Materialism: only material processes exist, mind is just the “sub-

jective side” of brain processes but has no reality whatsoever. All
“living” and “mental” processes can be reduced to changes in
matter fully determined by the laws of physics (reductionism ).
– Epiphenomenalism: what really goes on, are material brain pro-
cesses; our subjective experiences are real but logically redundant
(nothing is lost if we forget about mental phenomena).
– Monism (or identity theory ): only one “substance” exists which,
however, expresses itself in two ways: it has two “sides” or as-
pects: mind and matter. Mind and matter, so to speak, are the
two sides of one and the same sheet of paper. This theory has
been proposed by philosophers as different as Baruch Spinoza and
Bertrand Russell. Closely related is
– Panpsychism: everything which is material (an atom or an elec-
tron, say), has also some mental or psychic aspect, however small.
In living beings, and still more, in the human mind, these men-
tal aspects are getting more and more coordinated. This theory
seems quite strange at first glance; it is proposed, however, by
some of the greatest philosophers, such as Leibniz and Alfred
North Whitehead. It is, essentially, also implied in dialectic ma-
terialism, as well as (by definition) in the monism of Spinoza and
Russell.
– Idealism: the opposite of materialism: everything that exists is
an idea in a mind. The tree in front of me exists if I look at it:
then it is in my mind. If I don’t look at it, it does not exist,
1.2. BRAIN AND MIND 11

unless my wife happens to look at it. Since this is a rather queer

world if trees would jump into and out of existence depending on
someone’s looking at them, George Berkeley (1685–1753) discov-
ered a simple reason why the objects of this world continuously
exist: because God is always looking at them. (The reader should
not be misled by the author’s occasional sarcasms which are by
no means expressions of disrespect: Berkeley was a great philoso-
pher!) This view is expressed in a limerick by Ronald Knox,
quoted from (Russell 1945, article on “Berkeley”):
There was a young man who said, “God
Must find it exceedingly odd
If he finds that this tree
Continues to be
When there’s no one about in the Quad.”

REPLY

Dear Sir:
Your astonishment’s odd:
I am always about in the Quad.
And that’s why the tree
Will continue to be,
since observed by
Yours faithfully,
God.

Plato, Kant, Fichte, and Hegel are also considered idealists, al-
though in a somewhat different sense: they regard mind as a
primary concept, and matter a concept derived from mind.
This view is not so far from opinions of some modern physicists
who argue as follows. It is very difficult to define matter directly.
For a physicist it is natural to say that matter is what satisfies the
laws of physics. Now these laws are expressed by mathematical
formulas, which are certainly mental rather than material struc-
tures. Similarly, the entities satisfying mathematical formulas are
mathematical functions, which are equally mental. Hence matter
is a mental construction.
Such an argument can be refuted, as almost all philosophical ar-
guments can. Nevertheless it seems to incorporate at least a spark
of truth. Finally, we shall discuss
12 CHAPTER 1. THE HUMAN BRAIN

– Dualism: matter and mind are essentially different concepts or,

as some philosophers say, different substances. It is certainly the
commonsense view. If I cut my finger (which is a material object),
I feel pain (mentally). Here, clearly, matter acts on mind .
This also works in the opposite direction. If I see an apple in
front of me and wish to eat it (desire is mental), I take and eat
the apple (a material object). Here, so to speak, mind acts on
matter .
Philosophers have wondered how “different substances”, matter
and mind, can act on each other. The first example, a cut in the
finger causing pain, is so common–sense that no one uses to ask
many questions about such a “direct causation”. The opposite,
a mental volition causing a bodily movement (taking the apple)
has given, and is still giving, rise to innumerable discussions on
“downward causation ”, mind acting on matter. We shall consider
various aspects of this problem in sec. 4.1 and elsewhere in this
book.
Dualism in a normal, not extremist, view thus seems to be quite
natural, and we shall use it in this way. It becomes a problem
only if it is exaggerated to mean two “absolutely different” sub-
stances that are “absolutely separate” and hardly able to interact
with each other. This view has been introduced into philosophy
by René Descartes (Cartesius, 1596–1650) who is considered the
founder of modern philosophy, besides having been a great math-
ematician (Cartesian coordinates!).

The danger with philosophy is that natural language is adapted to

everyday use, but it has not been designed for philosophy. So philos-
ophy must use every–day words, making them more precise. If this
alleged precision is over–stretched and the new meaning of an old word
is taken too literally, then we may overdo an argument and come to
exaggerated conclusions. Alfred North Whitehead speaks of the “fal-
lacy of misplaced concreteness ”. This may happen, for instance, if we
consider mind as an indestructible substance, from which immortality
necessarily follows. (I have no objection against immortality, on the
contrary, but it is perhaps not quite so simple.)
Some followers of Descartes used the concept of God to mediate
between the “noninteracting” substances, matter and mind. It seems
that the concept of God is sometimes invoked to serve as a “deus ex
machina ” if a philosopher, overdoing his reasoning, has reached an
1.2. BRAIN AND MIND 13

impasse (see also the limerick above). It appears doubtful whether

such a concept of God as a “Universal Metaphysical Problem Solver”
who repairs fallacious reasoning of philosophers, is really adequate.
Thus it seems necessary to admit some interaction between body
and mind (causation and downward causation as mentioned above).
Thus body–mind dualism should actually be an interactionism.
This is the thesis impressively argued in the book (Popper and Ec-
cles 1977), which is a fundamental (though not uncontroversial) refer-
ence on the mind–body problem. The authors also treat, extensively
and rather fairly, the other alternatives mentioned above: materialism,
epiphenomenalism etc. Another excellent review is provided by the
papers by Sperry, Dewan and others in (Globus et al. 1976).
From p. 75 of (Popper and Eccles 1977) we take an argument against
materialism concisely formulated by J.B.S. Haldane in 1932: “If ma-
terialism is true, it seems to me that we cannot know that it is true.
If my opinions are the result of the [physical and] chemical processes
going on in my brain, they are determined by the laws of [physics and]
chemistry, not of logic.” (Insertions between brackets are by H.M.)
Remark: In order to be true, a proposition must be determined by the
laws of logic.
This argument, which goes back to Epicurus (around 300 B.C.),
is not entirely cogent, and it was later retracted by Haldane himself.
(Determination by the laws of physics and determination by the laws
of logic are not incompatible, as any electronic computer shows.) It is,
however, a typical and elegant example of a philosophical argument.
These arguments will be taken up again in sec. 6.4.
A more cogent argument against materialism is the fact that usually
brain processes are considered on the basis of classical physics only (cf.
Churchland 1988, otherwise an excellent introduction!). This, however,
is totally inadequate since the microscopic definition of matter must be
in terms of quantum mechanics. Here, however, matter and mind seem
somehow to be intimately interrelated as we shall see in sec. 3.5.
At any rate, concepts such as materialism, idealism or dualism can-
not be discussed in the narrow context of brain–mind interaction only.
Therefore, we shall come back to these concepts later in this book,
especially in sec. 5.1.
Metaphysics and ontology. At the present point, it might be useful
to give a first, rather preliminary explanation of some very frequently
used philosophical terms. Metaphysics studies philosophical questions
which are beyond the reach of natural sciences, “beyond (Greek: meta )
physics”. (Some people regard as “metaphysical” even concepts or the-
14 CHAPTER 1. THE HUMAN BRAIN

ories that explain rather than merely describe phenomena in physics

itself.) Traditional metaphysical questions are the relation between
God and world, the human soul, “what the world really is” and other
“ontological” questions. Ontology (already touched upon above) stud-
ies the various forms of “being” or “existence”, e.g. the nature and
existence of matter, mind, mathematics, etc.
The validity of metaphysics, including ontology, is denied by pos-
itivism: only observed data and logical conclusions are meaningful.
(These definitions are deliberately simplified: interpretations differ and
there are many shades and nuances.)
Suggested additional readings. The problem of mind and brain in
the context of quantum theory is discussed in several excellent books.
If one takes (Penrose 1989) together with (Stapp 1993) and (Lockwood
1989), one should be pretty much up to date. It is recommended,
however, to postpone reading these books until quantum phenomena
have been treated in sec. 3.5. A more “classical” but highly regarded
text is (Edelman 1989). An interesting update on (Popper and Eccles
1977) is (Eccles 1994), whereas (Hofstadter and Dennett 1981) provide
a fascinating counterpoint. A nice anthology is (Rosenthal 1991). For
ontology, there is the wonderful classic (Gilson 1972), but it is probably
too difficult at this early stage.

1.3 Human perception

You don’t see a pretty girl, you just see
some colored patches.
Anonymous

Auditory perception
Although our main emphasis will be on visual perception, the sense
of hearing also presents great interest. We listen to the voice of a friend,
or to a piece of music.
The ear is an organ of extreme sensitivity. If it would be only a little
more sensitive, we would constantly hear an unbearable background
noise produced by the thermal motion of the air molecules. The inner
ear contains a beautifully designed resonance mechanism which, so to
speak, performs a very precise and detailed harmonic analysis of the
auditory signal, a sound wave. The individual frequencies are heard
distinctly and separately. If we strike a chord of four simultaneous notes
1.3. HUMAN PERCEPTION 15

on the piano, we hear all four frequencies distinctly in their true pitch
(if we have an absolute ear) or in their correct relation of frequencies
(which is enough for the average musician and music–lover). Thus if we
strike on the piano simultaneously the notes c and e, we shall hear both
of them distinctly and not “averaged” to an intermediate note d, say.
This is important because in visual perception precisely this “averaging
effect” occurs.
On the other hand, auditory perception in man does not give very
precise information on the direction from which the sound comes, and
on the location of the source, except when these faculties are especially
trained in blind people. (Animals such as bats can “hear” directions
very precisely!)
It is also worth noting that each half of the brain receives informa-
tion from both ears.

Visual perception
This is the classical form of perception, on which philosophers tra-
ditionally have laid the greatest emphasis. It is well known that each
eye acts like a small photographic camera. A lens generates a picture
of the outer world on the retina, again usually a very precise picture,
which is then transmitted by its optical nerve to the cortex. The main
relay stations of both “optical tracks” are the two lateral geniculate
bodies (LGB), one in the right and the other in the left part of the
thalamus, cf. Fig. 1.1 on p. 4. The left hemisphere of the brain (or the
left half–brain) receives its visual information from both eyes, but from
the right half of the visual field of each eye! It is similar for the left
visual field of each eye, from which information goes exclusively to the
right LGB and the right visual cortex.
Optical processing occurs mainly in the retina, the LGB’s and the
visual cortex of both hemispheres.
Slight differences in the direction of the optical axes lead to small
differences in the respective images on both retinas, called parallaxes,
and they provide the third dimension, depth. This is called stereoscopic
vision.
In view of the high state of present electronic image processing, it is
tempting (but not cogent) to compare the processing of visual data of
the eye with our image processing technologies, and stereoscopic vision
has been employed already for several decades by photogrammetry for
the same purposes, namely to construct a three–dimensional model
from two–dimensional images.
16 CHAPTER 1. THE HUMAN BRAIN

Color perception. The eye’s retina consists of closely packed recep-

tors, some 107 cones and 108 rods. The rods are responsible for vision
in conditions of faint illumination; they do not recognize colors. (Try
to distinguish colors in a moon–lit landscape!) The cones are active in
normal daylight. They consist of three types, one type having maxi-
mum sensitivity for a wave length of 430 nanometers (blue ), the second
type for 530 nm (green ) and the third type for 560 nm (red ).
We remember that visible light is electromagnetic radiation between
wave lengths from about 400 to 700 nanometers (1 nm = 10−9 m).
If all wavelengths are uniformly represented, we speak of white light.
If some wavelengths are prominent, we see a color corresponding to the
average of these wave lengths. Thus the eye does not perform such
a sophisticated and detailed harmonic analysis as the ear does with
sound waves. That also could not be expected because then we should
have a full harmonic analyzer (corresponding to one ear) at each of the
107 cones which is patently impossible. On the other hand, the retina
provides precise two–dimensional and even, by the joint stereoscopic
effect of both eyes, three–dimensional spatial position.
Any cone has maximum sensitivity for one particular wavelength
and decreasing sensitivity for neighboring wavelengths. Every wave-
length stimulates each of the three cone types to a certain extent, and
the net impression is precisely the given wave length.
If the incident light contains several wavelengths, we thus see only
one “average” color . It is as if simultaneously striking the two notes
c and e on the piano we would hear only the intermediate note d!
This fundamental difference between sight and hearing has already been
mentioned at the beginning of this section; it means that there are no
“color chords” in the sense in which there are musical chords (never-
theless, artists and critics speak of “harmonious colors”!). Thus music
and painting are essentially different arts.
This averaging (or rather “mixing”) of two wavelengths by the eye
even produces a color to which no natural wave length corresponds:
purple as the result of mixing blue and red. The natural spectrum
reaches from blue over green, yellow and orange to red; it is a segment of
a line. Purple “closes” this (physical) segment to become a (subjective)
circle!
Thus, at least in color vision, there is an enormous simplification of
wavelength information: a light ray that contains many, even infinitely
many, wavelengths gives only one color. Color vision is a “many–to–one
relation”.
Oriented lines, moving objects, grandmother cells. Also in other
1.4. EVOLUTIONARY THEORY OF KNOWLEDGE 17

respects, vision does not simply provide us with a one–to–one, so to

speak, photographic image of the external world. There are cells in
the cortex that react only on lines or stripes of a certain orientation:
horizontal, vertical, or inclined at 45◦ . Other neurons react only if the
object is moving. It is even said that there are single neurons that react
only on a certain individual shape, such as the face of your grandmother
(“grandmother cells”).
Suggested additional reading: it is sufficient to mention two recent
standard works written by outstanding neuroscientists in easily acces-
sible language and treating different aspects: (Hubel 1988) and (Young
1987). We also mention the September 1992 issue of Scientific Amer-
ican dedicated to the present problem and in German: (Singer 1990,
1994).

1.4 The evolutionary theory of knowledge

Wär nicht das Auge sonnenhaft,
die Sonne könnt es nie erblicken.
Johann Wolfgang von Goethe

The “non–photographic” characteristics of vision mentioned at the

end of the preceding section can be explained by the fact that the
sense organs, as well as other parts of our body, have been shaped by
biological evolution, passing through virtually all stages of the animal
kingdom.
Hearing in animals was not designed for enjoying Beethoven’s music.
Hearing served to inform animals of approaching enemies or victims:
briefly, to help them survive. With sight it was similar: vision in an-
imals did not serve primarily to make them aware of the beauty of a
landscape, but to help them recognize and distinguish food and danger .
For this purpose it is essential if the object is moving or not, and also
outlines and horizontal or vertical strips are important, as well as direct
recognition of relevant objects: friends or foes (“grandmother cells”, cf.
end of sec. 1.3).
So the use of “sense data” for abstract purposes such as art, sci-
ence, and philosophy came at a very late state of human development.
Rather than blaming sense data for not providing us directly with all
desired scientific information about the world, we should be surprised
and grateful that sense data and their analysis by the human mind have
taught us so much already and will (hopefully) continue to do so.
18 CHAPTER 1. THE HUMAN BRAIN

A priori and a posteriori knowledge

In the morning we go to our car, open it and turn the ignition key.
Why? Because we know that, as a rule, the car will start and that we
can then drive to our office.
The expectation we have before turning the ignition key is some kind
of a priori knowledge (a priori means “beforehand”). If the car engine
really starts after turning the key (which may not happen on a very
cold winter morning), then we know “empirically” or a posteriori that
our expectation has become true (a posteriori means “afterwards”).
A priori information is generally believed to be provided by logic
and mathematics: 3 + 2 = 5, regardless whether we speak of apples or
of human persons. It is also believed to be always true (although one
cloud plus one cloud may coalesce to form another cloud: 1 + 1 = 1 for
clouds; this may be a joke, but not a trivial one!).
A posteriori information is obtained by empirical observation:
watching the car start in our example.
As this example shows, a priori and a posteriori elements interact
in almost every human activity, and clearly also in science: theories (a
priori ) are tested by experiment (a posteriori ).
What was discussed in philosophy already before Kant but was
placed in the center of philosophy by Immanuel Kant (1724–1804) was
the respective role and certainty of the a priori and a posteriori ele-
ments.
Universally recognized by philosophers is only that logic is a priori
and observation is a posteriori. Logically true propositions (tautologies)
are also called “analytic ”, everything else (especially propositions about
empirical facts) is called “synthetic ”.
Thus all analytic propositions are a priori, and all empirical propo-
sitions are a posteriori. Now Kant’s question was “Are synthetic propo-
sitions a priori possible?” Take for instance the theorems of mathe-
matics. They are independent of empirical observations, and were in
fact considered synthetic a priori by Kant. Bertrand Russell (1872–
1970) thought he could deduce all of mathematics from the axioms of
pure logic and consequently believed mathematics to be a priori but
analytic. Other contemporary mathematicians rather incline towards
Kant’s opinion.
Kant believed not only mathematics, but also our geometrical and
physical three–dimensional space to be given synthetically a priori , as
well as other “categories” such as time and causality.
He thought this to be true absolutely, not only approximately, by the
1.4. EVOLUTIONARY THEORY OF KNOWLEDGE 19

structure of our mind . He believed to have performed a “Copernican

revolution” in philosophy by putting the very basic laws of physics
not into nature but into our mind: We simply cannot think of space
differently than as three–dimensional and Euclidean.
Since Einstein’s General Theory of Relativity (or even since the
non–Euclidean geometries of Gauss, Bolyai and Lobačevsky) we know
that Kant was not right in this respect. Nevertheless the discussion
of a priori and a posteriori continues to the present day. See also our
outline of Kant’s philosophy in sec. 5.4.
Eddington’s example. Eddington (1939, p. 16) gave a nice illustra-
tion of a scientific a priori . A marine biologist is exploring the life of
the ocean. In order to get specimens of animals living in the sea, he
throws a net and examines the catch. He discovers
(1) No sea animal is less than 5 cm long.
(2) All sea animals have gills.
In order to make sure his discovery is correct, he repeats this experiment
several times at various places. His laws are always confirmed, so he
concludes that they are universally true.
It is obvious that at least Law (1) is not an objective law of na-
ture, but a consequence of the experimental setup. It would have been
different had he used a net of a smaller (or larger) mesh size.
This illustrates that we can hardly avoid subjective, “a priori”, el-
ements. Sense perceptions thus are as much given (“sense data”) as
fabricated by our brain’s highly complex processing (Eddington’s net!).

Evolutionary epistemology
Epistemology is nothing else than “theory of knowledge ”. The fa-
mous biologist Konrad Lorenz (1903–1989) was the last professor on
the philosophical chair of Immanuel Kant in Königsberg. Studying
Kant’s philosophy, he recognized that Kant’s a priori conditions for
human knowledge could be identified with man’s perceptual structures
(eye, ear, etc.) developed in the course of human evolution. Man’s
world view is largely, though by no means exclusively, conditioned by
his physical and mental constitution. For instance if, instead of per-
ceiving, in the electromagnetic spectrum, only the wave lengths from
400 to 700 nm, we would be able to visualize also other frequencies,
then our visual representation would be quite different. (We do per-
ceive infrared, but as heat.) Or think of the other extreme, a blind
person.
20 CHAPTER 1. THE HUMAN BRAIN

This theory beautifully explains why our perceptions agree so well

with nature (e.g., sight, hearing, and touch give consistent results). The
old philosophical question of “adaequatio mentis ad rem ” (correspon-
dence of mind to nature) is thus simply explained: if an animal’s sense
impressions do not correspond to reality (e.g., if it hunts after a nonex-
isting prey), it will not survive since it does not conform to Darwin’s
principle of “survival of the fittest”. It and its potential successors will
have long ago been eliminated from the process of evolution.
The animal’s a priori is its perceptual apparatus fit for survival, so
to speak, the set of its “working hypotheses for survival”.
In his important book “Filozofija znanosti i humanizam ” (Philos-
ophy of Science and Humanism, SNL, Zagreb, Croatia, 1991, p. 85),
Ivan Supek shows that the evolutionary theory of knowledge goes back
to the famous Austrian physicist Ludwig Boltzmann (1844–1906).
Science. Science is man’s extension of his perceptual apparatus;
primarily also for survival, later, since the times of the ancient Greeks,
with the emergence of “intellectual curiosity”, for satisfying the de-
sire to know for its own sake. Theories are no longer (only) “working
hypotheses for survival”, but “working hypotheses for understanding
nature”. The primary a priori is, of course, constituted by logic and
mathematics, on one hand, and by the structure of our sense percep-
tion, on the other hand. In which way logic and brain structure are
related, is still almost entirely controversial. Our perceptual apparatus,
on the other hand, has certainly been provided to us by evolution.
A secondary a priori for scientific research are the theories of
physics, chemistry, biology, etc. In contrast to Kant, however, they
are no longer regarded as absolute truths, but as working hypotheses
of varying degree of certainty, and always subject to “falsification” or,
at least, to revision; cf. sec. 6.5.
The bucket and the searchlight. Modern philosophers have empha-
sized to active role of the subject in information gathering. The logi-
cal positivists around Bertrand Russell considered material objects as
“logical constructions from sense data”, these data being more or less
passively connected in a “bucket”. (The expression is Popper’s (1979,
Appendix 1).) This view certainly suffers from Whitehead’s “fallacy of
misplaced concreteness”, oversimplification. It is simply not true that
we perceive “a red patch surrounded by green”, no, we immediately
recognize it as a rose. (It needs a high–level philosophical abstraction
to see a red patch where everyone else sees a rose.)
The modern theory regards the perceptual apparatus as a “search-
light” to discover what we presume, expect, hope for, or fear, such as
1.4. EVOLUTIONARY THEORY OF KNOWLEDGE 21

a hunter watching for a lion. As we have mentioned at the beginning

of the present section, this is also the reason why movement detectors
in the visual retina are so important.
But scientific theories and hypotheses, too, serve as such searchlights
for exploring nature. We hope to verify them by experiment, or if we
follow the severe ways of Karl Popper, we must try to falsify them, like
a merciless professor who tries to fail all students except the very best
ones (sec. 3.9).
Let us briefly recapitulate our main points. Logic and mathematics
seem to be absolutely a priori. The scientific theories are also used
in an a priori way, as working hypotheses, but they are subject to
refutation or revision by experience. (See also Quine’s model at the
end of sec. 4.5.)
The logical–positivist theory of sense data received more or less
passively, from which the external objects are obtained by “logical con-
struction”, does not correspond to reality. The a priori elements, from
primitive fears and expections to the most advanced scientific theories,
work as searchlights for discovering and understanding our external
world.
Already animals perceive their foes or preys directly as external ob-
jects (in however rudimentary a way). They certainly do not construct
their enemies mentally from sense data: they would be dead long before
having completed the logical construction. Also man perceives directly,
not sense data, but external objects. This is at least what man believes,
and the “hypothetic realism ” of “evolutionary epistemology” asserts he
is basically right.
If I remember correctly, Alfred North Whitehead said that we would
not even notice an elephant unless we would expect to meet this ani-
mal. This is another example of the searchlight thesis. To give a more
everyday example: if one hikes through nature with a botanist or an
ornithologist, one is surprised how much can be seen with their help
which otherwise would pass unnoticed.
Additional reading. As an elementary and excitingly written in-
troduction we recommend (Ditfurth 1976), as well as (Young 1987).
On a medium level is the classic (Lorenz 1973). Philosophical books
on evolutionary epistemology are (Popper 1979) and (Vollmer 1990).
The subject is treated from somewhat different angles by Maturana
and Varela (1987) and Piaget (1970). The best book, combining read-
ability and great depth, on all problems of philosophical epistemology,
including a fine chapter on Kant, is (Hartmann 1965).
The view of logical positivism is beautifully described in the book-
22 CHAPTER 1. THE HUMAN BRAIN

let (Russell 1929). We recommend it because a person interested in

philosophy must also be able to understand and appreciate a different
point of view.
We shall come back to the philosophical aspects of this problem re-
peatedly later in this book. So at this point, only Ditfurth and Young
can be understood fully. We also point out once more that the present
book should, in principle, be understandable without additional read-
ing, which is of course recommended for those desiring to know more
about certain problems.
Chapter 2
Logic and mathematics

2.1 Elements of symbolic logic

You are not thinking:
you are just being logical.
Niels Bohr

Set theory
The notion of set should nowadays been known from school. (In the
present treatment, we shall pursue an intuitive approach, like in school,
sacrificing full logical rigor to simplicity.) A set consists of elements. It
can be discrete or continuous (Fig. 2.1).

Figure 2.1: A discrete set (A) and a continuous set (B) in the plane

A set is equivalent to a property. Consider the poet Shakespeare.

Some of his properties are
(M ) He was a person (a human being).
(A) He lived in England.
(B) He was a poet.
(C) He was born in 1564.

23
24 CHAPTER 2. LOGIC AND MATHEMATICS

Instead of property (M ) we may also say: he is an element of the

set of all persons; instead of (A): he belongs to the set of all beings
who lived in England; instead of (B): he is an element of the set of all
poets, etc.

Figure 2.2: Shakespeare (s) belongs to the intersection of the sets M

(persons), A (English), B (poets), and C (persons born
in 1564). (Discrete sets are conveniently represented by
continuous sets in the plane!)

This may be symbolically represented as follows (Fig. 2.2). Shake-

speare, denoted by s, is an element of the set M , symbolically

s∈M . (2.1)

Obviously also s ∈ A, s ∈ B, and s ∈ C. The set A is a subset of the

set M , symbolically
A⊂M . (2.2)
Obviously also B ⊂ M and C ⊂ M . (The sign “∈” holds for elements
and “⊂” for subsets!)
We also need the concepts of union

A∪B (2.3)

and intersection
A∩B (2.4)
(Fig. 2.3).
From Fig. 2.2 we see that

s∈A∩ B , s∈B∩C , s∈ A∩C ,

2.1. ELEMENTS OF SYMBOLIC LOGIC 25

Figure 2.3: Union and intersection

s∈A∩ B ∩ C , s ∈ M , s ∈ M ∩ A∩ B ∩C ,
A∪B ⊂ M , A ∩B ⊂ M , A∪ M = M ,
A∩M =A ,
etc.
We also need the concept of complementary set (with respect to some
comprehensive set M ), see Fig. 2.4. In an obvious notation we may
write
A=M −A . (2.5)

Figure 2.4: The complement of a set

If M denotes the set of persons (human beings), we may also form

the complement M with respect to some more comprehensive set N ,
say the set of animals: M = N − M . We may read M as “not M ”.
For instance
n∈M
if n denotes the Roman emperor Nero (however inhumanly he behaved),
but
n ∈ M or n ∈ /M
(read: n is not an element of M ) if n denotes Nero as the name of a
dog.
We also need the concept of empty set ∅, e.g. the set of all triangular
squares. Thus if T is the set of all triangles and S the set of all squares,
we have
S∩T =∅ ,
26 CHAPTER 2. LOGIC AND MATHEMATICS

the intersection of S and T is empty.

The calculus of set is also called Boolean algebra (G. Boole, 1815–
1864).
Besides sets (or, equivalently, properties), there are also relations,
e.g.
aRb . (2.6)

The two individuals a and b are connected by the relation R, e.g.

a = 2, b = 3, R =< (smaller than): 2 < 3;

a = son, b = father, R = younger than .

In the latter case “aRb ” means “a son is younger than (his) father”.
There are many excellent introductions to set theory, e.g. (Halmos
1960).

Numbers
Let us try to define the integers 1, 2, 3, . . . Consider the number 2.
Obviously, “2” may be considered the property which is common to all
pairs:
2 = set of all pairs .

This set–theoretic definition (maintained, e.g., by the famous logician

and philosopher Bertrand Russell) appears counterintuitive, because
the set of all pairs is clearly infinite, but remember that “set” has been
seen to be equivalent to “property”, and “common property of all pairs”
is intuitively acceptable. But does not the concept of “pair” presuppose
the number “2”? The definition seems to be circular (circulus vitiosus,
vicious circle).
We may, however, proceed “recursively” as follows. Let S1 be a set
of one element only, or to avoid using “one”, a set of consisting of any
individual. For individuals a, b, c, . . . , S1 consists only of a or only of
b or only of c, . . . S1 is called a one–set here (the name “unit set” is
used more frequently).
Then
2.1. ELEMENTS OF SYMBOLIC LOGIC 27

1 = set of all one–sets;

2 = set of all two–sets, i.e., of all sets which
become one–sets by removing one element;
(2.7)
3 = set of all three–sets, i.e., of all sets which
become two–sets by removing one element;
etc.
Those readers who do not like the expression “set of all . . . ”, may
replace it by “property of all . . . ”. (For a more rigorous version of the
argument cf. (Carnap 1958, p. 71).)
Thus we can recursively define all positive integers. Negative inte-
gers and zero can be defined as pairs of positive integers:

−2 = 1 − 3 = (1, 3) = (5, 7) = · · ·

since −2 = 5 − 7 etc. Similarly,

0 = 1 − 1 = (1, 1) = (2, 2) = · · ·

Rational numbers are also defined as pairs of integers:

3 9 3
 
= [3, 7] = [6, 14] = [9, 21] = ! etc.
7 21 7
Irrational numbers are defined as infinite sets of rational numbers, e.g.,

π = {3, 3.1, 3.14, 3.141, 3.1415, 3.14159 . . .} ; (2.8)

this is a set of rational numbers, e.g.,
314
3.14 = ,
100
which better and better approximate π. Thus arithmetic is reduced to
logic (set theory). Through analytical geometry in Cartesian coordi-
nates (see also sec. 2.6), geometry can be expressed in terms of arith-
metic; the coordinates are positive or negative, rational or irrational
numbers.
In this way mathematics can be reduced to logic. This is at least
what G. Frege and B. Russell thought (around 1900). The standard
work is “Principia mathematica ” by B. Russell and A.N. Whitehead
28 CHAPTER 2. LOGIC AND MATHEMATICS

(1910–1913); it is, however, far too difficult to be included in the list of

suggested additional reading. We recommend (Carnap 1958).
This reductionist procedure is not without problems because the
concept of infinity enters; cf. (Weyl 1949) and (Barrow 1992).

Logic of propositions
A proposition, denoted by p, q, r, . . ., is a sentence or a statement.
E.g.
p ... It rains.
q ... Today is Friday.
r ... The street is wet.
Propositions can be true or false. In the first case, we assign the truth
value T or simply 1, in the second case, F or simply 0.
The following symbols are used:

p∧q p and q (both p and q)

p∨q p or q (either p or q or both)
∼p not p (2.9)
p⇒q p implies q (if p, then q)
p⇔q p is equivalent to q (p holds if, and only if, q)

There are so–called truth tables, e.g.

p ∼p ∨ 1 0 ←q ∧ 1 0 ⇔ 1 0
1 0 1 1 1 1 1 0 1 1 0
0 1 0 1 0 0 0 0 0 0 1 (2.10)
↑ -
p p∨q
It is clear that, if p is true, then ∼ p is false, and vice versa. This
explains the first truth table.
To understand the second table remember that, in our example,
p ∨ q denotes
“It rains or today is Friday.”
This is true if p has the truth value 1 (regardless of the truth value for
q). This gives the first line of the second truth table. If p has the truth
value 0 (false), then p ∨ q is true (1) if q is true (1) but false (0) if q is
false (0). This gives the second line of the table.
2.1. ELEMENTS OF SYMBOLIC LOGIC 29

The reader might wish to check this by means of the concrete ex-
ample just given.
The third truth table can be explained in a similar way; note that it
is formally identical to the multiplication table for the numbers 1 and
0. This was one of the starting points for Boole’s logical algebra!
The fourth truth table is almost self–evident: p ⇔ q is true if p = 1,
q = 1 or p = 0, q = 0 and false otherwise. (p = 1 is an abbreviation for
“the sentence p has the truth value 1”.) Other truth tables can easily
be established.
We have the basic laws (which should be checked by means of our
special example):
p∨ ∼ p There holds always p or not–p;
p is either true or false (nothing
else): law of the excluded middle,
“tertium non datur ”
∼ (p ∧ ∼ p) p and not–p cannot hold
simultaneously:
law of contradiction
∼ (∼ p) ⇔ p double negation = affirmation (2.11)
p⇒p∨q “If it rains, then it rains or
today is Friday”
p ∧ (p ⇒ q) ⇒ q “modus ponens ”
(we can put — ponere — q as true)
(p ⇒ q)∧ ∼ q ⇒∼ p “modus tollens ”
(we can remove — tollere —
p as false)
By means of these and similar laws, all logical conclusions (deduc-
tions, proofs) can be performed in a purely formal manner (e.g., also
by an automatic computer). Therefore we speak of formal logic, in our
case of propositional calculus.
We thus have a “mathematization” of logic, in a similar way as
set theory has provided a “logization” of mathematics. Formal logic
is understood to consist primarily of set theory and the calculus of
propositions.
There is a close connection between these two branches of formal
(or symbolic, or mathematical ) logic, expressed in a correspondence
between ∧, ∨ in propositional logic and ∩, ∪ in set theory:
30 CHAPTER 2. LOGIC AND MATHEMATICS

x∈A∩B ⇔ x∈A∧x∈B ,
(2.12)
x∈A∪B ⇔ x∈A∨x∈B .
Here x ∈ A (x belongs to the set A) is regarded as a proposition, say
p, and x ∈ B is another proposition, say q.
A final note on terminology: logical truths such as (2.11) are called
analytical truths or tautologies. They hold always, independently of the
special meaning of p or q.

Logical atomism versus holism

Symbolic logic works well if we deal with a finite number of well–
defined and well–separated objects, such as apples in a basket. Thus
it is no accident that Bertrand Russell (1929, p. 12) represents a “log-
ical atomism ” according to which the world consists of such discrete
objects. Clearly, 1 apple + 1 apple = 2 apples, so that ordinary arith-
metic holds.
But much in the world is continuous, with gradual transition, with-
out sharp boundaries (cf. sec. 2.4). Consider, in a blue sky, two clouds
which gradually coalesce into one cloud. Here we may say: 1 cloud +
1 cloud = 1 cloud, so that 1 + 1 = 1 here! Generally, we may consider
everything on earth related to (or correlated with) everything else; this
is called “holism ”: the world is one interconnected whole; the whole
is more than the sum of its parts. (The Greek word “holos ” means
“whole”.) Following meteorology, it may well be that a butterfly flying
over a meadow in Austria may cause a tornado in Florida . . . This is Ed-
ward Lorenz’ “butterfly effect” well known from chaos theory (sec. 3.2).
If holism were absolutely and exclusively true, then we could not
speak of anything without dragging in the whole universe. (If we think
of Mr. Smith, we must also consider the house in which he lives, gener-
ally his surroundings, his city, his country, the planet Earth, the solar
system, . . . !) So the truth seems to be somewhere in the middle be-
tween logical atomism and holism.

Logic and linguistics

Some modern philosophers have considered the object of logic, and
the task of philosophy in general, to analyze ordinary language (“lin-
guistic analysis”). They identify human thinking with (internal) speak-
ing. This identification seems to be questionable, however: mathemati-
cians “think” in terms of indistinct images and structures (Hadamard
2.1. ELEMENTS OF SYMBOLIC LOGIC 31

1945; Penrose 1989, p. 423), and what about people who are equally
fluent in several languages? In which language do they think? My
answer, based on introspection, is: in none. I also think generally in
images and indistinct structures (unless I prepare explicit formulations
in a certain language for lectures or for writing).
It seems, however, that linguistic formulations play, in philosophy
or history, a much larger role than in mathematical sciences. This may
be one of the reasons why English is used as a lingua franca almost
universally in natural sciences but much less so, e.g., in history. Phi-
losophy is often difficult to translate, and poetry is frequently almost
untranslatable.
We do no wish to downplay the importance of language for human
thinking. It has been claimed that a child who has not learned to speak
one (any) human language by the age of about 12 years, will forever be
unable to think as a normal adult (Cohen and Stewart 1994, p. 355).
Language is indeed an indispensable prerequisite for human thinking.
Thus there is no doubt that linguistic analysis is important in logic
and philosophy, but it seems to be a tool rather than a goal in itself.

Logic and intellectual activity

Reading some philosophical books, one could get the impression
that human intellectual activity is “pure thinking” expressed in objec-
tive language, logic being “the science of correct thinking”.
A moment of reflection is sufficient to see that this is by no means
the case. Saying “Of course, you are right!” may express almost ev-
erything, from objective agreement to furious sarcastic disagreement,
depending on the way we are saying it. A poem expresses much more
than any grammatical analysis can show. emotions pervade almost ev-
erything we can say or think. Our limbic system (sec. 1.1) is constantly
active indeed!
Much of our intellectual activity is nonconscious or subconscious.
Assume we give a speech without using a manuscript. During the mo-
ments when we express an argument, our subconscious keeps “thinking
ahead” on what we are going to say next. At the same time, we are try-
ing to suppress our anger about a letter which we just received, helped
by the (largely unconscious) expectation of a good lunch after the lec-
ture . . . So many lines of “thinking” are going on simultaneously: our
mental activity is indeed a “highly parallel processing”, to use computer
language which is now, more than ever, seen to be totally inadequate
for describing the activity of our brains . . .
32 CHAPTER 2. LOGIC AND MATHEMATICS

Our expression need not be verbal. Getting a red face may express
our embarrassment better than any words; and furiously walking out
of the room could be our final argument. The behaviorist psychologists
may not be so wrong after all!
Earlier we have talked about nonverbal thinking in mathematics.
What about the performance of a Beethoven sonata by a pianist? In
sec. 1.1 we have seen that the cerebellum is heavily involved. Does this
mean that the performing pianist is not active intellectually? If you
think so, don’t tell it to her; otherwise you may get a painful nonverbal
answer!
These examples from elementary psychology show that “verbal
thinking” is indeed only a small fraction of our total intellectual activ-
ity. They are not meant to discredit logic but to place logical thinking
into its proper, lofty but highly abstract, place.

2.2 The axiomatic method

What is truth?
Pontius Pilate

Axioms are basic propositions from which all true statements of a

certain branch of science of mathematics can be derived by a purely
formal procedure (also by an automatic computer!).
The first and best–known axiom system is Euclid’s axiom system for
elementary geometry (around 300 B.C.). A complete and rigorous ax-
iom system for this purpose was given, however, only by David Hilbert
in 1899.
Geometrical statements are fully proved only if they are derived in a
purely formal way from the axioms, without using intuition or figures.
Figures, etc., are to be considered only as “heuristic” aids in guessing
mathematical theorems or making them plausible; a rigorous deduction
from the axioms must then follow. A purely formal way of deduction
is frequently called an algorithm.
An axiom system may satisfy the following requirements:
(1) Consistency: this is the absence of internal contradictions; it is
absolutely necessary.
(2) Completeness: all true statements that can be formulated in our
logical system, can be derived from the axioms. This requirement
is often satisfied but there are also incomplete axiom systems.
2.2. THE AXIOMATIC METHOD 33

(3) Independence: the axioms should not be redundant. This require-

ment is desirable but not absolutely necessary.
A complete axiom system gives, so to speak, “implicit definitions” of
the concepts which it contains. For instance, “points”, “straight lines”,
or “planes” are those mathematical entities which satisfy the axioms
of geometry (they are by no means identical with small dots or “lines”
made with chalk on the blackboard!).
Examples for incomplete axiom systems are Peano’s axiom set for
integers or the axioms of group theory (this remark is only for the
specialist).
The axiomatic method accounts for the great abstractness of mod-
ern mathematics (e.g. the famous French school of “Bourbaki”). Ax-
iomatization is also a goal for other “exact” sciences such as physical
theories. Many axiom systems can be found in (Carnap 1958).
However, axiomatization is a final goal, but never the beginning of a
science. In physics, but also in parts of mathematics such as differential
geometry, theorems and whole theories are first derived intuitively or
heuristically, making use of figures, additional assumptions, etc. The
same holds for differential and integral calculus, where the basic theo-
rems were first derived intuitively, in an inexact way (using “infinitesi-
mal small quantities”). Only at a later stage they were made rigorous
by limit processes. The most advanced theories of physics (string theo-
ries, supersymmetry, Feynman integrals) are still largely at a heuristic
stage.
Even the “simplest” mathematical discipline, arithmetic or number
theory (the theory of the properties of natural numbers 1, 2, 3, . . . )
cannot be fully based on a single axiom system. This is Gödel’s theorem
to be treated in the next section.
34 CHAPTER 2. LOGIC AND MATHEMATICS

2.3 Logical paradoxes and Gödel’s theo-

rem
Even one of their own prophets has said:
“Cretans are always liars.”
St. Paul’s Epistle to Titus

Logical antinomies
Already around 1900, Russell, analyzing work by Gottlob Frege
(1848–1925), met with the first difficulties in reducing mathematics to
logic (more precisely, to set theory). He found antinomies or paradoxes,
that is, logical contradictions.
Russell’s antinomy: Let M be the set of all sets which do not contain
themselves as elements. Does M contain itself? Answer: it contains
itself as element, if and only if it does not contain itself! Thus the
concept of the set M is obviously contradictory.
This requires some abstract thinking. A more concrete form has
been given to it also by Russell: A barber in a village shaves exactly
those of its male inhabitants who do not shave themselves. Does the
barber shave himself? Yes, exactly then if he does not shave himself!
Russell’s antinomies and similar paradoxes show that the concept of
“set” in its “naive” form may be contradictory. Various methods have
been tried in order to avoid such antinomies:
(1) Limitation to simple logical systems in which it is impossible to
even formulate such antinomies.
(2) Axiomatic set theory: the axioms should exclude “dangerous”
sets. However, such attempts so far have been largely “ad hoc”. Even
after having excluded known antinomies, one cannot be sure that later
on one cannot find new antinomies in some remote corner of the system.
So far, the consistency of axiomatic set theory has not been proved!
Hilbert’s program of “formalism ” (around 1920) was intended to
give formal calculi for logic and mathematics whose consistency was
to be proved according to mechanical rules (as realized, e.g., by an
automatic computer).

Gödel’s theorem
Hilbert’s program received a death blow in 1931 when the Austrian
logician Kurt Gödel proved that such a program is impossible.
2.3. LOGICAL PARADOXES AND GÖDEL’S THEOREM 35

In particular, Gödel proved the following: In a certain formal (“com-

puterizable”) system which comprises logic and arithmetics, a propo-
sition G (or logical formula) can be constructed, which asserts its own
unprovability. Is G provable? If so, then G is true and therefore, ac-
cording to what it asserts, unprovable! Thus both G and ∼ G are
provable within the system.
Now, if both a formula (G, in our case) and its negation (∼ G) can
be derived, the underlying axiom system is inconsistent (i.e., logically
self–contradictory). If arithmetic is consistent, then neither G nor ∼ G
can be derived from the axioms.
Now comes the big surprise: by “meta–mathematical” reasoning
outside the logical system based on our axioms, it can be shown that
G is true! We, so to speak, manage to get at a level above the formal
system, and “looking down” we recognize that G must be true. (This,
of course, is not so easy as it sounds!)
What have we found? If arithmetic is consistent, then G is true and
hence unprovable (remember that G asserts its own unprovability!).
There is a true statement G in the system which cannot be derived
from the axioms! According to the definition of completeness given in
sec. 2.2, arithmetic, if consistent, is incomplete!
Now the consistency of arithmetic has been proved by other means
(“transfinite induction”, Gerhard Gentzen, around 1936). Hence arith-
metic is incomplete; it cannot be fully described by a (finite) system
of axioms! This reveals a fundamental limitation of the power of the
axiomatic method.
Because of the essential incompleteness of arithmetic, it has been
considered possible that some well–known unsolved problems such as
Goldbach’s conjecture (“Every even number can be represented as the
sum of two odd primes”) or even Fermat’s last theorem (“The formula
xn + y n = z n has no solution for integer x, y, z, n except for n = 2”)
may be a consequence of the fact that the presently known axioms for
arithmetics are incomplete and hence insufficient to derive them.
An important consequence of Gödel’s and related theorems is that
important areas of mathematics (and hence physics and other sciences
involving mathematics) cannot be completely characterized by axiom
systems, they are no “calculi” or “algorithms ” computable by machines.
Human reason transcends computability in an essential way.
Gödel’s paper is extremely difficult. Fortunately, there exists a won-
derfully lucid presentation (Nagel and Newman 1958).
The antinomy of the liar . Gödel’s proof reminds one of the “anti-
nomy of the liar” which goes back to ancient Greece:
36 CHAPTER 2. LOGIC AND MATHEMATICS

“What I am now saying is false”

(or simpler, “I am lying now”). Is this sentence true or false? You
have guessed correctly: it is true if it is false and it is false if it is true
(Fig. 2.5).

Figure 2.5: The “vicious circle” in the paradox of the liar.

This paradox, attributed to the Cretan Epimenides, was known to

Aristotle and Cicero, and is even alluded to St. Paul’s letter to Titus
(1,12–13). It is said that some ancient “philosopher” lost his sanity by
meditating on this sentence, which becomes plausible on looking at the
“vicious circle” of Fig. 2.5 which could really drive one crazy.
Gödel replaced “false” by “unprovable” and hence avoided Epi-
menides’ trap by a hair’s breadth. He converted the explosive energy
contained in Epimenides’ paradox into a powerful argument which by
logicians is considered the most important single achievement of 20th
century logic.

Self–reference
The common feature of all these paradoxes is self–reference: Rus-
sell’s set contains itself, the liar asserts something about what he himself
is just saying, etc.
It seems that self–reference has an importance far beyond some
esoteric paradoxes: its applications reach from human thinking (“the
thinking thinks the thinking”) and artificial intelligence to cosmology.
Hofstadter (1979, 1985) has written two remarkable heavy volumes
about it. Self–reference also seems to lie at the basis of dialectic think-
ing. Thus we shall meet it frequently in our book.
Let us here mention the delightful little book (Smullyan 1980),
which is a veritable treasure of paradoxes.
The logical antinomies and Gödel’s theorem seem to indicate that
even formal logic and mathematics cannot be made arbitrarily precise.
We may speak of a Gödelian uncertainty of logic and mathematics as
2.3. LOGICAL PARADOXES AND GÖDEL’S THEOREM 37

an analog to the well–known Heisenberg uncertainty relation in physics

(sec. 3.5).

Foundations of mathematics
We have already met two ways of founding arithmetic or the theory
of natural numbers (positive integers), which may then serve as a basis
of other branches of mathematics:

(1) Logicism: arithmetic can be reduced to logic (set theory), cf. (2.7)
in sec. 2.1.

(2) Formalism: logic and mathematics can be constructed together

by “mechanical rules” (Hilbert’s program), see above in the
present section.
There is a third approach:

(3) Intuitionism, which goes back to the Dutch mathematician Dirk

Brouwer (since 1907). It by no means implies that formal math-
ematical proofs are to be replaced by intuitive reasoning, quite
on the contrary: The logical requirements of intuitionistic proofs
are so stringent that a large part of contemporary mathematics
cannot be proved by intuitionistic methods and would have to be
sacrificed.
Gödel’s proof affects logicism and formalism. It does not affect
intuitionism since Gödel’s argument could not even be formulated
within intuitionism.
This does not mean that intuitionism is complete, it is even much
more incomplete than “classical arithmetic” (logicist or formalist
approaches)! Therefore it is a historic curiosity rather than a
method applied by contemporary mathematics (as embodied in
the Bourbaki school).

The main reason for all the problems and paradoxes in logic and
mathematics is infinity. Russell’s set is “highly infinite”; self–reference
implied “zero distance”: the barber shaves himself (he has “zero dis-
tance from himself”), the liar says something about which he is just
saying (both statements are coincident or have “zero distance”), and
“zero”, or “infinitely small”, is also an aspect of infinity, at least in the
present context (remember that one frequently bluntly puts 1/0 = ∞,
although mathematicians may frown on it).
38 CHAPTER 2. LOGIC AND MATHEMATICS

The problem why consistency proofs in set theory are so difficult

is that “strongly infinite” sets occur. Brouwer’s intuitionism tries to
avoid this problem by admitting only potential infinities, such as the
series 1, 2, 3, . . . which potentially goes to infinity, but forbidding actual
infinities such as the set of all infinitely many integers.
At any rate, many complicated systems of logic and set theory have
been constructed in order to avoid known paradoxes, but we can never
be sure that, some day, we may not run into a new antinomy.
It is true that few working mathematicians care much about these
“esoteric” difficulties: usually “nothing happens”. The great mathe-
matician Weyl (1949, p. 235), however, writes about the arbitrariness in
present logico–mathematical systems: “How much more convincing and
closer to facts are the heuristic arguments and the subsequent system-
atic constructions in Einstein’s general relativity, or the Heisenberg–
Schrödinger quantum mechanics.”
Modern references include (Rucker 1982) and (Barrow 1992).

2.4 Inexact concepts, “fuzzy logic”

Fuzzy logic makes better washing machines.
Advertisement

Inexact concepts
Exact
√ are certainly the concepts of mathematics: numbers such as
1, 2, 3, 2, π; geometrical points, straight lines, circles, spheres, etc.
Rather exact are empirically defined physical concepts such as “pre-
cisely” drawn lines, circles, “physical points” realized, e.g., by the in-
tersection of two empirical lines (+) or as the center of small circles (◦)
found on any illustration in a book on geometry. Rather exact are also
the concepts of “logical atomism” (see the end of sec. 2.1): Mr. Smith,
his house, a lamp or a book on his table, Mr. Smith’s dog, an apple,
and a huge amount of similar “well–defined” concrete objects.

Figure 2.6: Distance between two points

“Rather exact” are also physical, astronomical, or geodetic measure-

ments. Consider a distance d between two “points” A and B (Fig. 2.6).
2.4. INEXACT CONCEPTS, “FUZZY LOGIC” 39

Clearly, even the “points” A and B are not defined physically with ab-
solute precision: on paper, they are defined not much better than to an
accuracy of 0.1 mm; in nature, perhaps by a cross (+) on a stone to an
accuracy of 1 mm; most precisely, by marks engraved on metal or glass
to an accuracy of about 0.001 mm. At any rate, the distance d cannot
be measured to an accuracy higher than the accuracy to which physi-
cal points are defined. Let the result of our measurement be 23.281 m.
If the points are defined to 1 mm only, the result may equally well
be 23.280 or 23.282 m. In geodesy, we may write 23.281 m ± 1 mm,
there the “standard error” is denoted by ± 1 mm. This is not precisely
equivalent to the statement that the measurement lies in a “confidence
interval”, say between 23.280 and 23.282 m, but we may disregard this
distinction for the present purpose; cf. also sec. 4.4.
Here we already see that classical principles of logic such as the law
of contradiction or the law of the excluded middle (sec. 2.1, formulas
(2.11)) no longer hold with empirical concepts: p ∨ ∼ p, “d is either
23.281 or not”, is violated, there may as well be d = 23.281 m or d =
23.2812 m (6= 23.281 m). The same holds with the law of contradiction:
∼ (p ∧ ∼ p): d = 23.281 m (p) and d = 23.2812 m (not–p) may hold
simultaneously.
Inexact are many concepts of ordinary life, starting from our body:
nose (where in our face is the boundary line between our nose and an
adjacent cheek?), finger (where on our hand does it begin?), but also
a tall building (how high?), a hiking trail in the mountains (which to
his embarrassment, the hiker frequently finds very difficult to follow),
smoke, a cloud (see also the end of sec. 2.1), or the Earth’s atmosphere
in general (at its “edge” there are less and less molecules in a cubic
meter; if there are only a few stray molecules or ions, is this still atmo-
sphere or already empty space?).

Fuzzy sets
Many sets in ordinary life are not very precisely defined, e.g., the
Earth’s atmosphere just mentioned, considered as a set of molecules.
Such sets with ill–defined boundaries are called “fuzzy” (Fig. 2.7).
Obviously, a cloud is a fuzzy set of water droplets. But also more
abstract sets in ordinary life are “fuzzy”. Think of the set of honest
people. Who belongs to it? Saints, yes; thieves, no; but what about
myself, whose behavior frequently is less than perfect? Or consider the
set of red apples: which shades of red are permitted? Is an apple still
red or already closer to orange? What about a red apple with some
little yellow spots?
40 CHAPTER 2. LOGIC AND MATHEMATICS

Figure 2.7: Precisely defined and fuzzy sets

Since, as we have seen in sec. 2.1, sets are equivalent to properties

(“honest”, “red”), also properties of everyday life are frequently fuzzy:
think of “tall”, “cold”, “fast” and very many others.

Fuzzy propositions
The same “fuzziness” also applies to propositions (statements) of
everyday life. A few examples: “It is hot today” (which temperature?),
“You are stupid”, “Mahler’s symphonies are great”.
Among the most popular unprecise statements are weather fore-
casts. “There will be light rain showers tomorrow.” In the U.S.A.,
newspapers are frequently more explicit: “There is a probability of
25 % for rain tomorrow”. Intuitively this is pretty clear, but what does
it mean precisely? Will rain fall in 25 % of the next day (consisting of
12 h, say), that is, will we have 25 % × 12 = (25/100) × 12 = 3 hours
of rain tomorrow? If instead it rains for 4 hours, will we cease to trust
our newspaper or the weather forecast in general?
Obviously not, we have to live with unprecise statements and try
to interpret them reasonably, guided by experience. Much more diffi-
cult is the task to “precisely” formulate these “unprecise” statements,
especially if we want to simulate human thinking by a computer. Fre-
quently it does not work to consider some facts as precisely true and
contradictory facts simply as false, as we ourselves unfortunately do
often enough, following the example of small children (“I am right!”
“No, I am right!”)

Fuzzy logic
There are various cases and possibilities of formalizing these “fuzzy”
concepts. Probably the oldest one (Gauss and Legendre, around 1800)
is
2.4. INEXACT CONCEPTS, “FUZZY LOGIC” 41

(1) Least–squares adjustment used in geodesy, astronomy, physics,

and other disciplines. This may be illustrated by a simple example. In
a triangle we measure the three angles

α = 75◦ 130 3700

β = 49◦ 260 1300
γ = 55◦ 200 1600 (2.13)

sum = 180◦ 000 0600

Now we know that the sum of the angles in a plane must be 180◦ . Thus
there is a contradiction.
The simplest way is to leave α and β and to change γ to 55◦ 200 1000 .
This would correspond to the frequent human attitude, mentioned
above, to considering some facts as precise and disregarding others.
However, it contradicts our sense of fairness and justice: why not treat
all observations in the same way, “adjusting” them equally? Thus the
“misclosure” of 600 above 180◦ is divided evenly among all angles, di-
minishing each by 200 :

α = 75◦ 130 3500

β = 49◦ 260 1100 (2.14)
γ = 55◦ 200 1400
so that now α + β + γ = 180◦ . A general procedure for such prob-
lems is adjustment by least–squares estimation . Being quite an elabo-
rate method, it is beyond the scope of the present book (see, however,
sec. 2.6). What is important is that we have a well–defined procedure
which can be implemented on a computer.
(2) Subjective probability . According to classical logic (sec. 2.1),
propositions are either true (truth value 1) or false (truth value 0).
Now many statements are only more or less probable, as the example
of a weather forecast has shown.
More about the concept of probability will be said in sec. 3.3, but
even here it appears plausible to “interpolate” between the truth val-
ues 0 (impossibility) and 1 (certainty), defining as probability P of a
sentence any number
0≤P ≤1 . (2.15)
It is, of course, often difficult to assign, to any proposition under con-
sideration, a precisely defined number P , in much the same way as it is
42 CHAPTER 2. LOGIC AND MATHEMATICS

often difficult to assign a definite grade G (in Austria: 1 ≤ G ≤ 5, with

“1” denoting the best grade and “5” meaning “failed”) to a student in
an examination, especially if it is required to give non–integer grades,
such as 1.3. Still, decimal places are useful if average grades etc. are
computed. In the same way, probabilities of rain of 0.2 (or 20%) or 0.9
(90%) are more telling than “some rain may be expected” or “rain is
very probable”.
If the probability of certain basic propositions (axioms) are known
or assumed, then the probabilities of derived propositions can be com-
puted (cf. Carnap 1950; Jeffreys 1961).
The choice of initial probabilities is often not so crucial; what is
essential, is that “subjective logic ” (so called because it refers to propo-
sitions which are considered subjective, rather than to objective facts)
can be handled by a fixed algorithm. Thus, in much the same way as
least–squares adjustment, it can be implemented on a computer.
(3) Theory of fuzzy sets. This theory, somewhat similar but not iden-
tical to probability theory, was developed since 1965 by L.A. Zadeh and
others, mainly in the U.S.A. The practical breakthrough, however, was
made in Japan, in the early ’80s, where it was applied in a technolog-
ically innovative way. “By 1985 Hitachi had installed the technology’s
most celebrated showpiece, a subway system in Sendai, about 200 miles
north of Tokyo, that is operated by a fuzzy computer” (TIME, Septem-
ber 25, 1989). It gives an astonishingly smooth ride and uses less energy
than conventional systems.
Since then, “fuzzy control methods” are used in Japan and else-
where to improve auto–focus cameras, washing machines, fuzzy shower
systems that avoid too cold and too hot temperatures, etc.
As an example, let us mention a hypothetic automatic car braking
system based on the following very “fuzzy” rules:
1. IF the road curvature is LOW
AND
IF the speed is LOW to NORMAL
THEN the braking should be SOFT.
2. IF the road curvature is NORMAL
AND
IF the speed is HIGH,
THEN the braking should be NORMAL.
3. IF the road curvature is HIGH,
AND
2.4. INEXACT CONCEPTS, “FUZZY LOGIC” 43

IF the speed is VERY HIGH,

THEN the braking should be STRONG.
It is indeed astonishing that efficient and reliable control mechanisms
can be developed on the basis of such vague rules.
It is perhaps not surprising that in an Eastern country like Japan,
in which thinking is more inclined to “holism” (see end of sec. 2.1)
should have exhibited such a spontaneous interest in “fuzzy thinking”,
whereas the West, in which “logical atomism” comes more natural, has
at first been rather reluctant.
All types of methods based on inexact knowledge are comprehen-
sively treated in the book (Spies 1993), from which we also have taken
the example of “fuzzy automobile braking”. The book (Kosko 1993)
reads particularly well.
It seems that also the neural network of our brain, with its huge
number of intricately connected neurons, may be well equipped to han-
dle fuzzy information. Is precise thinking related to predominantly
digital functioning of our brains, and fuzzy thinking to analog compu-
tation also going on in the brain (cf. sec. 1.1)? There may well be some
relation of this kind, but it is almost certainly not that simple. See also
end of sec. 5.2.

Informal reasoning
The arguments we hear in daily life, in quarrels and disputes, even
in university lectures, are hardly capable of being expressed in the sym-
bolism of formal logic as, for instance, mathematical proofs may be.
Curiously enough, the same holds also for the “informal reason-
ing” in philosophical arguments. Even philosophical concepts such as
causality or determinism, matter and mind, freedom and law, are by no
means sharply defined. They subtly change their meaning during a dis-
cussion. Sometimes this is intentional, sometimes it passes unnoticed.
Lucas (1970, p. 58) speaks of “chameleon words ”.
In philosophical and other discussions, statements are not in general
simply true or false. There are arguments pro and con, some carry great
weight, some arguments are rather weak. Discussions may be intended
to prove the opponent wrong, but this seldom happens. Mostly the
participants in discussions and the readers of philosophical books are
invited to follow the arguments, to appreciate their strength, validity,
and cogency, and finally to form their own opinion.
Nobody could imagine to replace, in a discussion, the two opposing
philosophers by opposing computers. So far, no philosophical book has
44 CHAPTER 2. LOGIC AND MATHEMATICS

been written (authored, I mean) by a computer. (Of course, philoso-

phers use text processing, but this is a completely different matter . . . )
The informal reasoning which we admire in good philosophical
books, is, by no means, “purely logical” as an innocent reader might
think. As Niels Bohr has said (motto of sec. 2.1), thinking is much
more than “just being logical”. Only read a page of Sir Karl Popper or
J.R. Lucas and analyze the structure of their arguments. They differ
from dull logical deduction as a piano sonata by Beethoven differs from
a monotonous finger exercise.

2.5 Dialectic thinking

Not–being is a form of being.
Plato

Introduction
We may define dialectics as that aspect of human thinking that
transcends “algorithmic reasoning” which could as well (or better) be
performed by a computing machine. A beautiful example has been
pointed out by J.N. Findley in his article “The contemporary relevance
of Hegel” in (Findley 1963). It is Gödel’s theorem.
Formalized thinking based on axioms and, in principle, performable
by a machine, is called an object language, or symbolic language or for-
mal language. The informal language in which we speak about the op-
erations of formal logic, is called a metalanguage (considerations about
mathematics are called metamathematics!). In the present book, the
metalanguage is (Austrian) English. It is a useful convention to con-
sider the metalanguage to be of higher level than the object language.
Using this terminology, we may describe Gödel’s proof (sec. 2.3) as
follows.
In a certain symbolic object language, a proposition G is constructed
which asserts its own formal unprovability. By informal reasoning,
however, it can be shown that G nevertheless is true. Let us quote
Findley (1963):
But the unproveable sentence at the same time soars out of this logico–
mathematical tangle [of the formulation of G in the object language] since
the proof of its unproveability in one language [the object language] is itself
a proof of the same sentence in another language of higher level [the meta-
language], a situation than which it is not possible to imagine anything more
Hegelian.
2.5. DIALECTIC THINKING 45

(Italics are Findley’s, explanatory insertions between brackets [ ] are

mine.)
Never mind if you did not get all of the argument at first reading,
just go ahead! This example, trying to show that highly respectable
contemporary logical concepts are behind dialectics, was mainly in-
tended to counteract the preconceptions against dialectics due to its
dogmatic over–use (and sometimes misuse) in some contexts.
Already Plato identified dialectics with philosophy as the highest of
all sciences, even above mathematics.
Its explicit use in classical German philosophy started with I. Kant
(1724–1804), reached a first culmination with J.G. Fichte (1762–1814),
especially in his Wissenschaftslehre of 1804, continued with F.W.J. von
Schelling (1775–1854), and reached its final culmination with G.W.F.
Hegel (1770–1831). For details cf. (Hartmann 1960), (Gulyga 1990) or
(Kuznecov 1981), a very readable introduction is the booklet (Müller
1974).
Dialectics is also basic in Eastern philosophy, cf. (Capra 1975) and
(Smullyan 1977).
Dialectics is primarily a logical discipline and has been so under-
stood by all philosophers from Plato to Hegel. Karl Marx and Friedrich
Engels tried to apply it to “nature” itself. This is not necessarily wrong:
also modern physics applies logic and mathematics to nature. We shall,
however, consider here dialectics as a logical discipline, independent of
our metaphysical background (idealism, materialism, dualism, etc.; cf.
sec. 5.1).
After this lengthy introduction let us try to describe dialectics first
in the usual simplified manner. It is the triadic scheme

thesis −→ antithesis −→ synthesis (2.16)

or better
thesis −→ antithesis
& .
synthesis
The synthesis (S) is not simply a compromise between thesis (T ) and
antithesis (A), but S is a higher–level standpoint from which both T
and A become understandable and even compatible. Thus, in the triad,
the synthesis S, so to speak, lies on a higher level than both T and A.
(It would be absolutely wrong to regard A as the simple logical opposite
∼ T (non–T ) in the straightforward sense of formal logic, cf. sec. 2.1.
46 CHAPTER 2. LOGIC AND MATHEMATICS

Dialectics is not that simple! Contradictions and opposing views do

play a role, as we shall see, but not of the simple type p versus ∼ p.)

Figure 2.8: S lies on a higher level than T and A

It is important to remark right away that the usual scheme thesis

— antithesis — synthesis is well suited for introductory and didactic
purposes, but it constitutes by no means the essence of the dialectic
method. Much more important is rising to a higher level , cf. Gödel’s
argument mentioned at the beginning of this section and Fig. 2.10 on
p. 52. In short, dialectic thinking is metathinking.
Let us illustrate this by examples from science.
Example 1.
Light as wave (T ) — light as particles (photons) (A) — quantum
theory incorporating both aspects (S); cf. sec. 3.5.
Example 2.
The universe has always existed (Aristotle) (T ) — universe has
started at a certain time (Bible, “big bang”) (A) — both proposi-
tions unified by a mathematical transformation (t → log t, Milne,
sec. 3.7) (S)
Example 3.
Geocentric world system (T ) — heliocentric system (Coperni-
cus) (A) — equivalence by general relativity (Einstein) (S). This
equivalence is a result of Einstein’s principle of general covari-
ance, according to which all reference systems are theoretically
equivalent; cf. sec. 3.4.
Example 4.
Logical atomism (T ) — holism (A) — actual world probably uni-
fies both aspects (S); cf. secs. 2.1 and 2.4.
2.5. DIALECTIC THINKING 47

Example 5.
A geodetic example: measurement of all three angles α, β, γ in a
triangle (sec. 2.4).
Thesis T : α, β, γ have been measured.
Antithesis A: α + β + γ 6= 180◦ (contradiction!).
Synthesis S: least–squares adjustment as described in sec. 2.4.
As we have seen above (Fig. 2.8) the important property is that the
synthesis lies on a higher level. This is also illustrated by an example
from everyday life.
Example 6.
A discussion in which two participants defend two apparently
contradictory positions (T and A). A skilful chairman manages to
convince both partners that their opinions, if seen from a proper
perspective, are really compatible (S). If all three participants
have good will, this usually succeeds, often surprisingly, giving all
a profound feeling of satisfaction, especially if the initial positions
T and A have appeared quite contradictory.
Let us return to empirical science. All measurements are affected
by measuring errors, and no physical theory can be expected to hold
absolutely. We shall now give an example for a measurement and one
for a theory (cf. secs. 4.4 and 3.4).
Example 7.
T : A measured distance d has the value 18.85 m.
A: This cannot be true because of measuring errors, rounding–
off, etc.
S: d = 18.85m ± 0.007m. This shows that it is an empirical value
and at the same time estimates its accuracy.
Example 8.
T : Newtonian mechanics holds exactly.
A: No, since relativity theory gives better results.
S: Newtonian mechanics holds to a certain accuracy which is
defined by relativity theory (v/c  1, cf. sec. 3.4).
The dialected process can sometimes be iterated in order to come closer
and closer to reality, as the following two examples show.
Example 9. √
A trivial example from mathematics. What is the value of 2?
√ √
T : 2 = 1.4 A: 2 = 1.5
48 CHAPTER 2. LOGIC AND MATHEMATICS

Both answers are wrong. A better value is

√
S: 2 = 1.41 .
If this is not precise enough, we may iterate the procedure, taking
(S) as a new thesis T1 .
√ √
S = T1 : √2 = 1.41 A1 : √2 = 1.42
S1 = T2 : √2 = 1.414 A2 : 2 = 1.415
S2 : 2 = 1.4142
etc.
√
Note that the synthesis is not the “cheap” solution 2 = 1.4145,
the arithmetic mean between T2 and A2 . There is no “automatic”
way of finding the synthesis. This always requires creativity or
superior knowledge.
Such an iteration may be illustrated by Fig. 2.9, the right–hand
figure showing convergence particularly well.

Figure 2.9: Illustrating different levels

Example 10.
Iterative solution of an equation x − φ(x) = 0, assuming φ(x) to
change only slowly with x.
T : A certain approximate value x0 gives a solution.
A: x0 − φ(x0 ) 6= 0 (contradiction!)
S = T1 : x1 = φ(x0 ) is a better solution
A1 : x1 − φ(x1 ) 6= 0 (contradiction)
S1 = T2 : x2 = φ(x1 ) is again better
A2 : x2 − φ(x2 ) 6= 0 (2.17)
S2 : x3 = φ(x2 )
etc.
2.5. DIALECTIC THINKING 49

This is again illustrated by Fig. 2.9, especially by that on the

right–hand side.

These primitive examples (except perhaps Example 6) have only

served to illustrate the machinery of dialectics. Were it only for them,
however, we should not need dialectics! There must be deeper reasons.

Characteristic features of dialectics

(1) Opposites and contradictions are an essential feature in nature
and human life. They cannot be removed; on the contrary, they serve
as driving forces. Examples:

freedom — law
accident — necessity
content — form (e.g., in art)
analysis — synthesis
self–preservation — care for others
conservation — progress
subject — object
matter — mind
justice — love
continuous — discrete
theory — practice
simplicity — complexity
competition — cooperation
order — chaos
and many others.

Human thinking is full of contradictions. A well–known example from

psychology: you get a high reward if in the next ten minutes you will
not think of a blue elephant. You would never in your life think of
a blue elephant, but when you desire not to do so, you will certainly
think of a blue elephant!
Following the physicist and Nobel Prize Winner Niels Bohr (1885–
1962), such pairs of natural opposites are called complementary. The
term “complementarity” comes from quantum mechanics, cf. Exam-
ple 1 above and sec. 3.5. Bohr extended the principle of complemen-
tarity to other phenomena in nature and in the human spirit and has
contributed essentially to making dialectic thinking respectable to sci-
entists. Examples of complementarities in biological and other systems
can be found in (Haldane 1939) and (Holzmüller 1984, p. 118).
50 CHAPTER 2. LOGIC AND MATHEMATICS

(2) Human language (including logic and mathematics) is too

crude and too inflexible to represent complex situations by a single
linguistic perspective. Various different perspectives must be used.
Example 11.
“This leaf is green.” Is it really green?
T : “The leaf is green.”
A: “The leaf is not green.”
S: “The leaf is green with small yellow and brown dots.”
In a discussion it is frequently possible to bluntly assert the contrary
of what the predecessor has said (even if it is not a “profound truth”
in the sense of Bohr, see below). This may help render the discussion
more precise; in many cases it can cause effective blank surprise. (Try
it if you are sure of yourself!)
Human language is usually adequate for simple “discrete” objects
such as apples. Elementary language, and even more so symbolic logic,
however, may run into difficulties already with “simple” natural prop-
erties such as “green”. The “green” of the leaf is not the ideal “green”
of formal logic. Therefore even in Example 11, A is not ∼ T in the
formal sense since “green” in T and A have slightly different meanings.
Remember what we have said on “fuzzy” concepts and “chameleon
words” in sec. 2.4.
Language is even less adequate for the abstract concepts of phi-
losophy. This at least partly explains the multitude of philosophical
systems which may be considered attempts for looking at the world
from various angles, cf. sec. 6.8. Language and its inadequacy is a
central topic of Wittgenstein’s later philosophy; cf. sec. 5.4.
(3) Rising to a higher level . Another primary characteristic of
dialectic thinking is “thinking about thinking ”, higher–level thinking
about original thinking, above called “metathinking”. This is done
when discussing, in plain English (metalanguage), a computer program
performing a logical calculus (object language), or even in the informal
explanations in a book about symbolic logic such as (Carnap 1958).
Hegel’s dialectic corresponds to the sort of informal, non–formalizable pas-
sages of comment and discussion in a book like Principia Mathematica [by
Russell and Whitehead], rather than its systematic text, and it has the im-
mense importance of that interstitial comment.
(Findley 1963, “The contemporary relevance of Hegel”).
This “metathinking” is also of essential importance in the forma-
tion of the human personality. Who has never learned to question his
2.5. DIALECTIC THINKING 51

own actions and his own thinking, has missed the great opportunity
to become a human person. Formal “thinking” can also be performed
by machines and higher animals; only self–conscious “thinking about
thinking” is probably restricted to man only. We again remind the
reader what we have said about the dialectic nature of Gödel’s proof
at the beginning of the present section.
“Thinking about thinking” may involve several or even many levels
in the way of Fig. 2.9. A striking example, due to Poul Martin Møller,
may be found in (Bohr 1963, p. 13). A Danish student says:
I get to think about my own thoughts of the situation in which I find myself.
I even think that I think of it, and divide myself into an infinite retrogressive
sequence of “I”s who consider each other. I do not know at which “I” to
stop as the actual, and in the moment I stop at one, there is indeed again
an “I” which stops at it. I become confused and feel dizziness as if I were
looking down into a bottomless abyss, and my ponderings result finally in a
terrible headache.
According to Rosenfeld, this is a “delightfully humorous illustration of
Hegelian dialectics” (Folse 1985, p. 54). It also strongly reminds of
Fichte (Versuch einer neuen Darstellung der Wissenschaftslehre 1797,
II(2); Werke Band I, p. 526, simplified and modernized):
When you say that you are conscious of yourself, you distinguish your think-
ing “I” from the “I” about which you are thinking: the “I” as a subject
(S1 ) from the “I” as an object (O1 ). But in this process you are necessarily
regarding the subject S1 as the object O2 of a new subject “I” (S2 ), and so
on: S1 = O2 , S2 = O3 , S3 = O4 , etc. ad infinitum.
(4) The geometry of dialectics. If we represent syntactic “thinking
about objects” by a movement in a plane, then the dialectic triad (T ,
A, S) certainly does not correspond to a triangle lying in this plane.
The synthesis S rises above it (Fig. 2.8). To use the same metaphor,
dialectics corresponds to a third dimension, rising above the basic plane
of formal logic. This procedure may even be iterated, cf. Fig. 2.9.
The triad, or triangular motion, is by no means a necessary char-
acteristic of dialectic thinking, as we have already mentioned. In-
stead of the triad (T − A − S) Fichte frequently uses a 5–term pentad
(T − A − T A − A T − S), where T A is a partial synthesis with the
accent on T , and A T similarly with the accent on A, and the mathe-
matician Speiser (1952) uses a 7–term process in his attempt at a mod-
ern reconstruction of Hegel’s “Logic ”. Weizsäcker (1992) uses the term
“Kreisgang” (circular movement), in which the observer, so to speak,
52 CHAPTER 2. LOGIC AND MATHEMATICS

walks around the object in a circle, to look at it from all perspectives,

at the same time rising higher and higher, like in a circular staircase.
Geometrically, this corresponds to a spiral line or helix (Fig. 2.10).
The projection of this helix onto the basic plane is, of course, a circle
(Fig. 2.11), on which the projection of S coincides with T although, of
course, S is on a higher level than T (Fig. 2.10).

Figure 2.10: The dialectic helix

Figure 2.11: The dialectic helix only appears to be a vicious circle

Thus the dialectic helix is not simply a “vicious circle” correspond-

ing to the paradox of the liar, cf. Fig. 2.5 on p. 36, in which we remain
in the same plane, so that S really coincides with T . The liar’s paradox
and Gödel’s proof correspond to each other very much as a circle and
a helix (or perhaps the part of the helix in Fig. 2.10 which goes from
T to S).
The dialectic helix represents one way to get out of a vicious circle:
by rising higher and higher. Another way in which a vicious circle can
be made to lose its viciousness is iteration: here going around continu-
ously decreases the radius; the circle becomes a spiral (Fig. 2.12), which
2.5. DIALECTIC THINKING 53

Figure 2.12: An iterative spiral

contracts to the desired solution P . If we regard the iterative solution

of the equation x − φ(x) = 0 (Example 10 above) as a dialectic process,
then we go higher and higher with each iteration (Fig. 2.9, right), but
this is not really essential because we avoid the vicious circle anyway
by decreasing the radius to zero. So the iteration procedure described
in Example 10 can be considered to remain on the same mathematical
level.
A completely different geometrical interpretation of dialectic logic
is found at the end of sec. 2.6.
Fundamental principles of dialectics
(I) Unity of contraries. Synthesis is the unification of thesis and an-
tithesis at a higher level, or more generally, the unification of conflicting
tendencies at a higher level. This principle has been known at least to
Plato, who uses it in his dialogue “Parmenides ” (Speiser 1959). At the
end of the middle ages, Cardinal Nicolaus Cusanus (1401–1464) formu-
lates the principle of “coincidentia oppositorum ” (coincidence of oppo-
sites). Long after Fichte and Hegel, the physicist and philosopher Niels
Bohr, when awarded a title of nobility for his scientific merits, chose
the inscription “contraria sunt complementa ” (contraries are comple-
mentary) as motto for his coat of arms. As a matter of fact, we have
met Niels Bohr above as the pioneer of complementarity.
Dialectic thinking came natural to Niels Bohr: “The opposite of
a trivial truth is falsehood. The opposite of a profound truth may
well be another profound truth.” Or even more concisely: “A deep
truth is a truth whose opposite is also a deep truth.” Both statements
are condensed versions of a longer statement in (Bohr 1958, p. 66);
the first follows (Globus et al. 1976, p. 271). (Again A is not simply
∼ T of formal logic, because “profound truths” are too complex to be
expressible by the formalism of symbolic logic.)
The most famous example from philosophy is Hegel’s fundamental
triad: the synthesis of being and nothing is becoming. It is based on
54 CHAPTER 2. LOGIC AND MATHEMATICS

Plato’s saying (in the dialogue “Sophist ”) that “not–being” is a form

of “being” (Whitehead 1933, Chapter XV). It is also a central topic of
Eastern Philosophy (e.g. the I Ching). Other examples of “profound
truths” in the sense of Bohr will be given later in this book.
We have already mentioned the case of psychology. The human per-
sonality is a synthesis of many conflicting and contradictory trends. The
personality is the stronger, the more it is capable of synthesizing con-
tradictory tendencies, conflicting desires, and contrasting experiences
into a harmonious whole, thereby rising to a higher level of experience:
“Was uns nicht umbringt, macht uns stärker”. (What does not destroy
us, makes us stronger.)
(II) Negation of negation. As we have pointed out at several oc-
casions, the antithesis A is not the simple operation ∼ T (not–T ) of
formal logic. Hence the negation of the negation does not simply re-
produce the thesis T , but gives a synthesis S on a higher level as we
have remarked repeatedly. (This distinguishes genuine dialectics from
the liar’s antinomy where in fact A = ∼ T holds.)
Many examples have been given. Here one more example from
everyday life: driving an automobile.

Example 12.
T : steering an automobile
A: skidding (on ice, etc.) (negation)
S: countersteering to restore the course (negation of negation).

“Negation of negation” does not simply restore the original situa-

tion but implies progress (improved driving skill as a result of experi-
ence). We have essentially the phenomenon thesis – antithesis (nega-
tion) – synthesis (negation of negation). The progress is represented in
Fig. 2.10 by the vertical distance T – S.
This law, negation of negation, may be considered the logical ba-
sis of progress or evolution in human life, human history, and human
science. Contradiction between theory and experience (negation) leads
to improved theories (negation of negation). Even Darwin’s “survival
of the fittest” also fits into our category. A danger (negation), success-
fully overcome (negation of negation), strengthens the personality as
we have already remarked above.
(III) Inversion of perspective, “dialectic reversal ”. The best–known
example is the Copernican revolution: placing the center of the universe
at the sun (heliocentric system) instead of having the Earth in a central
position (geocentric system) essentially simplified the law of planetary
2.5. DIALECTIC THINKING 55

motion and led to Kepler’s laws and Newton’s mechanics. (The syn-
thesis lies on almost too high a level: according to Einstein’s general
theory of relativity (sec. 3.4), all reference systems and all origins are
theoretically (!) equivalent, cf. Example 3 on p. 46.)
By directing the attention of the observer from the object to his
own perceptual apparatus (cf. sec. 1.4), Kant claimed to have provided
a Copernican revolution in philosophy. In fact, it was followed by a
period of intense flourishing of philosophy (Fichte, Schelling, Hegel).
By looking at the antinomy of the liar from an unexpected angle,
Gödel was able to prove extremely profound and far–reaching theorems
in logic (sec. 2.3). An unprovable proposition is recognized to be true.
A mathematical example which is less trivial than it looks, providing
important
√ generalizations, is as follows.
2 is not a rational number; therefore it was not considered a num-
ber by the ancient Greeks. It can, however, be approximated with
arbitrary accuracy by the rational numbers

x1 = 1.4
x2 = 1.41
x3 = 1.414 (2.18)
x4 = 1.4142
..
.
The sequence x1 , x2 , x3 , . . . does not have a rational number√ as a limit,
it “leads to nowhere”. The solution is Cantor’s definition of 2 as pre-
cisely the entire sequence {x1 , x2 , x3 , . . .}! This principle is frequently
used in mathematics, cf. formula (2.8) on p. 27.
In adjustment theory (secs. 2.4, 2.6, and 4.4), the troublesome and
annoying measuring errors are re-interpreted as stochastic variables
with an interesting mathematical theory.
Mathematically “ill–posed problems” have become central in the
recently fashionable “chaos theory” (sec. 3.2) and “inverse problems”
(sec. 3.8)
Illness, pain, and grief if regarded from the proper point of view,
may be recognized as positive factors in the development of human
personality, “making a virtue of necessity”.
In art, this is the principle of tragedy.
In philosophy, finally, the contrast between materialism (matter is
primary, mind is derived from matter) and idealism (mind is primary,
56 CHAPTER 2. LOGIC AND MATHEMATICS

matter is derived) is, to a large extent, resolved by inverting the per-

spective, cf. sec. 5.3.
(IV) Passage of quantity into quality, and conversely. A classical
example is the boiling or freezing of water, but any phase transition in
physics or chemistry exemplifies this principle.
Such a phase transition is also the emergence of life from inorganic
matter, and the transition from animals to man. The phase transi-
tions are discontinuous (qualitative) but may be reached continuously
(quantitatively). An example from geometry may illustrate that such
a behavior is indeed possible (Fig. 2.13). As a matter of fact, this is
just an illustration and by no means a proof!

Figure 2.13: Illustration for the emergence of life (symbolized by vari-

ous polygons) from inorganic matter (various lines), and
of the emergence of the human mind (circle)

A rectangle, however narrow, is qualitatively different from a line,

and a circle is different from a regular polygon, however small its sides
and however large their number may be. Nevertheless, all transitions
are continuous!
Does it surprise you that Cohen and Stewart (1994, p. 436–441)
have linked emergence with Gödel’s theorem?
This dialectical theory of phase transitions seems to be particularly
attractive to natural scientists; cf. (Haldane 1939, p. 26). The excellent
theoretical physicist Kaku (1994, p. 210) writes:
This is the essence of dialectics. According to this philosophy, all ob-
jects . . . go through a series of stages. Each stage is characterized by a
conflict between two opposing forces . . . When the conflict is resolved, the
object goes to a higher stage, called the synthesis, where a new contradiction
begins, and the process starts over again at a higher level. Philosophers call
this the transition from “quantity” to “quality”.
2.5. DIALECTIC THINKING 57

Dialectics and logic

Formal logic presupposes exactly defined concepts, such as tree,
house, green: concepts that denote distinct objects or properties that do
not change with time. It is impossible, however, to rigorously separate
a tree or a house from its environment, and the color “green” is rarely
found in pure form. Hence, exact concepts are idealizations, as we have
already frequently seen. For such idealizations the laws of formal logic
such as

law of contradiction ∼ (p ∧ ∼ p)
(2.19)
double negation ∼ (∼ p) = p

hold.
If dialectics is to apply to the real world, it must be wider than
formal logic, somewhat like the “fuzzy logic” of sec. 2.4. Then these
laws may not hold, as the examples of sec. 2.4 and Examples 7 to 11
of the present section have shown. Thus we may say:
Dialectic logic is a logic of rational approximation.
This is not the whole story, however. Take, for instance, the process
of acquiring knowledge. According to Fichte, the “I” (the subject, T )
confronts the “non–I”, the surrounding world (the object, A). The “I”
continuously takes some information from the surrounding world and
thus gradually increases its knowledge through a continuous sequence
of syntheses (S).
In this process, the “I” of one year ago (or one minute ago), is
identical and yet not identical to the “I” now since, in the meantime,
I have undergone a development. The law of contradiction no longer
holds.
This important point was very well elaborated by Havemann (1964,
pp. 48-49). Thinking begins with the dialectic (not formal–logic!) con-
tradiction between identity and difference. Throughout the years we
change: I am certainly different from what I was 50 years ago, never-
theless I feel that somehow I have retained my personal identity, and
also legally I have definitely remained the same person.
Thus we may also say:
Dialectic logic is a logic of temporal evolution.
In several places above we have seen the importance of the concept
of complementarity due to Niels Bohr. Thus we say:
58 CHAPTER 2. LOGIC AND MATHEMATICS

Dialectic logic is the logic of complementarity.

Another aspect is the following. Formal logic only deals with well–
defined and distinct objects. If we want to think about our own think-
ing, as we have remarked at the very beginning of the present section, in
the context of a re-interpretation of Gödel’s theorem, then the subject
enters and logical paradoxes and antinomies may occur, as exempli-
fied by the paradox of the liar (sec. 2.3). There arise logical structures
which are beyond the reach of formal logic, e.g. “The thinking thinks
the thinking ” (Plotinus, A.D. 204–270). Subject, predicate, and object
are identical in this sentence, yet it is not meaningless. Thus we may
say:
Dialectic thinking is thinking about thinking.
Returning to sec. 2.3, we see that these are structures of self–
reference, or reflexive structures. The well–known book (Hofstadter
1979) is full of such structures. Thus we may finally say:
Dialectic logic is reflexive logic.
In his 3 volume–work “The Science of Logic”, Hegel proceeds from
synthesis to synthesis, from level to level. His culminating chapter is
entitled “The absolute idea”. This crowning concept implicitly contains
Hegel’s whole system √ and, in principle, permits to derive it. This is
somewhat similar to 2 “containing” all subsequent approximations
1.4, 1.41, 1.414, 1.4142, etc.
This example shows that Hegel’s claim is not so extravagant as it
looks. It really seems that Hegel’s logic is capable of deriving itself .
To see what this means, consider an all–comprehensive scientific
theory, fashionable under the term TOE, “theory of everything”. If
TOE is to explain “everything”, it also must explain itself !
In a sense, Hegel’s logic thus is the first TOE. As a matter of fact,
much of Hegel’s reasoning is “fuzzy” and difficult to follow. Neverthe-
less, if, some time in the future, a true TOE will be developed (which
I doubt), it must, in a way, have incorporated Hegel’s logic, hence it
must admit self–reference. More about this will be said in sec. 6.6.
Bootstrapping. “Deriving itself” is strongly reminiscent of “boot-
strapping”, pulling oneself up by one’s own bootstraps to get out of
a swamp, like the famous Baron Munchausen did. Learning a for-
eign language on the basis of one’s own language may need consider-
able effort but is simple “in principle”. A child learning his/her first
2.6. GEOMETRY: DIMENSIONS TWO TO INFINITY 59

language, however, is performing a remarkable feat of bootstrapping,

learning “more language” on the basis of “some language” already ac-
quired (Hofstadter 1979, p. 294). Bootstrapping is now also popular in
elementary particle physics, cf. sec. 6.6.
Dialectics and symbolic logic
Dialectic logic operates on several “levels”, cf. Fig. 2.8 on p. 46,
whereas symbolic logic remains always on the same level. A classical
example of this is Gödel’s proof mentioned at the beginning of this
section.
Attempts to formalize dialectic logic have failed. Thus dialectic
thinking cannot be implemented on a computer. As we have seen at
the beginning, dialectic logic rather is a higher–level informal commen-
tary. Still, a geometric interpretation is possible; see “The geometry of
complementarity” in sec. 2.6 below.
A deplorable source of misunderstanding has been the dialectic con-
tradiction. It is, of course, not the straight negation of symbolic logic,
but something much more subtle (for a simple model see “The geome-
try of complementarity” below). Dialectic thinking seems to be a habit
of thinking that must be practiced, like mathematics. Some of the
greatest formal logicians (Gödel, Whitehead, Smullyan) at the same
time are superb dialectic thinkers. We also mention the mathematician
Andreas Speiser (1952, 1955, 1959) and the physicist Niels Bohr (1934,
1958, 1963); cf. also (Folse 1985).

2.6 Geometry: dimensions two to infinity

‘O ϑεòς α
’ εὶ γεωµετ %ει̃.
Plato
Euclidean geometry has been a prototype of a rigorous axiomatic
system, from its discovery by Euclid around 300 B.C. to the present
day. The great philosopher Spinoza in the 17th century formulated his
philosophical system in the style of Euclid, with definitions, axioms,
and theorems. Until a hundred years or so ago, geometry was taught
in schools according to Euclid’s elements. Small wonder if boys and
girls got to hate Euclid. — Only recently, David Hilbert improved
upon Euclid by giving a fully rigorous axiomatic system of Euclidean
geometry in the modern sense of sec. 2.2.
The father of Greek geometry was the half–mythical figure of
Pythagoras (around 530 B.C.). He founded a philosophical school with
60 CHAPTER 2. LOGIC AND MATHEMATICS

a mystical theory about numbers: “Number is the essence of things”.

This sounds surprisingly modern: in fact, contemporary physics is thor-
oughly mathematical. He did not find the theorem of Pythagoras,
c2 = a 2 + b 2 , (2.20)
between the three sides a, b, c of a rectangular triangle (Fig. 2.14). This
fact was known much earlier, but Pythagoras gave the first exact proof .

Figure 2.14: The theorem of Pythagoras

In ancient Greece, “geometry” was more or less synonymous with

“mathematics”. The greatest philosopher, not only of Greece but of
all times, Plato (428–347 B.C.) had a particularly high regard of math-
ematics. His school was called “Academy”, and the entrance of the
Academy showed the inscription “No one ignorant of geometry may
enter here”. When one of his students asked what the occupation of
God was, he answered “God always geometrizes” (in Greek the motto
above the present section). Only dialectics is higher: the mathemati-
cians go out hunting, and the dialecticians use their catch (Speiser 1955,
p. 61).
Other Greek mathematicians such as Archimedes made great dis-
coveries too. In the Middle Ages there was hardly any progress in
mathematics or geometry.

Figure 2.15: Cartesian coordinates in the plane

A basic new discovery was made by the philosopher René Descartes

(Cartesius, 1596–1650): the Cartesian coordinates, which permitted
2.6. GEOMETRY: DIMENSIONS TWO TO INFINITY 61

to treat geometrical problems by algebraic methods. A point P in the

plane may be defined by its coordinates x and y referred to a rectangular
coordinate system xOy (Fig. 2.15). The distance s of P from the origin
O is given, according to Pythagoras’ theorem:
s2 = x2 + y 2 . (2.21)
In our usual three–dimensional space (Euclidean space ) we have
three coordinates, called x, y, z or x1 , x2 , x3 (Fig. 2.16) and the dis-
tance s becomes
s2 = x2 + y 2 + z 2 = x21 + x22 + x23 . (2.22)

Figure 2.16: Cartesian coordinates in three–dimensional space

We can visualize two–dimensional space (the plane) and three–

dimensional space. A space of dimension 4 is space–time with coor-
dinates x, y, z, t; t denotes time. Space of higher dimension cannot
be visualized, but the Cartesian method makes them accessible to an
algebraic treatment which is not essentially more difficult than for di-
mensions 2 or 3.
For dimension 4 we have
s2 = x21 + x22 + x23 + x24 , (2.23)
for dimension 5
s2 = x21 + x22 + x23 + x24 + x25 , (2.24)
and generally for Euclidean space of dimension n,
s2 = x21 + x22 + x23 + · · · + x2n , (2.25)
generalizing (2.21) and (2.22).
62 CHAPTER 2. LOGIC AND MATHEMATICS

Such n–dimensional spaces are of fundamental importance as aux-

iliary concepts in physics and in other natural sciences.
We may let n become greater and greater: n = 10, 100, 1000, a
million, a billion, etc. We may even let n become infinitely great, we
let n go to infinity, in mathematical symbols n → ∞. Then (2.25) may
be generalized to

s2 = x21 + x22 + x23 + · · · to infinity . (2.26)

A necessary and sufficient condition is that the infinite sum (2.26) “con-
verges”, as mathematicians say, that is, that s results as a finite number.
This is Hilbert space; it is the generalization of n–dimensional Euclidean
space for n → ∞.
Hilbert space is the mathematical tool for quantum mechanics, as
we shall see in sec. 3.5.
Spaces of higher dimension and Hilbert space appear awesome con-
cepts for the non–initiated. The main (or even only) problem is psycho-
logical. The psychological barrier is best overcome by practical com-
putations. For instance, solving a linear system of equations with 5
unknowns means that we work in 5–dimensional Euclidean space. And
if your pocket calculator solves a system with 10 unknowns, you work
in 10–dimensional space! What could be easier than that?
“But I cannot visualize these spaces!”, you say. Let me tell you
a secret. Nobody can visualize a higher–dimensional space, but all
mathematicians speak fluently about points, straight lines, planes, or
subspaces in n–dimensional space. The picture they have in mind when
they speak in this way, is our usual three–dimensional space (or even
the two–dimensional plane). Computing in n–space and at the same
time visualizing 3–space may sound a little schizophrenic but it is al-
ways done in this way, and it always works! So don’t worry. This
schizophrenic way of visualization even works in infinite–dimensional
Hilbert space, as we shall see in sec. 3.5.
Differential calculus. It was invented simultaneously and indepen-
dently by the great philosopher Leibniz (1646–1716) and by the perhaps
even greater physicist Newton (1642–1727). This led to a terrible quar-
rel about priority: even great personalities are only human beings.
What we need here is extremely simple: the Pythagorean theo-
rem (2.21) for infinitely small (infinitesimal ) differences dx, dy, and ds
(Fig. 2.17); ds is called line element. We obviously have

ds2 = dx2 + dy 2 . (2.27)

2.6. GEOMETRY: DIMENSIONS TWO TO INFINITY 63

Figure 2.17: The Pythagorean theorem for an infinitesimal triangle

What means “infinitesimal”? Somewhat loosely speaking, it means

“very small”, “as small as you like”, the smaller the better. For some
purposes, 1 meter may be “infinitesimal” (for instance in a map); if you
don’t like this, take 1 centimeter, 1 millimeter, or anything as small you
like. (Mathematicians will give you a more precise definition.)
Similarly, we have in three dimensions,

ds2 = dx2 + dy 2 + dz 2 (2.28)

and in n dimensions

ds2 = dx21 + dx22 + dx23 + · · · + dx2n . (2.29)

At this point, you may say, somewhat disappointedly, “But this is

quite easy!”. Dear reader, it is easy, believe me and don’t waste your
time looking for complications.
The following application is slightly more demanding mathemati-
cally. You can, however, safely disregard the formulas if you wish.

Geometry of least–squares adjustment

Consider a system of linear equations

a11 x + a12 y = l1 ,
(2.30)
a21 x + a22 y = l2 .
The coefficients aij are known, the left–hand sides l1 and l2 have
been measured, and x and y are unknowns to be determined. For
clarification, let me give a numerical example:

2x − 3y = 10 ,
(2.31)
x + y = 30 .
The solution is x = 20, y = 10, as we immediately find on substitution.
It frequently happens that more observations li are made, in order
to check the determination and to improve its numerical value. For
64 CHAPTER 2. LOGIC AND MATHEMATICS

instance, let a third observation l3 = 8 be made, which gives the system

2x − 3y = 10 ,
x + y = 30 , (2.32)
x−y = 8 .

The solution of the second and third equations gives x = 19, y = 11,
which obviously is different from the solution of the first and second
equations, namely x = 20 and y = 10. We say that the three equations
(2.32) are inconsistent.
Let

a11 x + a12 y = l1 ,
a21 x + a22 y = l2 , (2.33)
a31 x + a32 y = l3 .

be such an inconsistent system. To find a solution, we must slightly

change the li , obtaining

l1 + v1 = a11 x + a12 y ,
l2 + v2 = a21 x + a22 y , (2.34)
l3 + v3 = a31 x + a32 y .

We now introduce the very reasonable principle that the changes

v1 , v2 , v3 should be as small as possible, more precisely

v12 + v22 + v32 =⇒ minimum . (2.35)

This is best written in vector–matrix symbolism which should be

understood immediately. Equation (2.34) thus becomes

l+v = Ax , (2.36)

where
l1 v1
   
x
 
l =  l2  , v =  v2  , x= (2.37)
y
l3 v3
are vectors and
a11 a12
 

A =  a21 a22  (2.38)

a31 a32
2.6. GEOMETRY: DIMENSIONS TWO TO INFINITY 65

is a rectangular matrix. The condition (2.35) is thus abbreviated as

vT v =⇒ minimum , (2.39)

where
vT = [v1 v2 v3 ] (2.40)
is the transpose of the vector v in (2.37).
The solution can be written in the matrix form
 −1
x = AT A AT l , (2.41)

where M−1 denotes the inverse of a “square” matrix M.

The advantage of this symbolic notation is that l can be any n–
vector, x some m–vector, and A a matrix of n “rows” and m “columns”.
The only condition imposed on the integers m and n is that m < n, so
that we have an inconsistent system with a “standing” matrix A:

A n

What is interesting here is that the solution (2.41) of the problem de-
fined by (2.36) and (2.39) is a projection of a vector l in n–dimensional
space onto an m–dimensional subspace.

Figure 2.18: Adjustment as projection

66 CHAPTER 2. LOGIC AND MATHEMATICS

Terrible, isn’t it? Fig. 2.18 immediately clarifies the situation. If

the system of equation were consistent, then l would lie in the subspace
symbolized in Fig. 2.18 by a double line. Since they are inconsistent,
l lies a little off the subspace and must be projected onto it by adding
the vector v. Condition (2.35) requires that the vector v represents the
shortest distance of the point P from the subspace (line, plane, etc.).
Thus v must be orthogonal to the subspace.
The determination of the unknown vector x by the condition
T
v v =⇒ minimum is called least–squares adjustment. “Least–squares”
because the square sum of the “corrections” vi (i = 1, 2, . . . , n) must
be a minimum, cf. (2.35). “Adjustment” means that we must adjust
the data li slightly and “democratically” (equal treatment for v1 , v2 ,
v3 !) in order to fit the system. See the example of eqs. (2.13) and (2.14)
on p. 41.
Note that we have illustrated a projection of a vector in n–
dimensional space onto a m–dimensional subspace by the simple di-
agram of Fig. 2.18 where n = 2 and m = 1. This is simple and repre-
sents all the essential geometry. There is no need to visualize a general
n–dimensional space which nobody can do anyway.
It is like in a discussion in which all participants have slightly dif-
ferent opinions. In order to reach general agreement, all opinions (li )
must be slightly “adjusted”, everyone has to make some concessions
(vi !). It is, however, desirable that these concessions should be as slight
as possible (vT v =⇒ minimum).
Remember this principle of adjustment! We shall meet it again. If
you could not quite follow our mathematical argument, never mind.
Just remember the comparison with a discussion and the “democratic
adjustment” of the participants’ opinions.

Figure 2.19: The plane Ω as the “union” of two orthogonal straight lines
(subspaces) X and Y .
2.6. GEOMETRY: DIMENSIONS TWO TO INFINITY 67

The geometry of complementarity

(This is only for the courageous. If you find it too difficult, simply
skip it.) Geometry gives an ideal model to understand dialectic opposi-
tion (complementarity) and ordinary logical contradiction. Orthogonal
subspaces are also called complementary. Consider two mutually or-
thogonal straight lines X and Y (Fig. 2.19). They span the plane in
which our figure lies. We say that both straight lines X and Y are
complementary (mutally orthogonal) subspaces of the plane Ω, or that
Ω is the generalized union of X and Y :

Ω=X∪Y .

Because of the orthogonality, the (generalized) intersection of X and

Y is zero:
X ∩Y =∅ .
Now we write
X⊂Ω, Y ⊂Ω
Y = X̄ (not X), X = Ȳ (not Y )

to indicate that the lines X and Y are complementary subspaces of Ω.

Now, clearly, the vector a lies in the plane:

a∈Ω ,

and it lies in the subspace X:

a∈X , a 6∈ Y

and similarly for the vector b:

b∈Y , b 6∈ X .

For the vector c in Fig. 2.19 we have, however,

c 6∈ X , c 6∈ Y .

Let us compare this situation with the set theory of ordinary logic
(sec. 2.1). Here Ω is a set (represented by a square) which is decomposed
68 CHAPTER 2. LOGIC AND MATHEMATICS

Figure 2.20: The set Ω as the union of the two sets X and Y .

into two subsets (rectangles X and Y (Fig. 2.20). Again we have

Ω=X ∪Y
X ∩Y =∅
X⊂Ω
Y ⊂Ω
a ∈ X , a 6∈ Y
b ∈ Y , b 6∈ X

as before, but now

c 6∈ X , c 6∈ Y = X̄
is impossible. The straight boundary is considered to belong to the set
X but not to Y . Hence even for a point c on the boundary,

c∈X , c 6∈ Y .

Denoting

Ω − X = Y = X̄ (not X)
Ω − Y = X = Ȳ (not Y )

we have the mutually exclusive alternatives

c ∈ X or c ∈ X̄
p or ∼ p
p∨ ∼ p ,

which is the law of the excluded middle, eq. (2.11) on p. 29.

2.6. GEOMETRY: DIMENSIONS TWO TO INFINITY 69

The situation of Fig. 2.19 is essentially different, although we have

purposely used formally identical notations. Here we have

c 6∈ X , c 6∈ Y = X̄ ,

so that the law of the excluded middle

either c ∈ X or c ∈ X̄ (p ∨ ∼ p)

does not hold.

Thus the logic of complementary subspaces (Fig. 2.19) is essentially
different from the ordinary logic of complementary subsets of sec. 2.1
and Fig. 2.20.
In fact, Fig. 2.19 is for quantum logic what Fig. 2.20 is for ordinary
logic.
Fig. 2.19 contains, in a nutshell, all the strange quantum phenom-
ena of complementarity; it is the simplest example for a dialectic logic.
Thus dialectic complementarity, Y = X̄ according to Fig. 2.19, is essen-
tially different from “ordinary” logical negation, Y = X̄ according to
Fig. 2.20. Thus the “dialectic complement” of the X–axis is the Y –axis,
whereas the “ordinary” complement of the X–axis would be the whole
plane excluding the X–axis. Actually, in quantum theory (sec. 3.5),
the plane of Fig. 2.19 should be an infinite–dimensional Hilbert space,
but we do not want to misuse the patience of the reader who has kindly
followed us so far.
The main purpose of this example was to show that dialectic nega-
tion is meaningful. Here Hegel (dialectics), Bohr (complementarity),
and Weizsäcker (quantum logic) meet.
70 CHAPTER 2. LOGIC AND MATHEMATICS
 Part B
Natural Science
Chapter 3
Physics

3.1 Classical mechanics and determinism

Nature and Nature’s Laws lay hid in Night:
God said, Let Newton be! and All was Light.
Alexander Pope

The great Greek philosopher Aristotle (384–322 B.C.) believed that

the velocity of a body is proportional to the force to which it was
subjected. Ordinary experience seems to confirm this view. A horse
carriage moves the faster, the stronger the horses are. A body lying on
the floor does not move unless some force is exerted to drag it along.
Only Galileo Galilei (1564–1642) recognized that matters are not
so simple. A body lying on a very smooth and plane ice surface will
continue to move with constant velocity and in a constant direction
even if the initial force has ceased to act. To be sure, this body will
gradually slow down and finally stop, but the cause is friction. If there
is no friction, the movement will be continuous and will never come to a
stop. A space ship in intergalactic space will forever move with constant
speed along a straight line after the rocket engines have been shut off.
Thus Aristotle and common sense have been deceived by friction.
The correct law of motion in the absence of friction was discovered
by Isaac Newton (1642–1727). It has the form
mẍ = F . (3.1)
Here m denotes the mass, and F is the force. The position vector is
x = [x, y, z] , (3.2)

73
74 CHAPTER 3. PHYSICS

the velocity vector is its time derivative:

dx
ẋ ≡ = [ẋ, ẏ, ż] , (3.3)
dt
and the acceleration is the second derivative:
d2 x
ẍ ≡ = [ẍ, ÿ, z̈] . (3.4)
dt2
Thus Newton’s law of motion (3.1) says that the acceleration is
proportional to the force, and not the velocity as Aristotle thought.
In order to fully define the movement, in addition to the differential
equation (3.1) we need initial conditions: at a certain instant t = t0 ,
the position and velocity:
x0 = x(t0 ) , ẋ0 = ẋ(t0 ) (3.5)
must be given.
Assume motion under no force, F = 0. Then (3.1) gives
ẍ = 0 . (3.6)
The solution of this differential equation is
x = at +b , (3.7)
where the constant vectors a and b serve as integration constants. To
understand this, differentiate (3.7) twice:
ẋ = a , (3.8)
ẍ = 0 . (3.9)
Thus (3.6) is satisfied, what was to be shown. If we put t = 0 in (3.7)
and (3.8), we get
a = ẋ0 , b = x0 . (3.10)
Taking for the initial instant t0 = 0, we thus have a very instructive
interpretation of the integration constants a and b: they are nothing
else than the initial conditions (3.5).
Newton’s law of gravitation. Besides Newton’s law of motion (3.1),
we also have his law of gravitation:
m1 m2
F =G . (3.11)
l2
3.1. CLASSICAL MECHANICS AND DETERMINISM 75

Two point masses (Fig. 3.1) attract each other with a force F of mag-
nitude F , proportional to the masses m1 and m2 , and inversely propor-
tional to the square of their distance l; this is the famous inverse square
law . Here G denotes a universal constant, the gravitational constant.
We also have the equality of action and reaction: the two forces F1 and
F2 in Fig. 3.1 are equal in magnitude and opposite in direction. The
magnitude of both F1 and F2 is given by (3.11).

Figure 3.1: Illustrating the law of gravitation

If Newton’s law of gravitation (3.11) is used in the equation of mo-

tion (3.1), then this differential equation, on integration, gives the Ke-
pler ellipses, along which the planets move around the Sun.
Principles of mechanics. If the motion is subject to constraints,
the simple Newtonian equation of motion is no longer applicable. For
instance, frictionless motion of a particle constrained to move along
a curved surface cannot be along a straight line, even if there is no
external force, F = 0. The “straightest” curve on a surface is a geodesic,
representing the shortest line between two points that wholly lies in the
surface. If the surface is a sphere, then the geodesic is a great circle.
Now it can be shown that frictionless and forceless motion along a
surface really is motion with constant velocity along a geodesic. Even
this simple but important case is not covered by (3.1).
So for motion on a surface Newton’s equation (3.1) is not satisfied,
that is,
mẍ − F 6= 0 . (3.12)
If the left–hand side cannot be zero, then let us try at least to make it
as small as possible:

(mẍ − F)2 =⇒ minimum , (3.13)

subject to the given conditions, for instance, motion on a surface. This
is Gauss’ principle of least constraint.
It is in full analogy to the principle of least squares discussed in
sec. 2.6, eq. (2.35) on p. 64. In fact, (2.36) says that
A x − l 6= 0 , (3.14)
76 CHAPTER 3. PHYSICS

and (2.39) is equivalent to

(A x − l)2 =⇒ minimum . (3.15)

The analogy between (3.12) and (3.13), on the one hand, and (3.14)
and (3.15), on the other hand, is obvious.
Thus it is not surprising that both principles are due to Gauss, who
also recognized the deep analogy between them.
It may be shown that Gauss’ principle applied to a free particle on
a surface, does give geodesic motion. For the sphere, motion along a
great circle is obvious, cf. Fig. 3.2.

Figure 3.2: A free particle describes a geodesic on a sphere

For many other simple and complicated cases, Newton’s elementary

law (3.1) does not directly apply. A pertinent example is the rotation of
a rigid body, because Newton’s equations are essentially valid for point
masses only and do not apply to rotation. With respect to orbital
motion about the Sun, the planets may be considered point masses,
but Earth rotation must be treated in a different way.
A number of other principles, more general than Newton’s laws,
were proposed in the 18th century by d’Alembert, Lagrange and oth-
ers. This is subject of analytical dynamics, of which a non–specialist
account can be found in (Lindsay and Margenau 1957, Chapter III).
We have briefly considered only Gauss’ principle and shall now outline
Hamilton’s method.
Hamilton’s equations. The Newton equation (3.1) is in reality a
system of three ordinary differential equations of second order :

mẍ1 = F1 (x1 , x2 , x3 ) ,
mẍ2 = F2 (x1 , x2 , x3 ) , (3.16)
mẍ3 = F3 (x1 , x2 , x3 ) ,
3.1. CLASSICAL MECHANICS AND DETERMINISM 77

x1 = x, x2 = y, x3 = z denoting the Cartesian coordinates which are

the components of the position vector x, and similarly F1 , F2 , F3 for
the force vector F.
Now we introduce the auxiliary quantities

p1 = mẋ1 , p2 = mẋ2 , p3 = mẋ3 (3.17)

called momenta. The coordinates x1 , x2 , x3 are now denoted by

q1 , q2 , q3 . Then (3.17) and (3.16) become with i = 1, 2, 3:

1
q̇i = pi ,
m (3.18)
ṗi = Fi (q1 , q2 , q3 ) .

Thus we have reduced the three differential equations (3.16) of sec-

ond order by 3 + 3 = six differential equations of first order.
This method is standard in the theory of differential equations and
not particularly enlightening.
What is significant, however, is the fact that William Hamilton
(1788–1856) was able to bring (3.18) to the form

∂H
q̇i = ,
∂pi
∂H (3.19)
ṗi = − ,
∂qi

with one function H only, instead of the 3 functions Fi ! Because of

their importance, they are called the canonical equations of mechanics,
and H is known as Hamilton’s function or, briefly, as Hamiltonian. By
the way, H is simply the sum of kinetic and potential energy. Any
quantities pi and qi satisfying (3.19) are called canonically conjugate
variables.
The true importance of the Hamiltonian equations (3.19), however,
is the fact that qi need not be Cartesian coordinates but can be any
generalized coordinates (parameters), and i need not be restricted to
1, 2, 3 but can assume so many values as we need parameters to fully
describe the dynamical system. For instance, for a rotating rigid body
we need 6 parameters qi (i = 1, 2, 3, 4, 5, 6): three translations (along
the x, y, z axes) and three rotations (e.g., around the same axes). If
we have r particles, then we need 3r parameters qi : 3 for each particle.
78 CHAPTER 3. PHYSICS

Let us assume that we have n generalized coordinates qi . Then we

have 2n differential equations (3.19), and we can solve them uniquely
provided we have the 2n initial values qi and pi at time t = t0 .
Laplace’s demon.
An intelligent being which, for some given moment of time, knew all the
forces by which nature is driven, and the relative position of the objects by
which it is composed (provided the being’s intelligence were so vast as to be
able to analyze all the data), would be able to comprise, in a single formula,
the movements of the largest bodies in the universe and those of the lightest
atom: nothing would be uncertain to it, and both the future and the past
would be present to its eyes. The human mind offers in the perfection which
it has been able to give to astronomy, a feeble inkling of such an intelligence.
This impressive statement was given by Pierre Simon de Laplace
(1749–1827); the “intelligent being” has become famous as “Laplace’s
demon”.
This is the classical expression of causality or determinism: given
the equations of motion and the initial conditions at t = t0 , the state of
the system is exactly known at all earlier (t < t0 ) and all later (t > t0 )
times. Determinism reigned supreme until about 1925, when quantum
theory started thoroughly to shake it (sec. 3.5).
Recently, however, determinism has come under attack even from
its very stronghold, classical mechanics. This has been achieved by the
theory of chaotic systems (sec. 3.2).
The principle of least action. Instead of differential equations, clas-
sical mechanics can also be expressed by an integral minimum principle
of form
ZB
L dt =⇒ minimum (3.20)
A

where an integral (the “action”) of a function L is to be minimized. The

Lagrangian L is related to the energy and also to the Hamiltonian H
in a way which is not necessary for the present argument. Least–action
principles have been given by several scientists starting with Pierre
Louis de Maupertuis (1698–1759) and Leonhard Euler (1707–1783).
From the integral principle (3.20) it is possible uniquely to derive
the differential equations (3.19). This is of considerable philosophical
importance, for the following reasons.
An integral principle (3.20), minimizing (or maximizing) some
“overall” quantity, has been interpreted as expressing a tendency of
3.1. CLASSICAL MECHANICS AND DETERMINISM 79

nature towards perfection, attaining some ideal: maximum or optimum

sounds better than minimum, but is essentially the same thing. It thus
expresses a finalist tendency, a “causa finalis” in the sense of Aristo-
tle, cf. sec. 5.4. Such finalism occurs especially in biology (sec. 4.1).
It has been opposed to the causal determinism as exemplified by the
differential equations of classical mechanics.
The deduction of the deterministic equations (3.19) from the final-
istic integral (3.20) shows that both principles can coexist peacefully:
the principle (3.20), so to speak, creates its own differential equations
(3.19).
In a similar way we shall see in sec. 4.1 that a thermostat, governed
by a “finalistic” principle of producing a desired temperature, will “gen-
erate” its own physical “deterministic” differential equations that help
achieve the goal.
Thus causality, characteristic for classical mechanics, and finalism,
considered typical for biology, are far less incompatible as they first
appear, cf. also (Thom 1975, sec. 12.1.A).
“Causality”, so to speak, is the answer to the question “For which
reason?”, whereas “finality” answers the question “For which pur-
pose?”.
The basic results of the present section will also be needed to discuss
geodesic motion in general relativity (sec. 3.4) and a generalization of
Hamiltonian methods to quantum theory (sec. 3.5).
But also taken in itself, classical mechanics has an incredibly rich
structure. It comprises:
– causality: basic property;
– chaos: sec. 3.2;
– final causation: just discussed;
– constraints: eq. (3.13); and even
– “software laws” in a rudimentary form: as initial conditions (see
also sec. 4.5).
Ideas will be needed rather than formulas, so the reader need not
understand all mathematical details. Interested readers may consult
any textbook on theoretical physics; particularly suited for the present
purpose is the treatment in (Lindsay and Margenau 1957, Chapter III).
We also mention (Margenau 1950) which is less mathematical and more
philosophical and which is still a classic.
80 CHAPTER 3. PHYSICS

3.2 Deterministic chaos

In the beginning . . . there was Poincaré.
E. Atlee Jackson

The deterministic paradise of classical mechanics, over which

Laplace’s demon (sec. 3.1) exerted a rigid but essentially benevolent,
orderly, and stable regime, began to show, on closer inspection, some
strange and irritating features.
The application of mechanics to gases and fluids consisting of an
enormous number of particles (molecules) led to the statistical theory
of heat. Heat was explained as the random and irregular, more or less
violent motion of these particles. In view of the enormous number of
these particles, it is practically impossible to describe the trajectory
of every particle by Newton’s laws (even assumed that this would be
theoretically possible). Instead, these particles were treated statisti-
cally, which led to statistical mechanics or statistical thermodynamics,
created by Josia Willard Gibbs (1839–1903), Ludwig Boltzmann (1844–
1906) and others. A brilliant success was the derivation of the basic
equations of thermodynamics from the principles of classical mechanics
combined with statistical considerations. Temperature was explained in
terms of the average kinetic energy of the molecules; it is the higher, the
greater the average velocity of the particles is. The important concept
of entropy was introduced, and Boltzmann found his famous equation,
formula (4.3) of sec. 4.3.
But here a problem arises. The equations of classical mechanics are
time–reversible. This means that these equations retain their form on
replacing time t by −t. On the other hand, the equations of thermo-
dynamics are typically irreversible: the entropy in a physical system
always increases, see eq. (4.4). This contradiction must be due to the
introduction of statistics, either because of the enormous amount of
particles, or because of the incredibly complicated, “chaotic”, shape of
the trajectories of the particles (or both). These controverses, in which
already Boltzmann was involved, led to very important advances in
physics, mathematics, and probability theory (sec. 3.3), known by the
name of ergodic theory.
The French mathematician Henri Poincaré (1854–1912) found al-
ready in 1890 that even relatively “simple” nonlinear dynamical prob-
lems in astronomy etc. may admit extremely complicated, irregular,
even “chaotic” trajectories. In his classical work “Les Méthodes nou-
velles de la Mécanique céleste” (1899) vol. III, p. 389 he wrote:
3.2. DETERMINISTIC CHAOS 81

Imagine the figure formed by these two curves and their infinitely many
intersections . . . ; these intersections form a kind of meshwork, tissue, or
infinitely dense network . . . One is struck by the complexity of this figure
which I do not even attempt to draw. Nothing is better suited to give us an
idea of the complexity of the three–body problem and in general of all the
problems of dynamics in which there is no uniform integral [of the motion]
...
The modern theory of general nonlinear dynamical systems is con-
sidered to start with Poincaré’s work. The subject then lay relatively
dormant, known only to a few specialists, until 1954 when the fa-
mous Russian mathematician Andrei Kolmogorov (1903–1987) and his
younger colleague Vladimir Arnold started with a general and system-
atic treatment of such strange trajectories. In 1963 there followed an
independent paper on an application to meteorology by the American
Edward Lorenz. Then the subject exploded. Currently it is probably
the most popular subject of mathematics, known to a broad general
public.
Let me try to explain what Lorenz did. He took the equations of
mathematical weather prediction, simplified them and studied the so-
lution numerically with the help of a computer. These solutions proved
to be extremely unstable: two solutions with almost identical initial
conditions started to diverge wildly (Fig. 3.3). Since the data of me-
teorology are unavoidably insufficient and inaccurate, the initial con-
ditions are not exactly known; small deviations result in completely
different behavior. This is the reason why it is hardly meaningful to
make detailed weather predictions more than a few days ahead. (In
astronomy, predictions are good for tens or even hundreds of years, in
spite of Poincaré . . . )

Figure 3.3: Two unstable trajectories

Let us repeat:
stability: small causes produce small effects;
instability: small causes produce large effects.
82 CHAPTER 3. PHYSICS

Classical causality implicitly presupposes stability. Stability is the

environment in which Laplace’s demon thrives.
Unstable systems are mathematically described always by nonlin-
ear differential equations. Therefore, as we have already mentioned,
mathematicians speak of general nonlinear dynamical systems. (Popu-
larly speaking, the difference between “linear” and “nonlinear” is essen-
tially the difference between a straight and a curved line; the function
y = 2 + 3x is linear, whereas the functions y = x2 and y = sin x are
nonlinear.) Unstable nonlinear dynamical systems are nowadays widely
known by the name of chaos theory.
We distinguish between conservative dynamic systems for which
the total energy is conserved (e.g., those described by Hamiltonian
equations (3.19)), and dissipative systems for which part of the energy
is dissipated as heat, e.g., through friction.
The nonlinear systems of celestial mechanics as investi gated by
Poincaré, Kolmogorov, and Arnold are conservative. The meteorologic
systems studied by Lorenz are dissipative, because the atmosphere con-
stantly receives energy from the sun and radiates it again into outer
space: otherwise “global warming” would be very rapid indeed. The
name, chaotic systems, is particularly appropriate for meteorological
and similar dynamic systems.
Chaos theory is an outstanding example of a theory as an instru-
ment for discovery, a “searchlight”: now chaotic phenomena are found
everywhere, from clouds to earthquakes, and from turbulent mountain
streams to human heartbeats. Deterministic chaos, so to speak, is an
example of chaos out of order . There is also an emergence of order out
of chaos; cf. the derivation of thermodynamics from statistical mechan-
ics and sec. 3.3. In fact, both cases are closely interrelated and related
also to the production of order out of order , cf. sec. 4.3 (p. 182).
The historian of science, W. Schröder, tells me that the well–known
German meteorologist H. Ertel has found instability as the reason for
the impossibility of weather prediction beyond a few days already in
1941. Ertel must therefore be considered a predecessor of Edward
Lorenz in meteorological chaos. This is also true already for Poincarè
(1908), as the quotation in sec. 6.3 (p. 243) shows.
Suggested additional reading. There is an incredible amount of
books and papers on chaos theory. An advantage of its popularity is
the fact that there are outstanding presentations for the general public,
of high level but without formulas. An extremely readable introduction
is (Gleick 1988); Stewart (1990) is a fascinating presentation of all the
details but without formulas; an authoritative and very readable intro-
3.3. PROBABILITY 83

duction is (Lorenz 1993); and Abraham and Shaw (1984) managed to

present the intricate geometry, which was even too much for Poincaré
as his quotation shows, in beautiful pictures which should be accessible
to everyone with an interest in science. Applications to biology and
medicine may be found in (Glass and Mackey 1988). Chaos theory is
very popular also because its geometrical structures (fractals, strange
attractors) are of a truly exotic beauty. Particularly remarkable is the
combination of beauty and readability in (Briggs 1992). For statistical
mechanics and thermodynamics and their philosophical implications,
(Lindsay and Margenau 1957) is still unsurpassed.

3.3 Probability
God does not throw dice.
Albert Einstein

Nor is it our business to prescribe

to God how He should run the world.
Niels Bohr

A simple and extremely instructive example of an unstable motion

is throwing a die. The die is supposed to be a perfect, absolutely
homogeneous cube, whose faces are numbered 1, 2, 3, 4, 5, 6.
If we throw it, it will come to rest showing, say, face 3. If we throw
it again, trying to repeat the first throw as accurately as possible, it
may show a 6 (Fig. 3.4). The initial conditions defined by the way
of throwing may be almost identical; nevertheless the results will be
quite different and practically completely independent: instead of a 6,
we might as well have got a 4 or a 2.
This is a characteristically instable situation: an arbitrarily small
difference of initial conditions will give completely different and in-
dependent results. This is the typical situation of a chaotic motion
described in sec. 3.2, Fig. 3.4 corresponding fully to Fig. 3.3.
Even if we replace the human hand by a dice–throwing machine,
the initial conditions will never be exactly the same, and the result
is practically unpredictable. Theoretically its motion is determined by
classical mechanics (if also the impact of the air molecules is considered
a classical phenomenon), but prediction is hopeless. Laplace’s demon,
after having worried about the imprecise initial conditions, is then ad-
84 CHAPTER 3. PHYSICS

Figure 3.4: Throwing dice

ditionally bothered by Maxwell’s demon (responsible for air molecules,

cf. sec. 4.3).
The result of the fight between the two demons is a completely
random distribution of the results of the die: face 1 is as probable as
any other face. We may say that all faces have equal probability
1
p1 = p 2 = p 3 = p 4 = p 5 = p 6 = . (3.21)
6
We see that Newton’s laws, though theoretically applicable, are prac-
tically useless. Exit Newton, and Symmetry steps in and produces
the result (3.21). More prosaically, determinism loses importance and
symmetry takes over, producing order out of chaos.
(This is the reason why probability is treated here rather than in
Chapter 2 where logically it would seem to be better placed.)
Had the die been loaded , then, of course, symmetry would have been
destroyed and the probability of the various faces would be different.
(We, of course, would never use such a dirty trick!)
Such assumptions of equal probability, based on symmetry, were
used by Blaise Pascal (1623–1662) and contemporaries for a mathemat-
ical theory of games of chance. This was the foundation of the math-
ematical theory of probability. Laplace has perfected this symmetry–
based theory.
Here the important concept of symmetry appears for the first time.
A cube is symmetric because its six faces are geometrically equivalent:
they can only be distinguished by marking them with dots, from one
to six. If the faces were unmarked, then one could not distinguish a
cube lying on face no. 2 from a cube lying on face no. 5. So much
about geometrical equivalence or symmetry. A cube is also physically
symmetric if it is made of a homogeneous material : this is what we
3.3. PROBABILITY 85

mean by an unloaded die. A coin is symmetric if we disregard the

inscriptions on the two sides: then we could not distinguishing a coin
showing “head” from a coin showing “tail”: we would in both cases see
identical circles. We shall meet symmetry again; see secs. 3.6 and 4.2.
We have said that throwings of various faces were independent
events. Statistical independence is a basic concept, though it by no
means always holds. We shall, however, assume independence unless
the contrary is asserted.
Throwing a 3 or a 5 has a probability which is the sum:
p(3 ∨ 5) = p3 + p5 . (3.22)

Throwing a 3 and then a 5 is the product:

p(3 ∧ 5) = p3 p5 . (3.23)
These formulas do not presuppose equal probabilities (3.21), but they
do presuppose independence.
Now we remember symbolic logic, eq. (2.9) on p. 28. The “logi-
cal sum” of two propositions was symbolized by “∨”, and the “logical
product” by “∧”. Now the probability of a logical sum is the sum of
probabilities (3.22), and the probability of a logical product is the product
of probabilities (3.23).
Probability 1 corresponds to certainty, and probability 0 to impos-
sibility, and
0≤p≤1 . (3.24)
Obviously
p1 + p 2 + p 3 + p 4 + p 5 + p 6 = 1 . (3.25)
Thus the probabilities may be considered generalizations of or in-
terpolations between the truth values 0 and 1, cf. (2.15) on p. 41. (Note
the conflict of notations: in sec. 2.1, “p” stands for “proposition”, here
it denotes “probability”. As a temporary compromise, we have in (2.15)
symbolized probability by “P ”, but “p” is generally used in probability
theory.)
This can also be nicely expressed in the language of set theory
(Fig. 3.5). Throw a small particle at random in such a way that it
lands on set A with probability p(A) and on set B with probability
p(B). Both events may be considered independent if the two sets are
disjunct. The union A∪B consists of both sets A and B taken together.
Then
p(A ∪ B) = p(A) + p(B) (3.26)
86 CHAPTER 3. PHYSICS

Figure 3.5: Two disjunct sets A and B

in analogy to (3.22); cf. (2.12) on p. 29. The corresponding relation

p(A ∩ B) = p(A)p(B) (wrong!) (3.27)
unfortunately does not hold since the intersection A∩B = 0 for disjunct
sets. Here p(A ∩ B) would mean the probability that the particle lands
simultaneously on A and B, which is clearly impossible, so that A∩B =
0 implies p(A ∩ B) = 0.
Remark on terminology. The terms “probabilistic”, “statistic”,
“stochastic”, and “random” have more or less the same meaning and
are frequently used interchangeably.
Relative frequencies. Let us take an even simpler example, tossing
a coin. For an ideally symmetric coin the probabilities p1 of head and
p2 of tail are clearly equal:
1
p1 = p 2 = . (3.28)
2
If we throw the coin, say, a thousand times, there should be roughly
500 heads and 500 tails. In a real coin tossing experiment we may get,
say, 484 heads and 516 tails. Thus the relative frequencies of heads and
tails are
484
f1 = = 0.484 ,
1000
(3.29)
516
f2 = = 0.516 .
1000
If we throw 10000 times, we might get

5032
f1 = = 0.5032 ,
10000
(3.30)
4968
f2 = = 0.4968 ,
10000
3.3. PROBABILITY 87

which is clearly closer to p1 and p2 . It may be expected that, in some

sense, for n → ∞ throws

lim f1 = p1 = 0.5 ,
n→∞
(3.31)
lim f2 = p2 = 0.5 .
n→∞

In the so–called frequency theory of probability proposed by Richard

von Mises since 1928, it was suggested to define probabilities empirically
by such a limit
p ≡ n→∞
lim f . (3.32)

This, however, meets with mathematical difficulties because the in-

finite limit does not obey one of the more common limit definitions of
mathematics, and furthermore, it is not possible to perform infinitely
many coin tosses or similar procedures.
It is mathematically simpler and more elegant to introduce the con-
cept of probability axiomatically. This was done by A.N. Kolmogorov
in 1933. Here the probabilities were introduced generally, without spec-
ifying their numerical values, but subject to axioms such as (3.22) and
(3.23). Only later, approximate numerical values for them are found a
posteriori as relative frequencies such as (3.30), unless they were not
anyway given a priori by symmetry considerations (dice, coins).
The mathematical theory of probability has been developed to a
high mathematical level, including random functions (stochastic pro-
cesses) and Hilbert space techniques. Such techniques are, for instance,
applied in geodesy to determine the irregular gravitational field of the
Earth. This is called least–squares collocation and consists in an ex-
tension of least–squares adjustment (sec. 2.6) to infinite–dimensional
Hilbert space. Only for curious specialists we mention as reference:
H. Moritz: “Advanced Physical Geodesy ”, Wichmann, Karlsruhe, 2nd
edition, 1990.
Interpretations of probability. In sec. 2.4 we have already briefly
introduced subjective probability, expressing a degree of reasonable be-
lief, or just a degree of incomplete knowledge or of ignorance. Of such
character are the “probabilities of rain” given by American weather
forecasts mentioned in sec. 2.4.
The classical Laplace interpretation is clearly intended to be ob-
jective. When I calculate my chance to gain in gambling to be 95%
(p = 0.95), then I am not satisfied with this nice abstract result of
mathematics: I expect to gain concrete money.
88 CHAPTER 3. PHYSICS

Are physical probabilities subjective concepts or objective features

of nature? Consider statistical mechanics. In principle, presupposing
the validity of classical mechanics, we could calculate the trajectories
of all molecules without needing statistics. Statistics is needed because
we cannot do this in practice. Hence we do introduce statistics just
because of our inability or ignorance? This would indicate that our
probabilities are more or less subjective.
On the other hand, statistical mechanics provides important “emer-
gent ” concepts such as temperature or entropy, and an elegant theory
of thermodynamics has been developed on an axiomatic basis, with-
out needing mechanics or statistics. It seems clear that temperature or
entropy are objective “integral” properties of nature, and if they are
derived by statistics, this statistics should be more or less “objective”
as well. By the way, the derivation of thermodynamics from statistical
mechanics is a beautiful example of the emergence of a macro–law from
a micro–law . This is another example of order out of chaos.
A hundred years after Boltzmann, these questions are still being
discussed. To be sure, the mathematical formalism and its results are
completely unaffected by these “philosophical” discussions. Most work-
ing physicists could not care less whether their probabilities are sub-
jective or objective. Weizsäcker (1985, p. 100) writes: “The concept of
probability is one of the most striking examples for the ‘epistemologi-
cal paradox’ that we can apply our basic concepts successfully without
really understanding them.”
Whether “deterministic chaos” on the basis of classical mechanics
“really” introduces an objective probabilistic element into nature, is
still an open problem under discussion. Every physicist, however, agrees
that quantum theory does introduce objective probability into physics:
quantum fluctuations form the basic substratum of our world.
Objective probability has been vigorously defended, also in quantum
theory, by Sir Karl Popper (Miller 1985, sec. 15). He calls it propensity
and interprets it in the sense of Aristotle’s potentialities (possibilities)
which are not all realized but are nevertheless properties of nature. In
the progress of time, potentialities become actualities.
Summary. Probabilities have different interpretations, which are
presumably all needed.
(A) Probabilities of sets. The current standard mathematical the-
ory of probability, based on Kolmogorov’s axiom system, is considering
probability as a measure of sets. Any system of numbers which satisfies
Kolmogorov’s axioms is a possible system of probabilities.
Actual estimations of probabilities are done in two principal ways:
3.3. PROBABILITY 89

(1) By symmetry considerations. This is easy in the case of dice or

coins, but may even be possible in complicated physical applications.
(2) By relative frequency. The toss of a coin regards the toss under
consideration as one case out of an ensemble of 1000 or 10000 tosses,
cf. equations (3.29) or (3.30) above. Similarly, in physics, our “real”
physical system may be considered one out of a fictitious ensemble of
possible “similar systems”. This is the basis of Gibbs’ approach to
statistical mechanics (Lindsay and Margenau 1957, sec. 5.5).
Concerning the physical reality of probability or statistical consid-
erations, there are two possibilities:
(a) Probability is only a function of ensembles of physical systems;
probability considerations, such as in statistical mechanics, are
only done statistically because a (deterministic) treatment is too
complicated for us. Probability is a mathematical tool rather
than a physical reality. Hence it may be regarded as subjective,
at least to a certain extent.
(b) Propensity: this type of probability is a physical property of a
single physical system, as objective as its mass, energy, or velocity.
(B) Probability of propositions. Mathematically they are very
similar to probabilities (A), because the logical calculus of propositions
is very similar to the logical theory of sets (sec. 2.1). “Subjective” or
subjectivist probabilities of Carnap and others are of this type. (The
forecast: “There is a 20% probability of rain for tomorrow” is a sen-
tence, or in logical terms, a “proposition”.)
Degrees of credibility. Not everything which we call probability
must have a numerical value, or must be capable of being expressed
numerically. If we say that all our knowledge is only probable, if we
believe that the theory of relativity is very probably an outstanding
theory, if I say that my train next day will probably run reasonably on
schedule, it is difficult if not impossible to assign numerical values to
the “probability” expressed by such statements. We instinctively act
on beliefs with a high subjective degree of credibility as if they were
absolutely true, and we disregard theoretical possibilities which are very
small. When I go to work by car I know that I may have an accident. I
take this into account in a reasonable way, by insuring my car, having
my papers in order, and driving carefully. Having done this, I act as if
this eventuality will not occur.
If I kept in mind all the possible events which theoretically might
happen, but with a very low probability, then I would “probably” turn
90 CHAPTER 3. PHYSICS

crazy or at least become a “professional worrier”. This presumably is

what Bishop Butler had in mind when he said that probability is the
guide of life (Russell 1948, Part V, Chapter VI, p. 398).
In real life there is no absolute logical certainty, in the same way
as there are no real mass “points”, ideal straight lines or ideally exact
measurements, cf. sec. 2.4.
Suggested additional reading. Probability, especially of the subjec-
tive type, frequently is treated together with induction (to be consid-
ered in sec. 3.9). Our standard reference (Lindsay and Margenau 1957)
is slightly out of date on this topic but nevertheless worth reading.
There are many excellent books on mathematical probability. An easy
and delightful brief introduction by the most outstanding Russia spe-
cialists is (Gnedenko and Khinchin 1962). Geophysicists will not want
to miss (Jeffreys 1961, 1973). A recent excellent discussion of all in-
terpretations and their philosophical aspects is (Cohen 1989). Almost
all aspects of probability in their historical development from Blaise
Pascal to Niels Bohr are discussed with relatively little mathematics
but with beautiful physical intuition in (Ruhla 1992). The remarks in
Weizsäcker (1985, Chapter 3; 1992, Chapter 4) are brief but profound.

3.4 The theory of relativity

Henceforth space by itself, and time by itself,
are doomed to fade away into mere shadows,
and only a kind of union of the two
will preserve an independent reality.
Hermann Minkowski

Special relativity
Einstein’s special theory of relativity deals with inertial systems.
An inertial system according to Newton’s theory is a system on which
no force acts, so that equations (3.1) and (3.7) (p. 73–74) hold:

ẍ = 0 ; (3.33)
x = at + b . (3.34)

Eq. (3.33) says that there is no acceleration, and (3.34) says that the
motion of an inertial system is uniform, that is with constant velocity
along a straight line. An example is a spaceship in intergalactic space
3.4. THE THEORY OF RELATIVITY 91

whose rockets have been shut off and which, of course, continues to
move with constant velocity to reach the nearest galaxy.
A system which moves uniformly with respect to an inertial system,
is also an inertial system. According to classical mechanics, the two
are related by a Galilei transformation:

x̄ = x − vt ,
ȳ = y ,
(3.35)
z̄ = z ,
t̄ = t .
These equations can be directly found by inspecting Fig. 3.6.

Figure 3.6: Two inertial systems in relative motion

Let a photon (particle of light) move with velocity c with respect to

the car to which system x̄ȳz̄ t̄ is attached, and let the direction of the
photon coincide with the direction of velocity v. Then the velocity of
the photon with respect to the original system xyzt would be

c0 = c + v > c (?) (3.36)

Now it has been extremely well confirmed experimentally that c is

the maximum possible velocity. This is the Michelson–Morley experi-
ment; cf. also secs. 3.8 and 3.9. It shows that the velocity of light in
vacuum is constant and the same in every inertial system. Hence (3.36)
cannot hold, there must be

c0 = c . (3.37)

This, however, is incompatible with the formulas (3.35) for a Galilei

transformation.
92 CHAPTER 3. PHYSICS

The correct transformation is

x − vt
x̄ = q 2
,
1 − vc2
ȳ = y ,
z̄ = z , (3.38)
t − v2 x
t̄ = q c 2 .
1 − vc2

This is the Lorentz transformation. These formulas were first pub-

lished by the Dutch physicist H.A. Lorentz in 1904 in connection with
Maxwell’s theory of electromagnetism and extended by Henri Poincaré
in 1905. Its general validity for all physics, including mechanics, was
recognized by Albert Einstein (1879–1955) in 1905. Thus classical me-
chanics was replaced by Einstein’s special theory of relativity.
Several observations should be made:

(1) The velocity of light is constant in all inertial systems related by

(3.38), that is, c is invariant with respect to a Lorentz transfor-
mation.

(2) For small velocities, v  c or v/c  1, the Lorentz transforma-

tion (3.38) practica lly coincides with the Galilei transformation
(3.35). Thus for “normal conditions”, special relativity practically
coincides with Newtonian mechanics.

(3) The Lorentz transformation is absolutely counterintuitive: it is

against “physical common sense”. This is perhaps the first exam-
ple of a counterintuitive physical theory (a second, even more
counterintuitive, example is quantum theory), and has conse-
quently been violently attacked, even by “conservative” physi-
cists. This was at the beginning, around 1920–1930. Now, of
course, special relativity is a fully established physical theory be-
cause

(4) Special relativity has been confirmed in all relevant cases with-
out any exception. It is perhaps the most accurately confirmed
theory of physics. This is particularly important for high veloci-
ties v almost = c, which occur with very fast–moving elementary
particles in high–energy accelerators (CERN in Geneva, etc.).
3.4. THE THEORY OF RELATIVITY 93

Thus, a physicist denying special relativity would be taken even less

seriously than a biologist denying the theory of evolution.
It is true that the mechanics of Galilei and Newton had also been
considered counterintuitive by the defenders of Aristotelian mechanics,
but a few very simple experiments were sufficient to recognize that
Galilei and Newton were right.
Now, however, we are faced for the first time with the fact that
modern physics contradicts common sense. This trend continues to the
present day: physical theories are becoming more and more abstract
and counterintuitive.
From a philosophical point of view this is extremely important:
it shows that reality is much more complicated than we would have
believed a hundred years ago. The positive outcome is that we get
new models, however strange they look, which promote a much deeper
understanding, not only of nature, but also of classical philosophical
problems such as the nature of time and space, the structure of matter
and even of mind. This particularly holds of quantum theory (sec. 3.5).
The line element. It may be shown that the space–time line element

ds2 = dx2 + dy 2 + dz 2 − c2 dt2 (3.39)

is also invariant with respect to a Lorentz transformation (3.38). This

is even more striking if we introduce a new “imaginary time coordinate”

x4 = i c t (3.40)
√
where i = −1 is the “imaginary unit”. Then, and on putting x1 = x,
x2 = y, x3 = z as usual, we may write the line element in the form

ds2 = dx21 + dx22 + dx23 + dx24 (3.41)

which is nothing else than Pythagoras’ theorem for dimension four,

cf. (2.23) on p. 61! This was first recognized by the mathematician
Hermann Minkowski (1864–1909) who wrote the beautiful statement
used as a motto at the beginning of this section.
Thus space and time are welded together into a four–dimensional
space, called space–time continuum or briefly, space–time (even written
spacetime ).
Note that the Pythagorean or Euclidean form (3.41) requires an
imaginary time coordinate x4 . Much nonsense about this was and is
written in science fiction: time is no more imaginary than space, time
94 CHAPTER 3. PHYSICS

is no more unreal than space. What is true is that the imaginary trans-
formation is simply a mathematical artifice to bring the line element
into the Pythagorean form (3.41).
Beautiful as the form (3.41) is, however, it must be admitted that it
may give a false impression that space and time are not different at all.
An imaginary transformation, though easy mathematically, is a serious
operation from the point of view of geometrical and physical reality. A
circle
x2 + y 2 = 1 (3.42)
becomes, on replacing y by iy,
x2 − y 2 = 1 , (3.43)
which is a hyperbola and thus a quite different geometrical figure. This
simple example is relevant for relativity, as we shall see in sec. 3.7; cf.
also (Moritz and Hofmann–Wellenhof 1993, pp. 181–195).
Anyway, it is now customary to use as time coordinate, not x4 , but
x0 = ct , (3.44)
which is a simple real transformation of time scale, so that (3.39) be-
comes
ds2 = −dx20 + dx21 + dx22 + dx23 , (3.45)
which is almost as simple as (3.41), but in which all quantities are
real . It is called a pseudo–Euclidean form (mathematicians call it non–
positive–definite).

General relativity
Surface theory. For introductory purposes, let us first replace four–
dimensional space–time by a two–dimensional space, that is, a plane or
a curved surface. The line element in the plane is, of course,
ds2 = dx2 + dy 2 (3.46)
in Cartesian coordinates x and y. In polar coordinates r, φ with x =
r cos φ, y = r sin φ it becomes
ds2 = dr 2 + r 2 dφ2 . (3.47)
In general curvilinear coordinates (u, v) it may be written

ds2 = E du2 + 2F du dv + G dv 2 . (3.48)

3.4. THE THEORY OF RELATIVITY 95

This is quite natural: it must be quadratic, containing the squares du2

and dv 2 and, in the general case, the mixed product du dv as well. The
quantities E, F , G, introduced by Carl Friedrich Gauss (1777–1855),
are functions of u and v.

Figure 3.7: A curvilinear coordinate system

A curvilinear coordinate system is shown in Fig. 3.7. You may think

of a grid of geographic coordinate lines on a map or on a globe.
In fact, (3.48) holds indifferently for curvilinear coordinates in the
plane or on a curved surface. The distinction between a plane and
a curved surface can be made only on the base of a certain highly
nontrivial expression, the Gaussian curvature K, formed of E, F , G.
If K ≡ 0, the surface is a plane, otherwise it is curved.
A more modern way is putting E = g11 , F = g12 = g21 , G = g22
and u = u1 , v = u2 . Then (3.48) may be written
ds2 = g11 du21 + 2g12 du1 du2 + g22 du22 (3.49)
or more shortly,
ds2 = gij dui duj . (3.50)
Here the summation convention is used: a repeated index means sum-
mation over this index. In the present case, both i and j are summation
indices running from 1 to 2.
Curved space–time. The form (3.50) is particularly important. It
holds unchanged if the summation goes from 1 to 4 or from 0 to 3.
Thus (3.50) represents the metric also for four–dimensional space time!
If the coordinates ui are now denoted by xi (i = 0, 1, 2, 3), then
ds2 = gij dxi dxj . (3.51)
For instance, for the line element (3.45), the gij form the matrix
−1 0 0 0
 
 0 1 0 0 
[gij ] = 
 0
 . (3.52)
0 1 0 
0 0 0 1
96 CHAPTER 3. PHYSICS

Similarly to (3.49), however, the gij may also be functions of the coor-
dinates xi . The matrix [gij ] is generally called the metric tensor .
Similarly to the case n = 2, the gij may define a flat or a curved
space–time. The criterion, similar to the Gaussian curvature, is now
the Riemannian curvature tensor Rijkl , which is again formed of the gij
in a rather complicated manner. If

Rijkl ≡ 0 , (3.53)

then we have “curvilinear coordinates” in the flat space–time of spe-

cial relativity. These coordinates correspond to rotating or acceler-
ating (non–inertial) coordinate systems. The above–mentioned space
ship now no longer moves uniformly, but the rockets are working and
changing speed, direction, or orientation in space of the space ship. A
non–constant metric tensor [gij ] for which (3.53) holds, thus implies
“inertial forces ” in flat space–time, to which the crew of the space ship
is exposed.
To give a more down–to–earth example, such “inertial forces” are
well known from riding in an automobile which accelerates, brakes or
sharply turns around a corner. To be sure, they are no “real” forces,
but “only” due to having chosen a non–inertial reference frame, but tell
this to a passenger who has been inconvenienced, endangered, hurt or
even killed by such “unreal forces”!
The case of a non–zero Riemannian tensor is particularly interesting;
we may write
Rijkl = gravitational field . (3.54)
Curved space–time is equivalent to the existence of a “real” gravita-
tional field , expressed (approximately) by Newton’s law (3.11) on p. 74.
This is the essential content of Einstein’s general theory of relativity.
Informally we may write (3.54) as

gravitation = space–time curvature .

Newton’s law has an equivalent in this theory. A certain quantity is

formed from Rijkl , called the Einstein tensor Gij , and Einstein writes

Gij = κTij , (3.55)

where κ is a constant and Tij is the matter tensor , more precisely the
matter–stress–energy tensor. (The reader will have surmised in the
meantime that “tensor” is any quantity that carries indices. He is
3.4. THE THEORY OF RELATIVITY 97

basically right, at least for the present purpose. Let mathematicians

protest against such a sloppy definition.)
Now comes a fact of true philosophical importance. The tensor gij
represents geometry, Rijkl represents curvature and hence also geom-
etry, and so does the Einstein tensor Gij , derived in a purely math-
ematical way from gij . Hence the left–hand side of (3.55) represents
geometry. Since the tensor Tij represents matter and κ is only a con-
stant needed to get the dimensions right, (3.55) expresses matter in
terms of geometry, more precisely in terms of space–time curvature.
What is more, this is the first logically impeccable definition of
matter! The mass m in Newton’s equations (3.1) and (3.11) was only
an ill–defined constant.
Materialists will certainly not be satisfied with this highly abstract
definition of matter by (3.55). This is not the type of matter we know,
which we can kick, take into our hands, or throw, such as a stone or
a ball. It is not, however, the only possible definition of matter as we
shall see in sec. 5.1: there are worse definitions to come.
Preferred reference systems. In analytical geometry in the plane or
in space, Cartesian coordinates are clearly preferable because of their
simplicity. In special relativity, the same holds for inertial systems.
In general relativity, Einstein started from the principle of covari-
ance: the equations of general relativity are covariant, that is, invariant
with respect to general coordinate transformations. They have the same
form in every coordinate system. In curved space–time, inertial systems
are theoretically impossible; practically, systems can be defined which
are very close to such inertial systems and thus form a kind of preferred
systems. In Orwellian terminology, even in the perfect democracy of
reference systems in general relativity, where all reference systems are
equal, nevertheless some systems are more equal than others . . .
By the way, this fact of the practical existence of inertial reference
systems even in general relativity makes the transition to Newtonian
mechanics possible. Thus general relativity is usually taken into ac-
count by applying, to the results of classical mechanics, some small
relativistic corrections. They are on the order of 10−8 and can thus be
measured by contemporary exact techniques. For instance, relativistic
corrections are being routinely taken into account in the processing of
precise geodetic satellite measurements, for instance in satellite laser
ranging or also in precise uses of GPS, the well–known Global Position-
ing System.
Separability of gravitation and inertia. Besides the “real” force of
gravitation, there are also “apparent” inertial forces due to a rotation
98 CHAPTER 3. PHYSICS

or acceleration of the reference system (see above). For instance, people

sitting in a car usually remain sitting comfortably because gravitation
holds them to their seats, but they may start moving involuntarily if
the car takes a sharp curve or brakes abruptly: seats are for gravitation
and seat belts are against inertial forces.
From the time of Newton it has been recognized that gravitational
and inertial forces act in very much the same way, so that they cannot
even be separated. The reason is the equality between “inertial mass”,
the m in Newton’s law of motion (3.1), and “gravitational mass”, the
m in Newton’s law of gravitation (3.11). This equality has been verified
experimentally to very high degree of accuracy, but theoretically it was
a mystery until Einstein. In his general theory of relativity, Einstein
cut the Gordian knot by boldly affirming that gravitational and inertial
masses are identical because gravitational and inertial forces are essen-
tially the same phenomenon. They arise when the coordinate system in
space–time is curvilinear rather than inertial.
Again we have a nice Hegelian triad: thesis — inertial mass and
gravitational mass are conceptually different; antithesis: but they are
numerically equal; synthesis: this is because they are, after all, concep-
tually identical in the general theory of relativity.
At a single point, gravitational and inertial forces act inseparably
together. In fact, ordinary gravity to which we are all subjected (and
which is used to level geodetic theodolites etc.) is the resultant of
gravitational attraction of the Earth and of the centrifugal force of the
Earth’s rotation. We do not notice the two different components of
gravity, and no physical experiment can separate them.
If we wish, for some reason or other, to have only gravitation rather
than gravity, we can calculate the centrifugal force, which is given by
a very simple formula, and subtract it from measured gravity. Nothing
is simpler than this.
If we wish to measure gravitational force in an airplane, however,
matters are essentially more complicated: the inertial forces are so
irregular that they cannot be directly determined or computed. The
measurement of the gravitational force in an airplane is called aerial
gravimetry.
When I got involved, around 1966, in the theory of aerial gravimetry,
which attempts to separate gravitation and inertia, I casually talked
to a physicist about this. He immediately said: “This is absolutely
impossible: Einstein proved it”. Then I looked up the current literature,
and all books and papers confirmed this. The only exception was a book
on general relativity by J.L. Synge, which had been on the market only
3.5. QUANTUM THEORY 99

for a few years. There I found eq. (3.54), and suddenly everything was
clear.
The Riemannian curvature tensor Rijkl does separate gravitation
and inertia, cf. (3.53) and (3.54). It is formed by the metric ten-
sor gij and its space–time derivatives of first and second order (non–
mathematicians, don’t get scared and forget it!). Such derivatives, how-
ever, need gij not only at one point, but also in a surrounding region,
however small. Thus gravitation and inertia cannot be separated at
one point only, but they can be separated in an arbitrarily small neigh-
borhood of it! (The reason is that inertia has a much more regular
space–time structure than gravitation.)
Components of the tensor Rijkl can be measured, at least in prin-
ciple, by instruments called gradiometers. Thus a combination of
gravimeters and gradiometers provides, at least theoretically, a rigorous
method for separating gravitation and inertia.
This is important, not only in aerial gravimetry, where gravitation
is the “signal” and inertia is the “noise”, but also in inertial navigation
which is now used in almost every airplane: here inertia is the signal and
gravitation is the noise. One person’s signal may be another person’s
noise, as every musician knows from experience with his (her) neighbors.
Suggested reading. There are good nonmathematical presentations
of the principles of relativity. My favorites are (Lanczos 1965) and
(Will 1986). An “easy” but mathematical introduction is (Moritz and
Hofmann–Wellenhof 1993). Remarkable as a very readable introduc-
tion to geometry in general is (Lanczos 1970). The treatment in (Lind-
say and Margenau 1957) is excellent as usual. Philosophical problems
of relativity are treated on a high level, also freely using physics and
mathematics where necessary, in (Treder 1974).

3.5 Quantum theory

Natura [non] facit saltus.
Latin proverb

In the old small grocery stores it was possible to buy sugar, etc.,
in any reasonable amounts, e.g. 250 or 700 grams. In the modern
supermarkets, sugar comes only in packs of 1 kg, say.
It seems that nature furnishes energy also only in fixed packages of
a (of course, much smaller) constant size, called quantum of energy.
100 CHAPTER 3. PHYSICS

This discovery by Max Planck (1848–1947) in 1900 was applied

to light by Albert Einstein in 1905. Light somehow behaves as if it
consisted of identical particles, called photons. The same holds for the
quanta of electricity, the electrons, for other elementary particles, and
for the structure of atoms in general, as found by Ernest Rutherford
(1871–1937) and Niels Bohr (1885–1962).
This was a heavy blow to classical physics, contradicting the old
principle “Natura non facit saltus ” (Nature makes no jumps).

The formalism of quantum mechanics

A (up to now) final theory of quantum phenomena was furnished in
1925–1926 by the almost simultaneous and largely independent work
of several scientists of whom Werner Heisenberg (1901–1976), Erwin
Schrödinger (1887–1961) and Paul Dirac (1902–1984) are best known.
They did not develop three different theories, but three different aspects
of the same theory.
The basic idea is the correspondence principle. A quantity in clas-
sical mechanics corresponds to a linear operator in Hilbert space.
About infinite–dimensional Hilbert space we have already spoken
in sec. 2.6. A linear operator is the exact equivalence of a matrix
in n–dimensional Euclidean space; it may be considered an infinite–
dimensional matrix. (If you are scared, just skip the mathematics: we
shall be finished with it in a few minutes.)
A (symmetric square) matrix can always be brought into a diagonal
form:  
λ1 0 0 . . . 0
 0 λ 0 ... 0 
 2 
 .. 
L =  0 0 λ3 . 
 . (3.56)

 . .. .. 
 . ..
 . . . .  
0 0 . . . . . . λn
The λi are called eigenvalues. For linear operators, we simply have
n → ∞.
Physically it is basic that the eigenvalues of a certain operator con-
stitute the possible measured values of the physical quantity represented
by our operator .
A link to Hamiltonian classical mechanics (sec. 3.1) is the fact that
any canonically conjugate variables qk and pk (generalized coordinates
and momenta) are, in quantum theory, replaced by operators Qk and
3.5. QUANTUM THEORY 101

Pk satisfying the commutation relations

h
Pk Q k − Q k Pk = . (3.57)
2πi
For
√ the mathematically minded reader we explain the symbols: i =
−1 is the imaginary unit and h is a fundamental physical constant
called Planck’s constant. The operators P and Q, like matrices, are
not in general commutative (such as ab = ba for ordinary numbers).
For h = 0, (3.57) reduces to commutative relations, Pk Qk = Qk Pk . In
this case, quantum mechanics reduces to classical mechanics.
Operators act on functions just as matrices act on vectors. The
equation
L e = λe (3.58)
defines the eigenvectors e1 , e2 , . . . , en of the matrix L, and the cor-
responding numbers λ1 , λ2 , . . . , λn are, by definition, the eigenvalues
occurring in the diagonal of the matrix (3.56).
Let now L denote a linear operator. The precise analogue of (3.58)
is
Lψ = λψ . (3.59)
The function ψ is called state function, and there is usually an infinite
set of such ψ–functions: ψ1 , ψ2 , ψ3 , . . . (called eigenstates of operator
L). The corresponding numbers λ1 , λ2 , λ3 , . . ., the eigenvalues of L, are
the possible outcomes of measuring the physical quantity represented
by the operator L as mentioned above.
The mathematics, strange as it looks at first glance, is in reality
quite simple. As we have already remarked, linear operators (according
to Schrödinger) can be represented as infinite matrices (according to
Heisenberg), and Schrödinger’s state functions as Heisenberg’s infinite
state vectors. This analogy is now well known to any student of theo-
retical physics, to many applied mathematicians (Fourier transform!),
and also to theoretical geodesists, cf. (Moritz and Hofmann–Wellenhof
1993, Chapter 6). Matters are far less simple with
Problems of interpretation
State functions. Such functions ψ were introduced by the Aus-
trian Erwin Schrödinger; therefore the symbol ψ quite appropriately
decorates the Austrian 1000 Schilling banknote showing his portrait.
This, however, is not the most important property of the ψ–function;
much more important is the fact that it is difficult to find two promi-
nent physicists or philosophers of science who agree on its meaning.
102 CHAPTER 3. PHYSICS

Schrödinger himself was dissatisfied with all current interpretations

throughout his life. This is another striking example of Weizsäcker’s
“epistemological paradox” mentioned in sec. 3.3: we can perfectly well
work with concepts whose real significance we do not fully understand.
Suppose we measure a quantity L and get the value λ3 . Then we
know that the physical state of the system is described by the state
function ψ = ψ3 . It will evolve with time according to an equation
found by Schrödinger, giving ψ(t). Let us further assume that ψ(t) is
“normed” to correspond to a unit vector .
Now we perform the measurement of another physical quantity, rep-
resented by the operator M with eigenvalues µk and normed eigenfunc-
tions ξk (t). Then one of the eigenvalues µk must be observed, but we
don’t know beforehand which. But now comes the principal point: we
know the probability that a certain µk is measured. This probability is
precisely
pk =< ψ, ξk > , (3.60)
where the symbol <> denotes the inner product of two state functions.
If you know what the inner products of two vectors is, think of ψ and
ξk as vectors, and you have it; otherwise let it be just another of those
crazy mathematical terms and forget it. Thus we do not know whether,
say, ξ2 or ξ9 is measured, but we know the probabilities for ξ2 , namely

p2 =< ψ, ξ2 > ,

or for ξ9 , namely
p9 =< ψ, ξ9 > . (3.61)
If tonight we listen to the weather forecast for tomorrow, it might
be: p2 = 80% probability for rain and hence p9 = 20% for absence
of rain. (We are purposely using p2 and p9 , which is better for the
quantum analogy, rather than p1 and p2 as we should normally do.)
Next morning we look out of the window: sun is shining brightly. Thus
the event corresponding to the probability p9 has been realized (maybe,
St. Peter wanted to show the meteorologists who was who).
To return to our quantum measurement. We perform the measure-
ment of M and the outcome is µ9 , and hence the new state function is
ξ9 , for the very same reason that after measuring L with result λ3 the
state function was ψ3 .
Thus, as the result of our measurement, the state function, having
been ψ3 before the M –measurement, has suddenly become ξ9 after-
wards.
3.5. QUANTUM THEORY 103

It thus turns out that measurement changes quantum states discon-

tinuously, from ψ3 to ξ9 . Natura facit saltus!
Popularly speaking, a measurement disturbs a quantum state, and
it disturbs it unpredictably. From a physical point of view this is quite
plausible. Observing an electron by a microscope, say (assuming this
possible), means interaction of the electron with light, that is, we have
a collision between an electron and a photon. The outcome is as un-
predictable as with the collision of two cars, which is also subject to
probabilities (on the basis of which insurance companies calculate their
premiums).
This is expressed in the famous Heisenberg uncertainty relation.
They are a direct consequence of the commutation relations (3.57).
Any two canonically conjugate variables p and q cannot be measured
accurately at the same time. There are unavoidable measuring errors
∆p and ∆q related by the uncertainty relation
.
∆p∆q = h (3.62)

where h is again Planck’s constant. If q denotes a coordinate x, and p

the corresponding momentum p = m v (mass times velocity), Heisen-
berg’s uncertainty relation states that, if we measure the position x = q
accurately (∆q = 0), then the momentum p becomes indeterminate
.
(∆p = h/∆q → ∞) and vice versa. In fact, the light quantum used
in the microscope to measure the position x = q of the electron col-
lides with it and thus unpredictably disturbs its momentum p (size and
direction of its velocity). The measurement of coordinate q and the
measurement of momentum p are complementary events. In a related
way we say that the wave picture and the particle picture mentioned
below are complementary. Complementarity, introduced by Niels Bohr,
is closely related to dialectics (sec. 2.5).
But let us return to the discontinuous change from the state ψ = ψ3
to ξ9 . The difference with the car collision is that the car accident is
subject to the laws of mechanics and can therefore be precisely and “de-
terministically” described by them (in principle!). In the case of the
electron–photon collision, there are no underlying deterministic laws;
probability is essential and cannot be reduced to some “deeper” deter-
ministic laws.
Thus we have two ways in which a state can change: a continuous
“deterministic” evolution according to an equation called “Schrödinger
equation”, denoted by U (for “U nitary transformation”) by Penrose
(1989, p. 250), and a discontinuous irreducibly random change by the
104 CHAPTER 3. PHYSICS

observational process, called R by Penrose (“Reduction of the wave

packet”).
The term, wave packet, refers to the fact that, in a way, ψ rep-
resents “matter waves”. Electrons, besides being particles, in other
circumstances behave as waves (the “matter waves” of de Broglie and
Schrödinger), in much the same way as light, usually appearing as
waves, in certain respects also has particle character; there correspond:

light waves and photons,

matter waves and electrons.

The matter wave picture also beautifully illustrates the Bohr–

Rutherford model of the hydrogen atom consisting of an electron orbit-
ing around the atomic nucleus (Fig. 3.8). A stable orbit must form a
“stationary wave” containing an integer number of wave lengths (orbit
1). Orbit 2 does not satisfy this condition and hence is not possible:
the corresponding matter wave is discontinuous.

Figure 3.8: Orbit 1 is stationary (closed) and hence possible, orbit 2 is

not closed and thus is impossible

Stable (stationary) orbits thus form a discrete set: only a finite

number of orbits correspond to stationary waves. Hence discreteness
arises out of the continuity of waves: another beautiful example of
dialectics as discussed in sec. 2.5!
If the light is so weak that there is only one photon, then the light
wave may be considered to describe the probability of incidence of a
photon, say in a photon detector. In the same way, a matter wave
as represented by a function ψ may be regarded as a probability wave
describing the incidence of an electron.
The role of the observer . We have seen that the quantum observa-
tion has thrown the quantum state ψ3 into the new state ξ9 (Penrose’s
transformation R). To return to our meteorological example, has our
3.5. QUANTUM THEORY 105

looking out of the window changed the weather suddenly to sunshine?

In meteorology this opinion would be considered insane, in quantum
theory it is the generally accepted view.
What is causing “the reduction of the wave packet R”? Any answer
given so far seems to raise problems, especially since the observer’s mind
seems almost always to be involved.
Schrödinger’s cat. Schrödinger gave a famous striking example, a
“thought experiment”. Consider a cat sitting in a black box, without
being capable of being observed from the outside. In the box there is
a radioactive atom which is known to decay at an unknown random
instant. Let it be known that the atom, with equal probability, decays
or not during the couple of hours the cat is in the box (well equipped
with food, of course). If the atom decays, it triggers a device which
smashes a phial containing cyanide, killing the cat. At the end of the
experiment the observer opens the box and finds the cat dead or alive.
According to the usual quantum interpretation, only this observation
is determining whether the cat is alive or dead; before opening the box,
the cat cannot be said to be either alive or dead!
By means of this paradox, Schrödinger tried to express his unsatis-
faction with current interpretations of quantum mechanics.
Ingenious solutions of the cat paradox have been offered, none of
them entirely convincing: otherwise it would not be discussed even at
the present time (it was published in 1935!). Alluding to the stan-
dard “Copenhagen interpretation” of quantum mechanics, largely due
to Niels Bohr, Schrödinger ironically remarked: “I think I must accuse
Bohr — though in actual fact he is one of the kindest persons I ever
came to know — of unnecessary cruelty for his proposing to kill his
victim by observation.”
The Copenhagen interpretation. This interpretation, developed
mainly by Bohr together with Heisenberg, has been adopted by the
majority of physicists, more or less consciously. It has already been
used implicitly in much what has been said above. The wave function
ψ incorporates an objective and a subjective aspect, which are related to
Penrose’s transformations U (a “unitary transformation” or a rotation
in Hilbert space: objective) and R (“reduction of the wave packet” by
the observation: subjective). The observer as a subjective element en-
ters essentially, at least because he designs the extremely complicated
apparatus needed, but many physicists (e.g. E. Wigner) also ascribe an
essential role to the mind of the observer to register and thus fix the
outcome of the experiment: this is precisely what Schrödinger is allud-
ing to in the “accusation” just quoted that Bohr “proposed to kill his
106 CHAPTER 3. PHYSICS

victim” (Schrödinger’s cat) by observation. A classical brief but ideally

competent and readable presentation of the Copenhagen interpretation
is found in (Heisenberg 1958, Chapter III).
One of the difficulties with the role of the observer is the ques-
tion how quantum theory applied to nature before the advent of man.
Have dinosaurs also been subjected to the laws of quantum mechanics?
The answer of the Copenhagen interpretation is that the question is
meaningless: quantum theory describes, not nature as such, but our
interaction with nature by means of experiments: dinosaurs probably
did not perform quantum mechanical experiments. As Weizsäcker put
it: nature was before man, but man was before quantum mechanics.
But then the question arises again: which law took the place of quan-
tum mechanics before quantum phenomena came to the attention of
man around 1900?
It seems that such objections do not arise, at least not so conspicu-
ously, in the various “objective” interpretations which we shall discuss
now.
Every interpretation proposed so far contains some strange feature.
The strangest probably is the

Figure 3.9: Illustrating Everett’s interpretation with n = 3: the uni-

verse branches out into three equally real worlds

Many–world interpretation. This interpretation, proposed by H.

Everett in 1957, tries to avoid the arbitrariness involved in the real-
ization of just one possibility: in the above example, of ξ9 . Why not
ξ3 or ξ15 ? Everett says that in fact all possibilities are realized si-
multaneously. Hence with n (or infinitely many) possible outcomes
of the measurement (number of eigenvalues), the world forms n (or
n → ∞) branches (Fig. 3.9). At each measurement, the branching out
is repeated (Fig. 3.10). Thus the world containing Schrödinger’s cat
branches into a world in which the cat is alive and an almost identical
world in which the cat is dead.
3.5. QUANTUM THEORY 107

Figure 3.10: Repeated branches, all being equally real

It has been shown that this interpretation is perfectly logical and

self–consistent. It is also possible from a physical point of view (con-
servation of energy). According to Bohr’s dictum quoted as the motto
of sec. 6.6, Everett’s theory may well be “crazy enough to be true”.
Whitehead once said that it was more important that a proposition
were interesting than that it were true. In fact, Everett’s many–world
interpretation is taken seriously by a small but nevertheless surprising
number of most outstanding physicists such as Feynman, Gell–Mann
(1994, Chapter 11), Hawking, and Weinberg. It avoids the “reduction
of the wave packet” of the standard interpretation by replacing it with
a perfectly “objective” interpretation. Thinking, however, that the
incessant branching of the “world lines” creates an infinite number of
parallel worlds of which many are similar except for some detail and
some are quite different, makes many people feel dizzy and ask for an
application of “Occam’s razor” to cut off all branches except one. This
is the purpose of the
Propensity interpretation advanced independently by the philoso-
pher Karl Popper (Miller 1985, sec. 15) and the physicist Robert Have-
mann (1964, pp. 96–103). Here the probabilities are “propensities”, ob-
jectively existing tendencies. The difference with respect to the many–
world interpretation is that only one of the n possibilities is realized, cf.
Figs. 3.10 and 3.11. The probability of the branch chosen is, of course,
given by (3.60), now interpreted as a propensity.
The problem here is that it might look strange to consider the
propensities as real physical properties of nature. It may, however,
be only a prejudice to consider propensities less physical than, for in-
stance, forces. Both concepts are equally “physical” or equally “meta-
108 CHAPTER 3. PHYSICS

Figure 3.11: Propensity interpretation: only one possibility is realized

at each branching point

physical”, as Popper says. On closer look, also concepts like mass or

force lose their intuitive meaning acquired from every–day experience
with manual labor.
As we have said in sec. 3.3, the philosophical foundation of Popper’s
view is the Aristotelian distinction between potentiality and actuality
(reality), putting “the propensities as potentialities into things”. This
is clearly related to ideas of R. Bošković, cf. sec. 5.2, p. 210.
The physicist Havemann arrives at almost the same interpretation
starting from the Hegelian dialectics of potentiality and reality. By the
way, the booklet (Havemann 1964) appears to me to be by far the best
application of Hegelian and materialistic dialectics to contemporary
physics.
The ensemble interpretation. For dialectic materialists in the for-
mer Soviet Union, the propensity interpretation seems to have been
not “materialistic” enough. There is, however, no logical difficulty in
replacing the propensity interpretation of probability by a frequency
interpretation (sec. 3.3), ascribing probability not to one system, as
Popper did, but considering an ensemble of similar systems. The little
but beautifully written book (Blokhintsev 1968) appeals to the physi-
cist but is also very instructive and highly pertinent to our philosophical
discussion.
The holistic interpretation. As recognized already by Schrödinger,
quantum mechanics has a strong holistic flavor: “the whole — Greek
holos — is more than the sum of its parts” (cf. sec. 2.1): a quantum
system cannot simply be decomposed into a sum of independent subsys-
tems. Holism comes natural to Eastern — Indian, Chinese, Japanese —
philosophy, as particularly well (Rabbi type 1) described in (Capra
1975) and (Moser 1989). The most profound presentation (Rabbi type
2) is (Bohm 1980). (For “Rabbi types” see the Preface.)
According to Bohm, the real world is much more complex than
3.5. QUANTUM THEORY 109

our three–dimensional space or four–dimensional space–time. Loosely

speaking, the “many worlds” of H. Everett are enfolded into a higher–
dimensional universe from which our space–time world is only a certain
projection down to four dimensions. Quantum effects can propagate
faster than light (without, of course, carrying information for which the
light velocity c is the limit after all). A very rough model, which does
not pretend in any way to be realistic, is shown in Fig. 3.12. Besides the
normal space–time route with maximum velocity c there is a shortcut in
the form of a “jump across hyperspace” AbB, to follow the terminology
of the well–known “Foundation Trilogy ” by the science–fiction author
Isaac Asimov.

Figure 3.12: A “fold” in the universe. Effects can propagate along the
“normal” space–time route AaB, but also “cut across”
the fold following route AbB

The shortcut AbB demonstrates the “nonlocality ” of quantum me-

chanics, which exists in any interpretation of quantum mechanics but
comes out particularly well in the holistic interpretation. It says that
the universe is essentially more complex than the four–dimensional
space–time of Albert Einstein. Einstein had many discussions with
Niels Bohr, in the course of which he invented, as a thought exper-
iment, the EPR (Einstein–Podolsky–Rosen) paradox , no less famous
than Schrödinger’s cat (remember, it was “killed by observation”).
The general quantum–mechanical fact underlying non–locality was ex-
pressed 1965 by John Bell in the Bell inequality, which was empirically
tested by Aspect and others in 1982 and several times since.
Bohm’s interpretation with its hidden enfolded reality reminds of
Kant’s “Ding an sich” (thing in itself) which remains forever unknown
(sec. 5.4).
Concluding remarks and further reading. Quantum theory and rel-
ativity, particularly Einstein’s special theory, are as well tested as clas-
sical mechanics to which they reduce for ordinary macroscopic systems
(h → 0) and small velocities (v/c → 0). So far, within the limits of
110 CHAPTER 3. PHYSICS

their validity, none of these theories have ever been contradicted by ex-
periment: Popper’s “falsification” (sec. 3.9) has still to come for them.
For the working physicist, the results hold independently of the
philosophical interpretation. The philosophical interpretation of quan-
tum theory is much more difficult and controversial (and therefore even
more interesting) than the interpretation of relativity. The interpreta-
tions of quantum mechanics are closely related to the various interpre-
tations of probability. All interpretations described above seem to be
logically impeccable; their preference may be a matter of taste. (I like
them all.)
The basically probabilistic character of quantum theory holds, of
course, independently of interpretation. The probabilities (3.60) of
transition from state ψ to state ξk exist and cannot be replaced by
something more definite. The radioactive decay (cf. the example of
Schrödinger’s cat) is intrinsically random: the exact instant in which
the radioactive atom decays is unpredictable; it is subject only to a
probabilistic law. Thus quantum theory seems to provide a universal
background of essentially random fluctuations (“vacuum fluctuations”).
Quantum theory particularly clearly shows the relevance of modern
physics to philosophical questions such as the nature of matter and
mind, reality, potentiality, and probability.
It shows, however, also the relevance of philosophy to physics: we
have met with ideas of Aristotle, Kant, Hegel, dialectic materialism,
and Eastern philosophy; and the overarching influence of Plato will
become evident later.
I cannot resist the temptation to call quantum theory a dialectic
synthesis of extreme (mathematical) simplicity and extreme (philosoph-
ical) complexity.
The practical relevance of quantum theory is remarkable. Among
many other achievements, it allows to reduce chemistry to physics.
Quantum mechanics explains the structure of the atoms of chemical
elements, the formation of molecules from such atoms, i.e., the nature
of chemical bonds, etc.
Besides the books already quoted, we mention some more general
works. It is recommended to first read (Davies and Brown 1986), a
wonderfully readable (Rabbi type 1) introduction and overview of the
major interpretations with interviews of the leading proponents. (Grib-
bin 1984) comes next in readability. Bohr’s (1958, 1963) books are basic
but rather difficult (Rabbi type 2). Interpretation problems of quantum
theory with emphasis on the Copenhagen interpretation are dealt with
extensively in the books by Heisenberg and Weizsäcker which are much
3.6. ELEMENTARY PARTICLES 111

more readable. Contemporary reviews are (Lockwood 1989) and (Stapp

1993), both a “must” for those who wish to penetrate more deeply into
the mysteries of quantum theory (Rabbi type 1 to 2). (Penrose 1989)
is excellent but rather difficult (almost type 2). (Schrödinger 1958) is
a classic, popular in the best sense.

3.6 Elementary particles

Three quarks for Muster Mark!
Sure he hasn’t got much of a bark
And sure any he has it’s all beside the mark.
James Joyce, Finnegans Wake

History and description

The first elementary particles were the electron (the quantum of
negative electricity) discovered by J.J. Thomson in 1897, the photon
(quantum of light) found by Einstein in 1905, and the proton (the
nucleus of the hydrogen atom) which appeared first implicitly in the
Bohr–Rutherford model of the hydrogen atom around 1910–1913. The
hydrogen atom was considered to have a nucleus consisting of a posi-
tively charged proton, around which the negative electron is orbiting in
various “permissible” orbits. The proton was discovered explicitly by
Rutherford in 1919. Curiously enough, the nucleus of the helium atom
consisting of two protons, had been isolated already around the turn
of the century and was known by the name of α–particle (β–particles
were electrons, and γ–particles are photons).
The mass of the positively charged proton was found to be about
1836 times as great as that of the electron. In 1932, the British physicist
James Chadwick discovered the neutron, having a very similar mass
(1840 electron masses) but being electrically neutral, hence its name.
All atoms were found to consist of protons, neutrons, and electrons.
Since all atoms could be reduced to these three particles, it was believed
that these three were the only ultimate constituents of nature, the only
elementary particles. They were described by the recently discovered
quantum theory, and together with relativity the physical explanation
of the universe appeared to be essentially finished. Physics had become
a routine job elaborating the details of atomic and molecular structure
by known experimental and theoretical methods. The end of physics
appeared to be in sight: its job had almost been done.
112 CHAPTER 3. PHYSICS

Such stages of “near completion” of a science in general and physics

in particular are well known from the history of science. They fre-
quently signal the transition to a new “paradigm” in the sense of Kuhn
(sec. 3.10). The mechanist world picture of Newton–Laplace seemed to
explain almost everything in physics, even heat as statistical mechan-
ics and electromagnetism as the mechanical oscillations of an elastic
“ether”. This mechanist picture reigned supreme: the discovery of rel-
ativistic and quantum phenomena came entirely unexpected.
When Stephen Hawking assumed the famous Lucasian chair in
Cambridge (previously held by Newton and Dirac) in 1979, the ti-
tle of his inaugural lecture was: “Is the End in Sight for Theoretical
Physics?” The new unified theories of supergravity and superstrings
seemed to indicate that we might be close to a final physical theory.
Now we seem to be farther away from this goal than ever before, cf.
sec. 6.6.
But let us return to the situation with quantum theory and atomic
and particle physics around 1930. Photons and the constituent particles
of atoms: electrons, protons, and neutrons were all known.
But in the early thirties, new elementary particles were discovered:
the elusive neutrino which practically does not interact with matter,
so that even the Earth is “transparent” to it, and it passes straight
through; and the positron, the positive analogue to the electron. Thus
in 1933, already 7 particles (proton, neutron, electron, positron, 2 kinds
of neutrino, and photon) were known.
The photon is the quantum of light, that is, of electromagnetic
waves. The electromagnetic force binds electrons to the atomic nu-
cleus. In 1935, the Japanese physicist H. Yukawa suggested that the
nuclear force which holds protons and neutrons together in the atomic
nucleus, also corresponds to a particle, called meson.
By 1947, some 14 elementary particles were known. Besides the
7 mentioned above, there were the “antiparticles” antiproton and an-
tineutron, and 5 kinds of mesons.
Some of them were found in cosmic radiation, but the decisive exper-
imental event was the construction of more and more powerful particle
accelerators such as cyclotrons and linear accelerators. In these ma-
chines, the particles (protons, electrons, charged heavier atomic nuclei)
are accelerated to extremely high velocities and correspondingly high
energies; they are made to collide with each other, and new particles
are produced in this high–energy collision. Famous in Europe is CERN
(European Council for Nuclear Research) in Geneva, Switzerland.
Now the number of known elementary particles started to increase
3.6. ELEMENTARY PARTICLES 113

exponentially. Instead of the atomic–particle triad around 1930 (pro-

ton, neutron, electron), we now know hundreds of elementary particles.
Scientists have been expelled from the original particle paradise to find
themselves in a big particle zoo full of strange and extremely expensive
animals.
Even now, most properties of the chemical elements, of their
molecules and of the atoms of which they are made up, can be explained
by the triad: proton, neutron, and electron. The other particles occur
naturally in cosmic radiation; most of them, however, are “manufac-
tured” artificially in the big accelerators. Nevertheless they are also
essential for at least two reasons:
(1) They may occur in the process of formation of stars.
(2) They are necessary to set up general unified theories of physics
and to test them experimentally.
New particles now are hardly discovered accidentally; they are predicted
by theory and then systematically searched for by experiment.
Antimatter . In 1928 Dirac found his famous equation for the elec-
tron, which was the first successful unification of quantum theory with
special relativity (a convincing unification of quantum theory with
general relativity is still an open problem). It was soon discovered that
Dirac’s equation admitted a second solution, identical to the electron
except for its electric charge. This is the antielectron, called positive
electron or briefly positron.
Soon it was found that each elementary particle has its anti–
counterpart, called antiparticle. There is a negative antiproton, even
an antineutron (also electrically neutral, but electric charge is not the
only criterion). Only the antiparticle of the photon is the photon itself.
When a particle meets its antiparticle, they mutually annihilate
each other, generating a large amount of energy, according to the fa-
mous equation of Einstein: E = mc2 , energy equals mass times the
square of the light velocity. Remote galaxies which are apparently in
violent explosion may be due to a galaxy consisting of ordinary mat-
ter colliding with a galaxy consisting of antimatter. At any rate, this
happens in science–fiction, in which antimatter is very popular. Fortu-
nately, in our world antiparticles usually do not occur naturally; this
unsymmetry or “symmetry breaking ” between particles and antiparti-
cles is rather mysterious.
Spin. An elementary particle possesses an intrinsic angular momen-
tum, called spin, which is not due to actual rotation in space but is an
114 CHAPTER 3. PHYSICS

abstract quantum phenomenon. The spin number of a particle may be

one of the values
1 3 5
s = 0, , 1, , 2, , . . . (3.63)
2 2 2
and their negative values. In fact, the angular momentum correspond-
ing to spin s is
h
s (3.64)
2π
which shows spin to be a typical quantum phenomenon (it becomes
zero if h → 0, in the transition from quantum mechanics to classical
mechanics). Mesons have spin 0, electrons, protons, and neutrons have
spin 1/2, photons have spin 1, and gravitons (the hypothetical particles
corresponding to the gravitational force) would have spin 2.
Elementary particles can be classified by their spin. Particles with
integer spin are called bosons, the others fermions. A different division
is between leptons and hadrons.
Leptons (Greek: particles of light weight). There are 6 leptons
which all have spin 1/2: the electron, and two similar but much heavier
and extremely shortlived particles (muon and tauon ) and three neu-
trinos corresponding to the first three negative particles (we are here
disregarding the corresponding antiparticles).
Hadrons (Greek: strong particles). They consist of three groups:

(1) Mesons: particles of medium mass, especially the pi–meson

(muons and tauons are sometimes also called mesons).

(2) Nucleons: proton and neutron.

(3) Hyperons: they are heavier than nucleons and extremely short-
lived (so–called Λ, Σ, Ξ, and Ω particles).

Nucleons and hyperons together are called baryons (Greek: heavy par-
ticles).
Hadrons consist of smaller particles called quarks (whimsically
called after the quotation by James Joyce which figures as the motto
of the present section). There are some 18 different types of quarks
with deliberately nonsensical names, of different masses and different
electric charges and other properties.
Mesons are formed by the combination of two quarks, and baryons
(nucleons and hyperons) are made up of three quarks. (Leptons seem
to be truly elementary, not made up of any smaller particles.)
3.6. ELEMENTARY PARTICLES 115

Orders of magnitude. A typical size of a molecule is 1 nm (1 nanome-

ter = 10−9 m), of an atom 10−1 nm, of an atomic nucleus 10−5 nm, of
a proton 10−6 nm, and of a quark less than 10−9 nm. Electrons and
quarks are practically pointlike. (Remember that the wavelength of
visible light is about 500 nm (sec. 1.3).) The size of an atom is about
10000 nuclear radii, and the size of a proton is at least 1000 times larger
than the size of a quark. (This is even greater than the ratio between
the orbit of the Moon and the size of the Earth, and between the orbit
of the Earth and the size of the Sun!) Thus both an atom and a proton
consist essentially of empty space in which, however, extremely strong
forces act.
The four forces. The basic forces occurring in nature are:
– gravitation,
– electromagnetism,
– the weak force and,
– the strong force.
Gravitation and electromagnetism are sufficiently known to us. The
weak force occurs in the interaction between leptons or more popu-
larly, in radioactive phenomena. The strong force is responsible for the
interactions between hadrons; it holds the nucleus together.
Exchange particles. The electromagnetic interaction between nu-
cleus and electrons may be regarded as an exchange of photons, as we
have already remarked. Similarly, a quantum theory of gravitation has
been seen to lead to gravitons. Yukawa looked for the meson as ex-
change particle for the strong force between nucleons. The weak force
(between leptons) similarly corresponds to what has been called W and
Z particles which are extremely heavy: 80–90 proton masses. Finally,
quarks are held together by gluons (English “glue” with a Greek touch).
Summary. This all sounds quite complicated, so what should you
really know? Most particles are fermions of spin 1/2: electrons, pro-
tons, neutrons (generally all leptons and baryons), as well as quarks.
All exchange particles are bosons:
electromagnetism photon spin 1
gravitation graviton spin 2
weak force W −, W +, Z spin 1
strong force gluons spin 1
[mesons spin 0]
116 CHAPTER 3. PHYSICS

(Yukawa’s mesons (pi–mesons or pions) have been considered respon-

sible for the strong force between neighboring nucleons (protons and
neutrons), whereas gluons act between quarks. Now Yukawa’s theory
may appear superseded by the gluon theory of the strong force, but its
pioneering role is universally recognized.)
Symmetry
The only way to bring order into this chaotic zoo of particles is
by means of various symmetries, which admittedly arise in a rather
abstract and not very intuitive way.
Symmetries are mathematically expressed by groups of transforma-
tions, acting on an object x (which may be some abstract vector). The
simplest group consists only of the identity transformation I, leaving
the object unchanged
Ix = x . (3.65)
This identity group is responsible for the identity of all particles of a
certain kind: all electrons are identical, so are all protons, etc. This
fact is by no means trivial, but it corresponds to physical intuition.
Rotation in two–dimensional real space, in the plane, curiously
enough, is related to electromagnetism, via its gauge theory, see be-
low. (This relation is absolutely counterintuitive!)
Rotation in three dimensions plays a great role in the study of
rotationally–symmetric atoms. This group is denoted by O(3). (This
is abstract but reasonably intuitive.)
Already ordinary quantum theory is forced to operate with (auxil-
iary) complex space,
√ in which the coordinates are complex numbers of
form a + bi, i = −1 (sec. 3.5). A unitary transformation is in complex
space what rotation is in real space.
The group of unitary transformations in two–dimensional complex
space is denoted by SU (2) (actually U (2) but we shall here always dis-
regard mathematical niceties; SU (2) means “Special U (2) ”). Curiously
enough, SU (2) is mathematically equivalent to the real rotation group
in three dimensions:
SU (2) =ˆ O(3) . (3.66)
The group SU (2) is responsible for spin and related properties. We
mention in passing that, in some sense, SU (2) is “generated” by the
four Pauli spin matrices:
1 0 0 1 0 −i 1 0
       
I= , σ1 = , σ2 = , σ3 = .
0 1 1 0 i 0 0 −1
(3.67)
3.6. ELEMENTARY PARTICLES 117

These spin matrices are closely related to quaternions which play a

great role in spatial rotations O(3). This is another instance of the
basic relation (3.66) (which is abstract and does not have an intuitive
geometrical interpretation!).
The unitary group in three complex dimensions, SU (3), is funda-
mental for the representation of hadrons in terms of quarks. Nobody
can visualize such a complex space; it is a completely abstract math-
ematical tool. Nevertheless SU (3), and later SU (4) and SU (5), have
proved to be extremely efficient as order principles for the particle zoo.
SU (3) has eight “generators”, 3 × 3 matrices similar to (3.67), and
was, therefore, by its discoverer Murray Gell–Mann (born 1929, Nobel
Prize in 1969) called the “eight–fold way”, half seriously after Buddha.
(It should be mentioned that the co–discoverer of SU (3) symmetry
was Yuval Ne’eman.) An example is the baryon octet (8 particles) of
Fig. 3.13. Here n and p denote neutron and proton; the other par-
ticles are strange baryons (hyperons), the superscript –, 0, + denote
electrically negative, neutral, and positive.

Figure 3.13: The baryon octet

The symmetry of Fig. 3.13 must not be taken literally in a sim-

ple geometric sense. This is also evident by the fact that there are 2
particles Λ◦ and Σ◦ , at the center! The correct expression for hadron
symmetry is nothing simpler than the group SU (3). If U is a transfor-
mation matrix pertaining to the group SU (2) or SU (3), the complex
“vector” x on which it acts, and the resulting vector y,

y = Ux (3.68)

are called spinors in the case of SU (2) and 3–spinors in the case of
SU (3); they have 2 and 3 complex components, respectively. Enough!
Symmetry and the Greeks. The theory of symmetry played a great
role in ancient Greece, starting at least with Pythagoras in the 6th
118 CHAPTER 3. PHYSICS

century B.C. In the plane, symmetric figures are known to everyone:

the equilateral triangle, the square, the regular pentagon, the regular
hexagon shown in Fig. 3.13 etc. There are infinitely many regular
polygons, one for every integer n.
In space this is quite different. There are no more than 5 regular
polyhedra! They are called Platonic solids because, in his dialogue
“Timaeus ”, Plato associates each regular polyhedron with an “atom”
of the four “elements”:
tetrahedron ... fire
octahedron ... air
cube ... Earth
icosahedron ... water
The most complicated regular solid, the dodecahedron, is, rather log-
ically, considered the image of the universe as a whole (its face is the
mysterious pentagon).
Depending on our relation to the history of our culture, we may
smile at the naivité of the Greeks or admire their marvellous intuitive
anticipation of the role of symmetry for elementary particles. I prefer
the second attitude.
Gauge theories. The word “gauge” is a misnomer. In 1918, trying
to establish a unified theory of gravitation and electromagnetism, Her-
mann Weyl introduced a scale of length (a “gauge” in the true sense)
that varies from point to point. This theory was the first attempt at
such a unification and found many followers, among them Einstein,
Eddington, and Schrödinger. Weyl’s particular approach was given up
soon, however. Much more successful was Weyl’s attempt in 1929 to
find a theory of the electromagnetic field in interaction with charged
particle fields. The linear change of scale was replaced by a circular
change, a variable rotation, and the theory worked: the first gauge
theory was found. It permitted an elegant derivation of Maxwell’s
equations for the electromagnetic field , which was thus recognized as
a gauge theory for plane rotation, O(2)=U ˆ (1) in the terminology used
above.
Weyl’s principle remained forgotten for 20 years, until C.N. Yang
and R.L. Mills developed a gauge theory for SU (2) in 1954, which then
turned out to be useful for the description of weak interactions. In
1960, Sheldon Glashow outlined a unified theory of electromagnetic and
weak interactions, using the “product group” SU (2)×U (1) as the basis
of a gauge theory. In 1967–1968, Steven Weinberg and Abdus Salam
perfected the theory, which became known as the standard theory of
3.6. ELEMENTARY PARTICLES 119

electro–weak interaction. Glashow, Salam and Weinberg received for it

the Nobel Prize in 1979: usually one gets the Nobel Prize for a theory
only after it has been confirmed experimentally.
After Ne’eman and Gell–Mann had discovered the quark structure
of hadrons also around 1960, described by means of SU (3), it “only”
remained to find a corresponding gauge theory for SU (3) to describe
the strong interaction. Then a gauge theory of U (1) × SU (2) × SU (3)
would be expected to furnish a “grand unified theory ” (GUT), at least
of three basic forces: electromagnetic, weak, and strong interaction.
The theory thus obtained works so well in practice that is has been
called Standard Model , but it contains some ad hoc assumptions which
render it not yet perfect from a theoretical point of view.
By far the most difficult obstacle is the fact that, of the four basic
forces, it is most difficult to incorporate gravitation. Ingenious theories
have been elaborated: supersymmetry, supergravity, string theory, su-
perstring theory. What seems to be missing, however, is a completely
new idea from which a unified theory would follow automatically. Some
scientists think that superstring theory might be such an idea, others
are strongly against. (There is little doubt, however, that gauge the-
ories, just like relativity and quantum theory, express a valid physical
principle which will somehow have to be incorporated in any future
“final” theory.) A fascinating account, providing an excellent introduc-
tion followed by interviews with leading proponents and opponents, is
found in (Davies and Brown 1988).
This is the best general reference for this section. Books on particle
physics become obsolete almost before they are published. So none of
them is recommended here; the interested reader will easily find the
newest books in bookstores or libraries. As general treatises involv-
ing also philosophical aspects (cf. sec. 6.6) we mention (Davies 1984),
(Barrow 1991), (Weinberg 1993), (Kaku 1994), and (Gell–Mann 1994),
each book having its own merits. On symmetry in general, (Weyl 1952)
is still a classic; a very readable modern treatment is (Mayer–Kuckuk
1989).
120 CHAPTER 3. PHYSICS

3.7 Space and time; cosmology

Time is the moving image of eternity.
Plato

Space–time structure
As we have seen in sec. 3.4, the theory of relativity implies a unifica-
tion of space and time. By the line element (3.45) of special relativity

ds2 = −dx20 + dx21 + dx22 + dx23 , (3.69)

(p. 94) the spatial coordinates x1 = x, x2 = y, x3 = z are related to
time t or the time coordinate

x0 = c t , (3.70)

c being the velocity of light, but x0 is distinguished essentially by the

minus sign in (3.69). Thus ds2 is not necessarily positive: we have

ds2 > 0 on a space–like line ,

ds2 = 0 on a light–ray , (3.71)
ds2 < 0 on a time–like line .

This is illustrated by the space–time diagram of Fig. 3.14 and in more

detail in Fig. 3.15. Time–like lines lie within the light cone; they rep-
resent possible world–lines of a person (from birth at A to death at B)
or of a material object (mass point). Space–like lines are usual lines in
space, for instance a plumb line, the edge of a ruler, or an electric wire.
These lines need not be straight, but for simplicity we shall use
straight lines.
In Fig. 3.14, the “past” contains all events (points in space–time)
which can influence the event P . The “future” consists of all events
that can be influenced by P . The “present” contains all events which
neither can influence P nor can be influenced by P since all effects
propagate with velocity smaller than the light velocity c. By choosing
x0 as time coordinate we have implicitly put c = 1, which accounts for
the angle of 45◦ in Figures 3.14 and 3.15. Thus the “present” looks so
big.
In fact, however, c expressed in metric units is a great number,
close to 300 000 km/sec, so that a realistic picture looks quite different
3.7. SPACE AND TIME; COSMOLOGY 121

Figure 3.14: Space–time diagram

(Fig. 3.16). The realistic “present” is extremely narrow; it is an exact

plane as c → ∞.
The situation in the curved space–time is locally entirely the same,
in the same way as a surface can locally be approximated by its tangent
plane.
Creativity and the block universe. It has been claimed that space–
time is a fixed four–dimensional manifold, in which an observer only
“creeps” passively along his world–line from birth A to death B, being
helplessly and irreversibly dragged along by time. This is the theory of
the block universe: space–time exists ready–made, time is an illusion,
the world simply is, it does not become. We are passive spectators,
observing the flow of time without being able to influence it. Freedom
of the will is an illusion. We need not worry about immortality: the
world line from A to B in Fig. 3.15 simply is, birth, development, decay
and death are all illusions. We are sitting in the cinema of life, watching
the events which go on around us. We cannot change them, exactly as
we cannot change the action going on in the television program we are
just watching (we cannot even turn off the “TV set of time”).
This picture is difficult to accept, but it cannot be refuted logically.
If the line element of space–time were “positive definite”,
ds2 = dx21 + dx22 + dx23 + dx24 , (3.72)
with time x4 real (in a mathematical sense) so that ds2 can only be
positive, the block universe would in fact be the only possible interpre-
tation. There would be no light cone separating past and present; time
x4 would just be an ordinary space coordinate, of the same character
as x1 or x2 .
122 CHAPTER 3. PHYSICS

Figure 3.15: Various lines in space–time

It is the minus sign which makes all the difference, creating the light
cone separating past, present and future. We can travel in space but not
in time (see below). It is true that the Lorentz transformation (3.38)
also transforms time, but the change is usually negligible in practice
since v  c; cf. also the “realistic” third picture in Fig. 3.16.
Nevertheless, if we wish the flow of time to be real (in a philosophical
sense), we should like time to have a unique direction, along which the
creative process of life and mind can proceed. The “present” should
consist of simultaneous events, which would require the two lines in the
third, “realistic” model in Fig. 3.16 to coincide exactly, forming a single
line separating past and present. In reality, however, the two lines in
this figure form an extremely small angle, instead of coinciding as they
do in classical mechanics: the present is “thin”, but not “infinitely
thin”.
Whitehead and, more recently, Havemann (1964, p. 92) have
pointed out a way out of this dilemma. Past and future must be rede-
fined in agreement with relativity: past is of what we already can have
knowledge, future is what we can still influence by our actions. Thus
creative action is possible. Time is not an illusion, history really goes
on, biological evolution exists, animals and men are born, live, and die.
We are actors and not merely passive spectators in the universe. Let
us call this model open universe.
3.7. SPACE AND TIME; COSMOLOGY 123

Figure 3.16: Conventional and “realistic” light cones

Block universe and open universe contradict themselves only if taken

as absolute models. Regarding the past, our universe is a block universe:
the past is past and can no longer be changed. Regarding the future, the
universe is open. Now, in 1994, we are open towards the year 2000, and
can take active and creative steps to ensure that mankind still exists
in 2000; in 2001, the year 2000 is in the past and can no longer be
influenced. For 2001, the universe up to and including the year 2000,
is a block universe.
Again we have a Hegelian synthesis: the creative open universe is
the thesis, the block universe is the antithesis, and the distinction:
block universe for the past, open universe for the future, may be the
synthesis.
We also may say that the block universe represents the “static” geo-
metric view, whereas the open universe expresses a creative “dynamic”
view.

Cosmic time and the evolution of the universe

Locally, general relativity behaves exactly like special relativity,
with the light–cone separating past, present, and future. The global
structure is a block universe, but an open universe with “real” creative
time is consistent with general relativity as well.
It is even more so because the basic model universes of general
relativity admit a uniquely defined distinguished direction of time. They
have a general form (Robertson–Walker metric; “metric” is the same
as “line element”):
!2
2 2 R(t)
2
ds = −c dt + (dx2 + dy 2 + dz 2 ) , (3.73)
R0
124 CHAPTER 3. PHYSICS

for a flat three–dimensional space (curvature 0); for a space of constant

curvature (positive or negative) the second term on the right–hand
side is slightly more complicated. (Curvature of 3–space is essential for
any rigorous treatment; our disregard is motivated only by reasons of
didactic simplicity.)
Here R0 is an irrelevant scale factor. In the special case R(t) =
R0 = const., eq. (3.73) reduces to the flat space–time of special rela-
tivity. Generally, R(t) may be intuitively regarded as the radius of the
universe, although this is another barely acceptable oversimplification;
cf. (Misner et al 1973, p. 721).
Without going into details, we can distinguish three possibilities,
shown in Figures 3.17 to 3.19. The interpretation is obvious. The static
model of Fig. 3.17 corresponds to a universe that does not change with
time. It is excluded by the empirical fact that the universe is expanding.
Remote galaxies are moving away from us (or we are moving away from
them) with speeds which are the greater the farther away the galaxy
is. The mysterious quasars which are extremely far away, are moving
away with a speed approaching that of light.

Figure 3.17: A static model

Fig. 3.18 is regarded at present to be the most realistic model.

The universe started, some 13 billion years ago, as a point singular-
ity A. This would correspond to the creation of the world, but there
is a growing tendency in science towards sober and even humorous
understatement (think of the name “quark” for the most important
3.7. SPACE AND TIME; COSMOLOGY 125

Figure 3.18: An expanding universe

Figure 3.19: A closed universe

sub–elementary particle); therefore physicists prefer the term big bang.

After that, R(t) increases: the universe expands.
Fig. 3.19 shows a model which starts with a big bang singularity at
A. The radius R(t) first increases: the universe expands, but then it
contracts again to another pointlike singularity Ω.
At present, we appear to be at a point P , at a phase of expansion.
It had not been decided whether our actual universe corresponds to
Fig. 3.18 or 3.19. This question can be settled, at least in principle,
by determining the mean density of the universe. This is very difficult
and controversial, however, mainly because of the existence of invisible
“dark matter” in the universe. The majority of cosmologists seem to
favor the expanding universe of Fig. 3.18.
126 CHAPTER 3. PHYSICS

The closed universe of Fig. 3.19 is used in Tipler’s (1994) “omega

point theory” which is an extravagant and highly controversial but fas-
cinating combination of physical cosmology, science fiction, computer
theory, and theology. The term “omega point” (Ω in Fig. 3.19) goes
back to Teilhard de Chardin (1955).
The beginning of time. This has been the subject of considerable
philosophical discussion. Aristotle regarded the universe as eternal:
it has no beginning. St. Thomas Aquinas had a hard time adapting
Aristotelian philosophy (which he basically adopted) to the biblical
view of world creation which implied a “big bang” singularity. See also
St. Augustine’s view in sec. 5.4.
Kant’s first antinomy states as a thesis that the world has a be-
ginning in time and as its antithesis that the world has been lasting
forever. Mathematics is providing a remarkably simple solution of this
antinomy. Think of a new time variable τ introduced by
τ = ln t (3.74)
where ln denotes the natural logarithm. Now t = 0 (big bang) corre-
sponds to τ = −∞, which means “no beginning” in terms of τ !
This shows that mathematics and physics can indeed help philoso-
phy solve some of its basic problems.
We can even take one step further. Take
q
r= x2 + y 2 + z 2 (3.75)

as a radial coordinate. Then we may put in (3.73)

R(t) = c t , (3.76)
which corresponds to linear expansion according to a particularly simple
model proposed by the British cosmologist E.A. Milne. A particular
galaxy moves away with constant velocity v:
r = vt . (3.77)
Now introduce a new radial variable ρ by
r
ρ = arth (3.78)
ct
where “arth” denotes “area tangens hyperbolicus”, a well–known ele-
mentary mathematical function.
3.7. SPACE AND TIME; COSMOLOGY 127

Now comes the surprise: for our galaxy, eqs. (3.77) and (3.78) give
v
ρ = arth = const. (3.79)
c
and for the “limit sphere”
R = rmax = c t (3.80)
which is the maximum r possible at time t since c = vmax , light velocity
as the maximum possible velocity. Now (3.79) gives
c
ρmax = arth = arth1 = ∞ . (3.81)
c
That means, all ρ’s of galaxies is constant, and space is infinite.
Thus simple mathematical equations transform an expanding uni-
verse of the type of Fig. 3.18 to a static model of type Fig. 3.17 (with
R = ∞).
But the expansion of the universe has been confirmed by observa-
tion! By which observation? By means of the Doppler shift of spectral
lines. If a police car, with its siren howling, approaches me, the sound
appears higher than when it has passed me and happily disappears in
the opposite direction. Thus for an approaching object, the frequency
appears higher than normal, and for a receding object, the frequency
appears lower. For light, red light has a lower frequency than blue light,
so the light of an approaching star would appear shifted towards blue.
What we observe, however, is a red shift which indicates that the stel-
lar object (star, galaxy, quasar) moves away from us, and the greater
the redshift, the greater the speed. This is the observation method by
which the expansion of the universe is measured.
But a frequency change could also be explained by assuming that
in the past, light has had a lower frequency than today. As we observe
more and more distant stars and galaxies, we also go back in time. A
star at a distance of one light year is observed as it was a year ago,
because light took one year to travel from the star to the observer. If
we observe a galaxy which is a million light years away, we see it, not
as it is now, but as it was a million years ago! Hence, if the observed
galaxy has a strong redshift, it means that it is far away in time (since
it is far away in space) and we see it as it was long ago when the clocks
(oscillating atoms or molecules are clocks!) were much slower than they
are now. Mathematically, this means that time is measured in terms of
τ rather than of t, and logically there is no difference between explaining
128 CHAPTER 3. PHYSICS

redshift in terms of an expanding dynamic universe and explaining it

in terms of increasing “speed of natural phenomena”. (In fact, it seems
that the speed of modern life appears much greater than that of life in
the “good old days” — please don’t take this argument seriously!)
Although most astronomers and physicists prefer describing the uni-
verse in terms of r and t to the description in terms of ρ and τ , we see
that “things ain’t that simple”. So a good knowledge of mathematics
may sometimes help regard philosophical controversies (beginning of
time, finite or infinite universe) with less emotion.
It is similar with the classical Greek paradox of Achilles and the
tortoise: though he is much faster than the tortoise, Achilles can never
reach it. Let us assume that Achilles is a hundred times faster than
the tortoise, and initially they are separated by 1000 meters. After
Achilles has covered this distance of 1000 m, the tortoise has moved
10 m, after Achilles has covered these 10 m, the tortoise has moved
0.1 m and so on. If both Achilles and the tortoise are pointlike, before
Achilles has reached the animal, the latter has always moved a little
farther: Unlucky Achilles will never reach the smiling beast. Modern
mathematics reduces the problem to the summation of the infinite series

1000 m + 10 m + 0.1 m + · · · =
1 1 1
 
= 1000 m 1 + + + +··· =
100 1002 1003
1000 m
= 1000 m(1 + q + q 2 + q 3 + · · ·) =
1−q
1000 m 100000
= 1 = meters . (3.82)
1 − 100 99

This is the distance at which Achilles overtakes the tortoise, to convert

it into a tasty turtle soup.
This example, which looks simple to us now, has, for some 2000
years, played a serious role in philosophic discussions about time!

Gödel’s universes and time travel

The logician Kurt Gödel, of Austrian origin but then, together with
Einstein, at the prestigious Institute of Advanced Studies at Princeton,
wrote perhaps 5 or 10 papers throughout his life, but each creating a
major sensation. In the Einstein Festschrift (Schilpp 1949), Gödel pub-
lished on pp. 555–562 a paper “A remark about the relationship between
3.7. SPACE AND TIME; COSMOLOGY 129

relativity theory and idealistic philosophy ”. There he described a math-

ematically perfectly valid cosmological solution of Einstein’s equations
of general relativity which contained closed time–like world lines.
Usually, world lines are not closed, cf. the line AB in Fig. 3.15 on
p. 122. Closed time–like world lines permit you to travel into your
own past (Fig. 3.20). This may have paradoxical consequences. You
travel from P via Q to point G where you kill your grandmother — not
intentionally as sometimes claimed in the literature, but by accident
when, as a small girl, she was playing in the street and you, driving too
fast, saw her only when it was too late. Anyway, no grandmother, no
mother, no . . . you!

Figure 3.20: A closed time–like world line

Paradoxical conclusions of this type may occur with closed time–like

lines. Hence, Gödel concluded that space–time must be “unreal”: the
passing of time is a mere illusion, and the universe is a block universe
(see above). In his reply to Gödel (Schilpp 1949, p. 688), Einstein
was worried and cautiously remarked: “It will be interesting to weigh
whether these [solutions found by Gödel] are not to be excluded on
physical grounds.”
In his excellent book, Reichenbach (1957, pp. 141–142) gave an
even more interesting example: you meet a man who claims to be your
younger self. This idea is used “iteratively” in an irresistibly funny
way in the Seventh Voyage of Ijon Tichy in the science fiction book
“The Star Diaries ” by Stanislaw Lem. The space ship, going through
strongly irregular gravity fields of Gödel type, keeps filling and crowding
with identical Ijon Tichy’s who are multiplying by going through time
130 CHAPTER 3. PHYSICS

loops of one day’s duration: the Ijon Tichy of Monday quarrels with
the Ijon Tichy of Tuesday and that of Wednesday and so forth.
The first major tale of time travel is the classical novel “The Time
Machine” by H.G. Wells published in 1895, where he anticipated several
ideas of the theory of relativity, especially the similarity of time and
space.
He made, however, a philosophically very important error. Taking
the identity of space and time too literally, his hero travels into the
future much more rapidly than the “ordinary rate of time”. Now, what
is the rate of time flow? What does it mean? What is the “speed of
time”? Ordinary “spatial” velocity is measured in meters per second,
m/sec, say (automobilists prefer km per hour). Thus, the “velocity” of
time would be sec/sec, and this is really the case. The velocity of time,
by definition, is
1 sec
=1 , (3.83)
1 sec
the “velocity” of time is dimensionless and always equal to unity ! It
simply cannot be 20 seconds per second. Still less can we travel back-
wards along our own world line, because that would involve negative
“velocities”.
This is an evident but essential distinction between space and time.
“Time travel” can only be along time–like world lines with “speed” equal
to unity.
All about time travel – physics, metaphysics and science fiction – can
be found in the comprehensive work (Nahin 1993) which also contains
an incredible number of references. Very nice is also (Rucker 1984).

The arrow of time

Why does time flow in one direction only? Why are we irresistibly
getting older, instead of being able to return to our youth which many
people (excluding the present author) wish?
The equations of mechanics, classical relativistic and even quantum
mechanics, are reversible in time, that is, they keep their validity if t is
replaced by −t.
There are, however, irreversible processes as we have seen in statis-
tical mechanics and in the measuring process of quantum mechanics,
see secs. 3.2 and 3.5. They somehow seem to involve statistics or, what
is the same, probability.
Consider, for example, a box consisting of two chambers, chamber A
filled with a gas, and B empty. The shutter S between them is closed.
3.7. SPACE AND TIME; COSMOLOGY 131

Figure 3.21: The left chamber is filled by a gas, the right chamber is
initially empty

Now open the shutter: gas will flow from A to B until the gas is equally
distributed in A and B. If we assume that the gas molecules move
according to the reversible laws of classical mechanics, the opposite
process should also occur, at least from time to time: the whole amount
of gas now in B should flow back into A, and B should be empty, at
least for a moment.
This should happen, but it never happens. The explanation is sta-
tistical: the back flow does happen, but so seldom that there is no
chance whatsoever to watch such an event. The probability of such an
event is so small that it is practically zero.
To see what a time–reversal means for our practical life, run a
motion–picture backwards. Autumn leaves fall upwards, glass splinters
put themselves together to form a new bottle, two cars badly damaged
by a collision separate and become nice undamaged automobiles. The
biographical film of a famous man starts by showing him dead in his
grave. He resurrects from the grave and becomes progressively younger.
He goes to university, then to secondary school, to elementary school,
then to kindergarten. He becomes a baby and loses most of his hair.
His teeth become smaller and smaller and finally vanish. The end of
the film is an ugly little thing in a cradle, admired by his mother and,
somewhat more hesitantly, by his father.
What is the explanation of the arrow of time? I think, ultimately we
just have to accept it, just as we accept ourselves and our surroundings.
Nevertheless, explanations will help get a deeper understanding of this
mysterious phenomenon, which only looks natural because we have got
used to it.
So to speak, we have five main “arrows of time”:

– the thermodynamical arrow,

– the biological arrow,

– the historical arrow,

132 CHAPTER 3. PHYSICS

– the psychological arrow, and

– the cosmological arrow.
The thermodynamical arrow is based on the remarkable irreversibil-
ity of statistical mechanics as described above (Fig. 3.21). It is charac-
terized by the steady increase of entropy (sec. 4.3).
The biological arrow , in a sense, has opposite character although it
points along the same direction. It is the arrow of biological life, from
birth to death. It is the arrow of evolution, from amoeba to man. It
means increase of biological information contained in the genes.
The historical arrow is closely related. It is characterized by the
increase of our information. The unknown future becomes present and
then history, to be studied by historians, archeologists, and paleon-
tologists. The increasing knowledge is stored in libraries of rapidly
increasing size. Universities are growing and becoming diversified at a
breath–taking pace.
The cosmological arrow corresponds to the expansion of the uni-
verse. There is also a cosmic evolution, starting from a probably rather
undifferentiated “fireball” of unimaginably concentrated energy after
the “big bang” and leading to galaxies, stars and planets. It is ac-
companied by a chemical evolution from hydrogen to all the heavier
chemical elements; cf. (Weinberg 1977).
The psychological arrow of time, our subjective feeling of the pas-
sage of time, is related to the biological arrow, to the historical arrow
(the accumulation of information in our brains), and possibly even to
the thermodynamical arrow: a computer processes information at the
expense of energy (high–quality electric energy is degraded into low–
quality thermal energy which must be removed by a cooling system),
and the brain should work in a somewhat similar way (Hawking 1988,
p. 147).
Wheeler’s example. In his Nobel–Prize lecture, R.P. Feynman
talked about a telephone conversation between the two great physicists
(we are following (Gardner 1982, p. 268)):
“Feynman”, said Wheeler, “I know why all electrons have the same
charge and the same mass.”
“Why?” asked Feynman.
“Because”, said Wheeler, “they are all the same electron.”
What does this mean? Consider Fig. 3.22, following Whitrow (1980,
p. 332). The picture on the left admits the following standard interpre-
tation. At A, a photon spontaneously “splits” into an electron and a
3.7. SPACE AND TIME; COSMOLOGY 133

Figure 3.22: A positron is an electron running backward in time

positron (energy and electric charge are conserved!). At C, the positron

meets another electron; the two particles annihilate each other, emit-
ting a photon. According to Wheeler’s interpretation shown in the
picture on the right, an electron starts at D, collides with a photon at
C, suffers a “recoil” which sends it back in time to A, where it collides
with another photon which makes it change direction in time again,
sending it to B.

Figure 3.23: Two electrons and a positron or one and the same electron
at three different places?

At a certain instant t0 , we thus have the picture shown in Fig. 3.23.

Thus, if we regard a positron as an electron running backward in time,
we have, at the same time t = t0 , the same electron in three different
places a, b, and c (Fig. 3.23, right)!
This can be iterated. Fig. 3.24 shows the world–line of a single
electron. It generates all electrons and positrons in the world.
Wheeler’s idea is extremely interesting, but it cannot be entirely
true: it would imply that there are as many positrons in the world as
there are electrons. This is manifestly not the case.
The philosopher A.N. Whitehead once remarked: “It is more im-
portant that a proposition be interesting than that it be true.”
134 CHAPTER 3. PHYSICS

Figure 3.24: A single electron generates all electrons (•) and positrons (◦)

Wheeler’s idea is so beautiful that it almost has to express a true

partial aspect of nature. Many physicists now think that the arrow of
time is a macroscopic “global” feature of the world. On a microscopic,
quantum–mechanical, “local” scale, time may well be reversible.
Additional reading. The basic reference on time is (Whitrow 1980),
unsurpassed in comprehensiveness, depth and readability. (Gardner
1982), and (Rucker 1984) are “popular” books of high level. (Čapek
1976) and the smaller book (Smart 1964) are anthologies of articles of
authors from Greek antiquity to the present time. (Reichenbach 1957)
is a very readable classic. The books by Hawking (1988 and 1993) are
readable, profound and modern. (Hawking and Ellis 1973) is the basic
mathematical reference on space–time models. It contains a detailed
mathematical treatment of Gödel’s model and, for the first time, shows
a very instructive geometrical picture, cf. their Fig. 31 on p. 169. Once
more we mention (Treder 1974).

3.8 Inverse problems

Most mathematical problems in science,
technology and medicine are inverse problems.
Gottfried Anger

Introduction
In a poem, addressed to a physicist, Johann Wolfgang von Goethe
ascribes to him the opinion (which he himself rejects):
3.8. INVERSE PROBLEMS 135

Ins Innre der Natur

Dringt kein erschaffner Geist
Glückselig, wem sie nur
die äussre Schale weist!

(“No created spirit penetrates into the interior of nature. We are

already lucky if nature shows us only its external shell.”)
This is exactly the situation which we are facing in geophysics if we
want to investigate the interior of the Earth. Our measurements are
restricted to the Earth’s surface, and we want to use them for finding
out the Earth’s interior structure. Such measurements are, above all,
observations of seismic waves originating from large earthquakes, but
also, e.g., observations of the Earth’s external gravitational field.
Physicians face very much the same problem if they want to inves-
tigate the interior of the human body , for instance in order to discover
a tumor. Of course, they can directly penetrate into the interior by
surgery, and they will do so if they know where the tumor is. But first
they will use indirect methods such as X–rays (which closely correspond
to the seismic waves of the geophysicist) or nuclear magnetic resonance
(NMR) tomography.
Example 1 . Consider the Earth’s gravitational field. Obviously, the
gravitational force is produced and determined by the distribution of
the masses of density ρ inside the Earth. Symbolically we may write

g = Af (3.84)

where g denotes the gravitational force at the Earth’s surface, f = ρ

represents the distribution of density ρ inside the Earth, and A is the
“Newton operator ” (a generalization of Newton’s law (3.11) on p. 74)
computing g if f is given. (“Operator” is here again used in the sense of
performing a mathematical operation.) An explicit form of the Newton
operator will be found in sec. 6.6, eq. (6.2) on p. 268.
Let us now consider g as given by gravity measurements; the prob-
lem is to determine the density distribution f inside the Earth.
Formally, the solution is simple:

f = A−1 g (3.85)

where A−1 denotes the inverse to the Newton operator. It is a task of

mathematicians to find it; this task is not impossible, but still surpris-
ingly difficult if we consider the simplicity of the direct problem (3.84).
Eq. (3.85) then expresses a solution of the inverse problem.
136 CHAPTER 3. PHYSICS

Symbolically, we may express the problem by the diagram of

Fig. 3.25. The operator A is a “black box ” which converts the input
(density f ) into the output (gravitation g). This is the direct problem.

Figure 3.25: An input–output diagram

The diagram behaves well if we follow it from left to right, in the

sense of the arrows, that is for the direct problem. For the inverse
problem (3.85), we have to go through the diagram in the opposite
direction, from right to left, against the direction of the arrows. This
causes difficulties.
The general structure expressed by Fig. 3.25 and equations (3.84)
and (3.85) also applies to many other cases. The operator A may be
linear (sec. 3.5) or nonlinear (sec. 3.2). The Newton operator is linear.
Example 2. Projection. Consider a photographer A taking the pic-
ture g of a pretty girl f . This example is immediately clear, and it is
directly relevant to the problem of imaging. Thus f denotes the object
(also the interior, e.g., in X–ray pictures), and A represents the pro-
jection producing the image g. The operator A may denote amateur
photography, but also an ordinary X–ray equipment, X–ray tomogra-
phy or NMR (nuclear magnetic resonance) tomography in medicine, as
well as seismic tomography in geophysics. In any case, g will be the
output data.
This mathematical model thus is very general; the operator A is
now called a projection operator.
Example 3. Human perception. This is a non–mathematical ana-
logue of Example 2. Here, the “equations” (3.84) and (3.85) must be
taken in a very general sense, as a symbolic expression of the diagram
of Fig. 3.25.
The symbol “f ” denotes an object, or reality, or nature. The “pro-
jection operator” A symbolizes sense perception, by means of the eye or
the ear, and the “output” g are the sense data, cf. sec. 1.3. To empha-
size the active role of perception (“searchlight” rather than “bucket”,
cf. sec. 1.4), A may also be called a search operator .
The “direct problem” thus is sense perception of nature. The corre-
sponding “inverse problem” is the determination of f (object, reality,
3.8. INVERSE PROBLEMS 137

nature) from sense data g. This is the basic problem of the theory
of knowledge, or epistemology. There is no need to point out its dif-
ficulty, which has kept philosophers busy from Socrates and Plato to
Karl Popper.
Classification. Our mathematical (or pseudo–mathematical) sym-
bolism permits a useful classification:

given to be determined classification

A, f g direct problem
A, g f inverse problem of first kind
f, g A inverse problem of second kind

Before we discuss these problems in more detail, we must define “well–

posed” and “ill–posed” problems.

Ill –posed problems

A problem is called properly posed , or well–posed , if the solution
satisfies the following three requirements:

(1) existence,
(2) uniqueness,
(3) stability.

This means that a solution must exist for arbitrary (within a certain
range) data, that there must be only one solution, and that this solu-
tion must depend continuously on the data. If one or more of these
requirements are violated, then we have an improperly posed, or ill–
posed problem. For a long time it was thought that only properly posed
problems are physically meaningful. In fact, deterministic processes, as
considered in classical mechanics, depend uniquely and continuously on
the initial data — this is the essence of causality — and thus correspond
to properly posed problems.
Only relatively recently it was recognized that there are important
problems that are not properly posed. There is now an extensive liter-
ature on improperly posed problems; cf. (Anger et al. 1993).
In fact, most inverse problems are ill–posed , an extreme example
being epistemology (or philosophy in general). But even some direct
problems are ill–posed, as we know from weather prediction and, more
generally, from chaos theory (sec. 3.2).
Physically, well–posed problems are stable, and ill–posed problems
are unstable . Laplace’s demon, embodying classical determinism and
138 CHAPTER 3. PHYSICS

causality, operates in a stable fashion; St. Peter (if regarded as the saint
responsible for weather) has a rather unstable character.

Inverse problems of first kind

As we have seen above, such problems consist in solving the equation
g = Af for f . A formal solution has been given by eq. (3.85).
In this inverse problem, g represents the measured data, and f rep-
resents parameters that describe the object.
The simplest and most important case thus is provided by Example
2: Projection. In this case, the data g are some “image” of “nature” f ,
the projection being defined by the operator A. This case ranges from
amateur photography to tomography in medicine and geophysics.
Thus, generally, the projection operator A works in the direction
from nature to observation, it projects from “object space” into “ob-
servation space”. It must be inverted to give A−1 , working from ob-
servation to nature and determining some feature of nature or reality
f from observations g. Many alternative formulations may be given,
as we have seen before. Since the apparatus A (X–ray, tomography) is
searching to get information g from nature f , A has also been called a
search operator to express its active role.
Let us try to understand this better by means of the concrete ex-
ample of tomography. Here the objective is to determine the inner
structure of the object f . In medicine we have X–ray and NMR tomog-
raphy; in geophysics we have seismic tomography. The mathematical
structure is similar in all three cases. The data g comprise a set of
X–ray images or NMR data or observations of seismic waves.
Now we shall return to our Example 3 : Human perception. As we
have already mentioned above, sense data g provide information about
reality f . The projection operator A here is not an X–ray or tomo-
graphic apparatus but the human organs of sense perception such as
the eye and the ear, plus the additional “hardware” as provided by our
nervous system which has been developed (at least to a large extent)
by biological evolution. Continuing this analogy, we may say that ex-
perience and learning have improved or even created the corresponding
“software” which may be considered some modern equivalent of Kant’s
a priori. This “evolutionary theory of knowledge” has been described
in sec. 1.4.
In this context, calling A a projection operator corresponds to
Russell’s theory of sense data acquired passively and then analyzed,
whereas the name search operator for A corresponds to Popper’s theory
3.8. INVERSE PROBLEMS 139

of human perception as an active “searchlight”. (This is really a gross

oversimplification justifiable, if at all, then only by its didactic purpose.)
Of course, human perception is not directly a mathematical prob-
lem: the operators A and A−1 are realized by biology or physiology,
with a built–in mathematical structure and the corresponding “neuro-
computational” systems.

Inverse problems of second kind

Above we have defined inverse problems of the second kind as the
determination of the operator A, considering both f and g as given
data. The basic mathematical structure is, of course, always g = Af .
Here it is useful to consider the operator A as a law which trans-
forms certain given or measurable parameters f into other measurable
parameters g. Hence f and g can be prescribed or measured; the law
A is then to be determined.
For the law A we may have a certain number of alternatives; in
simple cases this reduces to a statistical testing of hypotheses. Or we
essentially know the law, and only a few parameters are to be deter-
mined. A particularly interesting case is Newton’s derivation of his law
of gravitation A from the Keplerian laws of planetary motion, which
thus served as data for determining the “grand” law A.
Generally matters are not quite that simple. Laws are usually found
by “guessing” or, using a more respectable expression, by physical in-
tuition. Let A be such a hypothetical law; it must now be tested by
comparing it with experimental data f and g: this is a verification of
A (there must be g = Af ). It may “survive” all known tests; is this
sufficient to assert that it is true? This is the problem of induction.
Well, we can never be sure that another verification may not show
g = Af . Take the example of the “law” that the sun rises in the
morning and sets in the evening. This has been thoroughly confirmed
by mankind; there was never a single exception. Is it thus necessarily
true also for tomorrow? True with overwhelming probability, yes, but
not with necessity: The Sun may have exploded during the night and
may also have destroyed the Earth in this process.
Some contemporary philosophers of science, such as Sir Karl Pop-
per, have replaced verification by falsification. Since verification can
never be certain, it would be better to try to disprove the theory by
all possible means; if this is not successful (if one always gets g = Af ),
then we may, for the time being, accept the theory.
Matters are again not so simple, however. Because of measuring
errors, g = Af will never hold with absolute accuracy. A “neighbor-
140 CHAPTER 3. PHYSICS

ing” theory may hold as well within the limits of the accuracy of the
experiment.
For instance, the theory of special relativity reduces to classical me-
chanics for velocities v that are small as compared to the light velocity
c (in other terms, for c → ∞), and quantum theory reduces to classical
mechanics for h → 0 where h is Planck’s constant. Thus, in describing
ordinary (“macroscopic”) physical experiments, classical mechanics is
sufficient, but if we absolutely wish, we may describe the experiments
in a more complicated way also by special relativity or by quantum
mechanics: the results will be practically the same.
Thus experiments must be carefully and ingeniously devised in order
to distinguish between two theories, the so–called crucial experiments.
An example is the well–known Michelson–Morley experiment for spe-
cial relativity. It really permits to “falsify” classical mechanics in an
extreme case, cf. secs. 3.4 and 3.9.
Also the gravitational field of the Earth and of planets may be
correctly described by classical mechanics, if necessary with very small
“relativistic corrections”, although the description by Einstein’s theory
of general relativity is theoretically superior. This is a typical case of
a phenomenon pointed out by the well–known mathematician Henri
Poincaré: several different laws may fit equally well (if necessary with
small corrections). Thus the choice is “conventional” and may be done
by “esthetic” criteria such as “simplicity” or “mathematical elegance”.
This is called conventionalism.
The importance of measuring errors or “noise” in such considera-
tions is evident.
Thus the theory of inverse problems may provide a preliminary first
introduction to the theory of induction, verification and falsification
which will be considered in more detail in the following section. Con-
ventionalism will play a certain role in sec. 6.5.
Our standard Example 3 : Human perception, may also be consid-
ered from the point of view of inverse problems of second kind. Here
the operator A is the apparatus of human perception, called by Konrad
Lorenz (1973) the “mirror” (“Spiegel”) which mirrors our environment
for us. This “mirror” is human perception including neural “hardware”
and “software”, cf. secs. 1.2 and 1.3. Thus the investigation of this
“mirror”, looking with Konrad Lorenz at the “back side of the mirror”
(die Rückseite des Spiegels), may be regarded as an inverse problem of
second kind.
Thus “unreflected” sense perception is related to an inverse problem
of first kind, as we have seen above. On the other hand, physiological
3.8. INVERSE PROBLEMS 141

investigation of and philosophical reflection about the mechanism of

perception leads to an inverse problem of second kind.
System identification. Sometimes a mathematical or technological
system is really a “black box” A in the sense of Fig. 3.25, and we
wish to determine its internal structure “from the outside”, without
opening the “box”. This is relatively simple if the general structure of
the system is known and only a few parameters are to be determined.
They characterize the materials by which the system is physically re-
alized and are therefore called material parameters. For this purpose,
a given input f is fed into the system, and the corresponding output g
is measured. This is repeated for several f and g. In this way, equa-
tions may be obtained from which the parameters of the system A can
be determined. A correct determination of material and other system
parameters is of basic importance for sensitive systems which have to
work reliably, for instance, aircraft engines. The determination of these
system parameters is called system identification. The construction of
a model of a complicated system is called reverse engineering.
Neural networks and learning. Such networks are attempts to pro-
vide a simple mathematical model for the activity of neurons in the
human brain (sec. 1.1). A linear neural network represents an output
signal gi as a linear combination of the input signals f1 , f2 , . . . , fn :
gi = ai1 f1 + ai2 f2 + . . . + ain fn . (3.86)
The coefficients aij may be considered some kind of weights for the
influence of the input fj on the output gi . If we have n such equations
(i = 1, 2, . . . , n), then we may write explicitly:

g1 = a11 f1 + a12 f2 + . . . + a1n fn ,

g2 = a21 f1 + a22 f2 + . . . + a2n fn ,
..
. (3.87)
gn = an1 f1 + an2 f2 + . . . + ann fn ,
or in vector–matrix notation using boldface symbols:
g = Af . (3.88)
The problem is to determine the weight matrix A from given inputs
f and their corresponding prescribed outputs g. Since A contains n2
unknowns aij , we need n given input vectors f k and corresponding given
output vectors gk , k = 1, 2, . . . , n.
142 CHAPTER 3. PHYSICS

By determining A and realizing it physically or computationally,

the system provides the required reaction g to the input f . In other
terms, the system has “learned ” to produce the desired reaction g.
Thus learning is formally reduced to system identification or re-
verse engineering, cf. (Spies 1993, pp. 311–313). Learning is an inverse
problem!
More details on neural nets, especially the details of the determina-
tion of A and the introduction of non–linearities, may be found in the
literature on neural networks; cf. (Kohonen 1988) and (Spies 1993).
Applications to psychology. Let the system A be a person applying
for a job. He or she has to undergo a psychological test: questions f are
posed and answers g are obtained. Thus the knowledge of the person
A, his (her) reactions to stress, emotional stability, etc., generally his
(her) qualification for the job are tested. It is of basic importance
that the test (the input f ) is so designed as to provide a balanced and
objective information on the person A. This, of course, is a problem of
psychological capability and experience, combined with common sense,
of the person designing or directing the test, asking the right questions
f , etc.
You might also test the character of a friend A in this way, observing
his (her) reaction g to your behavior f . Do this tactfully, however,
otherwise you may lose a friend.
Examinations have the same character: the professor puts the ques-
tions f to the student A and listens to his answers g. The only “system
parameter” to be obtained in this way is the grade characterizing the
student’s performance in the examination.
These psychological applications share, with mathematical inverse
problems, the logical structure, but the use of mathematics in this field
is naturally rather limited.

Overdetermined and underdetermined problems

According to classical causality, the causes are necessary and suffi-
cient to fully and uniquely determine the effects. Thus we may have n
causes and n effects: the system is fully determined .
In quantum mechanics (sec. 3.5), the state after measurement is
not fully determined by the state before measurement: the system is
underdetermined .
In Whitehead’s (1933, Chapters XII and XIV) theory of actual oc-
casions, the past does not blindly and automatically determine the fu-
ture as it does in classical causality. Especially in a mental event going
on in one’s consciousness, there are many incompatible and conflicting
3.8. INVERSE PROBLEMS 143

“causes” fighting for realization. Think of what is going on in your mind

when you have to take a difficult decision. This has also a counterpart
in the neural structure of the brain: many nerve impulses arrive, some
excitatory and some inhibitory (sec. 1.1), and the total effect will imply
some internal adjustment. This seems to be an overdetermined prob-
lem. Sometimes there may also not be enough data to automatically
assure a unique solution; thus leaving room for the additional element
of creativity.
It thus seems that we meet in nature underdetermined and overde-
termined problems as well as the well–determined problems of classical
mechanics.
A simple mathematical model for overdetermined and underdeter-
mined problems. Consider our basic equation (3.84)
Af = g (3.89)
as a system of n × n linear equations (3.87) for n unknown parameters
f . Thus A is a n × n square matrix , assumed regular. The solution is
uniquely determined and given by (3.85),

f = A−1 g . (3.90)
This time we have a well–posed problem since the inverse of a regular
square matrix A exists and is unique.

Figure 3.26: The matrix A for a well–determined (a), overdetermined

(b) and underdetermined (c) system of linear equations.

In the well–determined case of Fig. 3.26(a), the matrix inverse A−1

is regular and uniquely defined. Here the inverse problem is well–posed.
In the other cases (b) and (c), the solution may also be written
in the form (3.90) but now A−1 is not uniquely defined. Both the
overdetermined and the underdetermined inverse problems are ill–posed.
144 CHAPTER 3. PHYSICS

In the overdetermined case (b), the system of equations (3.89) is

in general inconsistent. It is treated like the adjustment problem in
sec. 2.6. A small vector v is added to g, and the equation is solved
subject to the minimum condition

vT P v =⇒ minimum , (3.91)

generalizing (2.39) on p. 65. P is an arbitrary symmetric positive–

definite matrix. (If you do not know what positive–definite means, for-
get it immediately, unless you like high–sounding and pompous words.)
The solution is
f = (AT P A)−1 AT P g . (3.92)
In the underdetermined case the system (3.89) is consistent but has
infinitely many solutions. Using the minimum condition

f T Q−1 f =⇒ minimum , (3.93)

the solution is
f = Q AT (A Q AT )−1 g (3.94)
quite similar to (3.92).
The free choice of the matrices P and Q expresses the fact that the
solution is not unique. Once P and Q have been fixed, however, the
solution is unique.
The generalized inverse A−1 is the product of matrices preceding g
at the right–hand sides of (3.92) and (3.94), respectively.
If you are not able to follow the mathematical argument, don’t
worry. The main result is that, by assuming definite matrices P and Q,
even overdetermined and underdetermined linear systems get a unique
solution.
Obviously this model is quite flexible because P and Q can be cho-
sen rather arbitrarily. As a matter of fact it is not claimed that the un-
derdetermined quantum problem or Whitehead’s overdetermined prob-
lem should be treated in this way. It is always good to know, however,
that well–defined models for the solution of overdetermined and under-
determined systems exist and can be used if necessity arises.
The present considerations also show the basic role of probability for
the solution of ill–posed inverse problems. Since a unique solution does
not exist, one tries to find the “best” solution on the base of statistical
considerations. Both least–squares principles (3.91) and (3.93) have
a statistical background: the “weight matrix” P and the “covariance
3.9. INDUCTION, VERIFICATION, FALSIFICATION 145

matrix” Q represent, so to speak, our statistical information on the

problem under consideration.
The importance of additional information. We have just considered
statistical information. Most inverse problems are problems of applied
rather than pure mathematics, so the physical etc. structure of the
problem furnishes fundamental additional information. For instance,
geological information may make geophysical (gravimetric, seismic) in-
verse problems better determined; cf. the articles by G. Anger and H.
Moritz in (Anger et al. 1993).
The set of possible solutions. Letting the matrix Q vary in (3.94),
we may get something like a set of possible solutions of eq. (3.89). The
advantage of this way of looking at the solution of an underdetermined
problem is obvious: we first get a complete set of possible solutions,
from which we may then select a suitable solution by imposing addi-
tional conditions.

3.9 Induction, verification, and falsifica-

tion
To ask whether inductive
procedures are rational is like
asking whether the law is legal.
Jonathan Cohen

Induction simply does not exist.

Sir Karl Popper

Induction
Frequently, induction is considered the inverse of deduction. Deduc-
tion proceeds from the general to the particular, using a general law to
compute particular observable quantities which then may be compared
with actual observations. Induction is said to proceed from the partic-
ular to the general, using particular observed data to derive the general
law. Thus induction is the inverse problem, in the sense of sec. 3.8, of
deduction.
This definition of induction is acceptable if we keep in mind that it
has a logical status completely different from that of deduction. Deduc-
tion is a precisely and uniquely defined, straightforward logical process
which can be formalized in terms of symbolic logic (sec. 2.1) and may
146 CHAPTER 3. PHYSICS

well be performed by a computer. It is a well–posed problem in the

sense of sec. 3.8.

In the same sense, induction is an ill–posed problem. It is impossi-

ble to uniquely derive the general law from the data. When Sherlock
Holmes claimed that he detected the criminal by rigorous logical de-
duction from the data — the traces of the crime — don’t believe him!
He had to solve a problem of induction which in general is far from
having a unique solution: it may even have no solution at all, as police
statistics show.

The task of the scientist, as “detective of nature”, is similarly a

problem of induction, which is a difficult “inverse problem” with several
solutions (a problem that can be solved by classical mechanics may as
well be solved, in a more complicated way, by relativity or quantum
theory – but not vice versa) or possibly with no solution at all (finding
a “theory of everything”, cf. sec. 6.6).

The problem of induction has been one of the most famous and
most difficult problems of philosophy, from David Hume (1711–1776)
to the present day. Let us start with some simple examples.

(1) Succession of day and night. This has been observed since
mankind came into existence, and there was never a single exception
(see also sec. 3.8). Can we conclude that tomorrow the sun will shine
again — at least above the clouds? Pragmatically we all believe that
there will be another day, but this cannot be proved logically. Induction
is not a purely logical problem. If logical procedures such as deduction
are called analytic, induction is not analytic. It is a physical problem:
there will be no tomorrow if the Earth or the Sun explode during the
night, or if the Earth has been destroyed by the impact of a huge me-
teorite. But still we may consider that with high probability there will
be another day.

(2) All swans are white. Let us assume, for the sake of argument,
that, so far, only white swans have been observed. Can we say (a) that
the next observed swan will also be white and (b) that all swans are
white? Obviously we can expect event (a) to occur with much higher
probability than the general law (b) to be true. Even if zoology claimed
that all swans are white (which it does not), a black swan could still
3.9. INDUCTION, VERIFICATION, FALSIFICATION 147

occur: a student might have painted the swan black in order to fool his
professor.
It is sometimes said that induction works if there is a certain uni-
formity of nature. This certainly applies to Example 1: the laws of
Earth rotation guarantee the succession of day and night if there is no
perturbation by a collision with a large meteorite or by an explosion as
mentioned above. But will these laws also hold tomorrow?
(3) Russell’s chicken. We quote from (Russell 1912, Chapter VI):
“Domestic animals expect food when they see the person who usually
feeds them . . . The man who has fed the chicken every day throughout
its life at last wrings its neck instead, showing that more refined views
as to the uniformity of nature would have been useful to the chicken.”
However, expectations govern our daily life as well as science. When
I come home and see and smell a nice meal, knowing that my wife is a
good cook, I expect the food to be good and healthy. When a physics
professor prepares a particularly showy experiment, knowing that he is
a skilful experimenter and trusting the laws of physics, he expects the
show to be successful. When a botanist plants the bulb of a tulip she
expects that, under normal circumstances, she will get the appropriate
flower. If a pianist starts a concert with Beethoven’s Appassionata and
touches the keys accordingly, he and his listeners expect to hear the
magnificent sounds of the sonata.
Still, one can never be sure: some jealous colleague may have put
the piano out of order right before the concert, etc.
But usually our, often unconscious, trust in the uniformity of nature
or its lawful behavior is justified, especially if we use experience and
common sense. If we discover a certain regularity and we expect an
underlying general law, then we may be justified to assume that the
regularity will persist in the near future, and if the regularity continues,
we may reasonably assume that we have discovered the corresponding
general law. Every new experiment will confirm the law and increase
its probability. This may be called induction by analogy.
Bayes’ theorem. Let H denote a hypothesis, and p(H) its initial “a
priori ” probability, or prior probability . Let E denote an event that
confirms the hypothesis. Then

p(E|H)
p(H|E) = p(H) = α p(H) . (3.95)
p(E)

This is the simplest form of Bayes’ theorem, cf. (Cohen 1989, p. 68).
It gives the “a posteriori ” probability, or posterior probability , of H
148 CHAPTER 3. PHYSICS

after the confirming event E. Here p(E) denotes the probability of E

in general, and p(E|H) the probability of E if hypothesis H is assumed.
As an example, following Cohen, take H the hypothesis that it will
rain within an hour. Let E be the presence of dark clouds. Then p(E) is
the probability that dark clouds are present, and p(E|H) the probability
of dark clouds if it will rain within the next hour (if hypothesis H is
true). Clearly, the probability p(E|H) will be considerably greater
than p(E). In fact, p(E) is the probability of dark clouds in general
(just looking out of the window without knowing beforehand how the
weather is: there may be sunshine or rain). Hence p(E|H) > p(E) and
p(E|H)
α= >1 (3.96)
p(E)
and
p(H|E) > p(H) . (3.97)
Thus the “a posteriori probability ” of H if E is observed is greater than
the original (“a priori ”) probability of H. The observation of E has
increased its probability.
Put E = E1 , the confirming event, and assume p1 (H) = p(H|E1 )
as the new “a priori” probability. Take a new piece of evidence E2 ,
for instance, that the barometer has fallen considerably, and compute
p2 (H) = p1 (H|E2 ) which can be expected to be greater than p1 (H).
Consider a third piece of evidence E3 , for instance, the broadcast an-
nouncement that very probably it is going to rain within an hour. Use
E3 to improve p2 (H), obtaining p3 (H) = p2 (H|E3 ), which by now is
presumably already very close to 1.
The mathematical foundation of Bayes’ formula is the simple rela-
tion
p(H|E)p(E) = p(E|H)p(H) , (3.98)
well known as one of the most basic formulas in the theory of proba-
bility.
Bayes’ method can also be generalized to permit a decision between
several competing hypotheses, cf. (Jeffreys 1973, p. 31).
Induction and Bayesian inference are particularly useful to find a
simple “law” from observations. This “law” may simply be a prescribed
mathematical formula such as a polynomial
y = a 0 + a 1 x + a 2 x2 (3.99)
whose parameters a0 , a1 , and a2 are to be estimated by observa-
3.9. INDUCTION, VERIFICATION, FALSIFICATION 149

tion (“curve fitting”). Here Bayesian estimation has become a suc-

cessful competitor to least–squares estimation (sec. 2.6). This is the
case in geophysics (Jeffreys 1961, 1973), but also in geodesy which
has been a stronghold of least–squares estimation since Gauss, cf.
Karl–Rudolf Koch, “Bayesian Inference with Geodetic Applications ”,
Springer, Berlin 1990.
An objection against Bayesian methods is that it requires the in-
troduction of prior information, in particular, a priori probabilities.
However, even the use of vague prior information leads in many cases
to acceptable results.
Conclusion. Karl Popper, known for his pointed statements, said
(Miller 1985, p. 104): “Induction simply does not exist, and the op-
posite view is a straightforward mistake.” The theory of inverse and
ill–posed problems, however, permits to “tidy up” the problem of in-
duction in the sense of (Jeffreys 1961, p. 8).
Traditional mathematical problems were well–posed. In 1902 the
French mathematician J. Hadamard explicitly defined well–posed prob-
lems by means of the three conditions given in sec. 3.8: existence,
uniqueness, and stability. During the last decades, however, it turned
out that many meaningful problems of mathematics and mathematical
physics were ill–posed or improperly posed , for instance, inverse prob-
lems, of which some examples are given in sec. 3.8 and more will be
found in (Anger et al. 1993), and deterministic chaos which was treated
in sec. 3.2. In fact, such ill–posed problems not only have lost the neg-
ative connotation implied by their name and have become mathemat-
ically respectable: because of their very difficulty and their practical
relevance they are currently even quite fashionable in mathematics.
If induction is to be considered a “logically well–posed problem”
such as logical deduction of which a unique solution always exists, then
Popper is right. The solution of the problem of induction is usually
non–unique. Also, it is not a purely logical problem: a priori statisti-
cal, physical, etc. information will make the problem of induction more
definite. This is exactly analogous to mathematical inverse problems
(see end of sec. 3.8). Just as mathematical “ill–posed” problems are
mathematically respectable, fascinating, and practically important, in
the same way induction is logically respectable, fascinating, and prac-
tically important, cf. (Cohen 1989).
(This relation to ill–posed problems is the reason why induction is
treated here rather than in Chapter 2 where it might seem to have been
better placed, but induction is not a purely logical problem!)
150 CHAPTER 3. PHYSICS

Verification and falsification

The general theories of science have not been found by induction.
They are “intelligent guesses” based, of course, on observational data
but primarily on other criteria such as logical simplicity and elegance,
the theoretician’s experience with older theories and, to put it bluntly,
by genius and good luck, helped by the “unreasonable effectiveness
of mathematics” to be mentioned below. Many people have admired
Linné’s classification of zoology and botany, but it required the genius
of Charles Darwin to discover, on its basis, the theory of evolution.
Many people have looked on world maps and noted the surprising gen-
eral similarity of the coast lines of South America and Africa, but it
required Alfred Wegener (1880–1930) (according to W. Schröder, pre-
ceded by H. Wettstein in 1880) to discover continental drift which then
became plate tectonics as the basic contemporary “paradigm” for ge-
ology, geodesy, and geophysics (sec. 3.10). It was the genius of Isaac
Newton to discover the general laws of mechanics underlying Galilei’s
principle of inertia and Kepler’s laws of planetary motion. The Lorentz
transformation was known to Hendrik Lorentz and Henri Poincaré be-
fore Einstein recognized its general physical significance, founding the
special theory of relativity. Just by a logical extension of the prin-
ciple of relativity to general coordinate transformations, postulating
“general covariance” of the physical laws, Einstein discovered the gen-
eral theory of relativity, having, as the only empirical basis, the striking
equality between “inertial mass” and “gravitational mass” verified with
extremely high precision by Roland Eötvös.
Thus the great scientific theories are invented by genius, developed
fully by mathematical deduction, and only then subjected to empiri-
cal testing or verification. Special relativity and quantum theory have
been used for a huge number of applications, and each application has
confirmed the theories; there has never been a single case where these
theories would have been found wrong. The possible tests of general rel-
ativity have been fewer in number, but also here the theory passed with
flying colors. The first confirmation of general relativity was by observ-
ing a solar eclipse by an expedition of the British Royal Society to South
Africa in 1919. It was a truly ceremonial occasion when the expedition
returned and announced that Einstein’s theory was fully confirmed. A
dramatic description of this event may be found in (Whitehead 1925,
Chapter I).
Sir Karl Popper has maintained that no amount of verification can
ensure the validity of a theory, whereas a single falsification is sufficient
3.9. INDUCTION, VERIFICATION, FALSIFICATION 151

to overthrow it. Therefore, falsification is logically more important than

verification. (This was already known to Francis Bacon, 1561–1626.)
This has an important counterpart in mathematics: a single coun-
terexample will invalidate the most sophisticated mathematical proof,
showing that it must contain a logical error. Counterexamples are also
useful to find out the exact scope of a mathematical theorem, and are
often used in this way, both by active researchers and by professors to
give their students a feeling of the applicability and the limitations of
a theorem.
However, this holds for pure mathematics and pure logic. Actual
data are almost always affected by uncertainties and errors. Thus even
Popper’s falsification is not absolute: the falsification may only be ap-
parent, caused by a measuring error, whereas in reality, the theory is
true. This is not a theoretical speculation: many modern experiments
operate in the gray zone between error and reality: it may be difficult
to decide whether a certain small effect is “real” or due to measuring
errors.
Following Gauss, astronomers, geodesists, and geophysicists have
developed a strong feeling for the importance of measuring errors: they
have acquired a psychological habit of basic mistrust towards obser-
vations. This is not because they do not respect observations. Quite
on the contrary: they respect them so much that they want to know
exactly how reliable and how accurate they are.
Measurements should always be accompanied by their “root mean
square error ”, or “standard error ” which is a measure of their accuracy,
e.g. for a measurement of length l:

l = 124.327 m ± 2 mm .

Another principle is “Eine Messung ist keine Messung ”, one measure-

ment is no measurement unless it has been checked by another mea-
surement.
This is an immediate practical objection against falsification: one
cannot be sure that the one falsifying measurement is really correct.
Thus both verification and falsification are necessary: as soon as
a promising new theory has been published, it will immediately be
subjected to experiments, if possible of the “crucial” type (see below),
and every attempt of verification is also an attempt of falsification,
depending on the outcome of the experiment. So a theoretician need not
worry about verification or falsification of his theory: his experimental
friends will try to verify it and his opponents will be most happy to
152 CHAPTER 3. PHYSICS

falsify it. The more experiments are performed, the more the effect of
measuring errors will be reduced.
It is fundamental that physicists and natural philosophers put much
emphasis on Heisenberg’s uncertainty principle, but ordinary measuring
errors, which occur much more frequently and may be much larger, also
deserve their attention; cf. (Jeffreys 1961, pp. 13–14). Jeffreys’ books
merit particular respect because they were written by a scientist of
enormous experience with actual “dirty” data.
Many extreme logical or philosophical conclusions do not apply to
our real “fuzzy” world because of their very subtlety: they are used
as razors, not for splitting hairs (which would be appropriate) but for
cutting trees.
The results of deduction are logically true if the process of deduction
is done correctly. The results of induction, including verified or, better,
not–yet–falsified, theories, can, at best, be probable or acceptable on
a hypothetic basis. No physicist will consider relativity or quantum
theory “absolutely true”, in the same sense as 2+2=4. He regards them
as excellent and unsurpassed working tools, even as correct, but only in
the sense of exceptionally good approximations; cf. also sec. 6.5. In this
way, also Newtonian mechanics remains “correct”: for small velocities
and phenomena above the quantum level.
Crucial experiments. Nevertheless, there are crucial experiments
which really permit to decide between two theories or hypotheses. One
of the most famous crucial experiments is the Michelson–Morley ex-
periment which shows that the velocity of light is the same along all
directions on the moving Earth. This is incompatible with classical me-
chanics, eq. (3.36) on p. 91, but perfectly compatible with the special
theory of relativity.
It is instructive to study this case by means of Bayes’ formula. Gen-
eralizing (3.95) to the case of two competing hypotheses H1 (classical
mechanics ) and H2 (relativity ), we have

p(Hi )p(C|Hi )
p(Hi |C) = (3.100)
p(H1 )p(C|H1 ) + p(H2 )p(C|H2 )
= A p(Hi )p(C|Hi ) (3.101)

where A is a constant which is independent of whether H1 or H2 is

accepted; i denotes 1 or 2.
Now, p(Hi ) denotes the prior probability of hypothesis Hi , p(Hi |C)
its posterior probability after performing the crucial experiment C, and
3.9. INDUCTION, VERIFICATION, FALSIFICATION 153

p(C|Hi ) is called the likelihood which is the probability of C on the

basis of Hi . We may thus write

posterior probability is proportional to prior probability × likelihood

(3.102)
(Jeffreys 1973, pp. 30–31). It is presupposed that either H1 or H2 is
true.
Now p(C|H1 ), the probability of the result of the Michelson–Morley
experiment (the light velocity on Earth is the same in all directions) on
the basis of classical mechanics is very small; we put

p(C|H1 ) =  . (3.103)

If the Michelson–Morley experiment would be absolutely true, then

 = 0, but no experiment is absolute so we take  to be a small number
> 0. The more reliable the crucial experiment, the smaller is .
Similarly
p(C|H2 ) = 1 − δ (3.104)
since on the basis of H2 (relativity), C theoretically must be true, but
we take δ > 0 but small for the same reason as  > 0.
Let us first assume that the prior probabilities of H1 and H2 are
equal:
1
p(H1 ) = p(H2 ) = . (3.105)
2
At any rate
p(H1 ) + p(H2 ) = 1 (3.106)
since either H1 or H2 are considered true.
Then (3.100) gives
0.5  .
p(H1 |C) = = =, (3.107)
0.5 + 0.5(1 − δ) 1+−δ
0.5(1 − δ) 1−δ .
p(H2 |C) = = = 1 −  , (3.108)
0.5 + 0.5(1 − δ) 1+−δ

stating that H1 (classical mechanics) has become very improbable and

that H2 (relativity) has been confirmed, the better, the more reliable
the crucial experiment is.
Let us now assume that the prior probability of H1 is high:

p(H1 ) = 0.9 , (3.109)

154 CHAPTER 3. PHYSICS

so that
p(H2 ) = 0.1 (3.110)
in view of (3.106). Then (3.100) gives
0.9 9 .
p(H1 |C) = = = 9 , (3.111)
0.9 + 0.1(1 − δ) 1 − δ + 9
0.1(1 − δ) 1−δ .
p(H2 |C) = = = 1 − 9 . (3.112)
0.9 + 0.1(1 − δ) 1 − δ + 9
Thus, the crucial experiment has made to drop the probability of clas-
sical mechanics from 0.9 to 9. If  = 10−6 (the experiment is very
reliable), then
.
p(H1 ) = 10−5 , (3.113)
.
p(H2 ) = 1 − 10−5 , (3.114)

which shows that relativity is confirmed also in this case: p(H2 ) has
risen from 0.1 to almost 1!
In fact, the assumption p(H2 ) = 0.1 is not too high because, even
before the Michelson–Morley experiment it was known that the Lorentz
transformation (3.38) on p. 92 holds for electrodynamics: Maxwell’s
equations are invariant with respect to (3.38); this was precisely what
Lorentz showed. So in view of the universality of physics, there was a
certain a priori probability that H2 would hold also for mechanics.
We see that basically the same result is obtained for very different
prior probabilities. This indicates that the choice of the prior probabil-
ities is not very essential.
We shall have much more to say on the laws of nature in following
sections. There is a long way from “Russell’s chicken” to the “eter-
nal inexorable laws of nature” of romantic poets and happy–minded
scientists, and a still longer way to “Schrödinger’s cat” . . .
The unreasonable effectiveness of mathematics. Already at this
point, however, we mention, for the first time in this book, a fact that
has intrigued physicists from Kepler to Einstein. The great quantum
physicist Eugene Wigner has called it the “unreasonable effectiveness
of mathematics in the natural sciences”. Penrose (1989, p. 430) puts it
as follows (“SUPERB” theories are, e.g., special and general relativity
and quantum mechanics, cf. p. 260):
It is hard for me to believe, as some have tried to maintain, that such
SUPERB theories could have arisen merely by some random natural selection
3.10. THE STRUCTURE OF SCIENTIFIC REVOLUTIONS 155

of ideas leaving only the good ones as survivors. The good ones are simply
much too good simply to be the survivors of ideas that have arisen in that
random way. There must, instead, be some deep underlying reason for the
accord between mathematics and physics, i.e. between Plato’s world and the
physical world.
Summary. Induction does exist. In simple cases, induction by anal-
ogy can be used with some care: Bayes’ theorem may help, also in the
estimation of parameters for simple laws given by a function contain-
ing several parameters. More general laws such as relativity require
the creative mind of a great scientist; verification or falsification by
experiment are necessary but can be expected almost automatically
to be performed by the scientific community. Crucial experiments are
especially important.
Additional reading. A nice introduction is the chapter on induction
in (Russell 1912). If you read what Russell (1948), Carnap (1950, 1966),
Popper (1977; Miller 1985), and Jeffreys (1961, 1973: if you don’t need
them, disregard the formulas but do read the text) have to say about
induction, and if you conclude with (Cohen 1989), then you should
know almost all the relevant present views on the topic.

3.10 The structure of scientific revol-

utions according to Kuhn
Led by a new paradigm, scientists
adopt new instruments and
look in new places.
Thomas Kuhn

T.S. Kuhn (1970) has given a theory of the history of scientific

revolutions which has attracted general attention (but also some con-
troversy as usual in such cases). Scientific revolutions introduce not
only new theories but imply a change in the general scientific climate.
Old paradigms are replaced by new ones. A paradigm is more than
a scientific theory: it is a way of thinking, a way of looking (Greek:
theoria ) at nature.
Perhaps the most famous new paradigm was the Copernican revo-
lution. The Earth was no longer the center of the universe. It is one
of the planets that orbit around the Sun. Kepler’s laws of planetary
156 CHAPTER 3. PHYSICS

motion and the mechanics of Newton, Legendre, and Laplace were the
consequences.
Another revolutionary change of paradigm was biological evolution.
The static system of botanical and zoological classification of Carl von
Linné (1707–1778) was made into a dynamic theory of evolution by
Jean Lamarck (1744–1829) and Charles Darwin (1809–1882).
Immanuel Kant (1724–1804) performed a “Copernican revolution”
in philosophy, emphasizing the role of the subject. In this way, he
founded the great school of German idealism (Fichte, Schelling, Hegel),
but he also influenced the contemporary philosophy of science.
The logical discoveries of George Boole (1815–1864), Gottlob Frege
(1848–1925), Giuseppe Peano (1858–1932) and Bertrand Russell (1872–
1970) lead to analytical philosophy and the modern theory of science,
or philosophy of science.
In physics, of course, we have the new paradigms of relativity and
quantum theory, and most recently, chaos theory.
The recent development in biochemistry and molecular biology
started modern genetics. Reductionism (life can be reduced to chem-
istry and physics) is not a logical consequence of these developments,
but is rather generally accepted at least as a working hypothesis.
Cybernetics, system theory, catastrophe theory, complexity theory,
and synergetics also have influenced modern scientific thinking, closely
related to the advent of electronic computers. These concepts will be
explained later (sec. 4.2).
In geosciences, we now have the paradigm of plate tectonics. The
Earth’s surface consists of a number of continental plates which move
with a speed on the order of 5 cm per year. Colliding with each other,
they pile up mountain chains such as the Rocky Mountains and the
Andes, but also the Himalayas and the Alps, accompanied by earth-
quakes and volcanism. It started with Alfred Wegener’s continental
drift (sec. 3.9) published in 1915, but it was generally recognized only
in the sixties.
A scientific revolution does not occur only because what Kuhn calls
“normal science” has been proved wrong or, in Popper’s terminology,
“falsified”. The old epicycle theory of planetary motion by Ptolemy
(2nd century A.D.) could have easily been adapted to the increasing
measuring accuracy by adding one more epicycle or two; it simply grew
too complicated to retain credibility. Kepler was motivated not only by
his belief in the simplicity and harmony of the world (which, in some
way or other, is shared by modern scientists as well) but also by mys-
tical speculations: he related the planetary orbits to the five Platonic
3.10. THE STRUCTURE OF SCIENTIFIC REVOLUTIONS 157

solids (p. 118). Genius is inexplicable: reasonings are frequently quite

irrational and even the significance of the results is not always properly
understood by their discoverers.
Lorentz found the equations of the “Lorentz transformation”, but
only Einstein and Minkowski realized their revolutionary physical sig-
nificance: the special theory of relativity was created. Einstein tried
to generalize the invariance with respect to Lorentz transformations
to “general covariance”: the general theory of relativity was discov-
ered. Now, however, we put more emphasis on curved space–time than
on the democratic equivalence of all reference systems which general
covariance means (sec. 3.4).
In fact, general relativity is not the only space–time theory of grav-
itation; there are other theories that explain the phenomena equally
well. Still, few physicists doubt that general relativity, in a way, is the
“best” theory of gravitation, in view of its incomparable internal per-
fection. (Remember the “unreasonable effectiveness of mathematics”
of sec. 3.9!)
Still, the great French mathematician Henri Poincaré (1854–1912)
believed that Euclidean geometry would never be given up precisely
because of its internal perfection. In fact, gravitation can be handled
(and is handled practically) in many cases by a basic Euclidean geom-
etry (and Newtonian physics) plus “relativistic corrections”, but the
conceptual framework of general relativity is still recognized superior
to this way of putting gravitational “epicycles” onto Euclidean geome-
try.
Thus in many cases, scientific revolutions are not always “logically
necessary”: normal theory could still continue for quite a while, perhaps
by piling up “epicycles”, but finally, the edifice collapses and is replaced
by a new “paradigm”.
Like many revolutions, new paradigms tend to be dogmatic. Moser
(1989, p. 145) has pointed out that a molecular biologist would commit
scientific suicide if he would support some kind of vitalism (sec. 4.5)
instead of the current dogma of reductionism. A geophysicist who flatly
contradicts plate tectonics has little chance to get his work published
in a reviewed journal.
Physicists, especially renowned ones, are relatively free to express
also views that run counter to the prevailing opinion. This attitude
has perhaps been fostered in physics by people like Niels Bohr who
said that “a theory must be very crazy to be true”. In fact, mathe-
matical models that run so much counter to common sense as Everett’s
“many–world interpretation” of quantum theory or the theories of su-
158 CHAPTER 3. PHYSICS

persymmetry and superstrings are being seriously discussed in physics.

This is what makes the extravagant mathematical models and the in-
credibly imaginative thinking of modern physicists so attractive and
important philosophically: they show how far logical speculation can
go. The most imaginative and unorthodox great contemporary physi-
cist is probably John A. Wheeler; the book (Wheeler 1994) thus is
enormously important regarding possible philosophical implications of
modern physics, although it is not quite easy (Rabbi type 2 to 3).
This also seems the reason why great physicists rarely content them-
selves with the current school of analytical philosophy. Einstein was a
pantheist of Spinoza type, Gödel was an objective idealist who rejected
the reality of space–time, Schrödinger’s views were close to Buddhism,
Bohr was a natural dialectic thinker, Weizsäcker likes classical philoso-
phy as much as physics . . .
The contemporary science business and paper industry with its ref-
ereed journals is probably necessary. Still, one wonders how Max Planck
could have published his first paper on quantum theory in a refereed
journal . . . Nevertheless, even today, novel ideas get their chance, e.g.,
through presentation in scientific meetings.
The scientific climate prevailing today is very well characterized by
the following anecdote kindly communicated to the author by Professor
Paul Melchior from his personal collection:

Why God Failed to Receive University Tenure

– He had only one major publication.
– It was not published in a refereed journal.
– It had no bibliography.
– It was not in English.
– He did not even entirely write it Himself, but had a number of mostly
anonymous co–authors.
– It may be true that He created the world, but what has He done or
published since?
– His relations with His scientific colleagues are often strained.
– The scientific community is having a very rough time trying to redis-
cover His unpublished results.
Chapter 4
Systems, information and
evolution

4.1 Feedback, regulation, and downward

causation
A variation of one–half degree centigrade in
the body temperature is generally a sign of
illness, and a permanent variation of five
degrees is scarcely consistent with life.
Norbert Wiener

In sec. 1.1 we have seen that, in man and animals, the hypotha-
lamus serves, among other functions, as a thermostat regulating body
temperature in order to maintain this temperature at a constant level
within very narrow bounds. Feedback and regulation mechanism play
a fundamental though largely unconscious role in the working of the
human body, in movements such as walking, and may even be respon-
sible for what we call the action of mind on our body, such as reaching
for a book which we need when working at a problem. Such an action
from a higher level (thinking or willing) to a lower level (bodily move-
ment) is called downward causation. Also the activity of a computer
when we feed in a program and data, or instruct it to display a certain
information on the screen, are examples of downward causation (caus-
ing activity on the “lower” hardware level by providing input on the
“higher” software level).
An essential concept is feedback . A teacher gives a course, and
the students listen with more or less attention. If the teacher cares

159
160 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

about being understood, he needs information about how his teaching is

received by the students. This is an example of feedback. Feedback may
comprise questions put during or after the lecture, criticism provided
by the students, or even the average result of the examinations, as
well as watching the students’ behavior during the lecture, from bright
attentive faces to yawning and sleeping.
The direct “feed–forward” of information from teacher to student,
together with this feedback, forms a characteristic loop, the feedback
loop (Fig. 4.1).

Figure 4.1: A feedback loop

Figure 4.2: The feedback loop for automobile driving

Another example from every–day life: the driving of an automobile

(Fig. 4.2). The driver’s (physical) activity comprises turning the steer-
ing wheel, alternately depressing gas pedal and brake pedal, shifting
gears (in cars which are not yet equipped with automatic transmis-
sion), etc. The final goal is reaching a certain destination, the im-
mediate goal, however, is staying on the road and avoiding accidents.
This goal requires the driver’s constant observation of the momentary
situation, comparing the car’s movement to the movement desired by
the driver and correcting deviations by appropriate physical activities
such as turning the steering wheel, accelerating or braking. The new
4.1. FEEDBACK, REGULATION, DOWNWARD CAUSATION 161

course of the automobile is again fed back to the driver, deviations are
observed and corrected, etc.
This example also helps introduce two additional concepts. The
automobile acts as some kind of servomechanism for man, helping
him/her to cover large distances much faster than by walking. Ser-
vomechanisms in the narrower sense frequently replace human move-
ment and observation of deviations (from a goal) by automatic func-
tions. This happens, for instance, with a stabilized platform: deviations
from a fixed orientation are measured by gyroscopic sensors and acti-
vate servomotors which restore the desired fixed orientation. Modern
airplanes are frequently equipped with autopilots which, using inertial,
radar, altimeter, etc. information will relieve the pilot from consider-
able purely routine work, frequently do certain specific and well–defined
tasks more precisely and accurately than the human pilot could do.
A second important concept is the name, cybernetics or kybernetics,
for a whole new discipline: control and communication in man, animals,
and machines. This name was introduced in 1947 by Norbert Wiener,
cf. (Wiener 1961, p. 11). It is derived from the Greek word κυβερνη0 τ ης
(kybernetes ), which means steersman or pilot.
It shows that machines exist which are designed to perform a certain
task (Aristotle’s “final cause ”) through appropriate mechanisms oper-
ating according to the usual laws of causality in the physical sense (Aris-
totle’s “efficient cause ”). Thus the gap between physics (believed to be
subject to usual causality) and biology (where “goals”, “aims”, “pur-
poses” and similar “final causes” seem to govern the behavior of animals
and even the course of evolution) has at least partially been bridged.
Earlier, a “vital force” has been considered necessary to explain “fi-
nalistic” animal behavior, but, according to Wiener (1961, p. 44): “In
fact, the whole mechanist–vitalist controversy has been relegated to the
limbo of badly posed questions”.
Other examples of servomechanisms operated by man are bulldozers
and other excavating machines, replacing shovels and similar tools. In
fact, such tools and machines can be regarded as extensions of our body
to perform certain tasks much better than using our hands only.
Now it is basic that our hands, feet and other bodily organs may
also be considered instruments or tools (“servomechanisms”) directed
by our mind (or by the neural activity in our brains, if you prefer) to
perform certain operations, for instance reaching for a glass of water
and bringing it to our mouth. This operation is constantly controlled
by visual and other feedback, as we easily recognize if we try to perform
it in darkness.
162 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

Here we have an instructive case taken from (Wiener 1961, p. 95).

A patient tries to perform the task just mentioned. He will take the
glass and move it violently in a direction towards his mouth. He moves,
however, too far in an initial direction and tries to correct this by an
equally violent movement in an opposite direction, until his motion
becomes nothing than a futile and violent oscillation. Needless to say,
he will have spilled the glass of water before even having it brought
close to his mouth. What has happened?
The patient is suffering from what is called cerebellar tremor. The
feedback mechanism is abnormally strong, so instead of a steady motion
(with small oscillations, to be sure), we have oscillations which grow
indefinitely and never “converge” to achieve the desired goal.
Or consider a loudspeaker amplifier. If the amplification becomes
too high, the amplitude of the oscillation grows in an uncontrolled way,
resulting in a shrill and penetrating uniform noise.
Here we have met with an important concept: amplification, which
is basic in our context. An amplifier converts the weak voice of a speaker
to the majestic message coming from a loudspeaker. The excavating
machine enormously amplifies the signal input provided by the opera-
tor’s hands. The extremely weak electric and/or chemical nerve signal
coming from our brain must also be amplified enormously so that we
can lift a heavy load, etc.
Thus all our bodily movements are possible only through elaborate
processes of feedback and amplification. The muscles contain sensors
whose outputs are fed back to the brain. A wonderful example is walk-
ing. We usually are not aware how delicate and complicated these
processes are until we try to walk at night or move on an icy surface.
In fact, it is extremely difficult to construct a walking machine.
As a final simple example consider the working of a thermostat. Its
function is to keep the temperature of a room constant even if external
circumstances (outside temperature etc.) vary. The desired tempera-
ture serves as input. A thermometer measures the actual temperature.
The difference: desired minus actual temperature, activates the heating
or cooling system such as to counteract the change (negative feedback ).
Possible modes of behavior are illustrated by Fig. 4.3. Starting from an
initial temperature, a well–regulated thermostat will approach the de-
sired temperature through small oscillations whose amplitude rapidly
decreases to zero. This is the usual case of negative feedback .
Consider now the right–hand side of Fig. 4.3. The graph of “actual
temperature”, starting at the same initial temperature, moves in the
right direction, towards the “desired temperature”, but widely over-
4.1. FEEDBACK, REGULATION, DOWNWARD CAUSATION 163

Figure 4.3: Negative and positive feedback

shoots the mark, and then oscillates back and forth with rapidly in-
creasing amplitude, so that a person in the room would not even have
the choice between being heated and frozen to death: this would hap-
pen in rapid alternation. This corresponds to positive feedback which
clearly does not provide the desired stabilization of temperature, quite
similar to the cerebellar tremor mentioned above.
Mathematically, this is evident: negative feedback corresponds to
exponentially decaying cosine (or sine) functions

e−αt cos ωt , (α > 0) (4.1)

whereas positive feedback, expressed by

eαt cos ωt , (α > 0) (4.2)

gives exponential growth of the oscillation. It is clear that the behavior

of the above–mentioned patient suffering from cerebellar tremor, as well
as the shrill loudspeaker noise, are described by functions of type (4.2).
Negative feedback also provides a stabilizing element in amplifica-
tion, as every electronic engineer and every hi–fi fan knows.
Downward causation. In principle, an arbitrarily small signal can
be amplified with high fidelity to an almost arbitrarily high power.
But what about if the input signal has no power at all ? Examples are
“powerless” compact discs which reproduce all the original magnificent
volume of sound of the Alpensymphonie of Richard Strauss. Of course,
a compact disc player needs considerable outside power to activate the
loudspeakers etc., but the point is that the required outside power must
be available regardless of the kind of musical information stored in the
164 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

disc, that is, independently of the signal and its (zero) power. Input
(programs and data) contained in magnetic discs or tapes may cause a
computer to perform rapid mathematical and logical operations which
require considerable electric power (which is, however, again indepen-
dent of the size of the numbers in a particular kind of input) and its
degradation to thermic energy (remember the cooling towers of modern
supercomputers); and “powerless” books such as the Bible, the Koran
or, to a lesser extent, the works of Marx, Engels, and Lenin, have com-
pletely changed the face of the Earth.
This is possible through the use of some “reading equipment” such
as laser or magnetic scanners, or people reading those books, which
convert the “written” input into small electrical signals (including the
eye converting the optical information into nerve impulses!), which are
then greatly amplified and produce the results just described.
Electronic computation, hardware being activated by software, is
a beautiful example of downward causation: information on a higher
level (software) causes action on a lower level (hardware). Note that
this hardware motion is fully governed by physical laws, in particular
the laws of electronics! “Software laws” activate the appropriate “hard-
ware” (physical) laws; “final causes” act through activating the appro-
priate “efficient causes”. Software provides initial conditions causing
the computer to start working, as well as boundary conditions which
regulate its work during the computation. Generally, the term “bound-
ary conditions” is used as a compact synonym for both initial conditions
and boundary conditions in the narrower sense.
The interaction of mind with matter may well be of this kind. Ob-
jections that mind does not possess physical energy and cannot there-
fore act on matter, lose their force if we compare the action of mind
on matter to the action of software on hardware in a computer. For
more details cf. (Popper and Eccles 1977, Chapter E7), (Eccles 1994),
(Globus et al. 1976), and (Margenau 1984). Haken (1981, p. 196) dis-
cusses interaction at various levels.
A thorough discussion of mind–matter interaction can only be given
by quantum theory since quantum effects are believed to play a role
in neural activity. A comprehensive reference is (Stapp 1993); also
(Lockwood 1989) and (Penrose 1989) provide valuable insight. One
now even speaks of “quantum computers”!
Finalism in physics. Even in classical physics (mechanics and op-
tics) there are principles that seem to express some purpose and are
thus “final causes” in the sense of Aristotle. E.g., we have Fermat’s
principle: light moves from point A to point B along a path for which
4.1. FEEDBACK, REGULATION, DOWNWARD CAUSATION 165

the travel time is a minimum, or the principle of Euler–Maupertuis

which states that a mechanical system moves in such a way that a
certain integral quantity, the “action”, is minimized (principle of least
action), cf. eq. (3.20) in sec. 3.1, p. 78.

As we have already mentioned there, these “finalistic” principles

can, by a mathematical procedure called the calculus of variations, be
transformed into the ordinary differential equations in which classical
mechanics usually is formulated and which are the expressions of classi-
cal causality. Again, “final causes” automatically produce the required
“efficient causes”!

The principle of least action acquires a particular interest in the

light of Feynman’s formulation of quantum mechanics in terms of “path
integrals”. Such integrals are sums (integrals) over all possible paths
leading from A to B. The smaller the Planck constant h (sec. 3.5) is,
the more influence is concentrated at paths very close to the “classical
path” of least action: the effect of paths far from the classical path
almost vanishes because of destructive interference, cf. (Misner et al.
1973, p. 499). For the classical case, h → 0, only the path of least
action “survives”.

Does not this remind you of a “survival of the fittest” in the sense
of Darwin’s evolution? And is it too c extravagant to regard Darwin’s
principle (survival of the fittest ) as an optimum principle which is some
kind of biological analogue to the finalistic principles of Fermat, Euler,
and Maupertuis in physics?

And to continue these pretty crazy ideas, is the synergetic cooper-

ation in the sense of Maturana and Varela (1987) and Jantsch (1980),
complementary to Darwin’s “law of the jungle”, a biological counterpart
to the Gaussian principle (3.13) on p. 75 which “adjusts” Newton’s law
to a “different physical environment” of a curved surface, accepting the
environment and “cooperating with it”? Cooperation and “democratic
adjustment” (see end of sec. 2.6) rather than struggle for survival?

Anyway, finalism or “teleology” seems to act quite obviously in bi-

ological systems, as already recognized by Aristotle. The present con-
siderations (from thermostats to downward causation and least–action
principles in physics) may thus help bridge an important gap between
living and nonliving systems, perhaps even between mind and matter.
166 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

4.2 Self–organization
Self–constructibility is an emergent
property of a complex system.
Sir Alan Cottrell

Self–organization versus external organization. Beethoven’s Fifth

Symphony is being performed. The orchestra is playing, precisely fol-
lowing the instructions of the conductor and realizing every nuance
of his conducting. This performance of Beethoven’s symphony is an
example of external organization done by the conductor.
Four players perform a string quartet, perhaps also by Beethoven.
No conductor is in view, the four players are on more or less the
same level, and their equally precise performance is achieved through
skilful and sensitive cooperation. This clearly is an example of self–
organization by cooperation or “synergy”, to use a term, synergetics,
which is becoming almost as fashionable as “cybernetics” introduced
in the preceding section.
The distinction between external organization and self–organization
is not so absolute as it might look at first sight. The conductor also lis-
tens to the orchestra and tries to lead it in a cooperative way, making
use of its strengths, taking into account its weaknesses and even at-
tempting to compensate mistakes which the musicians may have made.
The orchestra players not only watch the conductor (although some
seem to pay no attention to him at all) but also listen to the play of
their fellow musicians. This may result in a performance of overwhelm-
ing delicate precision and passion.
On the other hand, in the string quartet, the first violinist, primus
inter pares, nevertheless plays a more or less leading role.
Staying with musical examples, there are orchestras without conduc-
tor which may be very good (though not so good as excellent orchestras
under a brilliant conductor). The organist leading the congregation’s
singing in church is not a dictator either, he subtly adapts to the singing
of the people, just as a good conductor will not conduct “against” his
orchestra. And a pianist playing against a singer, rather than accom-
panying her, will soon have lost his job.
Other examples. Already in the 17th century, Huygens noticed that
two pendulum clocks hanging on the same wall tend to synchronize,
and he suspected correctly that this phenomenon of entrainment was
caused by a coupling of the two clocks through the elasticity of the wall.
4.2. SELF–ORGANIZATION 167

The synchronized playing of the two hands on the piano is mainly

controlled “externally” through the player’s ears and brain, but the
coordination of two players performing a musical piece for four hands,
is clearly another form of self–organization.
Self–coordination and “cooperation” of light waves with electrons
in a ruby laser, generating almost perfectly monochromatic light, are
a favorite example of the founder of “synergetics” (Haken 1981, Chap-
ter 5).
Waves and convection cells. I have always wondered about ocean
waves. It is clear that they are driven by wind, but why are there
waves, and if there are waves, why are their crests where they are, and
not a couple of centimeters away? (See Fig. 4.4.)

Figure 4.4: Water waves (c) formed by “self–organization” from an un-

stable inclined plane surface (b)

A uniform wind will first slightly tilt the plane ocean surface
(Fig. 4.4 (b)). This tilted plane is unstable and hence breaks down.
Then the wave motion, which is a stable motion determined by pa-
rameters of wind and water and by atmospheric pressure, takes over.
What remains undetermined and “random”, however, is the position
of the wave crests (the “phase”): the same factors might as well have
formed a wave translated to the right (as shown by the broken line) or
to the left. But once the symmetry of the smooth plane surface has
been broken, the phase is determined.
Another well–known phenomenon are convection cells formed by
heating the lower surface of a fluid layer (Bénard problem), cf. Fig. 4.5.
The latter is clearly relevant for convection in the Earth’s viscous
mantle, causing the movement of continental plates in plate tectonics.
Also regular rolling cloud patterns, frequently visible in an otherwise
clear sky, are caused in a similar way.
Chemistry and biology. Self–organization also occurs in chemistry,
168 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

Figure 4.5: Convection (a) in a plane slab and (b) in a spherical shell

forming spatial and temporal wave patterns (chemical clocks, Belousov–

Zhabotinsky reaction, cf. (Haken 1981, p. 75)).
In biology, external organization, coming from the genetic
“blueprint” (sec. 4.3), is complemented by processes of self–
organization: cell differentiation, forming cells in different parts of the
body which serve for different purposes, and morphogenesis, the for-
mation of different organs such as heart or eye.

Figure 4.6: Two separated cells with equilibrium concentration of sub-

stance A (shaded region)

Figure 4.7: Interaction between the two cells leads to two new config-
urations of equilibrium)

Very little is known in this field, which furthermore is rather con-

4.2. SELF–ORGANIZATION 169

troversial. We mention now only one of the simplest models for cell
differentiation due to the famous mathematical logician A.M. Turing
(Haken 1981, p. 104). Consider two separated identical cells (Fig. 4.6),
in which a substance A is contained in a certain equilibrium concen-
tration. Now we admit an exchange of substances between the two
cells. This may make the initial concentration of A unstable, and a
new equilibrium is reached, in which either cell 1 or cell 2 gets a higher
concentration, both possibilities being equally probable (Fig. 4.7).

Figure 4.8: Potential energy and symmetry breaking. The cases (a),
(b), (c) correspond to Figures 4.6 and 4.7

Equilibrium corresponds to a minimum of potential energy, illus-

trated by a small sphere permitted to roll without friction on the “po-
tential curves” of Fig. 4.8. The lowest position of “rest” corresponds to
minimum potential energy.

Symmetry and symmetry breaking

Symmetric shapes frequently are characterized by minimal poten-
tial energy and seem — for this and other reasons — be particularly
favorized by nature; cf. the beautiful book (Weyl 1952) and its mod-
ern counterpart (Mayer–Kuckuk 1989). Some well–known symmetric
shapes are the sphere and the cube, the equilateral triangle, the square,
the regular pentagon (5 corners), and the regular hexagon (6 corners)
known from honeycombs. Flowers exhibit beautiful symmetries; and
ornaments, based on symmetry, are as old as human culture.
Raindrops are spherical, rotating planets are (nearly) ellipsoidal.
Snowflakes also show beautifully symmetric patterns. Crystals formed
by growth in an appropriate solution are “self–organized” (in contrast
to a cut diamond formed by the “external” organization designed by
man). Symmetric patterns in biology reach from honeycombs to the
double helix of DNA, regarded as an “aperiodic crystal ” or “aperiodic
solid” by Schrödinger (1944); cf. sec. 4.3. Crystals, having minimum
energy, are particularly stable, and this fact, according to Schrödinger,
may also account for the stability of genes.
170 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

Turing’s model for cell differentiation involves a particularly simple

case of symmetry breaking. Cf. Figures 4.6 and 4.8(a) (symmetry) and
Figures 4.7 and 4.8(b) and (c) (broken symmetry). Such a symmetry
breaking is a very typical and frequently occurring phenomenon.
Think of a completely symmetrically arranged round dinner table.
Should I take the napkin to my left or that to my right? I prudently
wait until the most courageous dinner guest takes the left napkin. Then
the symmetry is broken and everyone will take the left napkin. Had
the first guest taken the right napkin, everyone would then also have
had to take the right napkin.
It has been speculated that, at the beginning of the universe (af-
ter the “big bang”), also many symmetries were broken: the symmetry
between matter and antimatter, leaving only matter (sec. 3.6); the sym-
metry between past and future, creating the “arrow of time” (sec. 3.7);
the symmetry between the four basic forces (sec. 3.6), etc.; cf. (Jantsch
1980).
Models for complexity
We have already considered Turing’s simple model for cell differenti-
ation (Fig. 4.7). Turing in 1952 also gave the first simple mathematical
model for morphogenesis (Mayer–Kuckuk 1989, p. 212).
A very detailed mathematical theory for such processes has been
elaborated by the French mathematician René Thom. It has become
famous by the name of “catastrophe theory ” and is highly ingenious but
also quite controversial, partly also because of its provocative name (it
is really “mathematical morphogenesis”).
The last 50 years have seen a number of closely related theories
trying to deal with the phenomena of complexity, self–organization etc.
It is typical that all these theories attempt to treat biological systems
and inorganic systems, both natural (convection cells) and man–made
(machines, computers), by the same methods.
It thus is quite characteristic that Norbert Wiener, the founder of
cybernetics, has defined it as “control and communication in the animal
and the machine”.
Thus we have
– cybernetics: Wiener (1961), cf. sec. 4.1;
– catastrophe theory: Thom (1975);
– chaos theory: sec. 3.2; cf. (Abraham and Shaw 1992; Lorenz
1993);
4.2. SELF–ORGANIZATION 171

– synergetics: Haken (1981);

– complexity theory: (Waldrop 1992), (Lewin 1992), (Gell–Mann

1994).

Ilya Prigogine (“non–linear thermodynamics”, “dissipative structures”)

and Manfred Eigen (“hypercycles”) did not invent such catchy names,
but they at least got the Nobel Prize, cf. sec. 4.3.
Some characteristic features of complexity. It is very difficult, if not
impossible, to define complexity, because it comprises many features
and is a field without definite boundaries. Let us nevertheless try to
list some features of complex systems.

– A great number of elements seems to be necessary but by no

means sufficient. A heap of sand contains many grains, but this
does not yet make it a complex system. A biological organism
consists of many cells and is a prototype of a complex system
because:

– a complex system possesses a rich structure, an order which is

intrinsic rather than imposed from the outside, e.g., an animal
versus an automobile. Both kinds of order are, however, com-
plementary: the genetic structure provides a certain basic in-
formation “from outside”, which is to be supplemented by cell
differentiation (e.g., a blood cell vs. a liver cell) and morphogen-
esis through self–organization, cf. sec. 4.3. An adaptation to the
environment is frequently implied; Gell–Mann (1994) speaks of
complex adaptive systems.

– The intrinsic order of a complex system is dynamic rather than

static: it must always defend itself against chaos. Think of a
warm–blooded animal: it must permanently strive to keep its
bodily temperature constant, in spite of the usually colder en-
vironment with all its random temperature changes. Another
example is a person who must constantly endeavor to maintain
mental equilibrium in spite of many disturbing impressions and
experiences.

– A picturesque description of this situation is to say that com-

plexity lies at the edge of order and chaos (Waldrop 1992); a
dialectician might call it a synthesis of order and chaos. A typ-
ical complex system encompassing order and randomness is also
172 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

the terrestrial environment consisting of atmosphere and hydro-

sphere, which is governed by “orderly” laws but is subject to
chaotic fluctuations going as far as hurricanes. Perhaps the best
book on the interrelation between simplicity, chaos, and complex-
ity is (Cohen and Stewart 1994).
– The element of chaos is by no means only negative: frequently
it provides spontaneous novelty and creativity. Natural selection
in Darwinian evolution is based on random mutations that occur
spontaneously. (The “survival of the fittest” then restores order,
possibly on a higher level.)
– The antithesis to Darwin’s “struggle for survival” is “coopera-
tion for survival ”, e.g. between algae and fungi to form lichens.
Similar to cooperation is adaptation to the environment. An im-
portant example of self–organization, struggle, cooperation and
adaptation is the market economy which is, however, beyond the
scope of the present book. An exemplary case of cooperation
between natural scientists and socio–economists is the Santa Fe
Institute, cf. (Waldrop 1992). See also sec. 4.1.
– The concept of emergence of new unexpected features is essential.
Examples are physical and chemical phase transitions, e.g. from
ice to water and then to vapor; but also the emergence of life and
the emergence of mind, cf. Fig. 2.13 on p. 56.
– There must also be a relation between complexity and the con-
cepts of information or entropy (sec. 4.3). The more complex a
system is, the more information is needed to describe it. There
is also the concept of computational complexity. Details are still
controversial, cf. (Zurek 1990), (Weizsäcker 1985, Chapter 5), and
particularly (Gell–Mann 1994, Chapter 3).
Thus complexity is a “complex” collection of interesting ideas and
mathematical models rather than a unified scientific theory such as, for
instance, quantum mechanics. It is a field that contains many fascinat-
ing open problems.
4.3. ENTROPY, INFORMATION, EVOLUTION 173

4.3 Entropy, information, and evolution

Ein kleiner Ring
Begrenzt unser Leben
Und viele Geschlechter
Reihen sich dauernd
An ihres Daseins
Unendliche Kette.
Johann Wolfgang von Goethe

Entropy and the Second Law of Thermodynamics

For our purposes, we use the definition of entropy given by the great
Austrian physicist Ludwig Boltzmann (1844–1906):

S = k log W (4.3)

where S denotes the entropy, k is a universal constant (the Boltzmann

constant), log denotes the natural logarithm, and W is a quantitative
measure of disorder of the system under consideration. A precise def-
inition of W is rather technical, but we do not need it; the interested
reader will find it in any textbook of physics.
Left to itself (without external influences), a physical system will
always tend to increase its entropy:

dS
≥0 . (4.4)
dt
The standard example is a scientist’s desk on which papers and books
are accumulating. With normal use, the books and papers become
increasingly disordered (until the scientist gets disgusted and cleans
the desk). The disorder W and hence the entropy S increase.
Another example: if we take a glass of hot water and a glass of cold
water, and pour both into a larger glass, the temperature of the mixture
very soon settles at a uniform lukewarm state. The mixture is less
ordered than the originally separated amounts of cold and warm water
whose temperatures were different, exhibiting more structure than the
resulting uniform mixture. Again S has increased.
A similar case is the dissolution of a piece of sugar in a glass of
water: the sugar solution possesses higher entropy (less structure) than
the system consisting of the piece of sugar and the glass of water.
174 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

Thus, entropy always increases. The reverse processes never occur:

the books and papers never order themselves, the lukewarm water never
separates into a cold and a hot portion, and the sugar contained in the
solution never collects to form again a piece of sugar surrounded by
pure water (the case of crystallization of sugar in a saturated solution
is another case).
The fact that the entropy never decreases spontaneously, as ex-
pressed by eq. (4.4), is called the Second Law of Thermodynamics.
(The First Law is the conservation of energy.)
Negentropy and information
This process can only be reversed by applying some information.
The desk becomes ordered again by a scientist knowing in which order
the books and papers should be arranged.

Figure 4.9: The shutter M is operated by Maxwell’s demon

In the case of lukewarm water the situation is somewhat more com-

plicated but very instructive (Fig. 4.9). As is known, molecules in a
liquid or a gas perform a highly irregular and erratic movement. The
molecules have individually different velocities, but the average velocity
increases with increasing temperature. The higher the average velocity,
the higher the temperature; in fact, the temperature is defined by the
average velocity.
Let the container of Fig. 4.9 be separated by a wall with a very small
aperture A. Let originally be the temperature the same (lukewarm) in
both compartments L and R. The aperture can be shut or opened by
moving the shutter M , operated by a small intelligent being, called
Maxwell’s demon. The demon observes the speed of the molecules ap-
proaching the aperture A. If a “fast” molecule approaches A from the
right compartment R, he opens the shutter, and the molecule can pass
to compartment L. If a “slow” molecule comes from R, the shutter re-
mains closed. If a “fast” molecule approaches A from L, the shutter is
also closed, but if a “slow” molecule comes from L, the demon opens the
shutter. In this way, “fast” molecules will accumulate in the compart-
ment L, and “slow” molecules will accumulate in R. Thus, the average
4.3. ENTROPY, INFORMATION, EVOLUTION 175

velocity of the molecules in compartment L will become increasingly

larger, and the average velocity will decrease in R. In other terms, wa-
ter in compartment L will become warmer, and in compartment R, it
will become cooler. It seems that the Second Law of Thermodynamics
has been beaten!
The answer is that Maxwell’s demon needs information about the
molecules approaching the gate, especially about their speed, otherwise
he cannot operate. In fact, the entropy difference between the initial
uniform state S0 and the final state S after the action of Maxwell’s
demon (S < S0 ) can be considered as a measure of the total information
I received by the demon in order to be able to operate:

I = −∆S , ∆S = S − S0 . (4.5)

Thus, apart from a constant S0 , information equals a negative entropy:

information = negentropy , (4.6)

as expressed by L. Brillouin. An excellent biography of Maxwell’s de-

mon together with relevant original papers is (Leff and Rex 1990).

Genetic information
Life clearly also runs counter to the Second Law, both in an indi-
vidual living being and in the course of biological evolution. Think of a
newborn baby. It grows up, learns an enormous amount of things, and
becomes a highly developed man or woman. In this course, the “or-
der” or “organization” increases, entropy (disorder) decreases. Only
after death, the Second Law takes over again: the body decomposes,
disorder and hence entropy increase again.
Similarly, biological evolution has run counter to the Second Law:
animals and plants become more and more complex and highly orga-
nized, from amoebae and bacteria to roses and human beings.
All this is far from being fully understood. Certain facts appear
sure, however. First of all, the fact of evolution is not contested by any
serious scientist. What is controversial is whether evolution proceeds in
a completely random way, according to chance mutations and Darwin’s
“survival of the fittest”, or whether it follows more or less an existing
“cosmic blueprint” (Davies 1988). Probably the truth lies between
these extremes. As Penrose (1989, p. 416) has put it:
To my way of thinking, there is still something mysterious about evolution,
with its apparent “groping” towards some future purpose. Things at least
176 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

seem to organize themselves somewhat better than they “ought” to, just on
the basis of blind–chance evolution and natural selection.

The second fact is that the development of an individual is deter-

mined, at least principally, by giant molecules, called genes, which are
contained practically unchanged in every cell of the individual’s body.
Genes are the basic carriers of genetic information which, so to speak,
is coded , in very much the same way as literary information is coded in
printed books or the performance of a Beethoven symphony is coded
on a compact disc.
The complete genetic information is contained in a set of giant chain
molecules, called DNA (nobody says DesoxyriboNucleic Acid ); this set
is also called a genome. They are Schrödinger’s “aperiodic crystals”
mentioned in the last section, and have the famous structure of a double
helix .
We have the Central Dogma of Molecular Biology enunciated by
Francis Crick (one of the co–discoverers of the double–helix structure
of DNA) in 1970:

DNA =⇒ RNA =⇒ proteins .

What is RNA? It is short for RiboNucleic Acid . RNA is closely related

to DNA (DesoxyRNA): DNA, so to speak, contains less oxygen than
RNA. DNA consists of long chains or relatively simple molecules called
nucleotides, with their bases

A: adenine,
G: guanine,
C: cytosine, (4.7)
T: thymine.

Their exact composition is irrelevant for the present purpose, as well as

the names: one always speaks of A, G, C, T only. RNA has the same
constituents, but T replaced by

U: uracil. (4.8)

As an “aperiodic solid” (sec. 4.2), DNA is a very stable “original”, or

“template”, from which RNA is obtained by “transcription”, with the
bases transcribed as follows:
4.3. ENTROPY, INFORMATION, EVOLUTION 177

DNA: A G C T
⇓ ⇓ ⇓ ⇓ ⇓ (4.9)
RNA: U C G A

The RNA thus obtained serves as “messenger” sent out by “her

majesty” DNA in order to produce proteins which are well known to
play a basic role as enzymes in the functioning of the living being. A
good comparison is to regard DNA as the legislative, and RNA and the
enzymes as the executive of the cell. A gene is that part of a DNA
strand which is responsible for the production of just one protein:

gene =⇒ protein .

What is incredibly astonishing is that each gene, or each DNA,

consists of the same four bases (4.7) only, whether DNA belongs to an
amoeba, a bacterium, a rose, or any particular human person! (This,
of course, is an overwhelming argument for evolution.)
As a matter of fact, the more complex the plant or animal becomes,
the longer will in general be its DNA.
Proteins consist of amino–acids, which correspond to triplets (say
UAG) of bases in RNA. A short but comprehensive and not too tech-
nical review is (Holzmüller 1984).

Problems and some preliminary answers

DNA is the basic and stable source of information for the formation
of a plant or an animal. How this information leads to morphogenesis
and hence to the formation of the fully developed organism, is a problem
whose study has just begun. How does the same DNA give rise to such
different cells as a cell of the liver, a blood cell, and a neuron? Another
problem is the increase of information in DNA itself, leading to more
and more complex organisms in the course of biological evolution.
There is an answer to such problems which, above all, has provided
an attractive name and indicated future directions of research: self–
organization. We have already treated it to some length in sec. 4.2,
under the heading of “complexity”, and we shall say more about it be-
low, under the name “dynamic structures”. (Repetitions from a slightly
different perspective are intentional.)
How is the Second Law of Thermodynamics bypassed? Four partial
but important answers have been suggested:
178 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

(1) Metabolism: the animal eats food (other animals, plants) which
furnish not only energy but also (negative) entropy, i.e., infor-
mation, since the food already has a rich structure. The body
uses this information for its own organization and releases the
food after having extracted energy and information, in a highly
degraded (high–entropy) state into its environment (Schrödinger
1944).

(2) Non–equilibrium thermodynamics. The Second Law holds for

states near thermal equilibrium. Animals etc. are in a state far
fromthermal equilibrium, hence the Second Law does not imme-
diately apply (Prigogine and Stengers 1984).

(3) Enzymes (organic catalyzers) may act as some kind of “Maxwell’s

demons” (Monod 1970; Eigen 1987).

(4) It seems that already in inorganic chemistry, catalyzers, enabling

or accelerating a chemical reaction, may counteract the Second
Law (Waldrop 1992, pp. 314–315).

Let us consider these questions and answers in some greater detail.

(The rest of this section may be skipped.)

Nonlinear and dissipative structures

First of all, we must distinguish two types of structure: autonomous
structures and dynamical structures (Mayer–Kuckuk 1989, p. 213). Au-
tonomous structures, like crystals and DNA (Schrödinger’s “aperiodic
crystals”), are relatively permanent, their stability being assured by
quantum mechanics; it is a static permanence not very much depen-
dent on the external world. They are in a state of equilibrium.
Dynamic structures, on the other hand, change all the time, so
their stability is dynamic rather than static. Think of yourself: you
can only exist by the intake of food and liquids, which are used and, in
a degenerated state, leave your system as waste. Dynamic structures,
like convection cells in the inorganic world (Fig. 4.5 on p. 168), thus
need interaction with the environment: heat in the convection case,
and energy and negentropy in the organic case.
It is believed that these two cases, convection and life, have more in
common than a superficial similarity. They are dissipative structures,
far from equilibrium, which constantly absorb and dissipate energy
(equilibrium is a state of minimum potential energy, so there would
4.3. ENTROPY, INFORMATION, EVOLUTION 179

be no energy consumption there). Nonlinear dissipative systems fre-

quently are chaotic (sec. 3.2). For instance, there is a close relation
between heat convection (Fig. 4.5) and meteorological chaos of Lorenz
type.
The thermodynamics of nonlinear systems far from thermal equilib-
rium was in particular investigated by I. Prigogine, who got the Nobel
Prize in 1977. Synergetics (H. Haken) is closely related, and so is chaos
theory (E. Lorenz), cf. sec. 4.2. This complex of models, theories and
ideas seems to be responsible, at least partly, for the dynamic structure
of living organisms.
The stable autonomous structure as represented by the genome
(DNA) seems to serve as the memory, or library of permanent “leg-
islative” information, as we have seen above. Another mechanism must
be able to read and copy this information (DNA → RNA) and to con-
vert the genetic “blueprint” into cell structures (RNA → proteins →
structure). This “executive” is certainly a dynamic structure.

Figure 4.10: The hypercycle of Manfred Eigen (the number of cycles

need not be 6!)

Nonlinear catalytic and autocatalytic processes generally play a

great role. These processes are frequently “synergetic”, that is, co-
operative. An example is Manfred Eigen’s “hypercycle ” (Fig. 4.10).
180 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

A “cycle” may copy part of a nucleic acid (e.g. RNA) and produce a
corresponding protein acting as an enzyme for increasing the activity
of the next cycle. Thus the action of the hypercycle is one of mutual
support of the work of the individual cycles. If Cycle 1 increases the
production of its enzymes (Enzyme 1), this enzyme will stimulate the
activity of Cycle 2, whose Enzyme 2 stimulates Cycle 3, etc.
Thus the cycles mutually help to increase their activity. It can
be shown that also the stability is mutually reinforced. The cycles
operate in an autocatalytic way, and also “assist each other” in a cross–
catalytic manner. The hypercycle is a model for cooperation rather than
competition in the organic world; cf. also (Jantsch 1980) and (Kauffman
1993). We also recognize a feedback acting within the hypercycle. As a
matter of fact, the hypercycle takes energy and “raw materials” from
the environment.
The great physicist John Archibald Wheeler was so fascinated by
the hypercycle that he spoke of the “life machine” of Manfred Eigen
(Wheeler 1994, p. 180)!
How the enzymes act on the structure of the cells and contribute to
their self–organization is a complicated and controversial issue. Cat-
alytic processes and hypercycles seem to play a role also here.
It thus appears that the organization of a living organism is of
the conductor–orchestra rather than of the “democratic” string quartet
type (see beginning of sec. 4.2).
The human genome seems to play the role of the over–all conductor,
but the musicians must be highly qualified to fill in the details of the
musical piece by improvisation (self–organization).
The genome thus determines the general features of the person, e.g.
intelligence on the one hand and genetic diseases on the other hand.
We are not, however, slaves of our genes, but can do much to make best
use of our capabilities and reduce our genetic limitations. We cannot
just shed our responsibilities and blame everything on our genes.
It is tempting to regard the human genome as the material basis
of the self–identity of every human person. This may well be, but the
material aspect is not the only one.
The stability of a personal organism is thus partly static. In part,
stability in nonlinear systems may also be dynamic, that of a fixed
point or point attractor (Fig. 4.11). (There are also more complicated
“strange attractors” such as the Lorenz attractor well known from me-
teorology, cf. (Abraham and Shaw 1992), (Briggs 1992), (Lorenz 1993).)
4.3. ENTROPY, INFORMATION, EVOLUTION 181

Figure 4.11: A point attractor. Every trajectory (path of development)

ends at this “stable” point.

Metabolism and entropy

We now take up Schrödinger’s ideas as elaborated by Prigogine. We
are following the book (Schrödinger 1944) and the modern treatment
of Schrödinger’s ideas in (Götschl 1992, Part 2).
According to the Second Law of Thermodynamics, the entropy of a
material system never decreases, cf. eq. (4.4):

dS ≥ 0 . (4.10)

In a living being, however, the internal order increases, so that entropy

decreases:
dSi < 0 . (4.11)
There seems to be a contradiction.
In fact, we must consider the total system, organism(s) plus envi-
ronment. The total entropy St must increase:

dSt = dSi + dSe ≥ 0 . (4.12)

A decrease according to (4.11) is possible if the entropy of the environ-

ment, Se , increases even more strongly so that (4.12) is satisfied.
This is achieved by metabolism. Living organisms maintain their
organization by “extracting order from the environment”. (Quotations
are from (Schrödinger 1944).) This is done by consuming food which
already has a high degree of organization (plants, other animals).
The high order of an organism, as exemplified by its DNA, has an
analogue already in the inorganic world: a crystal whose structure, as
determined by quantum–mechanical laws, is already highly ordered.
So a crystal, so to speak, is an inorganic forerunner of organic DNA,
which Schrödinger recognized as an “aperiodic solid”, of course of an
incomparably richer structure.
182 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

This is a mechanism which produces order from order : higher de-

gree order from lower–degree order. Living systems “feed upon negative
entropy”. Animals feed upon other animals or plants, but what about
plants? They take order from organic rests contained in the soil (hu-
mus) but, above all, from the sunlight which furnishes energy but also
has a rather low–entropy structure, cf. (Penrose 1989, pp. 319–321).
As Prigogine has shown, such processes occur far from equilibrium:
think of a warm–blooded animal in winter. If this animal were in
equilibrium with its environment, it would be frozen and hence dead.
Energy, much energy is needed to maintain life, keeping the organism
away from thermal equilibrium. (Consider only the high energy con-
sumption of the modern world!)
Even inorganic systems can decrease their entropy (increase their
order) through self–organization provided:

– they are far from thermal equilibrium;

– they are open, i.e., they can react with their environment;

– there is enough supply of energy and raw materials;

– there are auto–catalytic and cross–catalytic reactions such as

Eigen’s hypercycles (which are essentially nonlinear systems!).

Thus, just as DNA has its inorganic predecessor in crystals, metabolic

processes have their inorganic forerunners in non–equilibrium ther-
modynamics and self–organization such as convection cells and the
Belousov–Zhabotinsky reaction mentioned in sec. 4.2. Convection cells
(Fig. 4.5 on p. 168) are a particularly clear example of order produced
through the supply of thermal energy by heating.
It may even be that organic life is not an exception to the Second
Law of Thermodynamics, but that the Second Law comprises two af-
firmations: an increase of entropy for ordinary “simple” systems and a
decrease of entropy for “complex” systems including living organisms
((Waldrop 1992, Chapter 8), (Kauffman 1993, Chapter 8), for a related
idea cf. (Weizsäcker 1985, Chapter 5)).
Anyway, Schrödinger (as most physicists in (Küppers 1987)) be-
lieves that new laws are at work in the organisms of biology. This does
not mean a return to the old–fashioned “vital forces” but the new laws
are “software laws” in the sense of sec. 4.5 (just as the income tax law
serves as software law for the computer which calculated your income
tax, operating, of course, according to the laws of physics).
4.4. DATA AND ERRORS 183

Still, all these questions are extremely difficult, and the partial an-
swers given so far are rather controversial. Fortunately, many of the
pioneers have written excellent popular books: (Eigen 1987), (Eigen
and Winkler 1975), (Haken 1981), (Kauffman 1993), (Prigogine and
Stengers 1984), (Nicolis and Prigogine 1989), (Thom 1975), so that
the reader can judge for himself. The extremes are marked by (Teil-
hard de Chardin 1955) and (Monod 1970): evolution according to a
“holistic” plan, and completely random and mechanistic evolution. An
excellent comparative treatment of all these various tendences and di-
rections of research is found in (Moser 1989, Chapter 4). Balanced and
concise presentations are given in (Davies 1988) and (Mayer–Kuckuk
1989, Chapter 9). The little pioneering book (Schrödinger 1944) has
retained its charm and freshness for over half a decade and is my fa-
vorite. Off the beaten track are the passages on the genetic code in
(Hofstadter 1979) and in (Cohen and Stewart 1994), other favorites of
mine. The presentation of Sheldrake (1981) is unconventional but very
readable and interesting. An excellent general introduction into evolu-
tion is (Edey and Johanson 1989), combining readability with a high
level. If you look for open and controversial problems, consult (Duncan
and Weston–Smith 1977). A monograph representing the current state
of research on complexity is (Zurek 1990), a publication of the Santa Fe
Institute, whose work is excitingly described by Waldrop (1992; gen-
eral introduction) and Lewin (1992; emphasis on biology). Cohen and
Stewart (1994) show how complex evolution really is.

4.4 Data and errors

Errare humanum est.
Latin proverb

The only quantity that is exactly measurable is a number of indi-

viduals, which is a positive integer (or zero): 0, 1, 2, 3, . . . In fact, if a
basket contains 5 apples, it will be exactly 5 and not 4.9937.
All measurements of quantities which are permitted to assume a
continuous range of values and which are expressed by real numbers, are
only approximate. Even the simplest geometric quantity, the distance
between two points in “our” Euclidean three–dimensional space, can be
measured only approximately, as we have discussed at length in sec. 2.4.
The reason is that precise definitions and exact laws exist only in
mathematical logic and in mathematics (as far as Gödel permits!). In
184 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

nature, concepts are always inexact. In physics, the way out is to make
the experimental arrangement as exact as possible, and then, by a leap
of faith, idealize the situation and postulate that the measured values
obtained in this way are absolutely accurate. This may or may not
work.
One of the most accurately known physical constants is the speed
of light c in vacuum. Its value is
c = 299 792 458 m s−1 , (4.13)
that is, meters per second. The measuring accuracy is so high that the
last three numbers might be 459, but not, for instance, 464. We say that
c has been measured to an accuracy of ±1 m s−1 . This is the standard
error or r.m.s. (root mean square) error σ which is defined statistically
and known to everyone who has ever handled empirical data. Now,
σ = ±1 m s−1 does not mean that only 457 or 459 may be possible
outcomes of measurement; even an error of 3σ (possible outcomes 455
or 461) may occur, but with rather small probability. (Even larger
deviations are theoretically possible, but extremely improbable.)

Figure 4.12: The Gaussian error curve

This is a consequence of the famous Gaussian error curve shown in

Fig. 4.12. The equation of this curve is
1 x2
f (x) = √ e− 2σ2 (4.14)
σ 2π
where π and e are standard mathematical constants. (Actually f is not
a probability but a probability density, but this distinction, important
as it is, is irrelevant to the present discussion.)
This symmetric figure clearly shows that positive and negative er-
rors are equally possible, and that “outliers”, say x = 5σ, are theo-
retically possible although their probability is so small that they are
considered practically impossible.
4.4. DATA AND ERRORS 185

Important is the concept of relative standard error

σ 1 m s−1 .
= = 3 × 10−9 (dimensionless!) . (4.15)
c 299 792 458 m s −1

This accuracy is considered almost the highest accuracy that can con-
ceivably be achieved. So why not consider it “absolutely accurate”?
The chief limitation of the accuracy of
length
c= (4.16)
time
is the limited accuracy of measuring length or distance, that is, the
definition of the meter . The limited accuracy is due to the difficulties
mentioned at the beginning of sec. 2.4 regarding point definition and
distance measurement. The meter was defined as a certain multiple
of the wavelength of monochromatic light of a certain frequency, and
it was simply not possible to increase the accuracy of the frequency
definition. On the other hand, time definition by means of an atomic
clock is considerably more accurate.
So, some twenty years ago, it was in fact decided to give up the
old definition of the meter and define length as time × c. Thus the
defining constants were taken the second and the velocity of light, with
c according to (4.13) errorless and fixed once and for all. The meter
is now a derived quantity: it is the distance covered by light in c−1
seconds,
1
1m = ˆ s . (4.17)
c
Thus, in certain rare cases, it is permitted to consider an accurately
measured value as absolutely errorless, but then it may be necessary
to redefine other quantities in such a way that logical (and numerical!)
contradictions are avoided.
It would not be permissible to regard all three measured angles of
a triangle as errorless, since this results in a numerical contradiction:
the sum of the three angles in a triangle must be 180◦ or π:

α+β+γ =π , (4.18)

and this condition is not in general satisfied by the measurements. Of

course, one could avoid this difficulty and measure only the angles α and
β, computing γ from (4.18). Similar procedures are applied frequently
enough, but may not be quite fair for two reasons:
186 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

(1) Measuring α, β and γ provides a useful check to avoid gross errors,

e.g. α = 74◦ 130 3500 instead of the value (2.14).
(2) All three angles should have “equal rights” to be measured, which
is a principle of symmetry (or democracy if you prefer).
Thus, honestly, we should measure all observable quantities and re-
move any numerical contradictions by a least–squares adjustment as
mentioned in sec. 2.4 on p. 41.
This “honesty” is rewarded: the study of measuring errors, due to
R. Bošković (1711–1787), A.M. Legendre (1752–1833) and, above all,
to C.F. Gauss (1777–1855), the “princeps mathematicorum”, helped
found mathematical statistics. Least–squares adjustment was also elab-
orated by Gauss and led to the beautiful geometrical interpretation in
terms of normal projection in a higher–dimensional space mentioned in
sec. 2.6.

4.5 Complexity and reductionism

If physicalism is correct, then even a family of
monkeys in a tropical forest is “in principle”
a solution of the Schrödinger equation.
Carl Friedrich von Weizsäcker

We are resuming the study of complexity, started in sec. 4.2, from

a different perspective.
In biology there have been essentially two opposite opinions:
(A) Vitalism: A living organism is not determined by the laws of
physics only; there exist special “vital forces” which cause the purpose-
ful behavior of living organisms, their special structure, their ability to
heal wounds and even to regenerate lost organs (this is in particular
conspicuous in lower animals such as polyps or starfish), etc.
(B) Reductionism: a living organism is nothing else than a very com-
plex and well–structured system, which is completely governed by the
ordinary laws of physics and chemistry. Since chemistry, through the
laws of quantum mechanics, is thought to be reducible to physics, also
the laws governing the apparently so special behavior of living organ-
isms are reducible to the laws of physics. This is reductionism or phys-
icalism.
The main empirical data are clear:
4.5. COMPLEXITY AND REDUCTIONISM 187

(1) The behavior of animals and plants is completely different from

any mechanisms or similar man–made automata.
(2) All physical experiments performed with living organisms or with
living tissue have never indicated any measurable deviations from
the ordinarily known laws of physics and chemistry.
(3) There seems to be no sharply defined boundary between highly–
organized macromolecules and the most elementary organisms,
between chemistry and biology.
It is safest and least controversial to consider living organisms as very
elaborate and highly organized complex systems. (Is a beautiful moun-
tain flower only a complex physical system? Between ourselves, my
answer is “no”, but don’t tell this to anybody!)
Complex systems
It is instructive first to consider a highly complex modern electronic
computer. Is it governed by the laws of physics? Undoubtedly, yes.
Is it fully governed by the laws of physics only? No. Let me try to
explain.
Even in classical mechanics we need not only Newton’s laws, ex-
pressed by ordinary differential equations of the second order, but we
also need initial conditions (position and velocity at a time t = t0 ), to
get a well–determined motion. Boundary conditions are essentially the
same as initial conditions.
For a computer, the initial and boundary conditions are nothing
else than its input: program and data, in other terms, the software.
The computer with zero input constitutes the hardware (thus, for the
present purpose, hardware also comprises fixed programs built into the
computer).
With no input, the computer will not work in the sense of producing
a useful output. As soon as the software is introduced, the computer
starts to work and produces a useful output.
The input may consist of a complicated program which in itself may
contain a law : for instance a sequence of highly complex mathematics.
Or the program for computing income tax may contain a mathematical–
logical form of the income tax law (Davies 1988, p. 144). Hence we may
well speak of software laws.
Thus, the operation of a computer is governed by the physical hard-
ware laws and the non–physical software laws! (Nobody would claim
that the income tax law is derivable from the laws of physics . . . )
188 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

Thus the laws governing the behavior of living organisms may well
be regarded as some kind of software laws describing biological complex-
ity, following again (Davies 1988, p. 142). (This book is my favorite
reference for the present problems.)
Thus the “biological laws” define the initial or boundary conditions
for the work of our “biological” systems.
This comparison between animal and computer, imperfect as it is,
shows clearly not only the possibility, but even the necessity of laws
other than physics to describe the activity of a computer and, a for-
teriori , of a living organism. Note that in sec. 4.1 we have used the
computer in a similar way, as a model for mind–brain interaction.
The old fight between vitalists and reductionists is thus raised to an
objective and unemotional level. Any molecular biologist who speaks
of “vital forces” would commit scientific suicide, but if he (she) speaks
of “software laws”, nobody will pay attention to the fact that, possibly,
the same thing is expressed in two different forms.
Software contains information. Thus information must play a deci-
sive role for biological systems. Perhaps oversimplifying, we may thus
say
life = matter + information (4.19)
(Küppers 1987, p. 17). This is probably true, but what is information?
Unfortunately it is by no means a clearly and unambiguously defined
concept. The relation between entropy and information outlined in
sec. 4.3 is rather generally accepted, but neither gives a complete defini-
tion nor is understood by all scientists in the same way, cf. (Weizsäcker
1985, Chapter 5). It is one of the basic paradoxes of science and phi-
losophy that the most fundamental concepts such as matter, mind, or
information are so ill–defined.
By regarding a living organism as a complex system governed by
the laws of physics and “software laws” containing information we have
given the problem a simple (though probably oversimplified) structure
which may serve as a basis for more detailed investigations. The related
concept of self–organization (sec. 4.2) may serve for similar purposes.
Using the terminology of Niels Bohr, physical and biological laws
(if you don’t like this term, speak of software laws) are complemen-
tary (very much in the sense of complementary subspaces which are
orthogonal to each other, cf. Fig. 2.19 on p. 66 and Fig. 6.1 on p. 246).
This has been drastically formulated as follows (“Bohr’s para-
doxon ”): in order to determine whether a cat is fully governed by
physical laws, it is not sufficient to determine its weight or its bod-
ily temperature. One must use an X–ray equipment, which has to be
4.5. COMPLEXITY AND REDUCTIONISM 189

very powerful to determine the cat’s exact internal structure, so pow-

erful that it may well kill the cat or damage it irreversibly. This is
not sufficient, however: to get other physical parameters, we must im-
plant physical equipment in the cat’s body, and finally we must dissect
it. By then, the cat is surely dead. Thus, life and a “full” physical
examination are incompatible with each other!
A living organism is an individual “whole”. This is what the concept
“holism ” means: the whole is more than the sum of its parts. Since
we have already used this concept before (secs. 2.1 and 3.5), it may
be appropriate now to remind the reader again that “holism” is not a
mutilated version of “wholism” but comes from the Greek word “holos ”
which means “entire” or “whole” (after all!).
Thus “holism” and “reductionism” are complementary terms; this
is beautifully expressed in the drawing in (Hofstadter 1979, p. 310 and
the subsequent “ant fugue”).

On reductionism
As we have said at the beginning of this section, it seems that the
laws of physics hold also for living organisms. Let us consider this
question more closely. Physical laws have an inherent inaccuracy as
we shall see in sec. 6.5. Experiments so far have shown that they
are satisfied in plants and animals within the measuring accuracy; this
assertion we shall call moderate reductionism. Furthermore, as we have
just seen, more precise physical measurements may interfere with life
(of a cat, say).
So the question whether physical laws are “really” and “absolutely”
the same in living and nonliving matter may well be meaningless. Nev-
ertheless, this has been frequently asserted, so we shall call it strong
reductionism. If someone asserts this, we can immediately retort: re-
duction to which physical laws? The “real” ones if they exist? The
laws found in our books of physics? But does Nature, or God, read
those books? Questions are becoming cynical, so let us stop them.
Strong reductionism is asserted mainly by biologists. Biologists of-
ten have a much stronger faith in physics than physicists themselves.
(Similarly, physicists have a much stronger faith in mathematics than
mathematicians, cf. the end of sec. 2.3. This seems to be due to inside
knowledge: it is said that Napoleon’s valet had much less respect of him
than most other people.) The book (Küppers 1987) contains highly in-
teresting articles by great contemporary physicists on this topic which
are very relevant to our discussion. Great physicists are usually rather
190 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

cautious regarding reductionism. An eminent physicist such as Walter

Elsasser speaks of “biotonic laws”.
The old opinion of Whitehead (1925, p. 115–116) is still of interest
in this respect:
The concrete enduring entities are organisms, so that the plan of the whole
influences the very character of the various subordinate organisms which en-
ter into it. In the case of an animal, the mental states enter into the plan
of the total organism and thus modify the plans of the successive subordi-
nate organisms until the ultimate smallest organisms, such as electrons, are
reached. Thus an electron within a living body is different from an electron
outside it, by reason of the plan of the body. The electron blindly runs either
within or without the body; but it runs within the body in accordance with
its character within the body; that is to say, in accordance with the general
plan of the body, and this plan includes the mental state.
This is holism expressed in beautiful English.
Whitehead’s position does not need any “vital forces” but may
rather be regarded as imposing either constraints in the sense of sec. 3.1
(cf. eq. 3.13 on p. 75) or boundary conditions in the form of “software
laws”. To fully demystify “biotonic laws”, compare them to the income
tax law as mentioned above . . .
But let us assume now that even strong reductionism is true. Thus,
all physical laws are perfectly valid even in a living organism. This
includes the laws of quantum mechanics, and the “orthodox” Copen-
hagen interpretation (sec. 3.5) involves the mind of the observer. Thus,
via quantum mechanics, life, in being reduced to physics, brings mind
in again, especially according to the view of Eugene Wigner (p. 105).
This is probably the last thing which reductionists want, but it gives
a nice dialectic loop (Fig. 4.13) This shows where logical reasoning
may lead. I do not necessarily agree, but this may be a profound
truth in the sense of Bohr, of which the opposite is also a profound
truth (sec. 2.5). Here again, we should beware of oversimplification, of
Whitehead’s “fallacy of misplaced concreteness”!
Reductionism versus constructionism. Reductionism says that if we
investigate living beings as regards their physical properties, we shall
find that the usual laws of physics, as we know them, apply to this
process. Quite another problem is whether we can construct an animal,
say, from the laws of physics only. Thus, constructionism, so to speak,
is the inverse problem (sec. 3.8) of reductionism.
To clarify this, let us consider the following example. A small child
takes apart an old–fashioned mechanical watch and announces proudly
4.5. COMPLEXITY AND REDUCTIONISM 191

Figure 4.13: The self–reference loop of reductionism according to E.

Wigner

that the watch is nothing but cogwheels, springs, screws, pinions, and
other simple mechanical parts. The father is upset and tells the child
to put these “simple” mechanical parts together again to obtain the
original watch. The child is unable to do this. “Taking apart” was
a reductionist procedure, and “putting together” is a constructionist
problem which is much more difficult: the inverse problem is usually
essentially more difficult than the direct procedure.
In 1972 Philip Anderson asserted that “the reductionist procedure
does not by any means imply a ‘constructionist’ one: The ability to
reduce everything to simple fundamental laws does not imply the ability
to start from these laws and reconstruct the universe” (quoted after S.S.
Schweber, Physics Today, November 1993, p. 36).
H. Primas formulates it even more strongly: “Every machine relies
for its operation on the laws of physics and chemistry, but the machine’s
design is a higher–order principle. In this sense machines are irreducible
to physics” (quoted from Moser 1989, p. 137).
A rather generally prevailing compromise might be formulated as
follows. Physical laws hold throughout nature, living and lifeless alike.
What is characteristic for biology are not the laws but the boundary
conditions, which are responsible for the complexity of living organ-
isms, the physical laws being quite simple. You may call these bound-
ary conditions “biotonic” laws or “software laws”, depending on your
preferred way of thinking. Even the freedom of the will may be related
to boundary (or initial) conditions, cf. sec. 6.4 and Fig. 6.1.
192 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

Reductionism as a research program

Reductionism has been extremely fruitful as a method of research.
Mathematics has been reduced to logic with considerable success
(sec. 2.1). Chemical valence and the properties of the periodic sys-
tem of chemical elements can be derived from quantum mechanics; in
this sense, chemistry can be reduced to physics (sec. 3.5). Physics and
chemistry are absolutely essential for the functioning of a plant or of an
animal. In this sense, life is reduced to physics and chemistry. Signals
in the brain are transmitted partly electrically, or electrochemically,
along the neuron, and partly chemically, by neurotransmitters, across
the synaptic cleft (sec. 1.1). In this way, thinking may be said to be
reduced, at least in part, to physics and chemistry. Last but not least,
the study of atoms and molecules is reduced to the theory of elementary
particles and their gauge theories: the Standard Model and, perhaps,
superstring theory as a Theory of Everything (secs. 3.6 and 6.6).
It can safely be said that most research activity in the natural sci-
ences is concerned with reductionist work. Thus reductionism has been
immensely successful as a research program; cf. (Popper 1982) and the
recent books (Cohen and Stewart 1994) and (Gell–Mann 1994).
Is reductionism the final philosophical answer to all problems of
natural science? What we have just tried to show is that the answer is
a clear “No”; Popper (1982, pp. 131–132) confirms this. A reductionist
analysis may lose basic information, cf. “equation” (4.19).
Equally important conceptually are complexity theory, holism, and
constructionism. Thus we have three antitheses of complementary con-
cepts:
simplicity and complexity: (Cohen and Stewart 1994),
(Gell–Mann 1994),
holism and reductionism: (Hofstadter 1979, pp. 310–336),
reductionism and constructionism: discussed above.
A research program much smaller than reductionism but of increas-
ing importance is complexity theory, cf. sec. 4.2. Reductionism goes
“down” from the complex to the simple, whereas complexity theory
goes “up”, finding perhaps a new higher simplicity emerging from com-
plexity, “order out of chaos”, thermodynamics emerging from statistical
mechanics, the DNA of life emerging from organic chemistry by Eigen’s
“life machine” (Fig. 4.10 on p. 179).
Holism is less well regarded by science, which is analytical rather
than synthetic, and vitalism is completely out of fashion at the present
4.5. COMPLEXITY AND REDUCTIONISM 193

time. (The famous biochemist and Nobel laureate A. Szent–Györgyi

once wrote: “When a molecular biologist calls you a vitalist it is worse
than when an FBI man calls you a Communist” (quoted after W.M.
Elsasser’s autobiography).)

Quine’s model
Some positivists may tend to consider logic as “true in all possible
worlds” (Leibniz), mathematics is reduced to logic, physics is expressed
in mathematical laws and hence reduced to mathematics, chemistry is
reduced to physics, and biology is reduced to physics and chemistry,
and hence to physics. This throughgoing reductionism” has been seen
to be endangered by logical paradoxes, Gödel’s theorem, Heisenberg’s
uncertainty relations, and by inaccuracies of both measurements and
theories in general, not to forget self–reference (Fig. 4.13).
Quine (1961, p. 42) has given a somewhat different and probably
more realistic picture (Fig. 4.14).

Figure 4.14: Quine’s model

Empirical observations may show that existing theories may require

modification. Then it is advisable to start changes at the periphery, in
biology. If necessary, then chemical laws may be changed, then physical
laws. If still necessary, however, even mathematics and, as a last resort,
logic are not immune against change.
194 CHAPTER 4. SYSTEMS, INFORMATION, EVOLUTION

Thus, according to Quine, even logic and mathematics are not given
a priori once and for all, as gifts from heaven, perfect and changeless.
In fact, Weizsäcker (1985, p. 313–319) has worked on a “quantum
logic” in order to improve our understanding of quantum theory (cf.
end of sec. 2.6), and also Weyl’s opinion (the sentence following the
quotation at the end of sec. 2.3) fits well into this scheme:
A truly realistic mathematics should be conceived, in line with physics, as a
branch of the theoretical construction of the one real world, and we should
adopt the same sober and cautious attitude toward hypothetic extensions of
its foundations as is exhibited by physics.
 Part C
Philosophy
Chapter 5
Philosophy for scientists

5.1 Realism, idealism and dualism

Words, words, mere words, . . .
William Shakespeare, Troilus and Cressida

Now we come to the difficult and treacherous area of philosophical

“isms”. These terms are by no means clearly and uniquely defined.
They are frequently loaded with emotions and used, not only for ob-
jective classification, but also to fortify one’s own position and to de-
preciate the other’s opinion. Especially the words “materialism” and
“idealism” are used for such purposes. The argument partly repeats
sec. 1.2 in a somewhat broader context.

Realism and idealism

Naive realism. We see the world as it is. If we see a house over
there, we can be sure that it is there. If we see a car approaching us
at high speed, we better give way in order not to be hit. This naive
realism is so obvious, so appropriate to our daily life, that it appears
unnatural to doubt it, and it also leads to science: physics, chemistry,
and biology.
Now, however, we get a “dialectic reversal ” (sec. 2.5). Science tells
us that we do not see the house as such, but the walls of the house
reflect electromagnetic waves (light coming from the Sun), and these
waves, after having hit our eye, produce an image on our retina which
is processed by our brain (sec. 1.3). So what comes from the external
world, is only an image on our retina, which is a mental phenomenon.

197
198 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

Hence what is primary is our mind , and we do not see the house, we
see some mental image, which could also be produced in another way,
by a motion picture or by a television program. Thus this leads to
Idealism. The primary perceptions are affections of the mind, sense
data, from which by thinking we reconstruct an external world which
may not even exist in the way it appears.
Now, however, we ask: where do these sense data come from? Unless
we suffer from an illusion, sense data must come from the external world
(of which also cinemas and television sets are part).
After naive realism as thesis and idealism as antithesis, we thus
arrive at a synthesis:
Critical realism, or scientific realism. Through our senses we get
information on the external world, which is certainly partial and im-
perfect (we do not see ultraviolet and infrared light, for instance) but
is essentially true. If this information is not true (illusions, cinema) we
at least understand why it is not true: science (physics, evolutionary
theory of knowledge, psychology, and psychiatry) tells us about this.
This is the scientific world picture, which essentially gives the same
results as naive realism, but in a refined way. It is maintained by
practically all scientists and philosophers, even by those who do not
admit it.
Two remarks are in order, however. This dialectic game could be
continued, getting a refined idealism, an even more sophisticated re-
alism, etc. (This is beautifully shown in Fichte’s Wissenschaftslehre
of 1904, cf. sec. 5.3.) We could stop at any stage of realism or ide-
alism, and many philosophers did stop at an idealistic position. We
shall be satisfied, however, with scientific realism, at least as a working
hypothesis.
The second remark is the following. Neither realism nor idealism,
at any stage, can be proved or refuted. If we take an extreme idealistic
position, we arrive at
Solipsism. Only my own ideas are real, my sense perceptions are
nothing but illusions, I live exactly like in a dream. (“Solus ipse ”
means: only I exist, all the rest is illusion.)
It is impossible to refute this position. (When I heard this for the
first time at the age of 16 years from our excellent philosophy teacher,
I wanted to test it. Waking up one morning, I decided that “in reality”
I continued to sleep and only dreamed that I had woken up. I per-
sisted in this attitude, behaving normally but considering everything a
dream: going to school, studying, taking examinations, playing etc. It
worked perfectly and was absolutely self–consistent. After a few weeks,
5.1. REALISM, IDEALISM AND DUALISM 199

however, I decided that I had done my solipsist homework and that I

had grown tired of it, and continued to live normally.)
As I said, it is difficult to refute the solipsist position, but it is still
more difficult to maintain it. Bertrand Russell gave a nice example: at
a philosophical congress, a solipsist philosopher criticized a colleague,
asking him why he was not also a solipsist. (This, of course, is self–
contradictory, if not in theory, then certainly in practice.)
Another self–refuting attitude is universal
Skepticism. The skepticist doubts of everything. If his skepticism is
true, then the proposition “My skepticism is true” cannot be subject to
doubt, making impossible the skepticist’s intention to doubt everything.
This “dialectic reversal” is of course dangerously close to the paradox
of the liar. For a comprehensive discussion of this and related matters
cf. (Stegmüller 1969).

Materialism and science

For those who want to eliminate the last trace of thinking or mind
from the world, realism becomes materialism. Only matter is real,
mental activity is only an illusion. (Here I do not include dialectic
materialism, which will be discussed below.)
The problem is only to define what matter is. An obvious answer
is obtained by kicking a stone: a solid object.
Modern physics gives a more sophisticated answer. As we have seen
in sec. 3.6, a hydrogen atom consists of a nucleus (proton) of about
10−15 m, surrounded by the orbit of an electron which has a distance of
about 10−10 m. To understand this proportion, assume that the nucleus
has the size of the Earth. Then the distance to the electron is greater
than the distance from the Earth to the Sun! So matter consists almost
exclusively of empty space . . . If you kick a stone, you are kicking empty
space or rather, an electromagnetic field: only the strong atomic and
molecular forces provide the appearance of solidity.
It was shown only by modern physics how difficult a definition of
the concept of “matter” is. It is furthermore – explicitly or implic-
itly – dependent on one’s philosophical views: “Tell me what is your
philosophy, and I will tell you how you must define matter” (Weizsäcker
1985, p. 165). The only generally accepted definition is very abstract:
“Matter is what satisfies the laws of physics” (Weizsäcker 1977, p. 586).
Let us illustrate this by classical physics. Newton’s law of motion

d2 x
m =F , (5.1)
dt2
200 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

where m denotes the mass (the “amount of matter”), F represents

the force, and d2 x/dt2 (second derivative of the position vector x with
respect to time) is the acceleration.
Now we could argue: this is a mathematical equation connecting
mathematical symbols, and mathematics belongs to some “ideal world”.
So physics must be interpreted in the sense of idealism. This argument
has been advanced, especially with the much more difficult and abstract
equations of relativity and quantum theory.
It is easy to give a counterexample: 3 apples + 2 apples = 5 apples,
and it does not follow that apples are “unreal” mathematical concepts.
Mathematics does apply to the real world, so it does not permit us to
use it for deciding between realism and idealism.
Nevertheless we have seen in secs. 3.4 and 3.5 that matter is not at
all a simple concept. It is related to space–time curvature or satisfies
Schrödinger’s equation in “infinitely–dimensional” Hilbert space. Thus
it has become very abstract. If we adhere to a scientific realism (as
the majority of physicists do, including the present author), the stan-
dard “Copenhagen interpretation” somehow introduces the mind of the
observer, and hence an element of idealism. (This is not so terribly sur-
prising if we remember that, at the beginning of the present section, we
have obtained “scientific realism” as a higher–level synthesis of “naive
realism” and idealism.)

Figure 5.1: A particle as a wave packet

Anyway, matter has become very abstract indeed. Quantum me-

chanics regards a “particle” as some kind of “wave packet” (Fig. 5.1).
Such a wave packet is not an exact point (where the “particle” is sup-
posed to be), it has a certain finite though very small size. The impor-
tant thing, however, is that the wave function is concentrated at the
particle, but it has a non–vanishing, though very small, amplitude at all
5.1. REALISM, IDEALISM AND DUALISM 201

points of space. So to speak, a particle is everywhere! The Newtonian

belief in strictly local material points, or even strictly localized extended
material objects, is called by Whitehead (1925, p. 72) the “fallacy of
simple location ”. Whitehead arrived at this conclusion by other argu-
ments, some philosophical. Another physical argument against simple
location comes from general relativity: mass points are singularities of
curvature of space–time, but change the curvature also at other points.
In sec. 3.4 we have seen that eq. (3.55) on p. 96 provides a logically
particularly beautiful, though practically hardly useful, definition of
matter in terms of geometry.
The mathematical structure of physics is the same, whether we ad-
here to materialism or idealism. Thus, from a higher logical point of
view, the distinction between materialism and idealism may mainly be
one of words or of language, highly colored by emotions (if I hurt my
foot by kicking a stone, I may not be very satisfied by the considera-
tion that anyway, only “one mathematical equation has kicked another
mathematical equation” . . . ). Bertrand Russell calls the philosophi-
cal basis of physics a “neutral monism”, being neither materialism nor
idealism (or both).

Monism and dualism

Besides realism (or materialism) and idealism, we have another im-
portant pair of more or less opposed “isms”: monism and dualism.
Dualism, in its clearest and sharpest form, goes back to René
Descartes (1596–1650). He considers matter and mind as two to-
tally different and almost completely unrelated “primitive concepts”
(philosophers give them the lofty name of substances ). So to speak,
this is the Christian or rather Platonic idea of “body” and “soul” in
its extreme philosophical abstraction. By radically separating matter
and mind, Descartes cut himself off from any relation between these
two “substances”. In this way he artificially created a problem whose
solution has kept busy generations of philosophers. It is not surprising
if an artificial problem invented by a genius has given rise to even more
artificial, though ingenious solutions.
The greatest problem was the essential difficulty of interaction
between the two “substances” (“noninteraction” or “independence”
was considered a main property of a philosophically respectable “sub-
stance”). One of the most curious solutions was the following. Compare
matter and mind to two parallelly running clocks. The easiest way to
keep them synchronized is to connect them in some suitable manner.
This solution was excluded by the supposed impossibility of interaction
202 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

of the two “substances”. Thus there are two other possibilities: either
God has created both clocks so perfectly that they always remain abso-
lutely synchronized, or God is constantly watching them to keep them
continuously synchronized, immediately correcting any deviation. This
is a fine example of where “sharp” or “exact” philosophical reasoning
can lead to. (God as the “Universal Metaphysical Problem Solver”, cf.
sec. 1.2.)
The old philosophical fallacy was the belief that one has to start
from “clear”, “true”, “evident” and “precise” principles (“premises”)
and then follow sharp logical reasoning wherever it would lead you.
Only a slight error in the premises might lead to large errors in the
result (“ill–posed problem”, see below). Whitehead (1925, p. 75) calls
this unjustified sharpening of concepts (“noninteracting substances”!)
the “fallacy of misplaced concreteness ”. Ivan Supek (quoted in the
introduction) is more direct:
The old metaphysicians got caught in the trap of some absolutized words
and concepts, categories or principles; and wishing to construct a consistent
[philosophical] system, they locked themselves into a lifeless ivory tower.

Figure 5.2: Classical philosophy and the axiomatic method

Relation to the axiomatic method. This is formally similar to but

in reality quite different from the modern axiomatic method (sec. 2.2).
Here one also starts from premises (axioms) which, however, are not
necessarily considered “true” and “evident” but assumed tentatively.
Then, again, a logical edifice (theory) is built by rigorous derivation
from the axioms. Then, however, it is tested whether the theory fits
experience (horizontal connections in Fig. 5.2(b). If it does not fit,
other axioms are tried, etc.
5.1. REALISM, IDEALISM AND DUALISM 203

Thus the basic difference between these two approaches is

(a) Premises are considered “absolutely true” and all logical conse-
quences (however absurd) are accepted.
(b) Premises (axioms) are accepted preliminarily and the logical con-
sequences are tested by comparison with reality.
It is clear that method (b) can be much better fitted to reality than
method (a), which, already from the picture (Fig. 5.2(a)), appears much
less “stable”.
In the language of sec. 3.8, approach (a) may be compared to an ill–
posed (unstable) problem, whereas (b) may be regarded as a well–posed
and hence stable problem.
After this digression let us now return to dualism and monism.
Modern dualism or interactionism. This point is well argued and ex-
pressed in (Popper and Eccles 1977). There is no reason why mind and
matter (our body) should not directly interact. A bad tooth (material)
may cause pain (mental). This causal connection is almost trivial. But
mind can also act on matter: a toothache can cause me very quickly to
see a dentist, which otherwise I should have done only half a year later.
Considering mind as a higher level than matter, this is an example of
downward causation (sec. 4.1).
Monism. The doctrine that there is only one substance goes back
to Parmenides (5th century B.C.), was represented by the Neoplatonist
Plotinus (204–270) and revived by Baruch Spinoza (1632–1677). Albert
Einstein is known to have been a great admirer of Spinoza.
Spinoza taught that there is only one substance (Greek: monos =
one): deus sive natura, God or (equivalently) nature. This doctrine is
also called pantheism (all is God).
This one substance has two “sides” or “attributes”: matter and
mind (and many others since God must be infinite in every respect).
This view solves the problem of interaction of mind and matter, be-
cause, substantially, both are the same.
Spinoza’s view, as those of Parmenides and Plotinus, was typically
holistic: all essentially is one.
Russell’s neutral monism. The general attitude of Russell is logical
atomism, the very opposite of holism. Nevertheless to him, the (essen-
tially atomistic in the sense of discrete) sense data are primarily neither
mental nor material: They are “monadic structures” (see sec. 5.3) and,
depending on the arrangement one gets subjects or “minds” (formed
of all data collected by a person) and “material” objects (formed by
204 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

all data that are coming from a given object). This definition, seem-
ingly circular, can be made logically acceptable, breaking the “vicious
circle”, in much the same way as in Russell’s definition of natural num-
bers (positive integers, cf. sec. 2.1).
Materialism and idealism both belong to the category of monism.
Therefore both Hegel (a dialectical idealist) and the dialectic materi-
alists thought very highly of Spinoza’s monism, which could easily be
considered a forerunner of materialism (in deus sive natura, consider
“God” just an example of extravagant terminology, retain nature and
call it matter!). On the other hand, explicit idealism and especially
dualism are much less acceptable to dialectic materialism, because the
supposedly Christian notion of “immortal soul” may be hidden below.
Panpsychism is another modern form of monism in the sense of
Spinoza. Mind is simply the inside, and matter the outside of every-
thing. This doctrine has been more or less explicitly favored by so
different philosophers as Whitehead, Teilhard de Chardin, and also in
a sense by Russell as we have seen. It appears very plausible if regarded
from the point of view of human perception and thinking: mind “looks”
from the inside at the outside material world. It seems to run into dif-
ficulties at the atomic level: does every atom (in a stone, for instance)
also have an “inside” of a mental character? Leibniz (1646–1716) af-
firms just this and calls these “mental atoms” monads (cf. sec. 5.3).
A similar but more modern version of such a theory was held by the
early Russell and greatly elaborated by Whitehead. To be sure, it looks
rather extravagant, but is fascinating and may contain an important el-
ement of truth (we already know that Whitehead once said: “It is more
important that a proposition be interesting than that it be true”). The
possibly false theory of monads is indeed more interesting than the
certainly true relation “1+2=3” . . .
Panlogism (Greek: all is logic) is a label frequently put on Hegel
because his monumental three–volume work “The Science of Logic ”
contains his basic philosophy. He tried to derive everything by means
of his dialectic logic, which is an idealistic approach.
A quite different form of “panlogism” was proposed by the great
physicist John Archibald Wheeler (Misner et al. 1973, pp. 1211–1212).
The basic “building blocks” of the universe, even more fundamental
than elementary particles, are logical propositions p, q, r, . . . of the form
defined in sec. 2.1 (“Logic of propositions”)! This is the most radical
form of panlogism, much more extreme even than Hegel’s dialectic.
5.1. REALISM, IDEALISM AND DUALISM 205

Later on, Wheeler modified this view by retaining only the “truth
values” (1 or 0) of propositional logic (sec. 2.1). In informatics, each
alternative, 1 or 0, represents a bit (bi nary unit ). So the world consists
of “bits”, or according to Wheeler (1994, p. 296): It from Bit. Remark-
ably enough, Wheeler’s bits are much the same as Weizsäcker’s (1985,
p. 392) Ure (Uralternativen, basic alternatives). The same idea seems
to have been found independently. (I am sure that both Weizsäcker
and Wheeler are much too great personalities as to start a quarrel of
priority.)
Dialectic materialism. As we have already mentioned in sec. 2.5,
Marx, Engels and Lenin applied dialectics to nature itself, considered
to consist of matter only. The dialectic contradictions lie in nature
(matter) itself , they are in our thinking only because we think about
nature. Mind is an emergent property of matter, it does not exist
independently of matter. Engels’ well–known example of a dialectic
process in nature may appear somewhat farfetched: a wheat grain falls
to the ground and ceases to exist as such (negation ), giving rise to a new
plant which, on dying (negation of negation ) gives rise to the grain “at a
higher level”, i.e., greatly multiplied (cf. also New Testament, St. John
12:24).
Hegel’s dialectic idealism appears more refined and sophisticated,
but dialectic materialism (Engels’ dialectic of nature) may be more
concrete, “down to earth”. The difference may, at least partly, be ver-
bal rather than real: whether one calls the basic substance “spirit” or
“matter” has enormous emotional significance, but is of little actual rel-
evance if the logical structures are identical. For instance, Lenin was an
enthusiastic and highly intelligent reader of Hegel. Both Hegel and the
dialectic materialists claim the monist and “pantheist” thinker Spinoza
as their intellectual ancestor. In fact, monism (one basic substance)
seems to be what matters most.
True, Findley (1958, p. 58) speaks of “the Marxists, who try to op-
erate Hegelian machinery with quite alien and unsuitable fuel”, and the
above example of the grain of wheat seems to confirm this opinion. On
the other hand, Findley (1958) says also that Hegel “is more nearly a
dialectic materialist than most Hegelians have realized”. Again: “There
is, however, as much materialism in Hegel as in Marx, since matter is
for him certainly a stage in the ‘Idea’. (Just as there is certainly also a
strong strain of teleological idealism in the supposedly scientific mate-
rialism of Marx.)” (Findley 1958, p. 23). (The historic and economic
theories of Marxism are not a subject of the present book.)
Let us look more closely to the question whether dialectics is a
206 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

property of nature or a property of our thinking about nature. This

question has an analogy in mathematics applied to nature. Does a
“SUPERB” physical theory in the sense of Penrose (sec. 6.5), such as
the general theory of relativity or quantum mechanics, reflect a property
of nature itself, as Penrose seems to think, or is it only a law of our
thinking about nature, according to Kant and the neopositivists? I
think the answer is neither a clear “yes” nor a clear “no” (this would
be another instance of Whitehead’s “fallacy of misplaced concreteness”)
but seems to be somewhere between these extremes. My own very
personal view is that a SUPERB theory of physics seems to be closer
to being a property of nature, whereas a principle of dialectics seems
to be more related to our thinking. At any rate, one can very well be a
“dialectic dualist”! Remember also Spinoza’s “Ordo et connexio rerum
idem est ac ordo et connexio idearum ” (sec. 6.5), which emphasizes the
supposedly identical logical structure of our mind and of nature.
According to classical philosophy, logical structure is not all that
matters. Also important is ontological status. Ontology (sec. 1.2) stud-
ies the kind of being or existence in philosophy. In this sense, materi-
alism and idealism, monism and dualism are indeed different.
Pluralism. It is doubtful whether, if we decide to have more than
one substance, the number 2 (dualism) is sufficient. Already Descartes
and his followers had an additional fundamental substance: God.
In science, we have at least three levels: inorganic matter, life, and
mind. Whether these are different “substances” or “emergent proper-
ties of matter” (as dialectic materialism would have), appears to me a
matter of terminology rather than a fundamental distinction. Of es-
sential importance is only the interaction between the various levels by
upward or downward causation.
So, as a working hypothesis, I suggest that we might accept a hy-
pothetical pluralism, without excluding a reduction to a “pluralistic
monism” if it should become appropriate.
It is clear that all these “isms” are rather vague and by no means
do justice to the many–faceted work of the great philosophers. As we
have just mentioned, Hegel was in many respects a dialectic materialist
rather than an idealist or “panlogist”, and to dispose of Whitehead
simply as a panpsychist is about as enlightening as regarding Richard
Strauss as a late romantic composer.
The concept of a set of possible solutions considered in the context of
inverse problems (see end of sec. 3.8) may also be applied to philosophy:
rather than trying to find a unique philosophical system, let us look for
a set of possible philosophical theories that are all compatible with
5.2. THE THREE–WORLD MODEL 207

the given data. In this sense we may tentatively define: Philosophy is

the set of all possible (true, interesting) theories (models, perspectives)
regarding the universe.

5.2 The three–world model of Popper and

Eccles
In verity, an independent world
Created out of pure Intelligence.
William Wordsworth, The Prelude

This model was explicitly introduced by Popper and Eccles (1977)

and has become rapidly popular, because it is a useful “reference model”
also for those who disagree with it.
World 1 consists of all physical processes, it is the “universe of phys-
ical entities”, making up our usual physical, chemical, and biological
world, the “objective” world as explored and described by the natural
sciences.
World 2 consists of our mental processes, sense perceptions,
thoughts, ideas, emotions, pains and joys, in short, it is the world of
our subjective experience.
World 3 consists of the “objective” results of man’s scientific, liter-
ary, artistic, musical etc. activity. It encompasses, for instance, math-
ematics and logic, the theories of physics, chemistry, and biology (as
found in the pertinent textbooks), Shakespeare’s dramas, Goethe’s po-
ems, Beethoven’s symphonies, computer programs, and, of course, all
languages ever spoken by humanity. We may say, it is everything “spir-
itual” that transcends the mental life of a human individual. The dif-
ference between World 2 and World 3 is that between subjectivity and
objectivity (or, if you prefer, intersubjectivity). It is basically identi-
cal to Plato’s world of ideas, and is subject to the criticism which has
been leveled against Platonism, first by Plato himself (in his dialogue
“Parmenides ”) and then almost by every philosopher from Aristotle to
the present time. Nevertheless, it has been indispensable at least as a
level of reference. Alfred North Whitehead has characterized Western
philosophy as “a set of footnotes to the work of Plato” (cf. sec. 5.4).
It is easy to criticize the terminology. World 2 and World 3 are not
worlds in any geographical or astronomical sense, worlds to which you
can travel on cheap tourist–class tickets.
208 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

There is, however, related terminology in general use. The term

“universe of discourse” may be considered a subset of World 3, and the
“realm of fancy” belongs partly to World 2 (subjective daydreaming)
and partly to World 3 (fairy tales known to many people). The “World
of art and letters” is also a subset of World 3.
This “world” terminology is very catchy and therefore useful, even
though (or because) it is rather controversial.
Least controversial, of course, is World 1, the world of everyday life
and of science. Only solipsists and some extreme idealists will deny it.
Materialists may deny the reality of World 2: mental processes, they
say, are nothing but neural operations in our brain. This may be so, but
nevertheless terms such as “tooth–ache”, “laugh”, “anger”, or “noise”
exist, and, in good Russellian manner, define World 2 as the set of all
these terms. (Anyway, the bad headache I get when writing these lines,
is for me a somewhat painful proof of the existence of World 2.)

World 3
Most controversial is the “reality” of World 3. Let us think of
mathematics, considered as a prototype of this world already by Plato.
Do mathematicians “invent” or “discover” their theorems? Most
non–mathematicians would speak of “inventing”, most mathematicians
speak of discovery. Only a year ago (June 1993), there was a math-
ematical sensation: a proof of Fermat’s Last Theorem (xn + y n = z n
is unsolvable for integers x, y, z, n, except for n = 2, cf. sec. 2.3) was
finally found, after Pierre de Fermat had formulated this theorem in
1637 and claimed to have found a proof but not given it. Since then,
generations of mathematicians have tried in vain to find such a proof.
(I just — April 1994 — learned that even the recent “proof” contains
a serious flaw.) Thus, the general feeling of mathematicians is that the
theorem is “already out there”, just waiting for a proof (or disproof).
Mathematicians are constantly busy to prove the “existence” of a solu-
tion to a complicated equation. Thus, in some way, the mathematical
objects exist, just waiting for being discovered. Primarily, π is not “in
the sky”, but in the strange but obviously real mathematical universe
(Barrow 1992).
All great mathematicians have believed in the reality of this mathe-
matical universe, from Pythagoras to Penrose (1989). Bertrand Russell,
who in his philosophy stressed empiricism, was a Platonist in logic and
mathematics. Gödel emphasized the need of Platonism in logic and
philosophy.
5.2. THE THREE–WORLD MODEL 209

Of all contemporary sciences, mathematical physics is outstand-

ing for the imaginative richness of its structures, combining exotic
strangeness and intricate complexity with elegance and conceptual sim-
plicity. This is almost entirely due to the use of elaborate and very
abstract mathematical structures. It is difficult, if not impossible, to
a “real” mathematician to deny that the strangely beautiful world of
mathematics does not exist “somewhere out there”, and to accept that
this is merely the effect of the more or less random firings of neurons
in our brains.
Obviously, prime numbers (2, 3, 5, 7, 11, . . . ) exist in some sense
objectively, otherwise we would not have Bernhard Riemann’s theo-
rem on the distribution of prime numbers. This theorem, to me the
most beautiful theorem of all mathematics, expresses the distribution
of prime numbers (which are real integers ), by means of an integral
formula involving a certain analytic function of a complex variable,
Riemann’s zeta function. Of this function, we have Riemann’s conjec-
ture that all of its zeros lie on a certain vertical line in the complex
plane, but this has not yet been proved. In a way, the proof of this
conjecture would be much more important than the proof of Fermat’s
theorem mentioned above. The famous mathematician David Hilbert
was asked what his first question would be if he fell asleep and woke
up after 200 years. His question is reported to have been: “Has the
Riemann conjecture been proved?”. Where would this zeta function
exist if not in World 3?
To me it is particularly impressive that Karl Popper, otherwise a
sober empiricist and by no means an admirer of Plato, forcefully and
(to me) convincingly argued for the independent existence of World 3:
mathematics is more than the (material) set of all books on mathemat-
ics, Beethoven’s 5th symphony is more than the set of its printed scores
and innumerable performances and recordings up to now.
It is clear that “existence” has for World 3 a completely different
meaning than for World 1 or also World 2. In a dualistic philosophy,
World 1 and World 2 together constitute our “real” world in the every-
day sense, whereas World 3 “exists” in a completely different way.
“Existence” is a highly ambiguous and problematic word anyway,
but this word “exists” : we may look it up in any dictionary.
So even if you do not like the “existence” of World 3, consider it
simply as a name for the set of all mathematical, logical, and other
abstract terms, such as all terms contained in the usual dictionaries of
English and of other languages.
Since ontology studies the various kinds of being or existence, the
210 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

three worlds differ mainly in their ontological status; see also secs. 1.2
and 5.1.
At any rate, the Three–World terminology is very useful. As we
have already said above, materialists may consider World 2 a subset of
World 1; Spinozists may regard World 1, on the one hand, and World 2
+ World 3 together, on the other hand, as the two “attributes” of the
one divine substance; Eastern philosophers together with Schrödinger
(1958) who hold that there is only one universal mind of which we
all participate, will put World 2 ≡ World 3, Russell, in his “neutral
monism”, may put World 1 ≡ World 2, etc.
To the medieval philosophers, the world of ideas, our World 3, was
the mind of God ; this terminology is alluded to in the title of the book
(Davies 1992). Hegel calls World 3 the “objective spirit” (not a bad
terminology, after all).
Less bombastically, we may say that World 1 is governed by “fuzzy”
logic and “fuzzy” mathematics (sec. 2.4), whereas World 3 is the realm
of exact logic and mathematics. This point merits further elaboration,
which will be done in the next subsection.
R. Bošković (1711–1787) gave an ingenious interpretation of
World 3 as a mathematical space which is governed by potentiality in
the sense of Aristotle (sec. 5.4). World 3 is the seat of the human mind,
whereas matter and the human body belong to World 1 (World 2 may
perhaps be considered a subset of World 3). The interaction between
World 3 and World 1 explains why nature is governed by mathematical
laws. Not all mathematical structures are realized in nature (not ev-
erything that is potential , i.e., possible, becomes actual , i.e., real), but
theoretical physics can make free use of the inexhaustible treasure of
mathematics that is contained in World 3. Furthermore, if the human
mind is located in World 3, the direct access we have to mathematics
according to Penrose (see below) is explained.
Another application of these ideas is the interpretation of quantum
mechanics in terms of potentiality and actuality (sec. 3.5). An excel-
lent book about Bošković is Ivan Supek: “Rudjer Bošković ”, Croatian
Academy of Sciences and Fine Arts, Zagreb, 1989. More about Bošković
will be found in sec. 6.6.
5.2. THE THREE–WORLD MODEL 211

How is exact thinking possible?

Consider mathematical reasoning. Logical and mathematical think-
ing are proverbially rigorous. How can our brain perform exact think-
ing?
To see the problem, take any mathematical theorem about a circle,
e.g., its definition: the circle is the geometrical locus of all points whose
distance from a given point is constant; in other terms, the circle is a
curve of constant radius.
From the times of Euclid, millions of schoolchildren have learned
this and some of them even understood the definition and could work
with it.
Now comes the paradox: nobody, not even the greatest mathemati-
cian, has ever seen or drawn a mathematical circle. Nobody (I really
mean nobody ) has ever seen or marked a point, and I dare say that
probably nobody will ever by able to do so.
What is the reason? Logical, mathematical, and other axiomatic
systems (secs. 2.1 to 2.3) are rigorous, that is, absolutely accurate, at
least in principle. For instance, 2 + 1 = 3 and not 2.993. Logical and
mathematical objects belong to World 3. The fact that a mathemati-
cian, whose mind belongs to World 2, is able to perform a rigorous
logical deduction or find a rigorous mathematical proof which is rec-
ognized as such also by his fellow mathematicians, is very remarkable
indeed. Mathematicians have discovered all properties of and theorems
about a circle, without ever having been able to construct one on paper.
But what about the circles constantly used in illustrations in books
on geometry etc.? They are not exact circles, as one easily sees by
looking at them with a magnifying glass or under a microscope. At
best, they are “fuzzy” realizations of exact, or “real”, circles!
Some mathematicians write books full of geometric theorems and
proofs, which do not contain a single figure. All theorems must be
derivable from the axioms by logical deduction only. It is true that most
such books do contain figures, but only as an aid to better visualize the
geometric situation.
Thus logicians, mathematicians etc. appear to be capable of exact
thinking, of dealing with World 3 objects directly. Thus there seems to
be an intimate relation between World 3 and World 2. In a way, exact
circles, being objects of World 3, can be transferred directly to World 2 .
Now comes the surprise. Circles cannot be transferred directly to
World 1 ! Realizations in World 1 of abstract World 3 objects such
as points, straight lines, or circles are always approximate only! (This
212 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

holds at least for continuous objects; one might argue that integers
occur in World 1 more directly: a basket containing five apples may be
considered an “exact” realization of the integer “5”; cf. sec. 4.4.)
Thus we have the following scheme of objects:
in World 3: exact,
in World 2: exact (at least in principle),
in World 1: fuzzy (at least in general).
This seems to be a clear indication that World 1 and World 2 are
essentially different.
Analogy with computers. Now it is well known that by computers
(which belong to World 1), also exact logical and mathematical opera-
tions can be performed. The underlying computer programs (software),
however, belong to World 3. At the danger of overstretching the anal-
ogy, we may say that the program adapted so as to be able to serve as
input for a particular computer, “belongs to World 2 of that computer”.
Below we shall see that the computer analogy cannot be perfect.
Still we may say that World 1 and World 2 are essentially different, at
least as much as hardware and software in machine computation.
Mind vs. computer . It is claimed, however, by the most eminent
mathematicians that, in a way, they have direct access to World 3
which goes beyond formal (algorithmic) computation or logical deduc-
tion. This is most strongly and convincingly argued in (Penrose 1989,
pp. 416–423). From p. 418 we quote
Mathematical truth is not something that we ascertain merely by use of
an algorithm. I believe, also, that our consciousness is a crucial ingredient
in our comprehension of mathematical truth. We must “see” the truth of
a mathematical argument to be convinced of its validity. This “seeing” is
the very essence of consciousness. It must be present whenever we directly
perceive mathematical truth. When we convince ourselves of the validity of
Gödel’s theorem we not only “see” it, but by so doing we reveal the very
non–algorithmic nature of the “seeing” process itself.
What Penrose alludes to here, is the fact that in Gödel’s proof
(sec. 2.3) we can “see” by informal reasoning that Gödel’s basic propo-
sition G, though algorithmically unprovable, is nevertheless true. Cf.
the quotation of Findley at the beginning of sec. 2.5.
Lucas’ proof of free will . Lucas (1970, §25) has used Gödel’s theorem
in a similar way. The essentially non–algorithmic character of human
thought as exhibited by Gödel’s theorem shows that our thinking can-
not be the activity of a deterministic “thinking machine” because such
5.2. THE THREE–WORLD MODEL 213

a machine can only work algorithmically. More about this will be found
in sec. 6.4.
Thinking vs. acting. For those who believe that thinking is simply
an activity of the brain or the nervous system, essentially similar to its
activity in directing our bodily movements, we have a surprise in stock.
In its thinking function, our brain works (or is able to work) exactly,
as we have seen. Thus, by reasoning, we can establish rigorously valid
theorems about circles.
Now let us use the brain as a “steering unit” for a bodily movement,
e.g., the actual drawing (on paper or on the blackboard) of a circle. If
we hope to be able to draw a circle with equal exactness, we shall be
badly disappointed as we have seen above. Nobody is able to trace an
exact circle, cf. Fig. 6.2 on p. 255.
The reason is, of course, the inherent “fuzziness”, the inevitable
random background in World 1, cf. secs. 4.4 and 6.5.
Now comes the essential question. If this random background affects
our brain and nervous systems with respect to bodily movements, why
does it not affect our brain equally in its thinking activity? Comparing
the brain to a parallel–processing digital computer helps, but is only a
more or less perfect analogy because of the non–algorithmic character
of human thinking and because of the fact that the digital firing of
the neurons does not yet imply that the brain is a digital computer
(sec. 1.1).
As we have remarked at the beginning of sec. 4.4, only integers can
be determined exactly, whereas continuous quantities can only be de-
termined approximately. In computer language, this is the distinction
between digital and analog computers. It is thus tempting to relate
exact thinking to some digital operation of the brain; some analog
(continuous) operations of the brain would then be responsible for the
inexact bodily movements. There may be an element of truth in this
comparison, but almost certainly matters are not that simple. At any
rate, the “exactly thinking I” may be considered as a “center of com-
mand” for logical operations, which has no direct analogue in World 1
which is always “fuzzy”.
To me, this is another strong argument for the independent char-
acter of World 2: this “center of command” seems to be identical to
Penrose’s “consciousness” (1989, pp. 409–413) and to the “self” of Pop-
per and Eccles (1977) and to be essentially non–material in order not
to be subject to random fluctuations.
214 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

5.3 Subject and object

The belief in an external world independent of
the perceiving subject is the basis of all natural science.
Albert Einstein

Philosophical problems of perception

Subject and object are basic categories of the theory of knowledge,
or epistemology. Nothing seems to be more natural and straightforward
than my (subject) seeing a tree (object) and listening to the song of a
bird (object) sitting on one of the tree’s branches.
Neurophysiology has shown us how complicated the problem of
sense–perception is (secs. 1.3 and 1.4). Also from a philosophical point
of view, however, human perception raises some fundamental problems.
(1) Action of object on subject. How can a tree which may be a
hundred meters away, cause me, the subject, to see it? The obvious
answer is that it is an empirical fact that I do see it. How this is done
is explained by science: the tree (partially) reflects the light of the sun,
which strikes my eye, is focussed by the eye’s lense and thus produces
an image on the retina. This image is then processed by the brain, as
outlined in sec. 1.3. Similarly the sound waves produced by the singing
bird hit my ear, are captured by it, analyzed harmonically etc.
(2) Possibility of illusion. Optical illusions are frequent, ordinary
mirrors even more so. How do I know that the tree really exists and
is at the exact spot where I see it? Or am I just dreaming? Or am I
constantly dreaming, and all the surrounding world exists only in my
imagination?
This is the extreme position of solipsism, which we have already
discussed in sec. 5.1. As we have seen, it cannot be refuted. However,
it cannot be proved either. From a pragmatic point of view, a consistent
solipsism is not a practical way of life. If my surroundings are not real
anyway, I may as well cross a busy street without paying attention to
the traffic. Sooner or later I shall then dream that I am in a hospital,
or my dreaming will have ceased altogether. So an honest solipsism
is not an advisable strategy for survival, and this may be why any
formerly existing solipsist animal or human being has been eliminated
by evolution (sec. 1.4).
Descartes, starting with his “Cogito, ergo sum ” (I think, therefore
I am), from a solipsist position, argued that God, in His goodness,
5.3. SUBJECT AND OBJECT 215

cannot have admitted that I am constantly deceived by illusions, so the

surrounding world must be real after all. The history of philosophy has
shown how many philosophers, after having argued themselves into an
impossible situation, have called for God as a “deus ex machina ” to
save them, cf. also sec. 1.2. Therefore we become wary of looking at
God as the Universal Metaphysical Problem Solver, and we consider
Descartes’ argument with some reserve.
Anyway, if we accept the existence of an external world at least as a
working hypothesis (scientific realism, see sec. 5.1), we are in a position
to rationally explain also dreams, optical and other illusions, abnormal
perceptions, etc. It also increases our chances of survival.
(3) Knowledge of non–knowledge. If we do not know something,
how do we know that this “something” is out there, waiting to be dis-
covered? The answer is that our knowledge increases gradually, starting
from things already known and proceeding by hopes, expectations, ex-
periences, predictions, working hypotheses, etc. When, driving a car,
we see a ball rolling across the street, we watch out for the child fol-
lowing it and step on the brake. As Whitehead has said, we should
not even see an elephant unless we expect some animal of this kind.
A person may sit in a room, complaining about the terrible noise to
which he is exposed, and may not notice that it is Beethoven’s Seventh
Symphony. The bird lover walking in a forest recognizes dozens of birds
whereas the accompanying person does not notice anything.
(4) A related problem is knowledge a priori and knowledge a pos-
teriori . According to Kant, logic (analytic) and mathematics (syn-
thetic a priori) are tools by which we explore our surroundings. Well–
established theories are also knowledge a priori. Kant regarded Eu-
clidean geometry, too, as absolutely true a priori knowledge. General
relativity, however, has taught us that this is not warranted in an abso-
lute sense, though to an extremely good approximation. So our present
theories of physics are considered excellent a priori knowledge, but al-
ways subject to replacement by a better theory or by a more appropriate
paradigm (sec. 3.10); see also Quine’s model at the end of sec. 4.5.
(5) The problem of truth. How do we know that our knowledge
corresponds to reality? One answer is: by testing it through obser-
vations and experiments (verification and falsification). Such an ex-
perimental verification can never be absolute, not only because of ex-
perimental errors (sec. 4.4). For instance, take as law our standard
example: every day, the Sun rises at a precisely predictable time. It
may fail to do so tomorrow because, during the preceding night, the
216 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

Sun may have exploded or the Earth destroyed by a giant comet. Not
even a falsification necessarily proves a law wrong: the experimental
setup may have been inappropriate or measuring errors may have pro-
vided the false impression that the theory was incorrect. This is the
problem of induction already discussed at some length in sec. 3.9.
All this is relatively trivial. A deeper question is whether reality
really “is” as it looks. For instance, is space “really” three–dimensional?
It may be five–dimensional or ten–dimensional, only we may not be
capable of perceiving more than three dimensions. Are there aspects of
reality which are inaccessible to science? Are there Kantian “things–
in–themselves” which are unknown as a matter of principle (p. 230)?
As a partial answer, let us consider the spectrum of electromagnetic
waves, with wave lengths from several kilometers down to 10−13 m. The
visible spectrum, from red down to violet, is on the order of 5 × 10−7 m,
above there are infrared and longer electromagnetic waves; below there
are ultraviolet waves, X–rays and gamma–rays. We know very well and
use also electromagnetic waves above and below the visible part of the
spectrum: they are accessible to indirect physical observation.
So it seems that we know practically everything which in principle
can be observed by present–day physical, chemical, etc. experiments.
Other things, e.g., some elementary particles in physics, have been pre-
dicted by theory but not yet been discovered (this is a nice illustration
of “knowledge of non–knowledge” mentioned above). Also higher di-
mensions, which are inaccessible to direct observation, are mathemati-
cally fully understood (sec. 2.6) and are under consideration as candi-
dates for certain “unified theories”. Infinite–dimensional spaces are, of
course, standard tools in quantum mechanics (without implying that
“ordinary” space is more than three–dimensional). (See, however, also
Bohm’s (1980) “enfolded reality” and “implicate order”, cf. sec. 3.5.)
It would be very unwise indeed to exclude that there are physical or
other phenomena which are completely different from all that we know
today and which cannot be observed by contemporary science, and for
which we do not even know where to look . As an example, think of
quantum phenomena before 1900.
What we can say is that our knowledge of the physical world is
essentially correct in corresponding to some external reality; this does
not exclude the possibility or even probability that it is incomplete.
Think of trying to catch a fly. Why is it so difficult? The fly’s
knowledge of the external world is certainly extremely rudimentary: it
probably comprises mainly the (instinctive) knowledge of how to get
food, how to escape enemies, etc. This limited knowledge, however, is
5.3. SUBJECT AND OBJECT 217

nearly perfect. Perceiving a hand (not necessarily as a hand, but as a

dangerous object) approaching rapidly, the fly escapes even much more
rapidly in exactly the right direction! This ability was certainly acquired
through evolution: only the “most intelligent” flies survived. Within its
very narrow limits, the fly’s knowledge of the external world is perfectly
“true”: it is sufficient for performing exactly the right reactions.
So one answer why, in some way, man’s knowledge is “true”, has
been given: only those human beings who have a reasonably correct
relation to their environment, have been able to survive evolution. This
biological consideration cannot be expected to provide more than a
partial answer to the complex question of “truth”, but it seems to play
a certain basic role.
One of the best philosophical treatises on the theory of knowledge
is (Hartmann 1965), see in particular Chapter 6.

Monadic structures
Fig. 5.3 schematically illustrates a “monadic structure”, consisting
of a subject A and an object B. It is probably the simplest and most
fundamental structure in epistemology, relating an object, considered
to be at a location B, to a subject A (represented by the eye of the
observer). In order to avoid Whitehead’s “fallacy of simple location”
(sec. 5.1), we consider the object with respect to a certain location B
without excluding the possibility that the object is implicitly present
also at other locations, cf. Fig. 5.1 on p. 200.

Figure 5.3: A monadic structure

The concept of monad was introduced by the great mathematician

and philosopher Gottfried Wilhelm von Leibniz (1646–1716). A monad
is essentially an unextended “pointlike” subject A mirroring the whole
universe, that is, A as the center of perspective for all objects B. Thus
the monad is considered a pointlike “subject” together with all objects
B as seen from A. Symbolically we may represent a monad perhaps
somewhat as in Fig. 5.4.
218 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

Figure 5.4: A monad A

According to Leibniz but in modern terminology, there is a monad

situated at every point of space. Human persons (souls) are particularly
well–developed monads, and God is the greatest monad of all. All
monads are “windowless”, that is, closed to each other and without
interaction with each other except with God. By some kind of “pre–
established harmony”, all monads are kept in tune with each other by
God, like perfectly synchronized clocks.
Thus, according to Leibniz, monads are non–interacting “sub-
stances”, reminding of the two non–interacting substances of Descartes:
matter and mind, which may also be kept synchronized by some pre–
established harmony (sec. 5.1).
Just as Popper and Eccles have allowed these substances to interact,
Alfred North Whitehead has worked out a theory of monads (he calls
them “actual occasions”) which, in fact, do interact.
Whereas Leibniz’ souls are more or less permanent, Whitehead’s
“actual occasions” last only for a moment, they are, so to speak, space–
time points mirroring the universe. A subject is a “world line” (sec. 3.7)
of related actual occasions.
Russell’s position of “logical atomism” was also sympathetic to
monadic structures as we have already remarked in sec. 5.1.
If we think of A as a human subject (Leibniz) or as a point of the
world–line of a human subject (Russell, Whitehead), this looks reason-
able enough. If, however, we regard, with Leibniz, all points in space
(or all material atoms) or, with Whitehead, all points in space–time (all
actual entities) as monadic subjects with some kind of primitive “soul”
(Whitehead speaks of “mental pole”), we get a doctrine already known
to us as panpsychism: everything (Greek: pan ) has a soul (Greek psy-
che ), which seems hard to accept for many people.
Let us not forget, however, that Leibniz and Whitehead belong to
the greatest mathematicians, logicians and philosophers of their time.
5.3. SUBJECT AND OBJECT 219

If they held a doctrine which apparently contradicts common sense,

they had good reasons as everyone can see on reading Whitehead’s
books.
It is true that many philosophers use to criticize and even ridicule
their colleagues much more than we are used from other disciplines,
but we outsiders should exercise great restraint in order to avoid too
cheap criticisms. In fact, modern physics tells us one lesson: many
theories may be very far from “common sense” and still true, even un-
avoidable: think of special and general relativity, quantum theory, and
modern elementary particle physics, which are all extremely abstract
and “counterintuitive”. In comparison with these theories, Whitehead’s
“actual entities” are models of simplicity and common sense.
Again, Niels Bohr’s dictum comes to our minds: “Your idea is, of
course, crazy. The problem is only whether it is sufficiently crazy to be
true.”
According to panpsychism, “mind” is simply “matter” seen from
within, cf. (Teilhard de Chardin 1955) and the article by Globus in
(Globus et al. 1976).
Even quite apart from the question of panpsychism, monadic struc-
tures are important in relativity where A corresponds to a particular
reference system, and in quantum theory where A is the observer in
the standard Copenhagen interpretation; cf. secs. 3.4 and 3.5.
For the mathematical reader, this is very clearly expressed by the
simple matrix formula, well known from quantum theory,

l = eT L e (5.2)

where, in the observed average value l, the effect of the object (physi-
cal quantity) is the matrix L, and the effect of the observer is the unit
vector e (“state vector”); eT denotes the transpose of e. (For an even
more mathematical reader we shall not conceal the fact that the “ob-
servable” L represents an infinite matrix in the Heisenberg sense or,
equivalently, a linear operator in the Schrödinger sense; the vector e is
an infinite “state vector” or a “state function”.) The measured value l
thus is the projection of the “physical quantity” L onto the reference
system e of the observer. For more details see sec. 3.5.

Priority of subject or object?

This question is somewhat like the problem: which comes first, hen
or egg?
220 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

Philosophy quite naturally starts with the subject, that is, with
idealism. Equally naturally, science starts with the object, that is with
realism or materialism.
As we have already remarked in sec. 5.1, realism is not absolutely
identical with materialism. Realism only emphasizes the priority and
existence of the external world, but does not deny that other persons
may have minds, as materialism does (as a matter of fact, materialism
also denies the reality of mind in the subject).
In sec. 5.1 we have seen that scientific realism, in the form of
the mathematical laws of physics, contains “non–material” objects of
World 3. This synthesis of “realism” and “idealism” was seen to make
“matter less material and mind less mental” (Russell).
In fact, like in the hen–egg problem, subject and object are insep-
arably connected in human knowledge, as already Plato, especially in
his dialogue Parmenides, has emphasized (e.g. Speiser 1952, p. 13): no
subject without object, no object without subject.
In science, this connection is particularly strong in quantum theory,
and very weak in sciences like paleontology. There the scientific subject
may come millions of years later: nobody will question that dinosaurs
have existed even without human observers watching them with awe
and fear, but nobody will question either that human observers are
necessary for the science of paleontology. Nature may very well exist
without human observers, but philosophy certainly not, and not even
science.
Fichte’s iteration
Matters become difficult in introspection: the mind is subject and
object at the same time: the thinking thinks the thinking. This has
already been mentioned in sec. 2.5; see the quotation of P.M. Møller
which was so dear to Niels Bohr, and the subsequent closely similar
quotation from Fichte. In fact, this does lead to an infinite regress
of “I’s” but that may not really matter: there are also many kinds
of mathematical infinities, so that World 3 (sec. 5.2) is certainly not
subject to overpopulation as our small Earth is.
Still, there are the well–known paradoxes of self–reference which we
have already discussed in secs. 2.3 and 2.5. As Sir John Kendrew put
it (Duncan and Weston–Smith 1977, p. 207):
Perhaps the most fundamental of all the difficulties encountered in biological
research is that the investigator cannot detach himself from the system under
study because he himself forms part of that system . . . in psychology the
investigator is himself one of the subjects under study.
5.3. SUBJECT AND OBJECT 221

I believe that an iterative process given by Fichte in his Wis-

senschaftslehre of 1804 provides, if not a solution, at least an orderly
approach; cf. also sec. 5.1. Denoting realism by R and idealism by I,
we have an iterative approach which may be represented as a sequence
of syntheses in Fig. 5.5 (similar to Fig. 2.9 on p. 48). The first itera-
tion has already been discussed in sec. 5.1: R0 is naive realism, I0 is
scientific idealism, R1 is critical or scientific realism.
In his Wissenschaftslehre of 1804, Fichte has given 10 of such syn-
theses, according to M. Guéroult (“L’évolution et la structure de la doc-
trine de la Science chez Fichte ”, vol. 2, pp. 136–137, Strasbourg 1930).
We may even continue this process ad infinitum, in order to arrive at
what philosophers call the Absolute. Mysticists claim to have direct
access to the Absolute by meditation, rationalists like Fichte and Hegel
(“Wissenschaft der Logik”, 1812) prefer an iterative approach. Will the
iteration of Fig. 5.5 converge? No, in the sense that it does not converge
to any ultimate “ism”, in much the same way as the sequence (2.18) on
p. 55 does not converge √ to a rational number. Just as Cantor defined
the irrational number 2 as the sequence (2.18), we define with Fichte
and Hegel the Absolute by the complete infinite sequence of syntheses
of Fig. 5.5.

Figure 5.5: Fichte’s iteration

Fichte’s and Hegel’s works are of “Rabbi type 3” (see Preface),

which means dark, difficult and fascinating. For a clear description
of this process (Rabbi Type 2) cf. (Speiser 1959, Chapter II: Fichte’s
Wissenschaftslehre von 1804). At this point I should like to express my
appreciation for the books of Andreas Speiser who treated these difficult
problems as a lucid mathematician, and my thanks to the well–known
physicist Hans–Jürgen Treder (Potsdam–Babelsberg) with whom I had
222 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

brief but wonderfully enlightening discussions on Parmenides, Plato,

Fichte, Speiser, and unified theories of physics.
Plato, in his dialogue “Parmenides ” starts from the Absolute, con-
sidered as the One, and tries to get down in the opposite direction of
Fig. 5.5, which is much more difficult to understand in spite of Speiser’s
(1959) brilliant commentary.
At this point I should like to mention Weizsäcker (1971, Chapter IV,
secs. 5 and 6) who relates Plato’s “Parmenides ” to biology and quan-
tum theory in a most instructive way.
Starting with (Schrödinger 1958), relations between quantum
physics and Eastern mystical philosophy have become quite numer-
ous (Bohm 1980; Capra 1976; Moser 1989). This may be understood
also as a protest against the dominating influence of logical positivism
(sec. 5.4), to restore the balance between “atomistic” and “holistic”
thinking.
Fichte and Hegel provide a rational approach which is accessible
also to those who (like the present author) have not found a direct
access to mystical meditation.
As an ultimate synthesis, the differences between realism and ide-
alism vanish in the Absolute. Subject and object become One, which
Buddhists and other mystics try to achieve by meditation. An incom-
parable description in rational, though very difficult, language is found
in the final section of Hegel’s Logic, “The Absolute Idea”.
To put matters straight, let us mention that Fichte’s “I” is not his
personal individual “empirical” I, but the counterpart of the “I” in
World 3. It is related to the empirical “I” in precisely the same way as
an empirical “circle” drawn with a pair of compasses on the blackboard
is related to the mathematical concept of the circle (a World 3 object).
Fichte’s ideal “I” is essentially equivalent to Kant’s “transcendental
subject”, which also belongs to World 3.

Logical singularities
Fichte’s “Absolute” is clearly a logical singularity, as Fig. 5.5 shows.
This is a general situation, which is very well described by the quotation
from John Kendrew given above.
Let us recall some examples of such a subject–object interference:

(1) Mind : Thinking about thinking (Plotinus, Fichte, Hegel; just

discussed).
(2) Mind : Paradox of the liar, Gödel’s theorem (sec. 2.3).
5.3. SUBJECT AND OBJECT 223

(3) Mind : Paradox of introspection (introspection may disturb

the mental phenomenon under investigation: “blue elephant”,
sec. 2.5).

(4) Observer ↔ nature: observation may disturb the quantum phe-

nomenon under investigation (Heisenberg’s uncertainty relation,
sec. 3.5).

(5) Observer ↔ life (Bohr’s paradoxon): a detailed physical investi-

gation of a living organism may destroy it (sec. 4.5).
(6) Observation of human test persons: the behavior of a test person
may be changed by the fact that the person is aware of being
tested. This is a psychological or medical (placebo effect!) phe-
nomenon not unlike Heisenberg’s uncertainty relation (see Exam-
ple 4): observation disturbs the event under investigation. Intro-
spection (Example 3) is a special case.

Figure 5.6: Singularity O → S

A rather general principle is represented in Fig. 5.6. If the object

O approaches the subject S, so as to finally coincide with it, there is
a singularity. A conceptual gap has to be bridged, which is almost
impossible. Here it is appropriate to continue the quotation from John
Kendrew (Duncan and Weston–Smith 1977, p. 207):
At a more fundamental level the problem of the relationship between mind
and matter and of the nature of consciousness seems impossible of solution
because the investigator himself is conscious mind and there is a complete
conceptual gap between that mind and the physical objects in which it re-
sides.
Already Kant recognized that it is impossible to regard the universe
as an object to be studied, because the observing subject is necessarily
224 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

part of the universe. The way in which Fichte and Hegel approach the
problem may, as a variation of Fig. 5.5, also be described as a spiral
(Fig. 5.7), corresponding to looking at the singularity S in Fig. 5.6 from
above. Fig. 5.7 clearly corresponds to the dialectic spiral of Figures 2.10
or better 2.12 on p. 53, and hence to Weizsäcker’s “Kreisgang”.

Figure 5.7: The singularity is approached by a spiral

According to Kant, considering the infinite universe as an object

leads to antinomies (or paradoxes) regarding the beginning of the uni-
verse (t = 0 or τ → −∞, cf. sec. 3.7), the divisibility of matter (dis-
creteness vs. continuity), freedom and necessity, and the creation of the
world.

Figure 5.8: Wheeler’s self–reference universe

According to Wheeler (in Duncan and Weston–Smith 1977, pp. 31–

32), we have a “self–reference universe” (Fig. 5.8). The evolution of the
universe has finally given rise to man, whose observation and activity
give meaning to the universe. We thus have a “self–exciting circuit”
(the name comes from electromagnetic dynamo theory!):
5.4. HISTORICAL LANDMARKS 225

So to speak, the observer and the universe “bootstrap” each other

into existence. (In this light, Wheeler’s admiration (p. 180) for Eigen’s
“life–machine” (Fig. 4.10 on p. 179) is not surprising since Eigen’s “hy-
percycle” is also a catalytically “self–exciting circuit”.)
According to Bohr (1934, last sentence): (Let us not forget) “the
old truth that we are both onlookers and actors in the great drama of
existence”. A similar idea has been expressed by Maturana and Varela
(1987).

5.4 Historical landmarks

The safest characterization of the European philosophical
tradition is that it consists of a series of footnotes to Plato.
Alfred North Whitehead

Philosophy cannot be understood without some knowledge of its

historical development. A comprehensive history of philosophy is the
9–volume work of Copleston (1946). Extremely profound, sympathetic,
and readable is (Jaspers 1962–1993). At this point we also mention
the wonderful booklet “Introduction to Philosophy ” by the same au-
thor (Jaspers 1953) which, together with (Bocheński 1959) and (Russell
1912) forms a triad that furnishes an ideal, easy and yet comprehensive,
introduction to general philosophical thinking in all directions.
A relatively short and eminently readable (Rabbi Type 1) history is
(Russell 1945). This brilliant work is, however, somewhat biased by its
author’s strongly expressed viewpoint of logical positivism. Especially
German idealism (Fichte, Schelling, Hegel) is grossly misrepresented.
On the other hand, we have excellent monographs on German ideal-
ism: (Gulyga 1990) and especially (Hartmann 1960). A general concise
history of philosophy, particularly for mathematicians and physicists,
is (Kuznecov 1981). An excellent anthology of texts of various philoso-
phers is (Perry and Bratman 1993).
With (Russell 1945) we restrict ourselves to Western philosophy. It
would be tempting to include Eastern (Indian, Chinese) philosophers,
especially with a view to dialectics, but my ignorance does not permit
it; see, however, (Jaspers 1962–1993).

Greek philosophers of nature

We know names such as Thales, Anaximander , and Anaximenes
(around 560 B.C.), belonging to the Milesian school (city of Miletus
226 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

in Asia Minor). Their activity would today be considered some rudi-

mentary form of physics. They were concerned about elementary con-
stituents of matter: water, air, earth, fire.
More important for the present view of science is Pythagoras
(around 530 B.C.), the founder of mystical mathematics, whose ideas
on the harmony of the world influenced Plato as well as Johannes Ke-
pler. They even influenced the modern theory of oscillations including
light and matter waves, fundamental in quantum theory.
Parmenides (around 480 B.C.) was the first monist philosopher: all
is One, change is an illusion. He thus was the first representative of
a block universe (secs. 3.7 and 6.3). He exerted a strong influence on
Plato and on modern monists such as Spinoza and Hegel.
Heraclitus (around 500 B.C.) represented an opposite point of view:
panta rhei , all is flowing. It is impossible to step into the same river
twice. His universe is dynamic, and he may be considered the first
dialectic philosopher: he discovered the dialectics of identity and dif-
ference (sec. 2.5).
Democritus (around 420 B.C.) was the first atomist. All matter
consists of atoms. This hypothesis had to wait for its experimental
confirmation until Rutherford around 1900 (so Democritus would never
have got the Nobel Prize).
Many concepts of modern physics and philosophy were foreshad-
owed by these philosophers between 600 and 400 B.C. No wonder that
they have exerted a considerable attraction to modern atomic physi-
cists such as Heisenberg and Schrödinger (1954) (who did get the Nobel
Prize).
Plato
Plato (428–347 B.C.) was the founder of modern philosophy and is
regarded by many as the greatest philosopher of all times. According
to Whitehead (1929), the European philosophical tradition “consists of
a series of footnotes to Plato” (see the motto of the present section).
He is not a systematic philosopher and provides questions rather than
answers. He gave the theory of ideas (Popper’s World 3, cf. sec. 5.2)
and emphasized the importance of mathematics. He raised almost all
contemporary problems of philosophy (mind and matter, God, dialec-
tics, etc.). Rather than constructing one philosophical “system”, Plato
wrote dialogues of high literary rank, which permitted him to regard
a problem from various angles. Thus he is the first representative of
philosophical pluralism, which provided an inspiration to many differ-
ent philosophical directions.
5.4. HISTORICAL LANDMARKS 227

Aristotle
Aristotle (384–322 B.C.) was a student of Plato. He is the creator
of logic and systematic metaphysics. Substance consists of matter and
form. There are 4 causes:
– causa efficiens: causality in the modern sense (determinism),
physics;
– causa finalis: final cause, “downward causation”, biology (cf.
sec. 4.1);
– causa formalisis (the plan of a house); and
– causa materialis (matter, e.g. bricks of which the house is built).
The reader may wonder about these Latin names: they come from
medieval philosophy, especially St. Thomas Aquinas (see below). Of
course, one may also speak of “efficient cause”, “final cause”, “formal
cause”, and “material cause”.
The last two “causes” correspond to matter and form mentioned
above. In contrast to Plato, his emphasis is on classification and de-
scription rather than on mathematics. He was much more interested
in concrete science (physics, biology) and much more systematic than
Plato who was primarily a mathematical thinker. God, to him, is the
“prime mover”, the first cause. Aristotle was very influential in the
Middle Ages, where he was regarded as an authority in science as well
as in philosophy. The very apparent completeness of his scientific work
made it obsolete as soon as new discoveries were made which did not fit
into his system. His main importance in science today is his creation of
systematic logic, which was transcended only by symbolic logic (Peano,
Frege, Russell, Gödel etc.). Through the mathematization of modern
science, Plato won over Aristotle. With some pointed exaggeration we
may say that Plato is the philosopher of mathematical physics, whereas
Aristotle is the philosopher of biology.
Nevertheless, his distinction between potentiality and actuality is
important for the difficult problem of the relation between mathemat-
ics and the physical world (sec. 5.2, p. 210), for the interpretation of
quantum mechanics (sec. 3.5, p. 108), etc.

Neoplatonists
Plotinus (204–270 A.D.) took up and developed further the mystical
and mathematical aspects of Plato’s thoughts. His works are literary
228 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

masterpieces. Goethe studied Greek in order to read Plotinus. He

was not a Christian but exerted great influence on St. Augustine. His
statement “the thinking thinks the thinking” may have served as a
model for Augustine’s theory of the Holy Trinity.
St. Augustine (354–430) is most important for our purpose because
of his theory of time (“What is time? If nobody asks me, I know, but
if I want to explain it to someone, then I do not know.”) What did
God do in the time before the creation of the universe? A humorous
answer is: He prepared hell for people asking such foolish questions.
Augustine’s correct answer was: the question is meaningless because
before creation (the “big bang”) there was no time.
Boethius (480–524) was a high government official. Sentenced to
death for alleged treason, he wrote in prison his famous work “Con-
solationes philosophiae ” (Consolations of Philosophy). Here he makes
the first allusion to the problem of a block universe (secs. 3.7 and 6.3).
If God is outside of time, then He has the whole universe (today we
would say: the entire space–time continuum) before Himself. Thus He
also knows the future. If the future is thus determined, where is the
place for man’s free will? Boethius answers that God’s foreknowledge
(in His ever–present eternity) does not mean predestination, so that
man remains free and responsible for his actions. For St. Augustine’s
and Boethius’ views of time cf. also (Lucas 1973).

Medieval philosophy
The main miracle of Western philosophy in the dark time between
500 and 800 is its sheer survival: all the principal works of classical
antiquity have been preserved. With Johannes Scotus Erigena (around
810–877) original work in the Platonist tradition started again, and
with Anselm of Canterbury (1033–1109) it reached a first culmination.
With Anselm, the school of medieval scholastic started. It flour-
ished in Paris and at other universities, and it is a school in which
theological and philosophical questions were discussed in a professional
way, based on more or less generally accepted premises; sometimes also
these premises were subjected to discussion. This kind of school philos-
ophy has contemporary analogues in the schools of Marxist philosophy
in the communist countries, and of logical positivism in the West.
The most eminent scholastic philosopher was St. Thomas Aquinas
(1225–1274). Whereas his predecessors were Platonists, Thomas used
the work of Aristotle as the basis of his philosophy. This was his
strength and his weakness. He gained a rigorous and consistent sys-
tem, but he partly lost Plato’s inspiration. Thomas’ merit was also a
5.4. HISTORICAL LANDMARKS 229

strict separation between philosophy and theology. Until recently, he

was the official philosopher of the Roman Catholic Church.
Scholastic philosophy continued its systematic and detailed work.
Well–known is William of Occam (around 1280–1350) because of the
principle of Occams razor : “Entia non sunt multiplicanda praeter ne-
cessitatem ”: no unnecessary concepts should be introduced.
The last medieval philosopher, Nicolaus Cusanus (1401–1464), how-
ever, was an individualist. He was a typical dialectic philosopher. With
his principle of “coincidentia oppositorum ”, the coincidence of logical
opposites, he was a forerunner of Hegel and Bohr (cf. sec. 2.5).
The non–Christian (Arabic, Jewish) medieval philosophers Aver-
roes, Avicenna, Maimonides etc. should at least be mentioned.
Descartes
René Descartes (Latin: Cartesius, 1596–1650) was the first mod-
ern philosopher and mathematician. He invented analytical geometry
(Cartesian coordinates!). He started with systematic doubt about ev-
erything. His principle “Cogito, ergo sum ” (I think, therefore I am)
allowed him first to prove his own existence. (This is a typical ar-
gument of dialectic reversal : original doubt gives rise to higher–level
certainty! Cf. Gödel’s argument in sec. 2.5.) Descartes might have
ended as a solipsist, were it not by an appeal to God. In fact, theoret-
ically, what I experience and what looks real to me, might well be only
a dream. However, God, being the absolute Truth, will not permit that
I am deceived all the time in this way. So the external world must be
real.
Descartes maintained a strict difference between matter and mind,
considering them as two completely separate “substances”, whose inter-
action is difficult to explain without recourse to God as the “Universal
Metaphysical Problem Solver”. His dualism is thus subject to White-
head’s “fallacy of misplaced concreteness”; nevertheless it has been very
influential in philosophical thinking until today.
Spinoza
Baruch de Spinoza (1632–1677) is a monist, recognizing only one
substance, called nature or God (“deus sive natura ”). Matter and mind
are “attributes” or “modi” of this divine substance. This doctrine is
called pantheism (Greek: everything is God).
Spinoza’s one substance can easily be identified either with mind
or with matter , so he is claimed by idealists like Hegel, as well as by
dialectical materialists.
230 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

The pantheistic, religious, side of Spinoza has attracted Goethe as

well as Einstein, both of whom liked to see God in nature.
We may say that Spinoza is halfway between Parmenides and Hegel.
Leibniz
Gottfried Wilhelm von Leibniz (1646–1716) was also a mathemati-
cian and logician, besides being a great philosopher. Together with
Isaak Newton, he founded the differential and integral calculus. As a
logician, he wanted to invent symbolic logic, but fell short of his aim.
He also constructed a computing machine.
Leibniz’ main philosophical merit is the theory of monads, already
discussed in sec. 5.3, cf. Figures 5.3 and 5.4. All monads are “window-
less” and interact only with the highest monad, God.
The monadic theory, artificial as it looks, has nevertheless influ-
enced such diverse philosophers as Bertrand Russell and Alfred North
Whitehead. Whitehead’s “actual occasions” are interacting monads
in space–time; his theory has proved surprisingly fruitful in quantum
theory (Bohm 1980; Stapp 1993).
Kant
Immanuel Kant (1724–1804) is considered the greatest philosopher
after Plato (with the possible exception of Aristotle or Hegel). Im-
portant and influential has been his criticism of classical metaphysics,
which studies the existence of God, the immortality of human soul, the
freedom of the will, and the existence of the external world in the tradi-
tion from Descartes to Leibniz. Kant admits only empirical knowledge.
(According to this criterion, also Kant’s philosophy, being nonempiri-
cal, is metaphysical and, strictly speaking, should also be rejected! This
remark, however, should not be taken too seriously: this would be a
rather gross oversimplification.)
The “things–in–themselves ” (Dinge an sich) of the external world
are unknown as a matter of principle. What exists is only a generalized
subject, the “transcendental I”. Kant is therefore a “transcendental
idealist”. Human knowledge can be a priori and a posteriori . A priori
are logical identities (e.g. a man is a human person); they are also called
analytic. A posteriori knowledge is empirical , obtained through the
senses; it is also synthetic, which means something beyond analytical
tautologies (logical identities).
Kant now asks: Is also synthetic a priori knowledge possible, that
is, is there non–empirical and logically non–trivial knowledge? We thus
have the following scheme:
5.4. HISTORICAL LANDMARKS 231

Kant considers synthetic a priori our knowledge of mathematics,

of three–dimensional Euclidean space, and of Newtonian mechanics
(causality). By means of a very original argument he says that this
synthetic a priori is not a property of nature, but reflects the struc-
ture of our mind . This “Copernican revolution of philosophy ” (turning
from object to subject) makes our mind responsible for our basic knowl-
edge of mathematics, Euclidean geometry, and causality, which, being
imposed by ourselves, is by definition absolutely exact. This is best
understood by Eddington’s example given in sec. 1.4.
The presently prevailing view is that mathematics is analytic (re-
ducible to logic), and that the present fundamental laws of nature in-
deed reflect the properties of our mind, together with real properties
of nature. Our theories are perspectives from which we look at nature
(theoria = insight, a way of rega rding). A theory thus may be com-
pared to a telescope for looking at stars, or to a microscope by which
we study organic tissue. Just as a telescope or microscope can be more
or less precise, even out of focus, a theory can be more or less exact
(sec. 6.5).
Thus, with Eddington (1939) and Bohm (1980) we may accept syn-
thetic a priori knowledge, but in contrast to Kant, this knowledge is
not absolutely accurate. In fact, general relativity implies curvature of
space–time through gravitation, hence deviations from Euclidean ge-
ometry; and quantum theory predicts deviations from causality. Evolu-
tionary epistemology provides a contemporary biological interpretation
of Kant’s synthetic a priori; cf. sec. 1.4.
Modern science (relativity and quantum theory) has frequently,
though critically, evoked Kant. Logical positivism has been particu-
larly sympathetic to Kant’s rejection of metaphysics.
Kant’s dialectics, concerning the antinomies connected with clas-
sical metaphysics, on the other hand, has strongly influenced Fichte,
Schelling, and Hegel, who, in this way, developed a metaphysics of their
own. In fact, metaphysics has steadfastly refused to die, as philoso-
phers such as Whitehead (1929) and Hartmann (1965) show; cf. also
232 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

(Stegmüller 1969).
Fichte
Johann Gottlieb Fichte (1762–1814) started out as a student of
Kant, transforming his transcendental idealism and enormously devel-
oping his dialectic approach. Originally, in his Wissenschaftslehre of
1794, he emphasized the (transcendental) I, posing the dialectic triad
I (subject, thesis), non–I (object, antithesis) together with their syn-
thesis. This obviously led Russell (1945, p. 718) to write that Fichte
“carried subjectivism to a point which seems almost to involve a kind
of insanity”. This is a classical example of a gross misjudgment of
a philosopher by a colleague; such misjudgments unfortunately occur
rather frequently in the history of philosophy (and it would be pre-
sumptuous to claim that the present book is free of them).
Later, in the Wissenschaftslehre of 1804, Fichte started with the
object (Fichte’s iteration, sec. 5.3). So, if Fichte is called an idealist, by
the same token he could be called a realist. This is another example of a
profound truth in the sense of Niels Bohr (sec. 2.5). The mathematician
Speiser (1959) relates Parmenides, Plato, and Fichte in a remarkable
“synthesis”.
Hegel
Georg Wilhelm Friedrich Hegel (1770–1831) is considered the great-
est dialectic philosopher. We have already discussed his dialectics to
considerable length in sec. 2.5. Although he is usually regarded as an
idealist philosopher, his thinking has become basic also for dialectic ma-
terialism: Karl Marx (1818– 1883) and Friedrich Engels (1820–1895).
Vladimir Ilich Lenin (1870–1924), the Russian revolutionary, studied
Hegel in great detail.
Needless to say, philosophers like Kant and especially Fichte and
Hegel are very difficult to read. Nevertheless, Fichte’s Wissenschafts-
lehre of 1804 and Hegel’s logic, which provide two alternative ways of
“ascent to the Absolute”, are extremely rewarding and well worth the
reading effort, just as the Rabbi’s third speech mentioned in the Preface.
(It is preferable, however, to start with the secondary literature, e.g.
(Findley 1958).)
Logical positivism and analytical philosophy
Logical positivism, also called neopositivism, started at Vienna, in-
spired by the physicist Ernst Mach (1836–1916), with Ludwig Wittgen-
stein (1889–1951), Rudolf Carnap (1891–1970), and others, and by
5.4. HISTORICAL LANDMARKS 233

Bertrand Russell (1872–1970) in England, where it soon became the

dominant philosophy; also Wittgenstein emigrated to Cambridge in
1929.
It is based on mathematical logic (secs. 2.1 and 2.2) and on the
empirism of the British philosophers John Locke (1632–1704) and David
Hume (1711–1776), who had also influenced Kant, as well as on the
developments of modern science, cf. (Russell 1945). It basically says
that all knowledge seriously so–called consists of logic, mathematics,
and empirical science.
It has become the leading philosophy of science, analyzing basic
concepts such as axiomatics, deduction and induction, verification and
falsification, etc. In this respect, its work is recognized by probably all
serious philosophers. What characterizes logical positivism, in partic-
ular, is the claim that philosophy is nothing else than logical analysis
of science and of language, that metaphysics is meaningless etc. This
claim, of course, has been contested, cf. (Stegmüller 1969) and any book
on contemporary philosophy.
Its underlying “ontology” is logical atomism (sec. 2.1). Besides
metaphysics, it also rejects dialectics etc.
After 1945, analytically–minded philosophers in Great Britain and
elsewhere recognized that they didn’t have to be logical positivists.
Thus they are called analytical philosophers ; cf. (Cohen 1986).
Important philosophers who initially adhered to neopositivism later
went their own ways:
Kurt Gödel (1906–1978), perhaps the greatest logician after Aristo-
tle, emigrated to Princeton (U.S.A.) and worked exclusively on foun-
dations of logic (apart from a pioneering paper in general relativity,
cf. (Hawking and Ellis 1973)). In his later years, he was an idealist of
Parmenidean type and a defender of a static “block universe”.
Alfred North Whitehead (1861–1947) was a British mathematician
who collaborated with Bertrand Russell in writing the monumental log-
ical work “Principia Mathematica ” (1910–1913). In 1924 he went to
Harvard University (U.S.A.) and became one of the greatest philoso-
phers and metaphysicians of our century. In the present book, we have
encountered him on many occasions.
Karl Raimund Popper (1902–1994) originally was close to the Vien-
nese school of logical positivism, together with Carnap, Gödel, Reichen-
bach, and Wittgenstein. He has written basic books on many questions
of philosophy of science; cf. (Popper 1977, 1979). In his later years he
critically approached metaphysics (Popper and Eccles 1977); cf. also
sec. 5.2.
234 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS

Ludwig Wittgenstein (see above) later on became a linguistic

philosopher, concerned with language and its limitations. Like Gödel,
he was a strange but enormously respected and influential philosopher,
although his later philosophy was very remote from science. Wittgen-
stein’s later thinking is expressed in short aphoristic, apparently simple
but very profound statements which are often paradoxical and subtly
dialectical. The great physicist Hawking (1988, pp. 174–175) writes:
Philosophers reduced the scope of their inquiries so much that Wittgenstein,
the most famous philosopher of this century, said: “The sole remaining task
for philosophy is the analysis of language.” What a comedown from the
great tradition of philosophy from Aristotle to Kant!
I have quoted this passage not because I agree with it, but because it
shows how separate great science has become from great philosophy in
our times.
In fact, none of the great physicists of our century was a logical
positivist, rejecting metaphysics. Einstein was a follower of Spinoza
(pantheism), Wolfgang Pauli (the famous “Pauli exclusion principle” in
quantum mechanics) was a mystical thinker, Schrödinger was an ideal-
istic monist of Eastern type (all consciousness is one), and Heisenberg
and Weizsäcker continue the classical tradition of Western philosophy.
Finally, Niels Bohr was a typical dialectic thinker: his principle of com-
plementarity is for a physicist what dialectics is to a philosopher. (It
is very interesting in this respect to compare the physicist O. Hittmayr
and the philosopher E. Heintel; cf. (Heintel 1990, p. 232–234)).

Why do scientists need classical philosophy?

In mathematics and science, the names of great discoverers are men-
tioned but it is very rare that their works are read. Which mathemati-
cian has studied the works of Euler or Gauss? Which physicist has read
Newton’s “Principia ” or Kepler’s “De harmonice mundi ”? Students
of history of science, certainly, but practicing researchers, rarely.
Mathematics and science form a rather homogeneous block which
grows, or a stream which flows. New discoveries are added, older parts
are discarded. With more or less reason, scientists think that the im-
portant results of the classics have survived and are implicitly contained
in modern textbooks and monographs.
With philosophy it is quite different. Philosophy is the history
of philosophy. Philosophy has few well–established and uncontested
results; it provides questions, problems, approaches and perspectives,
5.4. HISTORICAL LANDMARKS 235

which have retained their validity even today. Some problems formu-
lated by Parmenides, Plato, Aristotle, Descartes, or Leibniz have not
lost their actuality to the present day. The whole intellectual personal-
ity of a philosopher, in a sense, is much more important than the per-
sonality of a scientist. Newton’s or Einstein’s discoveries have merged
into the general stream of science. Plato’s views are discussed as Plato’s
views, Hegel’s views are commented or condemned as Hegel’s views. In
his charming little book, Jaspers (1953) says that philosophy is not a
science with more or less well established and recognized results, but an
ongoing process of thinking: “Philosophy means to be on the road. Its
questions are more essential than its answers, and each answer becomes
a new question.” Sometimes, walking on the road of philosophy, one
sees wonderful new vistas, but the true philosopher is not satisfied: he
moves on.
Thus all important philosophers from Plato to Wittgenstein are for
us, so to speak, contemporary philosophers. This may be compared to
music: judging from concert programs, available CD–recordings etc.,
our real contemporary composers are Bach, Mozart, Beethoven, and
Brahms! Also the best book on Beethoven is no substitute for listening
to his symphonies, and the best book on Plato’s philosophy cannot
dispense us from reading his dialogues. On the other hand, although
Einstein wrote fine books on relativity theory, there are contemporary
works such as (Misner et al. 1973) which are more modern and more
comprehensive and thus, in a certain sense, do supersede Einstein’s
books.
Let me conclude with a quotation from C.F. von Weizsäcker (1970,
p. 202), which I owe to Viktor Gutmann (Vienna):
[The physicist does not notice] . . . that by rejecting professional philosophy
he did not free himself from philosophy but became himself a dilettante
philosopher. Unconscious philosophy, however, is in general worse than a
conscious one, and thus just the most profound thinkers of modern physics
invariably return to philosophy.
236 CHAPTER 5. PHILOSOPHY FOR SCIENTISTS
Chapter 6
Philosophical implications of
science

6.1 Matter and mind

Mein Kind! ich habe es klug gemacht,
Ich habe nie über das Denken gedacht.
Johann Wolfgang von Goethe

There is no doubt that matter and self–conscious mind exist in some

way. Different opinions only concern their ontological status (secs. 1.2,
5.1, and 5.2). The clear–cut Cartesian dualism of two independent sub-
stances (to a lesser degree the “interactionism” of Popper and Eccles)
may suffer from the “fallacy of misplaced concreteness” (Whitehead).
An emergent theory of mind in the sense of R.W. Sperry (Globus et al.
1976, pp. 163, 181) might be more acceptable to the present habit of
scientific thinking, which is perhaps oriented materialistically (physics)
rather than idealistically (philosophy). Also dialectic materialism had
an emergent theory of mind (cf. Fig. 2.13 on p. 56). The problem
with clear–cut materialism and idealism is that modern science has not
produced a satisfactory definition, neither of mind nor of matter. We
recall Russell’s statement that “modern science has made matter less
material and mind less mental” and Weizsäcker’s saying “Tell me what
is your philosophy, and I will tell you how you must define matter”.
In fact, the physical definition of matter (mass) as given by classical
mechanics (the constant m in eqs. (3.1) and (3.11)), general relativity
(eq. (3.55) on p. 96), and especially in quantum mechanics (sec. 3.5) are
extremely abstract; see also sec. 5.1, especially Fig. 5.1. In particular,

237
238 CHAPTER 6. IMPLICATIONS OF SCIENCE

materialist theories of mind that are currently very fashionable suffer

from the grave defect that they are based on classical physics. A con-
temporary theory of brain processes, however, requires a treatment in
terms of quantum mechanics, in which the mind of the observer seems
to enter in one way or other; see sec. 6.4.
A relatively safe, although crude and perhaps corresponding to cur-
rent fashion only, analogue for brain–matter and mind is the compari-
son with computer hardware and software as discussed in various parts
of the book. This analogy might even be used as an argument for the
“immortality of the soul ”: a computer program may survive even if the
computer has become outdated and is no longer used . . . Of course, this
is not a proof of immortality, only a possible model ; such “proofs” have
been given and refuted in philosophy since Plato’s dialogue “Phaedo ”.
(Generally, many “proofs” in philosophy and quite a few “proofs” even
in science are little more than more or less plausible analogies!)
It is not without interest that St. Thomas Aquinas, the most in-
fluential Catholic philosopher, has considered mind the “form” of the
human body, in agreement with Aristotelian terminology (sec. 5.4).
Thus mind “informs” the body to become a living human being. Cf.
expression (4.19 on p. 188) and (Tresmontant 1971). This is far less
“substantial” than Descartes’ mental substance!
Emotions. It is characteristic that most philosophers concentrate
on conscious thinking. Emotions and feelings are usually disregarded
or considered secondary. This disregard is essential if the brain is to
be treated as a computer. The limbic system (sec. 1.1), however, plays
a basic role in human consciousness. Feelings must be considered as
fundamental as (or even more fundamental than) logical thinking; cf.
(Whitehead 1933, Chapter XI) and the end of sec. 2.1.
Selected readings on the body–mind problem may also be found
in (Hofstadter and Dennett 1981), (Rosenthal 1991), and (Perry and
Bratman 1993).
6.2. MATERIALISM, IDEALISM AND THE OUTER WORLD 239

6.2 Materialism, idealism and the outer

world
Materialism and spiritualism,
which are only defined by concepts
taken from each other,
are two aspects of the same thing.
Niels Bohr

It is best to take the reality of the external world as something that

is immediately given (Gilson 1972, pp. 286–287), no less directly than
Descartes’ “Cogito” (sec. 5.4), in agreement with intuitive common
sense and with the philosophical tradition up to Descartes. It is only
with Descartes and Kant that philosophy has started with the subject
rather than the object.
Concerning materialism and idealism it is instructive to compare
Hegel with Marx and Engels. If we simply replace Hegel’s “spirit”
everywhere with “matter”, we more or less get dialectic materialism.
To a mathematician, what counts is the logical structure, and the logical
structure is largely the same in all dialectic philosophy.
Now one might object that “matter” ontologically is very differ-
ent from “mind” or “spirit”. This may well be, but the mathematical
equations which implicitly define matter are supremely indifferent con-
cerning the “nature” of the quantities which these equations relate.
For instance, “temperature” is only statistically defined, hence it may
well be considered a creation of the human mind. The quantum state
function ψ (sec. 3.5) expresses “matter waves”, which essentially in-
corporate our subjective knowledge. So are matter waves objective or
subjective? Mathematicians would hardly “mind”, for them only the
form of the equation “matters” (the reader will kindly forgive me this
pun). Let us recall Bohr’s saying which served as motto of the present
section:
Materialism and spiritualism, which are only defined by concepts taken from
each other, are two aspects of the same thing.
To return to dialectics, it may be said that “matter” must somehow
be different from “spirit”. However, the “matter” of dialectic material-
ism has many properties reminding of Hegel’s spirit: it is capable of cre-
ative evolution, from hydrogen originally present after the “big bang”
to intelligent man. This “matter” has all the properties of Spinoza’s
240 CHAPTER 6. IMPLICATIONS OF SCIENCE

“deus sive natura ”. So it is largely a question of words. As Findley

(1958, p. 23) says: “There is, however, as much materialism in Hegel as
in Marx, since matter is for him certainly a stage in the ‘Idea’.” More
about this was said in sec. 5.1.
Why, then, so much ado about “idealism” or “materialism”? There
is a strong emotional difference. “Matter” sounds “scientific”, “down–
to–earth”, and “practical”, for the “working class”. “Spirit”, on the
other hand, sounds “aristocratic”, “theoretical”, and even “religious”.
Concerning the reality of the outer world, common sense says that
the world exists independently of an observer such as myself. Classical
physics and relativity are consistent with this view, but not quantum
theory, at least not in the generally accepted Copenhagen interpreta-
tion. In fact, quantum theory expresses the interaction of the observer,
the measuring apparatus, with the external world rather than describ-
ing this world in itself (sec. 3.5).
We can disregard the observer, but then quantum theory gives only
statistical averages. If one is satisfied with these statistical averages, in
the sense of objective probability or propensity (sec. 3.3), then quantum
theory could perhaps be said to describe the external world as such.
If this were not the case, the question would arise what had hap-
pened before man appeared . Earlier, there was no observer to make
the quantum phenomena “definite”. Clearly, however, nature took its
well–defined course also before man. As Weizsäcker said, “nature was
before the appearance of man, but man was before natural science”.
So the dinosaurs did not have to worry about the course of nature,
even without performing quantum experiments. They didn’t even need
Berkeley’s God in order to exist (see the limerick in sec. 1.2 on p. 11).
For further discussion we refer the reader back to sec. 5.1.

6.3 Time, creativity, and the block uni-

verse
The creative advance of the world
is the becoming, the perishing, and the
objective immortalities of those things
which jointly constitute stubborn fact.
Alfred North Whitehead

As we have seen in sec. 3.7, there are two different doctrines of :

“open”, creative and the block universe. The first doctrine says that
6.3. TIME, CREATIVITY, BLOCK UNIVERSE 241

the passage of is real; at present, the future is not yet determined and
can still be influenced; thus there is room for creativity. The second
doctrine says that the block universe is essentially finished; everything
has been determined beforehand, and there is no place for freedom and
creativity; the passage of is only an illusion. According to Weyl (1949,
p. 116):

The objective world simply is, it does not happen. Only to the gaze of my
consciousness, crawling upward along the life line of my body, does a section
of this world come to life as a fleeting image in space which continuously
changes in time.

Both doctrines have prominent adherents. Curiously enough, the

representatives of each doctrine take their view as an absolute and self–
evident matter of fact and frequently do not even bother to argue with
the other party. Representatives of the block universe are Minkowski
(cf. the motto of sec. 3.4), Weyl (just quoted), Gödel (in Schilpp 1949),
and Gold (in Duncan and Weston–Smith 1977). The block universe
dates back to Parmenides (sec. 5.4).
The majority, however, takes an open, creative universe for granted,
also without much discussion: Whitehead, Weizsäcker, and many oth-
ers. (Einstein seems to have vacillated between the two doctrines.) In
fact, if we take entropy, biological evolution, and the freedom of the
will seriously, we have little choice. Whitrow (1980) and Popper (1982)
thus provide arguments for an open universe.
The first doctrine is perfectly compatible with general relativity
(e.g., relativistic evolutionary cosmology), although the block universe
would come somewhat more natural to the four–dimensional space– of
Minkowski and Einstein, considered as a mathematical manifold.
A synthesis between the two complementary doctrines may have
been pointed out already by Boethius (sec. 5.4). For us, the universe
is open and ready to be influenced by our actions, in which we are
basically free and for which we are responsible. For an observer who
lives in the very far future, immediately before the end of the universe
and of , the world appears a finished block universe. Only God sees the
universe in both complementary ways, according to Boethius.
A detailed physical discussion has been given in sec. 3.7. We remem-
ber that the minus sign in the metric (3.69) on p. 120, making light
cones possible, is an essential condition for an open universe and thus
for creativity and free will (cf. next section). With a positive–definite
metric (3.72), only a block universe would be possible. We also recall
242 CHAPTER 6. IMPLICATIONS OF SCIENCE

that “time travel” is impossible except in fictitious block universes of

Gödel type.

Indeterminism and the open universe

If the determinism of classical mechanics (Laplace’s demon, sec. 3.1)
were correct, then the past would completely determine the future, or
even the present would completely determine both past and future.
Thus everything is related by a rigorous determinism, there is no place
for novelty, and the block universe would be appropriate.
In the same way, an open, creative universe must have some features
of indeterminism, so that novelty can arise.
This is particularly evident in the beautiful theory of Whitehead
(1933, Chapters XII and XIV), see also sec. 3.8. Here an “actual occa-
sion” (a monad in space–time) has a physical and a mental pole. The
past determines the physical pole, but the mental pole provides for cre-
ativity and novelty. The action of mind on matter (action of World 3
and World 2 on World 1, sec. 5.2; downward causation, sec. 4.1) is thus
entirely natural and immediate.
If one does not accept Whitehead’s metaphysics, the one may fol-
low Popper (1982) who provides a comprehensive, profound, and (to
me) absolutely convincing argument for indeterminism and an open
universe. Popper here anticipates chaos theory: a first draft of his
book dates back to 1956. (By the way, Popper refers for instability
to fundamental work of Hadamard published in 1898, a reference that
is standard in the theory of improperly posed problems (sec. 3.8), but
much less quoted in chaos theory!) The following reasoning is based on
Popper (1982) and Penrose (1989), as well as on secs. 3.1 to 3.5. (As
usual, we do not presuppose knowledge of the references, but they may
be helpful later on for a deeper understanding.)
Classical determinism (Laplace’s demon, sec. 3.1) presupposes sta-
bility: small causes produce small effects; a small error in the initial
data causes only a small error in the result.
Chaotic systems (sec. 3.2) are unstable: small initial errors may
later produce extremely large errors. The classical example is Edward
Lorenz’ “butterfly effect” in weather prediction. The initial errors may
be arbitrarily small (not zero, of course), corresponding to the best
accuracy of our measurements. It is clear that any physical measure-
ment is affected by measuring errors: perhaps one part in a billion, but
nevertheless never zero.
Already Poincaré (1908) formulated the properties of unstable and
chaotic phenomena in an admirable and entirely “modern” way, even
6.3. TIME, CREATIVITY, BLOCK UNIVERSE 243

including the “butterfly effect”:

If we knew exactly the laws of nature and the situation of the universe
at the initial moment, we could predict exactly the situation of that same
universe at a succeeding moment. But, even if it were the case that the
natural laws had no longer any secret for us, we could still only know the
initial situation approximately. If that enabled us to predict the succeeding
situation with the same approximation, that is all we require, and we should
say that the phenomenon had been predicted, that it is governed by laws.
But it is not always so; it may happen that small differences in the initial
conditions produce very great ones in the final phenomena. A small error in
the former will produce an enormous error in the latter. Prediction becomes
impossible, and we have a random phenomenon.
Our second example will be very much like our first, and we will borrow
it from meteorology. Why have meteorologists such difficulty in predicting
the weather with any certainty? . . . The meteorologists see very well . . . that
a cyclone will be formed somewhere, but exactly where they are not in a
position to say; a tenth of a degree more or less at any given point, and the
cyclone will burst here and not there, and extend its ravages over districts
it would otherwise have spared. If they had been aware of this tenth of
a degree, they could have known it beforehand, but the observations were
neither sufficiently comprehensive nor sufficiently precise, and that is the
reason why it all seems due to the intervention of chance. Here, again, we
find the same contrast between a very trifling cause that is inappreciable to
the observer, and considerable effects, that are sometimes terrible disasters.
He then goes on to discuss the fundamental importance of these phe-
nomena for probabilistic systems, completely in line with the present
book!
Deterministic chaos is a rigorous mathematical consequence of clas-
sical dynamics. In a certain sense, “most” classical systems are un-
stable. This is shown in (Penrose 1989, pp. 174–184) for Hamiltonian
systems. Thus, in fact “classical mechanics cannot actually be true of
our world! ” (ibid., p. 183).
As a matter of fact, this statement, shocking as it is at first, cannot
surprise us. As we have seen in sec. 2.4, even points in space and their
distance (Fig. 2.6 on p. 38) cannot be defined with absolute precision.
This is true a forteriori for the point masses and similar concepts of
classical mechanics. Here we do not even use the fact that classical
mechanics is only an approximate limiting case (v/c → 0) of relativity
theory and (h → 0) of quantum theory!
Thus the mechanical determinism of Laplace’s demon simply cannot
be true for our actual world, no matter how much it appeals to our
244 CHAPTER 6. IMPLICATIONS OF SCIENCE

intuition shaped by our first childhood experiences with mechanical toys

and by unconscious brainwashing, unnoticed by teacher and student,
in physics courses in school and university (I give a university course
on classical mechanics myself!).
Mechanics, of course, works (almost) perfectly in the textbook ex-
amples of planetary motion, free fall, throwing a stone along a parabolic
trajectory, pendulum, etc. These simple examples may be supple-
mented by equally simple examples where it does not work : e.g. flipping
a coin or throwing a die as mentioned in sec. 3.3. Nobody would try in
practice to predict the outcome of dice–throwing (Fig. 3.4 on p. 84) by
a computation of the mechanical trajectory: the outcome is much too
unstable.
A reason why we do not spontaneously notice this discrepancy be-
tween throwing a stone along a parabolic trajectory (deterministic)
and throwing a die (indeterministic) may be that missile trajectories
are treated in courses and books on mechanics, and dice–throwing is
treated in courses and books on probability, situated in different water-
tight compartments, at least for many students. (This is why courses on
natural philosophy exist: to break down the walls between watertight
compartments . . . )
In the case of throwing dice, the outcome can only be predicted
probabilistically by symmetry considerations: all faces have equal prob-
ability pk .
Here we may still object that even in this case, precise trajectories
exist “in principle” (as theoretical possibilities without any practical
significance). Very similar practically, but very different theoretically,
is the case of quantum mechanics (sec. 3.5). Here only probabilities
pk (3.60) are meaningful (p. 102); it is meaningless and theoretically
wrong to speak of trajectories.
In spite of these theoretical differences, the outcomes of a quantum
experiment and of dice –throwing are remarkably similar: only prob-
abilities are meaningful. Einstein expressed his lifelong objection to
quantum theory by the famous aphorism: “God does not throw dice”
(our motto of sec. 3.3).
6.4. FREEDOM OF THE WILL 245

6.4 Freedom of the will

The experience of being free
is a real experience.
Euan Squires

The problem of free will has a close relation to the mind–body prob-
lem (“downward causation”, cf. sec. 4.1; action of World 2 on World 1,
cf. sec. 5.2) and to the concept of an open, creative universe (cf. sec. 6.3).

Immediate perception of free will

In our decisions, actions, and volitions we are influenced by external
circumstances, but we feel that in our ultimate decisions we are free.
Otherwise there would be no intentional action. Planning would be
useless. We would not be morally responsible for our actions. Punish-
ment would be a farce. Policemen and judges could retire immediately,
hoping not to be killed, together with other people, by criminals who
could act without moral scruples since they are not following their own
free will and hence are not responsible for their murders.
A consequent denial of free will, like solipsism (sec. 5.1), is difficult
to refute, but still more difficult to maintain in a credible way. By
the way, both views have in common that they consider something
“normal” as an illusion: free will and the external world, respectively,
are regarded as illusions.
Anyway, it seems that people who consistently and honestly deny
free will, have been and are being eliminated by biological evolution.
Trying to find such persons, the most probable places would be prisons
and mental hospitals.

Free will and boundary conditions

Consider Fig. 3.3 on p. 81. Two “trajectories” (possible lines of ac-
tion) which originally, at time t = t0 are very close, would subsequently
follow completely different courses. The difference at t = t0 may be
as small as the diameter of an atom; two neighboring trajectories will
finally diverge and lead to very different actions.
In this model, free will does not so much affect the deterministic
trajectory (any possible action), as the boundary conditions, in our
case the initial conditions at time t = t0 (Fig. 6.1). The “orthogonality”
between the trajectories and the initial line t = t0 is a nice expression of
the complementarity between freedom and necessity. It is comparable
246 CHAPTER 6. IMPLICATIONS OF SCIENCE

Figure 6.1: Selection (final cause) and action (deterministic cause)

also to the complementarity between “physical laws” (corresponding to

action) and “software laws” (corresponding to selection, cf. Fig. 6.1).
Randomness and fluctuation occur everywhere in nature, as we shall
see in sec. 6.5. The random fluctuations constantly going on in our neu-
ral and brain activity might possibly provide a mechanism for facilitat-
ing the free process of selection shown in Fig. 6.1. Random fluctuations
will probably save the life of Buridan’s ass, which, according to (Russell
1945, p. 213) “was unable to choose between two bundles of hay placed
at equal distances to right and left, and therefore died of hunger.”

Analogy with a thermostat

In sec. 4.1, we have considered a thermostat as a model for “down-
ward causation”, the action of mind on the body. There is no need to re-
peat the argument in detail: the thermostat “causes” the required tem-
perature, mind “causes” the willed bodily movement, software “causes”
the computer “hardware” to perform the desired computation. “Cau-
sation” here means “downward causation”, or Aristotle’s “final cause”.
These “final causes” generate the differential equations and other
“efficient causes” necessary to perform the desired “program” (in the
sense of computer language). “Efficient causes” in the sense of Aristotle
are simply the deterministic operations of usual causality as embodied,
e.g., in the differential equations of classical mechanics. This has been
seen to work even in classical physics, following the principles of Fermat
and Euler–Maupertuis (ends of secs. 3.1 and 4.1).
The thermostat case of automatic regulation is very instructive also
in another respect. In order to get appropriate behavior (left side of
Fig. 4.3 on p. 163), we must have “damped” solutions of form (4.1).
“Damping”, or friction, is a typical “dissipative” phenomenon, that
6.4. FREEDOM OF THE WILL 247

is, mechanical energy is not conserved . What happens? Part of me-

chanical energy is “dissipated” by converting it into thermal energy:
friction generates heat. So the total energy, mechanical plus thermal,
is conserved after all.
Conservation of energy. It is sometimes asserted that free will is
incompatible with conservation of energy. For this reason we have
in sec. 4.1 considered the case that the input signal (corresponding
to the action of mind) has no power at all , giving the example of a
compact disc. So the energy balance in the brain is not influenced
at all. (That the brain uses a great amount of energy for its work
is no more surprising than the large amount of energy used in a hi–fi
amplifier with its loudspeakers or the even larger amount used by a
modern supercomputer.)
To be still more concrete, let us consider the following example. If
I wish to pick up my pencil, this wish will cause the necessary bodily
movements. In the same way, if I put the thermostat of my room
to 22 degrees centigrade, this will cause the heating or cooling to act
so as to produce the desired temperature. The (very small) energy
I need to move the pointer in the thermostat is practically the same
whether I select 20◦ or 22◦ (the energy needed afterwards to produce
the selected temperature is something different!). In the same way,
the (extremely small) energy in the activity of my brain involved in
my decision (to pick up or not to do so) is practically the same (the
mechanical energy involved in actually moving my hand to pick up the
pencil is something different). So thermostats and free wills indeed do
not violate the principle of conservation of energy.
Anyway, as all laws, conservation of energy may be only approxi-
mate, and it is taken much more narrowly by biologists than by physi-
cists who know about the difficulties of energy conservation in general
relativity and quantum theory.
In terms of Fig. 6.1, “selection” corresponds to temperature selec-
tion by moving the pointer (almost effortless) and “action” corresponds
to the production of the selected temperature (involving considerable
energy).

Quantum theory
There is no doubt that quantum theory governs neural activity
rather than classical deterministic physics (Stapp 1993). Nonetheless
is it controversial whether the indeterminism of quantum jumps is di-
rectly responsible for free will, as some physicists think; cf. (Jordan
248 CHAPTER 6. IMPLICATIONS OF SCIENCE

1968, p. 338). The reason is that between quantum jumps, the quan-
tum state function evolves deterministically (according to Schrödinger’s
equation), and the quantum jumps themselves are governed by statis-
tical laws which seem to be no less rigid than deterministic laws, so
as to leave little room for freedom. At any rate, quantum fluctuations
provide an omnipresent random background which is important as we
have seen above.
Also, quantum theory mitigates the dualism between mind and
matter showing that the simplistic materialism of some modern “neu-
rophilosophers” and “neuroscientists” such as Churchland (1988), Den-
nett, and Edelman (1989) is inadequate: the adequate quantum theory,
in some way, always seems to involve mind; see sec. 3.5, (Lockwood
1989), (Margenau 1984), (Squires 1990), and (Stapp 1993).
Going an essential step further, the physicist F. Beck and J.C. Eccles
(Eccles 1994, sec. 9) have elaborated a detailed quantum–mechanical
model for mind–brain interaction. Such a model may not describe the
real situation, but it shows that such an interaction is possible.
This model is based on the plausible assumption that mind may
change the distribution of quantum probabilities of micro–events in the
brain, whose combined action produces the desired macro–event. This
is another example of amplification (sec. 4.1).

The argument of Lucas

In sec. 5.2 we have outlined an ingenious argument of Lucas (1970,
§25). He uses Gödel’s theorem as an additional argument that human
thinking is not determinable by any physical law: the human brain
cannot be a deterministic machine or computer because human think-
ing essentially transcends computability (sec. 2.3). As for almost all
philosophical arguments, there are also counterarguments (ibid., §26)
which can again be refuted (§27).

The argument of Epicurus

Following Lucas (1970, §21) we can also apply a variant of the
Epicurus–Haldane argument already given in sec. 1.2. If everything
is determined, then it is also determined that I believe that everything
is determined. I thus believe this not because I find by logical consid-
erations that it is true, but because the structure of my mind has been
determined by physical law so that I must think this way (regardless
of what logic says). Many things, however, are determined by physical
law which need not be true. Hawking (1993, p. 129) says:
6.4. FREEDOM OF THE WILL 249

Each week, my mail contains a number of theories that people have sent me.
They are all different, and most are mutually inconsistent. Yet presumably
the grand unified theory has determined that the authors think they were
correct. So why should anything I say have any greater validity? Aren’t I
equally determined by the grand unified theory?
What does Epicurus’ argument really show? It does not directly
refute determinism. It shows, however, that universal determinism
(“everything is predetermined by a grand unified theory” (sec. 6.6)) is
non–arguable. If someone claims that he has found a proof of universal
determinism, we can immediately reply that his reasoning is inevitable
if universal determinism is true, regardless of the argument being log-
ically valid or not. So his argument is simply worthless, as the above
quotation from Hawking shows very clearly.
The argument of Epicurus is particularly malicious and insidious. It
turns determinism against itself: it beats determinism at its own game.
By the way, Gödel’s proof uses self–reference (reflexive logic, see
end of sec. 2.5), and also the argument of Epicurus is self–referential.
Otherwise, both arguments have little in common, but we see again the
importance of self–reference in philosophic reasoning.

Indeterminism is not enough

As Popper (1982, pp. 126–127) has pointed out, “indeterminism is
not enough” for freedom of the will, “indeterminism is necessary but
insufficient ”. What does this mean?
As we have seen in sec. 6.3 (“indeterminism and the open universe”),
indeterminism rather than determinism seems to reign in the universe:
we have the instability of chaotic systems, not to speak of quantum
indeterminacy. The cause does not determine the result to any useful
accuracy; unmeasurably small variations (errors) in the initial data may
completely change the result: the “butterfly effect”, cf. Fig. 3.3 on p. 81.
Thus there is practical indeterminism, but how can we “harness ”
(Penrose 1989, pp. 172, 174) this indeterminacy to get free will? To
be sure, free will is incompatible with rigorous determinism, but inde-
terminism does not yet directly imply the freedom of the will: inde-
terminacy is not enough. Does the uncertainty of a weather forecast
define, or at least increase, my freedom in planning a weekend hike in
the mountains? This is an obvious absurdity.
What indeterminacy does, however, is to create a free space in which
the will can act. Indeterminism is necessary but insufficient: it provides
the necessary leeway, but only the action of mind is then sufficient for
250 CHAPTER 6. IMPLICATIONS OF SCIENCE

an act of free will. Mind belongs to World 2, partly even to World 3

(sec. 5.2), but our actions refer to the material World 1. For free will it
is therefore necessary that World 1 is open to World 2 (Popper 1982,
p. 113 ff.). This action of mind on matter has been called downward
causation (see above).
Let us give an example: unfortunately, tomorrow I have to attend a
faculty meeting. The meeting has been scheduled long ago, it is known
to all professors, and it has even been announced in the University Bul-
letin. Being objective or at least intersubjective, the scheduling of the
meeting belongs to World 3 (although Plato may not like to see it in
his lofty realm of ideas). I see the schedule and reluctantly make up my
mind to attend the meeting. This decision clearly belongs to World 2 :
World 3 thus has acted on World 2! The next day, I really attend the
meeting and walk to the meeting room. This movement of walking
obviously belongs to World 1. Thus we have the action: World 3 (ob-
jective schedule) → World 2 (subjective decision) → World 1 (physical
movement).
How can this interaction between World 2 and World 1 be realized?
We consider three possibilities, summarizing what we have already said.
(1) Whitehead’s solution. In Whitehead’s metaphysics (Whitehead
1933, Chapter XII), see secs. 3.8 and 6.3, an actual occasion (a monad
in space–time) has a physical pole and a mental pole, which are more
or less on the same footing. Hence elements of World 1 and World 2
are integrated in a most natural and spontaneous way.
(2) Thermostat analogy. As we have just seen, a thermostat causes
the room temperature to assume a prescribed value. The tempera-
ture selection (cf. Fig. 6.1 on p. 246) is made by moving a pointer on
the thermostat, which can be done with arbitrarily small energy (by
appropriate lubrication, amplification, etc.). The corresponding room
temperature change involves very considerable energy (heating or cool-
ing).
It would thus seem that we have the case of a small cause resulting
in a large effect, which, according to chaos theory, corresponds to in-
stability and uncontrollability. This is certainly not true in the present
case which is very precisely controllable!
The explanation is feedback. The temperature achieved is constantly
compared to the desired temperature (fed back to the thermostat), and
the temperature difference (“deviation” or “error”) is used to improve
the room temperature until the error is ideally zero.
Thus feedback allows “controlled amplification”, in contrast to the
“uncontrolled amplification” in unstable chaotic systems. A particu-
6.5. LAWS OF NATURE 251

larly impressive case is a hi–fi equipment amplifying the almost mi-

croscopic signal on a compact disc to produce the majestic sound of a
Beethoven symphony; again feedback is, of course, essential. All ser-
vomechanisms operate in this way, as we have seen in sec. 4.1.
A final example may illustrate this. I wish to walk from point A to
the nearest village. From point A I see the village and select the di-
rection accordingly. In the first case, there is a dense forest between A
and the village, so the best I can do is to maintain the initial direction
as precisely as possible. Not surprisingly, I will find out that I have
missed the village by a couple of kilometers. In the second case, there
is no forest and I can see the village all the time, so that I can contin-
uously correct my direction so as to arrive at the desired place. The
first case corresponds to instability (a small initial error is enormously
magnified), in the second case we have feedback (by seeing the village)
continuously stabilizing the direction of my walk.
Is there a more direct relation between automatic regulation (ther-
mostat, servomechanism), and indeterministic, chaotic motion? As we
have seen above, feedback must involve damping or friction. Friction,
however, converts mechanical energy into heat, which is nothing else
than the chaotic motion of gas or liquid molecules as described by sta-
tistical mechanics (sec. 3.2)! In sec. 4.1 we have seen that damping is
inseparably related to positive feedback which always occurs in auto-
matic regulation.
Thus, in Penrose’s terminology introduced above, automatic regu-
lation “harnesses ” chaotic motion. This equally holds for “indetermin-
ist” classical chaotic motion and for the “indeterminism” of quantum
mechanics. (Heat in solid bodies corresponds to vibration of molecules,
which is a quantum rather than a classical effect!)
(3) Quantum interaction. Above we have mentioned the model of
Beck and Eccles (Eccles 1994). The change of probability distributions
may be regarded as an analogue to moving the pointer of a thermostat.

6.5 Laws of nature

God has not made the original
equations available.
James Gleick

Take the example of Kepler’s three laws for the motion of planets
around the Sun. The first and most important law is that planets
252 CHAPTER 6. IMPLICATIONS OF SCIENCE

move along ellipses whose focus is the center of the Sun. It is easy (I
do it regularly in my introductory course on mechanics for students in
the second semester) to derive, from Kepler’s laws, Newton’s laws of
gravitation.
Inversely, Newton’s law of gravitation permits the derivation of Ke-
pler’s laws, with two important qualifications.
(1) Possible orbits are not only ellipses, but also other conic sections:
parabolas and hyperbolas. This is relevant for comets.
(2) If, in addition to the Sun and the Earth (say), the attraction of
other planets is taken into account, then the orbits are slightly
perturbed: they no longer are exact ellipses.
The second fact is particularly important: Kepler’s laws do not hold
exactly, due to the perturbation of other planets! If the measurements
of Tyho de Brahe, from which Kepler derived his laws, had been more
accurate, then Kepler would never have arrived at his simple laws, and
Newton’s laws would not have followed so readily. In this case, as Alfred
North Whitehead has pointed out, too accurate measurements might
have hindered the progress of science!
This is another nice instance of a Hegelian triad (sec. 2.5). Thesis:
Kepler’s laws; antithesis: Newton’s laws; synthesis = thesis on a higher
level: Kepler’s laws with corrections for the effect of other planets.
Even Newton’s law of gravitation, however precise (to 10−7 ), is
not absolutely exact: Einstein’s theory of general relativity is better
(sec. 3.4). Einstein’s theory comprises Newtonian mechanics as a spe-
cial case for “small” velocities and “weak” gravitational fields, which is
sufficient for most applications.
Einstein’s general theory of relativity is extremely elegant, general,
and beautiful. So far, it has never been refuted by experiment. Is it
absolutely true?
Quantum mechanics (sec. 3.5) is equally elegant and general; it has
been confirmed by experiments better than any other theory. Is it
absolutely true?
Quantum theory holds for very small distances: between nucleus
and electrons in an atom, and between atoms in a molecule. At these
small distances, quantum effects must occur: all physicists agree on
that.
Now, no quantum effects can be derived from general relativity.
Therefore, general relativity cannot hold for very small distances. Thus
it cannot be absolutely true.
6.5. LAWS OF NATURE 253

But many scientists agree that general relativity is perfect and

unique with respect to elegance and conceptual simplicity.
How can it happen that such a beautiful theory is not exactly valid?
Let me try to illustrate this by means of examples which we understand
better. The ancient Greeks thought that the Earth is a perfect sphere
because the sphere is the most perfect and symmetric of all surfaces.
In fact, a fluid planet without rotation must be an exact sphere. A
rotating planet, however, is slightly flattened and very similar to an
ellipsoid of revolution. But the actual figure of the Earth, the geoid,
is not an exact ellipsoid: it deviates from an ellipsoid by 100 m at
most and is rather irregular, because of the attraction of the irregular
“topographic masses”: mountains, hills, valleys, every feature which
makes our planet attractive and interesting. So symmetry (sphere,
ellipsoid) creates the basis, but irregular features are superimposed.
In the case of planetary motions, Kepler was so fascinated by the
beauty and simplicity of his laws, especially of the fact that the plane-
tary orbits were exact ellipses, that he never ceased to praise the divine
harmony of the Universe.
Nevertheless, Newton introduced perturbations to this perfect har-
mony, and Einstein replaced Newtonian mechanics by an even more
accurate, general, and beautiful theory.
What should we conclude? Elegance and internal perfection may
be a necessary condition for a good scientific theory; they are, however,
not sufficient. (Wolfgang Pauli exaggerated when he said: “Elegance is
only for the tailor”, implying “not for the scientist”.)
The very perfection of both general relativity and quantum theory
may stand in the way of their unification, but many physicists believe
that such a unification is necessary and also possible. In his extraordi-
nary book, Penrose (1989) has attempted to fathom the important and
far–reaching consequences which such a unification would have even for
general human thinking and for artificial intelligence. (Special relativ-
ity and quantum theory are, of course, unified in Dirac’s theory of the
electron.)
Let us come back to our question: What is a physical law? Does it
reflect the behavior of nature or only the structure of our mind and of
our perceptional apparatus? This question was considered already in
sec. 1.4 of our book, especially in connection with Eddington’s exam-
ple. The answer is: probably both, it has an objective and a subjective
component. The subjective component is particularly evident in quan-
tum theory (sec. 3.5), but no physicist really doubts that his laws do
have an objective basis in nature. What would happen if the engineer
254 CHAPTER 6. IMPLICATIONS OF SCIENCE

calculated bridges or high buildings according to formulas which do not

express real properties of nature?

Are natural laws exact?

Kant has considered Euclidean geometry and Newtonian mechanics
as a priori given and hence exact. We now know that even symbolic
logic (set theory) and mathematics are under suspicion (logical para-
doxes, Gödel’s theorem, see sec. 2.3). The objects of everyday life,
including physics, are subject to “fuzzy logic” (sec. 2.4). In the same
section, we have seen that measuring errors are unavoidable, and at-
tempts to idealize one concept (regarding the angles α and β as error-
less) may lead to a falsification of others (the angle γ). Furthermore, we
have just seen that even such apparently “perfect” laws as Kepler’s laws
or Einstein’s theory of general relativity cannot be valid absolutely.
The beautiful and important laws of thermodynamics (including the
famous Second Law of Thermodynamics) are a consequence of applying
mechanics to a system of a huge number n of particles or molecules
(statistical mechanics) and, strictly speaking, require n → ∞.
Properties such as temperature are averages over a large number of
particles (it would be meaningless to speak about the temperature of
a system that consists of 10 molecules only). At least thermodynamics
is only a statistical theory, an example of “order out of chaos”.
Quantum measurements are limited by Heisenberg’s uncertainty re-
lation (sec. 3.5), and complex mathematical and logical theories might
be affected by the limitations imposed by Gödel’s theorem.
Hence we find inaccuracies, errors, and limitations wherever we look.
It becomes increasingly more difficult to believe that physical laws are
exact if already the concepts which they relate, cannot be defined with
absolute accuracy.
Randomness and fluctuations occur over and over again in nature.
On the quantum level we have a universal background of essentially
random quantum fluctuations (see end of sec. 5.2). But also on the
classical level we have Brownian motion, the erratic movement (“ran-
dom walk”) of a small dust particle suspended in a fluid and of course,
the statistical mechanics of air molecules which we have just mentioned.
Even in our nervous system and in the brain we seem to have some
“randomizers”, similar to “random number generators” familiar from
computer work. For this reason, repeated precise measurements never
give the same result: there are random errors described by the Gaussian
curve of Fig. 4.12 on p. 184. Or try to follow a precisely drawn circle
by free tracing performed by your hand: you will never be able to stay
6.5. LAWS OF NATURE 255

exactly on the curve (Fig. 6.2). No matter how firm your hand is, it
will inevitably perform (very small) shaky movements. There seems to
be a randomizer in our nervous system which, however, works on the
physical level only and does not affect the mental level (sec. 5.2).

Figure 6.2: Tracing a circle with a shaky hand

The great contemporary physicist John Archibald Wheeler even

questions the unlimited validity of physical laws in space and time.
Are the laws of physics on Sirius the same as on Earth, and have the
laws governing the Big Bang been the same which we know today? If
they are (and so far we have hardly a reason to doubt this although it
appears to be short of a miracle to some people) we could never test
this by direct observation. In a certain sense, Wheeler (1994, p. 300)
even says: “No laws.” Also: “Law without law.”
What has been left of “the eternal laws of nature to which it is in-
exorably subject”? I don’t know. In the absence of man, are there still
laws of nature? Obviously. The same? Probably not. Why are “our”
laws so applicable and beautiful? What about Wigner’s “unreasonable
effectiveness of mathematics in natural sciences” (sec. 3.9)? Why can-
not we then have exact laws? Why “did not God make available the
exact equations”? Are there “exact equations”? Questions, more or
less legitimate ones, but no answers . . .
Generally, because of unavoidable measuring errors, it is never pos-
sible to verify or falsify our physical theories exactly! This seems to
have been frequently overlooked in discussions on empirical verification
or falsification of physical theories (cf. sec. 3.9).
As an example, take the empirical verification of Euclidean geom-
etry by measuring the three angles of a large triangle formed by light
rays such as done in geodesy. This idea was already proposed by C.F.
Gauss. Let us assume, for a moment, that we can define the points
ABC of a triangle and measure the angles α β γ with absolute preci-
sion. Will their sum be π or 180◦ ?
256 CHAPTER 6. IMPLICATIONS OF SCIENCE

Surprisingly enough, the answer is no! The reason is that general

relativity, which we have no motive to doubt in this context, states
that light rays are not straight lines but are slightly curved because of
the curvature of space–time due to gravitation. In this sense, empirical
geometry is not Euclidean. (This has been hard to accept for many
scientists such as Alfred North Whitehead who developed an alternative
relativistic theory of gravitation in 1922.)
An original idea concerning physical theories is due to Bohm (1980,
p. 3). The word “theory” derives from the Greek “theoria ” (same root
as “theater”!) which means “view” or “perspective”. Thus theory is
not so much a knowledge about the functioning of the world, but rather
an insight, a way of looking at the world. It is thus primarily an activity
of the human mind and as such a priori in the terminology of Kant.
It is not, however, absolute and errorless as Kant thought, but subject
to our errors and weaknesses. One theory, one perspective, may be
better than another (relativity is better than classical mechanics), but
no theory can be expected to be absolute and final: we cannot expect
to catch the whole universe in one perspective. Thus a theory is to the
mind what a microscope or telescope is to the eye. Regarding nature
equipped with a theory is similar to looking at a landscape through the
eyes of one’s favorite painter.
Also Whitehead (1933, Chapter VII) considers the possibility that
natural laws only hold approximately (somewhat as the criminal law
governs human societies). This comparison immediately raises the
question: If the laws are only “approximately respected”, who sets
the limits of possible deviation from the law and watches them? God
as the “policeman of nature”?

Conventionalism
Well, if the relativistic concepts differ so little from the correspond-
ing Euclidean concepts, why not assume Euclidean geometry to be
exactly valid, and treat the very small deviations (of order 10−8 ) as
“relativistic corrections”? In this spirit, why should we not say that
light rays are not Euclidean straight lines but are slightly curved, so
that α + β + γ = 180◦ + , where  is a small quantity which is not zero
because of curvature effects?
6.5. LAWS OF NATURE 257

This is the point of view of conventionalism. It says that the choice

of geometric and physical laws, to a certain extent, is arbitrary or con-
ventional. We know that modern physics considers the speed of light
c, given by (4.13) on p. 184, as a fixed errorless constant by defining
length in terms of c, cf. (4.17), that is by convention. So why not keep
Euclidean geometry and Newtonian mechanics and treat deviations as
corrections?
Logically, this is perfectly possible, and even practically one fre-
quently proceeds in this way, cf. (Moritz and Hofmann–Wellenhof 1993,
p. 255). The great mathematician and author of charming philosoph-
ical books, Poincaré (1902, 1908, 1958, 1963) thought that one would
never give up Euclidean geometry because of its unrivalled simplicity!
Even great scientists such as Poincaré and Whitehead can be wrong.
In fact, Einstein’s theory of relativity is such a marvellously perfect
theory that one has even preferred to give up Euclidean geometry in
order to keep it.
As a matter of fact, the general theory of relativity, in view of its
very perfection, has challenged many scientists to invent gravitational
theories of similar perfection and similar beauty. Some of them have
been refuted by experiment, but others have withstood all tests. Nev-
ertheless Einstein’s theory is preferred almost unanimously: because of
its unsurpassed perfection, beauty and internal consistency. Remember
Wigner and Penrose!
Hence conventionalism is right in the sense that for many theo-
ries it is possible to invent rival theories which are equally compatible
with experiment, within the limits of measuring accuracy. Here Kuhn
(sec. 3.10) holds out against Popper (sec. 3.9): falsification of a theory
is only possible within the limits of measurements; new “paradigms”
are introduced by other reasons (elegance, utility) in a more or less
conventional way.

Positivism
Hipparchus (around 146–127 B.C.) and Ptolemy (2nd century A.D.)
explained the motion of the planets as seen from the Earth, by a su-
perposition of epicycles. Mathematically this is a series of cosines of
form

x = a1 cos(ω1 t + β1 ) + a2 cos(ω2 t + β2 ) + a3 cos(ω3 t + β3 ) + · · · (6.1)

Here ai , ωi and βi are constants, and the geometrical explanation is

shown in Fig. 6.3.
258 CHAPTER 6. IMPLICATIONS OF SCIENCE

Thus it corresponds to a superposition of circles Ci of radii

a1 , a2 , a3 , . . . The angular velocity of the movement of the center of C2
around the basic circle C1 is ω1 , the angular velocity of the movement
of the center of C3 around C2 is ω2 , etc. Finally, the βi are appropriate
phase angles (i = 1, 2, 3, . . .). This is called a system of epicycles. It
gives merely a geometrical illustration and interpretation of eq. (6.1)
and does not correspond to the actual movement of the planets.
Nevertheless, these theories of epicycles described planetary motion
with respect to the Earth, with astonishing accuracy, the better, the
more terms (circles of radius ai ) were superimposed (also i = 4, 5, 6,
etc.)

Figure 6.3: A system of epicycles

Only Nicolaus Copernicus (1473–1543) placed the Sun, rather than

the Earth, at the center of the universe, and Johannes Kepler (1571–
1630) recognized that the planetary orbits are ellipses rather than cir-
cles. Finally, as we have seen at the beginning of this section, Newton
derived from Kepler’s laws his law of universal gravitation, eq. (3.11)
on p. 74.
It is now highly remarkable that if the laws of planetary motion,
under the influence of the Sun and of the other planets, are calculated
from Newton’s law (3.11), referred to the center of the Earth, the solu-
tion has exactly the form of eq. (6.1). Computationally, Newton brings
us back to Ptolemy! (By the way, from the convergence or divergence
of astronomical series of form (6.1), Poincaré discovered chaos theory,
sec. 3.2 . . . )
From a positivist point of view, the purpose of a physical law is to
describe (as accurately as possible) rather than to explain. To explain
using entities that are not directly observable is considered more or less
metaphysical (see end of sec. 1.2) and should thus be rejected.
Now the epicycle theory, expressed by eq. (6.1), furnishes a perfect
and rather simple mathematical description, and the planets have been
inaccessible to man until now, as the Moon was before the astronaut
Armstrong stepped on it in 1969. It can be argued that accessibility
can also be achieved through instruments such as telescopes, so that
6.5. LAWS OF NATURE 259

the planets are directly observable entities, but these arguments need
not be accepted by a stubborn opponent. Thus I should say that those
who regard the planets as metaphysical entities are still less in number
than those who regard in this way atoms, electrons or quarks (have you
ever seen a quark?), but the principle persists (have you ever been on
Mercury?).
In reality, of course, Kepler’s laws of planetary motion, and New-
ton’s and Einstein’s (sec. 3.4) theories of gravitation are universally
preferred by physicists. This shows that even physics contains unob-
servable quantities so that the positivist point of view is not carried
through completely in scientific practice.
Whitehead (1933, Chapter VIII) has given a beautiful example.
Earlier in this century it was remarked that a series (6.1) as known
before had to be supplemented by extremely small additional terms in
order to achieve best agreement with very precise contemporary ob-
servation. “Every Positivist must have been completely satisfied. A
simple description had been evolved which fitted the observed facts”
(ibid., p. 127). But, in fact, the American astronomer Percy Lowell
was not satisfied. He found out by computation that the new anoma-
lous terms corresponded to the attraction of an imaginary small planet,
an astronomer at the Lowell Oberservatory photographed in 1930 the
sky in this direction and found . . . the new planet Pluto! Whitehead
wrote
The civilized world has been interested at the thought of the newly discov-
ered planet, solitary and remote, for endless ages circling the sun and adding
its faint influence to the tide of affairs . . . The speculative extensions of laws,
baseless in the Positivist theory, are the obvious issue of speculative meta-
physical trust in the material permanences, such as telescopes, observatories,
mountains, planets, . . .
Thus positivism, with its justified emphasis on observability and
mathematical description, is important but should not be interpreted
too narrowly.
Here positivism, similar as conventionalism before, represents a cer-
tain general direction of thinking rather than a definite school of philos-
ophy such as logical positivism (sec. 5.4). Whitehead’s pointed remarks
should be understood in this sense.

Laws, theories, and mathematical models

The borders between these concepts are ill–defined and shifting, so
that only rough, “fuzzy” definitions are possible.
260 CHAPTER 6. IMPLICATIONS OF SCIENCE

A theory usually is comprehensive and well confirmed by experi-

ment, e.g., Newtonian mechanics, special and general relativity, quan-
tum theory, or biological evolution.
A law expresses an important and generally valid feature, formula,
or theorem of a theory, e.g. Newton’s law of universal gravitation, or
the law of conservation of energy. Often, “laws” is practically the same
as “theory” or “theories”, e.g., when we speak of the “laws of physics”.
A hypothesis is theoretically possible and interesting, but usually
not (yet) completely confirmed by experiment. One frequently speaks
of a working hypothesis. I should consider the reality of the external
world an example of an extremely important working hypothesis.
A mathematical model usually implies a simplification to make a
phenomenon accessible to mathematical treatment. Examples are the
sphere or a slightly flattened ellipsoid of revolution as models for the
shape of the Earth which, of course, is more complicated. A perfect ex-
ample of a mathematical model are Kepler’s laws for planetary motion
along ellipses which are mathematically very simple and useful but not
entirely precise as we have remarked at the beginning of this section.
Very important is the concept of reference model , which serves as a
basis for determining or defining a real phenomenon, thus
phenomenon = reference model + deviations .
For instance, an Earth ellipsoid serves as a reference for determining
the actual Earth surface (the geoid). Kepler ellipses serve as references
for determining deviations of planetary or satellite orbits from such
ellipses. (The Traffic Code is a reference model for actual traffic (in
order to be able to define and punish deviations), but this, of course,
is beyond mathematics and even science.)
Mathematical models reach from well–established, purposely sim-
plified mathematical laws to mathematical working hypotheses. Ex-
amples for the first type are Kepler’s laws and e.g., the ideal gas laws
in thermodynamics; examples for the second type are the quantum me-
chanical model of mind–brain interaction of Beck and Eccles (Eccles
1994, Chapter 9) and the overdetermined and underdetermined linear
mathematical models at the end of sec. 3.8.
Penrose’s classification. Penrose (1989, p. 152) gave a very useful
classification of physical theories. Such a theory may be
SUPERB,
USEFUL, or
TENTATIVE.
6.5. LAWS OF NATURE 261

SUPERB are, for instance, classical mechanics (within the usual limits
of applicability defined by the theories mentioned next), special and
general relativity, and quantum mechanics. USEFUL are, e.g., the
gauge theories in elementary–particle physics, in particular the “Stan-
dard Model” (sec. 3.6). TENTATIVE is, e.g., superstring theory as a
possible candidate for a “Theory of Everything” (sec. 6.6).

What is a law of nature?

What do we mean by a law of nature? We have asked this ques-
tion and found many partial answers in sec. 3.9 and elsewhere, and in
particular, of course, in the present section. We keep asking the same
questions over and over again!
Never mind, let us keep insisting: if the poet speaks of the “eternal
laws of nature”, if the reductionist wishes to reduce biology to the
“laws of physics”, what are these laws? All “professional” answers
we have found sound unconvincing. A theory confirmed by induction
or surviving falsification is a human construct rather than a law of
nature, not even of physics: it is a law of man containing, as we hope,
some objective elements. According to Kuhn, it is a currently accepted
paradigm, some kind of “scientific fashion”.
As we have said above, a law of nature, in the ideal case, has an
objective and a subjective component, and it is only approximate.
Probing obstinately further, we do not wish the subjective compo-
nent. We wish a law of nature to be a law of nature. Einstein objected
to quantum mechanics for this very reason.
Let us try to view the question from another angle. Can we suppose
that nature has its own laws? I think all scientists believe this, at
least subconsciously. Without this belief there would be no physics,
no induction, no verification, no falsification, not even a reasonable
paradigm. To state it bluntly, we believe that nature, or God (Spinoza:
deus sive natura ), does have its own laws, which are essentially the
same as they were before the appearance of man. In terms of James
Gleick’s motto at the beginning of the present section, we believe that
God does have the original equations, even if He does not provide them
to us. Nature is ruled by its own laws, and we try to find them, at least
approximately. Our laws of nature are approximations to God’s laws
of nature, and our laws of physics are approximations to God’s laws
of physics. If you do not like this very “unreductionist” terminology,
replace “God” by “nature”.
Is this belief justified? Clearly, it cannot be proved. Quantum the-
ory even seems to flatly contradict it. In sec. 3.5 we have seen, however,
262 CHAPTER 6. IMPLICATIONS OF SCIENCE

that “objective” interpretations of quantum mechanics may not be im-

possible. Following Kant, many philosophers (Neo–Kantians, perhaps
the late Wittgenstein) and scientists (Wheeler, Maturana, Varela) seem
to question the existence of an external world independent of the ob-
server.
So can we believe that purely “objective” laws of nature exist? I
think so. The main argument: We are part of nature, and our thinking
should be at least in basic agreement with nature (otherwise we might
have been eliminated by evolution long ago). More philosophically,
Spinoza says: “Ordo et connexio rerum idem est ac ordo et connexio
idearum ”, the order and connection of things are the same as the order
and connection of ideas. Again, we recall the quotation from Penrose
at the end of sec. 3.9.
Many things in philosophy, in daily life and in science must be
believed. So why not believe in the existence of an external world gov-
erned by laws? This belief cannot be proved, but it is not irrational:
it has never been contradicted by experience. Science, rather than
disproving our naı̈ve faith, has refined it: the external world may be
essentially (following Kant, for us) three–dimensional, but the correc-
tions due to quantum theory (possible subjective elements, nonlocality,
etc.) and perhaps to some higher–dimensional “super–theory” should
be accepted.

The complexity of the concept of physical law

It may be appropriate finally to compile some problems concerning
laws of nature, in particular of physics, which we have noticed so far.
(1) How is a law found?
– induction, verification, falsification (sec. 3.9)
– the “unreasonable efficiency of mathematics”: the role of
mathematical simplicity, symmetry, and elegance (secs. 3.6,
3.9)
– the role of conventional paradigms (sec. 3.10)
(2) The character of a law
– a priori vs. a posteriori (secs. 1.4, 5.3, 5.4)
– objective and subjective elements: objective truths vs. sub-
jective perspectives (secs. 3.5, 5.3)
– positivism or metaphysics: description vs. explanation
6.6. THEORIES OF EVERYTHING 263

– Do the laws remain unchanged throughout space and time?

– Are laws exact? Do they contain probabilistic elements or
uncertainties of Gödel or Heisenberg type?
– Are our laws at least approximations to unknown (and pos-
sibly unknowable) “exact laws”?
These and many other questions arise. Some may be answered, some
not. At any rate, there does not seem to exist a unique answer to
all these questions. A “law of nature” is a “chameleon concept” (J.R.
Lucas), it subtly changes its meaning during the course of a discussion,
and even more so during the course of a book like the present one. So to
speak, this is a “skeleton in the cupboard” hidden in almost any book
on science or philosophy of science, and it is best to open the cupboard
and study the skeleton.
Let reductionist biologists look for simple physical laws underlying
biological complexity: philosophers of science will continue to be con-
cerned about the conceptual complexity underlying the very meaning
of a “simple” physical law. Both parties treat meaningful problems in
meaningful ways.
A modern treatment of laws of nature, simplicity and complexity,
reductionism and emergence is found in (Cohen and Stewart 1994).
Also (Barrow 1988) can be recommended. Writings of great scientists
about laws of nature are always fascinating: (Eddington 1939, 1959),
(Einstein 1957), (Heisenberg 1955, 1958, 1973), and many others quoted
elsewhere in this book.

6.6 Theories of everything

Dear Reader, you have before you a Theory of
Natural Philosophy deduced from a single
law of Forces.
Rudjer Bošković

Your theory is crazy, but probably not crazy enough

to be true.
Niels Bohr

An old dream of theoretical physicists, essentially the dream of

“Laplace’s demon” (sec. 3.1), is to find the “world formula”, combin-
ing relativity and quantum theory and unifying the four basic forces of
physics:
264 CHAPTER 6. IMPLICATIONS OF SCIENCE

gravitation,
electromagnetism,
weak force,
strong force.

Gravitation and electromagnetism are more or less familiar. For the

rest, let us briefly recapitulate from the rather technical Section 3.6. A
free neutron disintegrates within several minutes, leaving a proton, an
electron, and a neutrino. For such and many other transmutations
of particles (“radioactive decay”), the weak force is responsible. An
atomic nucleus consists of protons and neutrons. The electromagnetic
repelling force between the positively charged protons must be over-
come in order to keep the nucleus together. This is done by the strong
force. The strong force similarly acts between the quarks that make up
protons and neutrons.
Already Einstein had dreamed of the unification of gravitation and
electromagnetism (the only forces known at that time) by modifying
the equations of general relativity. A very interesting theory was found
by Hermann Weyl as early as 1918; it served as the ancestor of modern
gauge theories.
Almost all famous theoretical physicists tried their hand at finding
unified theories. Einstein worked for most of his life on this problem,
unfortunately in vain, but also Eddington in 1923 and Schrödinger in
1950 were fascinated by this question.
Whereas these scientists based their efforts on general relativity,
Heisenberg in 1959 tried to proceed from quantum theory. Also his
attempt failed.
In the meantime, C.N. Yang and R.L. Mills had formulated the first
gauge theory in 1954, generalizing Weyl’s ideas. Since then, gauge the-
ories have become basic mathematical tools for describing elementary
particles.
The first unified gauge theory for electromagnetic and weak forces
by A. Salam, S. Weinberg, and S. Glashow brought its authors the
Nobel Prize in 1979. Contemporary unified gauge theories include, in
addition, also the strong force. Thus we have the Standard Model as
a preliminary but extremely useful approximation to an ideal GUT, or
Grand Unified Theory.
6.6. THEORIES OF EVERYTHING 265

Thus three of the four forces have been unified. The great excep-
tion is gravitation, which somehow refuses being included in an ordi-
nary gauge theory. An extension of the latter, supersymmetry, may
provide a possibility to include gravitation as well. A final refinement
(and complication) is achieved by regarding elementary particles not
as pointlike but as an extremely small (∼ 10−13 cm) curve, or loop, a
“string”.
These strings can be combined with supersymmetry to give super-
string theory. This theory (or its variants) is so complicated that the
mathematical consequences have not yet been elaborated. Thus it has
not yet been possible to really understand whether and how well this
theory applies to the real world. With superstring theory we are now
(1994) so far as with Heisenberg’s unified theory in the sixties.
Excellent Rabbi type 1 references on these fascinating topics are
(Barrow 1991), (Davies 1984), (Davies and Brown 1988), (Hawking
1988, 1993), and (Kaku 1994). We again refer also to sec. 3.6.
Rudjer Bošković (1711–1787). “One of the most remarkable and
neglected figures in the history of modern European science was Roger
Boscovich” (Barrow 1991, p. 17). A Jesuit from Dubrovnik, he worked
in philosophy, mathematics, physics, geodesy, and similar sciences
which at that time were not yet so separated and specialized as to-
day. Recognized by Niels Bohr and Werner Heisenberg, he nevertheless
is little known outside Croatia. He proposed a grand unified force law,
generalizing Newton’s law of gravitation to include all other physical
forces. He believed in atoms and elementary particles and tried to find
a single law encompassing gravitation as well as the forces that would
hold his hypothetical elementary particles together. “His continuous
force law was the first scientific Theory of Everything” (Barrow 1991,
p. 18).
Bošković is important not so much for his particular contributions to
science, but he was a visionary whose ideas influenced great physicists
such as Faraday, Maxwell, Boltzmann, and Lord Kelvin. Some of his
philosophical ideas are relevant even today, cf. sec. 4.4 (p. 186) and
sec. 5.2 (p. 210).
Philosophical problems. A future TOE (Theory Of Everything )
poses very difficult and sometimes contradictory problems which are,
however, profound and fascinating.

(1) In the ideal case, it would describe and predict all physical
processes, however complex and “chaotic”: laws together with
boundary conditions.
266 CHAPTER 6. IMPLICATIONS OF SCIENCE

(2) If chemistry and biology are, in fact, nothing but theories of par-
ticularly complex physical systems, a TOE must also fully explain
chemistry and biology, including “self–organization”.
(3) It must incorporate the uncertainty principles of Gödel and
Heisenberg.
(4) If mental brain activities were, in reality, only physico–chemical
processes, then TOE would also describe human thinking; the
“freedom of the will” would then be nothing but an illusion.
(5) The TOE, predicting everything, must also derive itself .
It is clear that all this appears quite improbable. We would then
have Laplace’s demon (sec. 3.1) again, in contradiction to chaos theory
and Heisenberg’s uncertainty relation.
As we have seen in sec. 4.5, a reduction of biology to physics may
work in this “analytic” direction. In the opposite (“synthetic”) direc-
tion it probably will not be possible because information is needed, in
agreement with the “equation”
life = matter + information .
As a TOE, at least as it appears now, does not contain this “vital”
information, “synthesis” and prediction appear more or less impossible.
So the freedom of the will does not seem to be in danger, not even
from a TOE. Even if theoretically, provided we knew the required ini-
tial conditions fully and with absolute precision, we could calculate
human thinking, this cannot be done practically because the amount
of calculation exceeds all that is humanly possible. Furthermore, if we
could exactly predict our future decisions, they would be known, and
this knowledge might change our future thinking and acting! This in-
genious argument was given by Hawking (1993, p. 135). In fact, this
“pre–cognition” would constitute an inadmissible time loop (sec. 3.7,
Fig. 3.20 on p. 129).
Finally, even if a TOE were able to derive itself, how would we know
that it is true? Of course, in this case, the TOE would be internally
consistent, but how do we know that it corresponds to external reality?
Hawking’s answer is Darwin’s theory of evolution: animals behaving
not in agreement with reality, and early man having wrong “theories”
about reality are unfit to survive and will be eliminated by evolution (cf.
sec. 1.4). Interesting as this argument is (we have used it frequently),
I should not consider it a philosophically fully satisfactory answer.
6.6. THEORIES OF EVERYTHING 267

This is related to the argument of Epicurus (sec. 6.4); and also the
argument of Lucas (ibid. ) is of direct relevance here.
Let us therefore ask again: “Is everything predetermined?” Assum-
ing that a TOE exists, Hawking (1993, p. 139) finally says: “The answer
is yes, it is. But it might as well not be, because we can never know
what is determined.” Isn’t this a beautiful example of a profound truth
(of which the logical contrary is also a profound truth) in the sense of
Bohr (sec. 2.5)?
A final, and presumably conclusive, argument against complete pre-
determination comes from the Heisenberg uncertainty relation. Since a
TOE must comprise quantum theory, it must also contain this uncer-
tainty relation which is an ineluctable consequence of quantum theory
(sec. 3.5). Cf. also (Gell–Mann 1994) and (Weinberg 1993).
Relation to dialectics. As we have seen, a true TOE must derive it-
self . This is another instance of the important concept of self–reference:
As we have seen at the end of sec. 2.5, the dialectic system of Hegel’s
logic provides such a self–reference or self–derivation; cf. also (Speiser
1952, p. 110). This is possible in Hegel’s informal thinking, but not so
readily in the formal structures of contemporary logic and mathemat-
ics. It would have to be a self–referential deductive system (Wheeler
1994, p. 309).
In the light of these remarks it seems unlikely that we shall ever
have a true TOE in the strict sense of the contradictory requirements
(1) to (5) given above. What can happen if no TOE exists at all?
Hawking (1988, p. 166) gives two alternatives.
(1) There is no final theory of the universe, only an infinite sequence
of theories which better and better describe the universe. This is the
present view of those physicists who do not believe in a universal TOE.
(2) There is no theory of the universe. The physical laws are valid
only to a certain amount of accuracy; below this accuracy, events occur
in a spontaneous and random fashion. This would be a general instance
of the dialectics of freedom and necessity (sec. 2.5); this possibility has
also been envisaged by Whitehead (1933, Chapter VII). We are also
reminded of Wheeler’s “no laws” (sec. 6.5).
Bootstrapping. Remarkably enough, superstring theory seems in
fact to be able to “derive itself” in the form of “bootstrapping”, cf. end
of sec. 2.5: provided the system of elementary particles (as predicted
by superstring theory) exists, it produces itself. This is the bootstrap
principle of self–consistency (Gell–Mann 1994, Chapters 10 and 14); cf.
also (Capra 1976, Chapter 18).
268 CHAPTER 6. IMPLICATIONS OF SCIENCE

Logical singularities
Self–reference and self–derivation may lead to logical singularities.
Examples have been given at the end of sec. 5.3 (especially Examples
1 to 3).
Logical singularities occur if the observer (or theoretician) A
(α) coincides with the object B or
(β) is part of the object B.
Case (α) has been seen to apply Plotin’s definition of human think-
ing: The thinking thinks the thinking (sec. 2.5). Case (β) corresponds
to B being the whole universe, of which A necessarily is a part (from
Kant’s antinomies to TOE). It is illustrated by Fig. 6.4.

Figure 6.4: Subject A lies (a) outside object B, and (b) inside uni-
verse B
(For specialists only.)A simple model for such a logical singularity is
provided by the mathematical singularity of the gravitational potential
of an extended body B, well known to physicists and geodesists. The
Newtonian potential V at point A is
ZZ Z
ρ
V (A) = G dv , (6.2)
l
B

where G is the gravitational constant, ρ is the density, dv is the volume

element, and l denotes the distance of dv from A; the integral is taken
over the whole body B (Fig. 6.5). Since in case (b), A may coincide
with dv so that l = 0, formula (6.2) has a singularity at l = 0 for an
internal point A, whereas it is regular for an external point, case (a).
In fact, both cases have quite a different behavior:

(a) ∂2V + ∂2V + ∂2V = 0 Laplace’s equation ,

∂x2 ∂y 2 ∂z 2

(b) ∂2V + ∂2V + ∂2V = −4πGρ Poisson’s equation .

∂x2 ∂y 2 ∂z 2
6.7. THE ABSOLUTE 269

Figure 6.5: The Newtonian potential (a) for an external point A, and
(b) for an internal point A.

This model comes entirely natural to geodesists and mathematical

physicists, so it is given here. Readers from other disciplines can safely
disregard it. (Specialists may, however, recognize equation (b) as the
classical analogue to Einstein’s equation (3.55) on p. 96, cf. (Moritz
and Hofmann–Wellenhof 1993, p. 232).)

6.7 The Absolute

I want to know how God
created this world.
Albert Einstein

Almost all metaphysical systems contain the concept of an Absolute,

from Plato to Whitehead. He may be the creator of the world, according
to the Jewish–Christian–Muslim tradition, he may be the Prime Mover
and the supreme “Final Cause” of Aristotle and Teilhard de Chardin, he
may be the metaphysical link between man and nature of Descartes, the
Supreme Monad of Leibniz, the great Mechanical Engineer of Newton,
the supreme Moralist of Kant, the Absolute Idea of Fichte and Hegel; or
the origin of order, structure and harmony, the Supreme Mathematician
of Plato, Kepler, Whitehead, and Hawking (1988, concluding sentence).
Generally we may distinguish in God
– transcendence, and
– immanence.
A transcendent God is fully above or outside the world (Jahwe, Allah).
An immanent God is fully inside the world; this is the doctrine of pan-
theism (Spinoza, Hegel, dialectical materialism: matter is the Absolute,
see above).
270 CHAPTER 6. IMPLICATIONS OF SCIENCE

The Christian God is partly transcendent (the creator, the Father)

and immanent (the redeemer, the Son, the “fellow sufferer who under-
stands” (Whitehead)).
The biblical concept of God has developed “from force to persua-
sion” (Whitehead 1933). This is evident in a famous passage from the
1st Book of Kings:
The LORD said, “Go out and stand on the mountain in the presence of the
LORD, for the LORD is about to pass by.” Then a great and powerful wind
tore the mountains apart and shattered the rocks before the LORD, but the
LORD was not in the wind. After the wind there was an earthquake, but
the LORD was not in the earthquake. After the earthquake came a fire, but
the LORD was not in the fire. And after the fire came a gentle whispering
breeze. When Elijah heard it, he pulled his cloak over his face and went out
and stood at the mouth of the cave.
Whitehead (1929, Part V, Chapter II) gives the following beautiful
antitheses:
It is as true to say that God is one and the World many, as that the World
is one and God many.
It is as true to say that the World is immanent in God, as that God is
immanent in the World.
It is as true to say that God transcends the World, as that the World
transcends God.
These are perhaps the finest examples of “profound truths” in the
sense of Niels Bohr. As a matter of fact, it is an established tradition
in mysticism to speak of God in antitheses (cf. Huxley 1970). This is
comparable to the paradoxes of the Infinite as outlined in sec. 2.3.
Curiously enough, bootstrapping (secs. 2.5 and 6.6) occurs also here,
namely in the so–called ontological argument for the existence of God.
It goes back to medieval philosophy, to Anselm of Canterbury (sec. 5.4).
With considerable oversimplification, it may be formulated as follows:
By definition, God is absolutely perfect in every respect. Since an exist-
ing being is more perfect than a non–existing one, absolute perfection
entails existence. In this way, profanely speaking, God “bootstraps”
Himself into existence.
The ontological argument has a colorful history. It was rejected
already by St. Thomas Aquinas but revived by Descartes, Spinoza and
Leibniz, rejected by Kant, revived by Fichte and Hegel, and rejected
again by Schopenhauer.
A modern analysis was performed by Findley (1963, Introduction)
and Ch. Hartshorne: If God’s existence is any way possible, then it is
6.7. THE ABSOLUTE 271

also certain that God exists. But lest believers rejoice too early, the
argument can be inverted: if God’s nonexistence is in any way possible,
it is certain that God does not exist!
This is indeed a logically fascinating paradox of the Infinite, and
the little book (Plantinga 1965) is as intellectually rewarding as (Nagel
and Newman 1958).
So far we have considered the concept, or concepts, of God mainly
from a philosophical and logical point of view. So far it has been
independent of revealed religion. This distinction between philosophy
and theology (religion) was made already by St. Thomas Aquinas. An
excellent, comprehensive and objective treatment of both points of view
(philosophy and religion) was given by Küng (1978); an outstanding
concise, clear and understandable summary is found in the last chapter
of (Bocheński 1959).
Science and religion. It is a commonplace that science and religion
do not contradict each other because they are dealing with different
subject matters. Nevertheless, they have a common boundary, and
wars are frequently waged about boundaries.
What should be rejected is the doctrine of two separate truths, one
for science and one for religion. In fact, the synthesis between science
and religion is not a cheap compromise, but requires considerations on
a rather high level.
Such high–level efforts can be found, for example, in the writings
of Whitehead and of Weizsäcker, and on the Catholic side of Teilhard
de Chardin (1955), and in particular of Karl Rahner. Cheap popular-
izations frequently do more harm than good. Tipler’s (1994) ingenious
and ingenuous attempt to develop a theology on the basis of physics
is probably a singularity, although an interesting one. (Davies 1983)
and (Ferguson 1994) are more realistic. (Kolakowski 1982) is a fine
counterpoint to (Küng 1978) by a former Marxist.
A chief stumbling–block has been the theory of biological evolution.
The theory that man has evolved from animal predecessors was consid-
ered incompatible with man’s immortal soul and hence with Christian-
ity. This apparent contradiction was used by both sides to fight the
other, with most unfortunate results.
In fact, even if man has evolved in this way (which no serious sci-
entist would doubt nowadays), man’s intellect is such that he is clearly
separated from the other animals. Thus, he seems to be the only ani-
mal capable of thinking about himself. This “thinking about thinking”
is self–consciousness, cf. (Eccles 1989).
Another really stupid stumbling–block has been the apparent con-
272 CHAPTER 6. IMPLICATIONS OF SCIENCE

tradiction between the creation in seven days according to the Bible

and the history of evolution which took billions of years. Most unfor-
tunately, theologians did not do their homework well: had they read
St. Augustine, they would have found that the biblical days should not
be taken literally: “With the Lord, a day is like a thousand years, and
a thousand years is like a day” (2nd Letter of St. Peter). In fact, thus
interpreted, we have a remarkable similarity of present–day scientific
thinking and the biblical report:

BIBLE SCIENCE
creation “big bang”
light (1st day) energy
Heaven and Earth (2nd day) stars, planets, Earth
oceans, oceans,
vegetation (3rd day) vegetation
Sun and Moon, day Earth rotation,
and night (4th day) formation of the Moon
fish and birds (5th day) animals
mammals, man (6th day) mammals, man

With the possible exception of the 4th day, the order is almost exact;
cf. (Weizsäcker 1973).
As we believe today, evolution is not restricted to biology, but
started right after the “big bang”. First hydrogen was formed, then
helium was synthesized, and heavier elements followed. Galaxies and
stars were formed. Stellar evolution is now standard in astronomy. The
evolution of the planetary system was discussed already by Laplace and
Kant.
The history of mankind and the story of the bible fit beautifully into
this picture of a thoroughly evolutionary universe. Christianity was
always proud of its historical evolution, from the first man (“Adam”
simply means “man”!) to Moses, King David, and Jesus. Not to see
that this historical development fits perfectly into general evolution
seems, in hindsight, an almost unbelievable shortsightedness. Teilhard
de Chardin (1955) was one of the first Catholics to work this into a
consistent picture, and he had to pay for it. Nowadays, of course, this
is no problem any more.
Can science help prove the existence of God ? When Napoleon asked
Laplace why he did not mention God in his famous book on celestial
mechanics, Laplace answered: “Sire, I do not need this hypothesis”. Of
course, he was right from the point of view of classical mechanics: God
6.7. THE ABSOLUTE 273

may have installed the giant mechanical clockwork of the universe, but
since then it continued to run automatically according to the rigorous
laws of mechanics. Divine interference was no longer needed in classical
mechanics.
Other scientists and philosophers thought differently: Kepler ad-
mired the divine harmony of the universe, and so did Newton. Leibniz
tried hard to reconcile science and religion. Kant deduced the existence
of God and the immortality of the soul from the moral law to which all
men are subject. Goethe, following Spinoza, sought God in nature and
art, Einstein in the wonderful mathematical laws of nature (“Science
without religion is lame, religion without science is blind” (Einstein
1954, p. 46). Heisenberg and Weizsäcker are Evangelic Christians.
Also to me it is impossible to believe that a beautiful mountain
flower or Riemann’s formula for the distribution of prime numbers are
pure products of “chance” or “natural selection”.
Biologists, discussing evolution, frequently speak of the “creativ-
ity” or the “inventiveness” of evolution. The physicist Freeman Dyson
(quoted from (Kaku 1994, p. 258)) writes:
As we look out into the Universe and identify the many accidents of physics
and astronomy that have worked together to our benefit, it almost seems as
if the Universe must in some sense have known that we were coming.
Here obviously to evolution and to the Universe we find ascribed prop-
erties of the Absolute . . .
It is generally believed that the increasing complexity of evolution
implies an increase of information (or a decrease of entropy). Does this
imply a continuous supply of information “from the outside”, some kind
of “continuous creation”? Tresmontant (1976) and others think so.
The old clash between religion and science was largely due to White-
head’s “fallacy of misplaced concreteness”. The exaggerated “spiritual-
ism” of religion seemed to be incompatible with the exaggerated “ma-
terialism” of science. As we have seen, materialism no longer reigns
supreme in modern physics. In relativity and quantum theory, science
has proved to be infinitely more complex and counterintuitive than the
most daring speculations of theologians and metaphysicists.
In fact, the main importance of modern science for religious thinking
may be the fact that it has provided highly counterintuitive and strange
models which make the old difficulties of religion (miracles etc.) look
harmless in comparison. No one any longer believes that God really
stopped the movement of the Sun in order to permit Joshua to win his
battle against the Amorites in daylight: the bible is a religious work and
274 CHAPTER 6. IMPLICATIONS OF SCIENCE

not a textbook on Earth rotation! Also the allegoric interpretation of

Scripture, already advocated by St. Augustine(!), is now being generally
accepted in theology.
There is one real contribution of the hard theological work of the
Fathers of the Church to contemporary thinking: the theory of the
Trinity. Plotinus’ statement “The thinking thinks the thinking” has
been applied to God as follows: The thinking (1st person) thinks (3rd
person) the thinking (2nd person). The Holy Trinity served not only as
a basic model for mutual immanence which plays a fundamental role in
the theory of internal relations (Whitehead 1933, Chapter X, sec. IV;
1925, Chapter XI). It also provided the prototype of a dialectic triad:
thesis, antithesis, and synthesis.
Questions of the purpose of man, of the meaning of existence, and
of ethics and care for coming generations have been thought to be be-
yond science. Now, in view of the nuclear threat, the destruction of
the environment, and the possibility of genetic manipulation we start
to think differently. I do not feel qualified to treat these supremely im-
portant questions here and refer the reader to the enormous literature,
e.g. (Fischer 1993) and (Koltermann 1994).

6.8 Pluralism
Die Wahrheit ist symphonisch.
Hans Urs von Balthasar

Scientific pluralism is the peaceful coexistence of (mildly) conflicting

theories such as classical mechanics, relativity and quantum theory, as
well as of their interpretations, which are particularly controversial in
quantum mechanics (sec. 3.5). Nevertheless we may speak of a general
consensus between scientists.
The first impression on reading philosophy, however, is a bewildering
number of apparently contradictory philosophical systems and “isms”.
There is practically nothing in common, say, between Heidegger and
Russell. Which of these systems is true?
The answer is: none and all (another “profound truth” in the sense
of Bohr!). In fact, every reasonable system of philosophy offers an
aspect or perspective of reality, a “theory” in the sense of Bohm: a way
of looking at our world. In Greek “theoria ” means a way of looking, a
perspective, as we have seen in sec. 6.5. Reality is far too complicated
6.8. PLURALISM 275

to fit into one single system, even too complicated to fit into a common
physical theory.
A high mountain looks different as we regard it from north, south,
east, or west. Which one is the “correct” perspective? Is the view from
north “true” and the view from east “false”? Should the observer from
south critize the viewer from the west?
In philosophy, however, it is customary for the adherents of one
system to critize their colleagues from the other systems. Such criti-
cism is necessary in order to make the reasoning logically stringent and
convincing. It should be an objective and fair criticism, preceded by
intense and sympathetic study of the other’s complete reasoning.
Superficial criticism based on sentence–by–sentence refutation is
easy and cheap. It is not difficult to condemn existentialism on the
basis of logical positivism and vice versa. A particularly shocking mis-
understanding regarding Fichte was mentioned in sec. 5.4.
It is quite improbable that great philosophical systems are com-
pletely erroneous. They at least contain some important truth or in-
sight.
My general experience with philosophical criticism is that positive
opinions of philosophers about other philosophers seem usually right,
and that negative judgments of philosophers about other philosophers
seem usually wrong. Why? I don’t know, it might be related to human
psychology.
All this speaks in favor of tolerance and scientific and philosoph-
ical pluralism. The plurality of philosophical systems is as necessary
for human understanding as the plurality of musical instruments is for
an orchestra. Figuratively speaking, Plato’s violin is, of course, om-
nipresent. But Fichte’s flute and Hegel’s trumpet add color, and the
cello of Kant and the contrabass of logical positivism provide a reliable
basis. The exotic instruments of Gödel and Wittgenstein are used if
the orchestra is to play Schönberg and Webern . . .
Thus Hans Urs von Balthasar says (see the motto of this section):

“TRUTH IS SYMPHONIC”.
276 CHAPTER 6. IMPLICATIONS OF SCIENCE
Selected additional reading

It is quite clear to me that you will probably neither be able to read all the
books given, nor even wish to do so. It is a rich menu, from which you are
invited to select at your convenience.
You are asking for a suggestion? If you can read only one book, I suggest
(Davies 1988), if you have time for two books, take also (Whitehead 1925),
and as a third, (Penrose 1989), which is tougher but extremely rewarding.
My next candidates on the reading list would be (Hofstadter 1979), (Cohen
and Stewart 1994), and (Popper 1982). Further recommendations are found
in the text of the book.
A triad of very readable and mutually complementary introductory book-
lets on “pure” philosophy is (Bocheński 1959), (Jaspers 1953) and (Russell
1912).
As far as available to the author, book titles have been given in English.
Several German and French books are available in an English translation
(and vice versa).
Abraham R H and Shaw C D (1992) Dynamics: The Geometry of Behavior,
2nd ed, Addison–Wesley, Redwood City, Cal.
Anger G, Gorenflo R, Jochmann H, Moritz H, and Webers W eds (1993)
Inverse Problems: Principles and Applications in Geophysics, Tech-
nology, and Medicine, Akademie Verlag, Berlin (mathematical).
Barrow J D (1988) The World Within the World, Oxford Univ. Press.
Barrow J D (1991) Theories of Everything, Oxford Univ. Press.
Barrow J D (1992) Pi in the Sky, Oxford Univ. Press.
Blokhintsev D I (1968) The Philosophy of Quantum Mechanics, Reidel,
Dordrecht, Holland (mathematical but also of general interest).
Bocheński J M (1959) Wege zum philosophischen Denken, Herder,
Freiburg.
Bohm D (1980) Wholeness and the Implicate Order, Routledge, London.
Bohr N (1934) Atomic Theory and the Description of Nature, Cambridge
Univ. Press.
Bohr N (1958) Atomic Physics and Human Knowledge, Wiley, New York.
Bohr N (1963) Essays 1958–1962 on Atomic Physics and Human Knowl-
edge, Wiley, London. (All three volumes reprinted 1987 by Ox Bow
Press, Woodbridge, Connecticut.)

277
278 ADDITIONAL READING

Briggs J (1992) Fractals: the Patterns of Chaos, Simon and Schuster, New
York.
Čapek M ed (1976) The Concepts of Space and Time, Reidel, Dordrecht,
Holland.
Capra F (1975) The Tao of Physics, Wildwood House, London.
Carnap R (1950) Logical Foundations of Probability, Chicago Univ. Press
(for specialists).
Carnap R (1958) Introduction to Symbolic Logic and Its Applications,
Dover, New York (for specialists).
Carnap R (1966) Philosophical Foundations of Physics, Basic Books, New
York.
Churchland P M (1988) Matter and Consciousness, 2nd ed., MIT Press.
Cohen L J (1986) The Dialogue of Reason, Oxford Univ. Press.
Cohen L J (1989) An Introduction to the Philosophy of Induction and
Probability, Oxford Univ. Press.
Cohen J and Stewart I (1994) The Collapse of Chaos: Discovering Sim-
plicity in a Complex World, Viking–Penguin, New York.
Copleston F (1946) A History of Philosophy, 9 vols., Burns and Oates,
London.
Davies P (1983) God and the New Physics, Simon and Schuster, New York.
Davies P (1984) Superforce: The Search for a Great Unified Theory of
Nature, Simon and Schuster, New York.
Davies P (1988) The Cosmic Blueprint, Simon and Schuster, New York.
Davies P (1992) The Mind of God: The Scientific Basis for a Rational
World, Simon and Schuster, New York.
Davies P and Brown J R (1986) The Ghost in the Atom, Cambridge Univ.
Press.
Davies P and Brown J R (1988) Superstrings: A Theory of Everything?
Cambridge Univ. Press.
Ditfurth H von (1976) Der Geist fiel nicht vom Himmel, Hoffmann und
Campe, Hamburg.
Duncan R and Weston–Smith M eds (1977) The Encyclopedia of Ignorance,
Pergamon, Oxford.
Eccles J C (1973) The Understanding of the Brain, McGraw–Hill, New
York.
Eccles J C (1989) Evolution of the Brain: Creation of the Self, Routledge,
London.
Eccles J C (1994) How the Self Controls Its Brain, Springer, Berlin.
Eddington A (1939) The Philosophy of Physical Science, Cambridge Univ.
Press.
Eddington A (1959) New Pathways in Science, Ann Arbor Paperback, Univ.
of Michigan Press.
Edelman G M (1989) The Remembered Present: A Biological Theory of
Consciousness, Basic Books, New York.
ADDITIONAL READING 279

Edey M A and Johanson D C (1989) Blueprints: Solving the Mystery of

Evolution, Penguin Books, New York.
Eigen M (1987) Stufen zum Leben, Piper, München.
Eigen M and Winkler R (1975) Das Spiel, Piper, München.
Einstein A (1954) Ideas and Opinions, Crown, New York.
Einstein A (1957) Mein Weltbild, Ullstein, Berlin.
Ferguson K (1994) The Fire in the Equations, Bantam Press, London.
Findley J N (1958) Hegel: A Re–examination, Allen and Unwin, London.
Findley J N (1963) Language, Mind and Value, Allen and Unwin, London.
Fischer H (1993) Plädoyer für eine sanfte Chemie, C.F. Müller, Karlsruhe.
Folse H J (1985) The Philosophy of Niels Bohr, North Holland, Amster-
dam.
Gardner M (1982) The Ambidextrous Universe, 2nd ed, Penguin, London.
Gell–Mann M (1994) The Quark and the Jaguar: Adventures in the Simple
and the Complex, Freeman, New York.
Gierer A (1985) Die Physik, das Leben und die Seele, Piper, München.
Gilson E (1972) L’être et l’essence, 2nd ed., J. Vrin, Paris.
Glass L and Mackey M C (1988) From Clocks to Chaos, Princeton Univ.
Press.
Gleick J (1988) Chaos: Making a New Science, Cardinal–Penguin, London.
Globus G G, Maxwell G, and Savodnik I eds (1976) Consciousness and the
Brain, Plenum Press, New York.
Gnedenko B V and Khinchin A Ya (1962) An Elementary Introduction to
the Theory of Probability, Dover, New York.
Götschl J ed (1992) Erwin Schrödinger’s World View, Kluwer, Dordrecht,
Holland.
Gribbin J (1984) In Search of Schrödinger’s Cat, Wildwood House, London.
Gulyga A (1990) Die klassische deutsche Philosophie, Reclam, Leipzig.
Hadamard J (1945) The Psychology of Invention in the Mathematical Field,
Princeton Univ. Press.
Haken H (1981) Erfolgsgeheimnisse der Natur: Synergetik, 2nd ed.,
Deutsche Verlags–Anstalt, Stuttgart.
Haldane J B S (1939) The Marxist Philosophy and the Sciences, reprint
by Books for Libraries Press, Freeport, New York.
Halmos P R (1960) Naive Set Theory, Van Nostrand, Princeton (mathe-
matical).
Hartmann N (1960) Die Philosophie des deutschen Idealismus, 2nd ed., W.
de Gruyter, Berlin.
Hartmann N (1965) Grundzüge einer Metaphysik der Erkenntnis, 5th ed.,
W. de Gruyter, Berlin.
Havemann R (1964) Dialektik ohne Dogma? Naturwissenschaft und
Weltanschauung, Rowohlt, Hamburg.
Hawking S (1988) A Brief History of Time, Bantam Press, London.
280 ADDITIONAL READING

Hawking S (1993) Black Holes and Baby Universes and Other Essays,
Bantam Press, London.
Hawking S W and Ellis G F R (1973) The Large Scale Structure of Space
Time, Cambridge Univ. Press (mathematical).
Heintel E (1990) Die Stellung der Philosophie in der “Universitas Litter-
arum”, Publishing House of the Austrian Academy of Sciences, Wien.
Heisenberg W (1955) Das Naturbild der heutigen Physik, Rowohlt, Ham-
burg.
Heisenberg W (1958) Physics and Philosophy, Harper & Row, New York.
Heisenberg W (1973) Der Teil und das Ganze, Piper, München.
Hempel C G (1966) Philosophy of Natural Science, Prentice–Hall, Engle-
wood Cliffs, N.J.
Hofstadter D R (1979) Gödel, Escher, Bach: an Eternal Golden Braid,
Basic Books, New York.
Hofstadter D R (1985) Metamagical Themas: Questing for the Essence of
Mind and Pattern, Basic Books, New York.
Hofstadter D R and Dennett D C, eds (1981) The Mind’s I, Basic Books,
New York.
Holzmüller W (1984) Information in Biological Systems: The Role of
Macromolecules, Cambridge Univ. Press.
Hubel D H (1988) Eye, Brain, and Vision, Scientific American Library, New
York.
Huxley A (1970) The Perennial Philosophy, Harper & Row, New York.
Jantsch E (1980) The Self–Organizing Universe, Pergamon, Oxford.
Jaspers K (1953) Einführung in die Philosophie, Piper, München.
Jaspers K (1962–1993) The Great Philosophers, Harcourt Brace Jo-
vanovich, New York.
Jeffreys H (1961) Theory of Probability, 3rd ed., Oxford Univ. Press (math-
ematical).
Jeffreys H (1973) Scientific Inference, 3rd ed., Cambridge Univ. Press.
Jordan P (1968) Der Naturwissenschaftler vor der religiösen Frage, 5th ed.,
Stalling, Oldenburg.
Kaku M (1994) Hyperspace: A Scientific Odyssey Through Parallel Uni-
verses, Time Warps, and the 10th Dimension, Oxford Univ. Press.
Kauffman S A (1993) The Origins of Order: Self–Organization and Selec-
tion in Evolution, Oxford Univ. Press.
Kohonen T (1988) Self–Organization and Associative Memory, 2nd ed.,
Springer, Berlin.
Kolakowski L (1982) Religion — If there is no God, Fontana Paperbacks,
London.
Koltermann R (1994) Grundzüge der modernen Naturphilosophie, Knecht,
Frankfurt a. M.
Kosko B (1993) Fuzzy Thinking, Hyperion, New York.
ADDITIONAL READING 281

Kuhn T S (1970) The Structure of Scientific Revolutions, 2nd ed., Chicago

Univ. Press.
Küng H (1978) Existiert Gott? Piper, München.
Küppers B O ed (1987) Leben = Physik + Chemie? Piper, München.
Kuznecov B G (1981) Philosophie – Mathematik – Physik, Akademie–
Verlag, Berlin.
Lanczos C (1965) Albert Einstein and the Cosmic World Order, Wiley–
Interscience, New York.
Lanczos C (1970) Space Through the Ages, Academic Press, London
(rather mathematical).
Leff H S and Rex A F eds (1990) Maxwell’s Demon, Adam Hilger, Bristol.
Lewin R (1992) Complexity: Life at the Edge of Chaos, Macmillan, New
York.
Lindsay R B and Margenau H (1957) Foundations of Physics, Dover Pub-
lications, New York (rather mathematical).
Lockwood M (1989) Mind, Brain and the Quantum, Basil Blackwell, Ox-
ford.
Lorenz E N (1993) The Essence of Chaos, Univ. of Washington Press,
Seattle.
Lorenz K (1973) Die Rückseite des Spiegels, Piper, München; English trans-
lation: Behind the Mirror, Methuen, London, 1977.
Lucas J R (1970) The Freedom of the Will, Oxford Univ. Press.
Lucas J R (1973) A Treatise on Time and Space, Methuen, London.
Margenau H (1950) The Nature of Physical Reality, McGraw–Hill, New
York.
Margenau H (1984) The Miracle of Existence, Ox Bow Press, Woodbridge,
Connecticut.
Maturana H R and Varela F J (1987) Der Baum der Erkenntnis, Scherz,
Bern.
Mayer–Kuckuk T (1989) Der gebrochene Spiegel, Birkhäuser, Basel (on
symmetry, symmetry breaking, and order in nature).
Miller D ed (1985) Popper Selections, Princeton Univ. Press.
Misner C W, Thorne K S, and Wheeler J A (1973) Gravitation, Freeman,
San Francisco (mathematical).
Monod J (1970) Le hasard et la necessité, Éditions du Seuil, Paris.
Moritz H and Hofmann–Wellenhof B (1993) Geometry, Relativity,
Geodesy, Wichmann, Karlsruhe (mathematical).
Moser F (1989) Bewusstsein in Raum und Zeit, Leykam, Graz.
Müller G E (1974) Dialektische Philosophie, A. Francke Verlag, München.
Nagel E and Newman J R (1958) Gödel’s Proof, New York Univ. Press.
Nahin P J (1993) Time Machines: Time Travel in Physics, Metaphysics,
and Science Fiction, American Institute of Physics, New York.
Nicolis G and Prigogine I (1989) Exploring Complexity, W.H. Freeman,
New York.
282 ADDITIONAL READING

Oeser E and Seitelberger F (1988) Gehirn, Bewusstsein und Erkenntnis,

Wissenschaftliche Buchgesellschaft, Darmstadt.
Ornstein R and Thompson R F (1984) The Amazing Brain, Houghton
Mifflin, Boston.
Penrose R (1989) The Emperor’s New Mind: Concerning Computers,
Minds, and the Laws of Physics, Oxford Univ. Press.
Perry J and Bratman M eds (1993) Introduction to Philosophy: Classical
and Contemporary Readings, 2nd ed., Oxford Univ. Press.
Piaget J (1970) L’épistémologie génétique, Presses Universitaires de
France, Paris.
Plantinga A ed (1965) The Ontological Argument, Doubleday Anchor
Books, Garden City, New York.
Poincaré H (1902) La science et l’hypothèse, Flammarion, Paris; reprint:
Science and Hypothesis, Dover, New York.
Poincaré H (1908) Science et méthode, Flammarion, Paris; reprint: Science
and Method, Dover, New York.
Poincaré H (1958) The Value of Science, Dover, New York (reprint).
Poincaré H (1963) Mathematics and Science, Dover, New York (reprint).
Popper K R (1977) The Logic of Scientific Discovery, 9th ed., Hutchinson,
London.
Popper K R (1979) Objective Knowledge: An Evolutionary Approach, 2nd
ed., Oxford Univ. Press.
Popper K R (1982) The Open Universe: An Argument for Indeterminism,
Routledge, London.
Popper K R and Eccles J C (1977) The Self and Its Brain, Springer, Berlin.
Prigogine I and Stengers I (1984) Order out of Chaos, Bantam Books, New
York.
Quine W V O (1961) From a Logical Point of View, 2nd ed., Harvard Univ.
Press.
Reichenbach H (1957) The Philosophy of Space and Time, Dover, New
York.
Rosenthal D M ed (1991) The Nature of Mind, Oxford Univ. Press.
Rucker R (1982) Infinity and the Mind, Birkhäuser, Boston.
Rucker R (1984) The Fourth Dimension, Houghton Mifflin, Boston.
Ruhla C (1992) The Physics of Chance, Oxford Univ. Press.
Russell B (1912) The Problems of Philosophy, William and Norgate, Lon-
don.
Russell B (1927) An Outline of Philosophy, Allen and Unwin, London.
Russell B (1929) Our Knowledge of the External World, Allen and Unwin,
London.
Russell B (1945) A History of Western Philosophy, Allen and Unwin, Lon-
don.
Russell B (1948) Human Knowledge: Its Scope and Limits, Allen and
Unwin, London.
ADDITIONAL READING 283

Schilpp P A (1949) Albert Einstein: Philosopher – Scientist, 2 vols., Harper

and Brothers, New York.
Schrödinger E (1944) What is Life? Cambridge Univ. Press.
Schrödinger E (1954) Nature and the Greeks, Cambridge Univ. Press.
Schrödinger E (1958) Mind and Matter, Cambridge Univ. Press.
Sheldrake A R (1981) A New Science of Life, Blond & Briggs, London.
Singer W ed (1990) Gehirn und Kognition, Spektrum Akademischer Verlag,
Heidelberg.
Singer W ed (1994) Gehirn und Bewusstsein, Spektrum Akademischer Ver-
lag, Heidelberg.
Smart J J C ed (1964) Problems of Space and Time, Macmillan, New York.
Smullyan R M (1977) The Tao is Silent, Harper and Row, San Francisco.
Smullyan R M (1980) This Book Needs No Title: A Budget of Living
Paradoxes. Prentice–Hall, New York.
Speiser A (1952) Elemente der Philosophie und der Mathematik,
Birkhäuser, Basel.
Speiser A (1955) Die geistige Arbeit, Birkhäuser, Basel.
Speiser A (1959) Ein Parmenides–Kommentar, 2nd ed, Koehler, Stuttgart.
Spies M (1993) Unsicheres Wissen, Spektrum Akademischer Verlag, Hei-
delberg.
Squires E (1990) Conscious Mind in the Physical World, Adam Hilger,
Bristol.
Stapp H P (1993) Mind, Matter, and Quantum Mechanics, Springer, Berlin.
Stegmüller W (1969) Metaphysik, Skepsis, Wissenschaft, 2nd ed., Springer,
Berlin.
Stewart I (1990) “Does God Play Dice?”, Penguin Books, London.
Teilhard de Chardin P (1955) Le phénomène humain, Éditions du Seuil,
Paris.
Thom R (1975) Structural Stability and Morphogenesis, Benjamin, Read-
ing, Massachusetts.
Thompson R F (1985) The Brain, Freeman, New York.
Tipler F J (1994) The Physics of Immortality, Doubleday, New York.
Treder H J (1974) Philosophische Probleme des physikalischen Raumes,
Akademie–Verlag, Berlin.
Tresmontant C (1971) Le problème de l’âme, Éditions du Seuil, Paris.
Tresmontant C (1976) Sciences de l’univers et problèmes métaphysiques,
Éditions du Seuil, Paris.
Vollmer G (1990) Evolutionäre Erkenntnistheorie, 5th ed, Hirzel, Stuttgart.
Waldrop M M (1992) Complexity: the Emerging Science at the Edge of
Order and Chaos, Simon & Schuster, New York.
Weinberg S (1977) The First Three Minutes, Andre Deutsch, London.
Weinberg S (1993) The Dream of a Final Theory, Vintage, London.
284 ADDITIONAL READING

Weizsäcker C F von (1970) Zum Weltbild der Physik, 11th ed., Hirzel,
Stuttgart.
Weizsäcker C F von (1971) Die Einheit der Natur, Hanser, München.
Weizsäcker C F von (1973) Die Tragweite der Wissenschaft, 4th ed., Hirzel,
Stuttgart.
Weizsäcker C F von (1977) Der Garten des Menschlichen, Hanser,
München.
Weizsäcker C F von (1985) Aufbau der Physik, Hanser, München.
Weizsäcker C F von (1992) Zeit und Wissen, Hanser, München.
Weyl H (1949) Philosophy of Mathematics and Natural Science, Princeton
Univ. Press.
Weyl H (1952) Symmetry, Princeton Univ. Press.
Wheeler J A (1994) At Home in the Universe. American Institute of
Physics, Woodbury, New York.
Whitehead A N (1925) Science and the Modern World, Macmillan, New
York.
Whitehead A N (1929) Process and Reality, Macmillan, New York (diffi-
cult!).
Whitehead A N (1933) Adventures of Ideas, Macmillan, London.
Whitrow G J (1980) The Natural Philosophy of Time, 2nd ed., Oxford
Univ. Press.
Wiener N (1961) Cybernetics, 2nd ed., Wiley, New York.
Will C M (1986) Was Einstein Right? Basic Books, New York.
Young J Z (1987) Philosophy and the Brain, Oxford Univ. Press.
Zurek W H (1990) Complexity, Entropy and the Physics of Information,
Addison–Wesley, Redwood City, California.
Index

a posteriori, 18, 19, 87, 147, 148, 270, 274

215, 230, 262 aperiodic crystal, 169, 176, 178
a posteriori probability, 147, 148 aperiodic solid, 169, 176
a priori, 18–21, 87, 138, 147–149, Aquinas, St. Thomas , 126, 227,
154, 194, 215, 230, 231, 228, 238
254, 256, 262 Aristotle , 73, 79, 88, 108, 110, 126,
a priori probability, 147–149, 154 161, 165, 227, 234, 246
Abraham R H , 83, 170, 180 arithmetic, 27, 30, 33, 35, 37
Absolute, 221, 222, 232, 269, 273 Arnold V I , 81, 82
actual occasion, 142, 218, 230, 242, arrow of time, 130–132, 134, 170
250 artificial intelligence, 3, 7, 36, 253
actuality, 88, 108, 210, 227 Asimov I , 109
adjustment, 41, 42, 47, 55, 63, 65, Aspect A, 109
66, 87, 143, 144, 165, 186 Augustine St. , 126, 228, 272, 274
algorithm, 32, 35, 42, 212 autocatalytic, 179, 180, 182
algorithmic, 4, 44, 212, 213 autonomous structure, 178, 179
amplification, 162, 163, 248, 250 axiom, 18, 32–35, 42, 44, 59, 87,
analytic, 5, 18, 146, 215, 230, 231, 88, 202, 203, 211
266 axiomatic method, 4, 32, 33, 35,
analytical dynamics, 76 202
analytical geometry, 27, 97, 229 axon, 6, 7
analytical philosophy, 156, 158,
232 Bacon F , 151
analytical truth, 30 Barrow J D , 28, 38, 119, 208, 263,
Anaximander , 225 265
Anaximenes , 225 Bayes’ theorem, 147, 152, 155
Anderson P , 191 Beck F , 248, 251, 260
Anger G , 137, 145, 149 Bell J , 109
Anselm of Canterbury , 228, 270 Bell inequality, 109
antimatter, 113, 170 Berkeley G , 11, 240
antinomy, 34, 36, 38, 58, 126, 224, big bang, 46, 125, 126, 132, 170,
231, 268 228, 239, 255, 272
of the liar, 35, 52, 54, 55 biotonic law, 190, 191
antiparticle, 112–114 black box, 105, 136, 141
antithesis, 45–48, 53, 54, 98, 123, block universe, 121, 123, 129, 226,
126, 172, 198, 232, 252, 228, 233, 240–242

285
286 INDEX

Blokhintsev D I , 108 Carnap R , 27, 28, 33, 42, 50, 155,

Bocheński J M , 225, 271 232
Boethius , 228, 241 Cartesian coordinates, 12, 27, 60,
Bohm D , 108, 216, 222, 230, 231, 61, 77, 94, 97, 229
256 catalyzer, 178
Bohr N , 44, 49, 51, 53, 59, 69, 90, catastrophe theory, 156, 170
100, 105, 107, 110, 157, causa
188, 219, 220, 223, 225, efficiens, 161, 227
229, 234 finalis, 79, 161, 227
Boltzmann L , 20, 80, 88, 173 formalis, 227
Bolyai J , 19 materialis, 227
Boole G , 26, 156 causality, 18, 43, 78, 79, 82, 137,
Boolean algebra, 26 138, 142, 161, 165, 227,
bootstrapping, 58, 59, 225, 267, 231, 246
270 causation, 12, 13, 79, 246
Bošković R , 108, 186, 210, 265 cell differentiation, 168–171
boson, 114, 115 cerebellum, 5, 32
boundary condition, 164, 187, 188, cerebrum, 4, 5
190, 191, 245, 265 CERN, 92, 112
Brahe T de , 252 Chadwick J , 111
brain, 3–10, 13–15, 19, 20, 43, 132, chameleon concept, 263
143, 161, 162, 167, 188, chameleon words, 43, 50
197, 208, 209, 211, 213, chaos, 30, 49, 79, 80, 82–84, 88,
214, 238, 246–248, 254, 137, 149, 156, 170–172,
260, 266 179, 192, 242, 243, 250,
brainstem, 4 254, 258, 266
Bratman M , 225, 238 chaos out of order, 82
Briggs J , 83, 180 chaos theory, 30, 82, 83, 137, 156,
Brillouin L , 175 170, 179, 242, 250, 258,
Brouwer D , 37 266
Brown J R , 110, 119, 265 chemistry, 13, 20, 56, 110, 156,
Brownian motion, 254 167, 178, 186, 187, 191–
bucket, 20, 136 193, 197, 207, 265, 266
Buddhism, 158 Churchland P M , 8, 13, 248
Buridan’s ass, 246 classical mechanics, 73–80, 83, 88,
91, 92, 97, 100, 101, 109,
Butler, Bishop , 90
114, 122, 131, 137, 140,
butterfly effect, 30, 242, 243, 249
143, 146, 152–154, 165,
187, 237, 242–244, 246,
canonical equations, 77 256, 261, 272–274
canonically conjugate variables, classical physics, 13, 100, 164, 199,
77, 100, 103 238, 240, 246
Cantor G , 55 closed time–like world line, 129
Čapek M , 134 cogito, ergo sum, 214, 229, 239
Capra F , 45, 108, 222, 267 Cohen J , 7, 31, 56, 172, 183, 192,
INDEX 287

263 cortex, 5, 6, 8, 15, 17

Cohen L J , 90, 147–149, 155, 233 cosmic blueprint, 175
coincidentia oppositorum, 53, 229 cosmic time, 123
commutation relations, 101, 103 cosmology, 120, 126, 241
commutative, 101 counterintuitive, 26, 92, 93, 219,
complementarity, 49, 53, 57–59, 273
67, 69, 103, 234, 245 creation of the world, 124, 126, 224
geometry of, 67 creativity, 121, 143, 172, 241, 273
complementary, 25, 49, 53, 67, 69, Crick F , 176
103, 165, 171, 188, 189, cross–catalytic, 180, 182
241 crucial experiment, 140, 151–155
complementary set, 25 crystal, 169, 176, 178, 181, 182
complementary subsets, 69 aperiodic, 169, 176, 178
complementary subspaces, 67, 69, curvilinear coordinates, 94–96
188 Cusanus N , 53, 229
completeness, 32, 35 cybernetics, 156, 161, 166, 170
complexity, 49, 170–172, 177, 183,
186, 188, 191, 192, 262, d’Alembert J , 76
263, 273 Darwin C , 150, 156, 165, 172, 175
complexity theory, 156, 192 David, King , 272
computation, 4, 7, 43, 164, 212 Davies P , 110, 119, 175, 183, 187,
consciousness, 212, 213, 223, 234, 188, 210, 265, 271
238, 241, 271 degree of credibility, 89
conservation of energy, 107, 174, Democritus , 226
247, 260 Dennett D C , 14, 238, 248
conservative, 82 Descartes R , 12, 60, 201, 214, 229,
consistency, 32, 34, 35, 38, 257 238, 239
constructionism, 190, 192 determinism, 43, 73, 78, 79, 84,
continental drift, 150, 156 137, 227, 242, 243, 249
contradiction, 29, 32, 34, 39, 41, deterministic, 79, 80, 82, 88, 89,
46–49, 54, 57, 59, 67, 80, 103, 137, 149, 212, 244–
181, 185, 186, 205, 266 248
dialectic, 59, 205 deus sive natura, 203, 204, 229,
contraria sunt complementa, 53 240, 261
convection cell, 167, 170, 178, 182 Dewan E M , 13
conventionalism, 140, 256, 257, dialectic contradiction, 59, 205
259 dialectic materialism, 10, 110, 199,
cooperation, 49, 165–167, 172, 180 204–206, 232, 237, 239,
Copenhagen interpretation, 105, 269
106, 110, 190, 200, 219, dialectic reversal, 54, 197, 199, 229
240 dialectic thinking, 36, 44–59
Copernican revolution, 19, 54, 55, dialectic triad, 45, 51, 53, 98, 232,
155, 156, 231 252, 274
Copernicus N , 258 dialectics, 44–46, 49, 51, 53, 54,
Copleston F , 225 57, 59, 60, 69, 103, 104,
288 INDEX

108, 205, 226, 231–234, electromagnetism, 92, 112, 115,

239, 267 116, 118, 264
geometry of, 51, 52, 67–69 electron, 8, 100, 103, 104, 111–115,
dialectics and logic, 57, 59 132–134, 190, 199, 252,
die, dice, 83–85, 89, 244 259, 264
differential equation, 74–79, 82, elegance, 140, 150, 209, 253, 257,
165, 187, 246 262
digital, 7, 43 elementary particles, 92, 100, 111–
digital computer, 7, 213 116, 118, 125, 192, 204,
Ding an sich, 109, 216, 230 216, 219, 261, 264, 265,
Dirac P , 100, 113, 253 267
Dirac equation, 113 Ellis G F R , 134, 233
direct problem, 135–137 Elsasser W M , 190, 193
dissipative, 82, 171, 178, 179, 246 emergence, 9, 56, 82, 88, 172, 263
Ditfurth H von , 21 emergent, 9, 88, 205, 206, 237
DNA, 169, 176–179, 181, 182, 192 emotion, 8, 31, 197, 201, 207, 238
Doppler shift, 127 Engels F , 45, 164, 205, 232
double helix, 169, 176 ensemble, 89, 108
downward causation, 12, 13, 159, entropy, 80, 88, 132, 172–175, 178,
163–165, 203, 206, 227, 181, 182, 188, 241, 273
242, 245, 246, 250 enzyme, 177, 178, 180
dualism, 12, 13, 45, 197, 201, 203, Eötvös R , 150
204, 206, 229, 237, 248 Epicurus , 13, 248, 249, 267
Duncan R , 183, 220, 223, 224, 241 epicycle, 156, 157, 257, 258
dynamic structure, 177–179 Epimenides , 36
Dyson F , 273 epiphenomenalism, 10, 13
epistemological paradox, 88, 102
Eccles J C , 7, 13, 14, 164, 203, epistemology, 19, 21, 137, 214
207, 213, 218, 233, 248, evolutionary, 3, 19, 21, 231
251, 260, 271 EPR paradox, 109
Eddington A , 19, 231, 263, 264 Erigena, Johannes Scotus , 228
Eddington’s example, 19, 231, 253 Ertel H , 82
Edelman G M , 14, 248 estimation, 41, 88, 149, 155
Edey M A , 183 Euclid , 32, 59
Eigen M , 171, 178, 180, 182, 183, Euclidean geometry, 19, 59, 157,
192, 225 215, 231, 254–257
eigenstate, 101 Euclidean space, 19, 61, 62, 100,
eigenvalue, 100–102, 106 183, 231
eigenvector, 101 Euler L , 78, 165
Einstein A , 19, 92, 98, 100, 111, event, 85, 89, 102, 103, 120, 122,
129, 150, 157, 158, 234, 142, 147, 148
235, 241, 244, 263, 264, Everett H , 106, 107, 109, 157
273 evolution, 4, 9, 10, 17, 19, 20, 54,
electromagnetic force, 112, 119, 56, 57, 93, 103, 122, 123,
264 132, 138, 150, 156, 159,
INDEX 289

161, 165, 172, 173, 175– fuzzy logic, 38, 40, 57, 210, 254
177, 183, 214, 217, 224, fuzzy proposition, 40
239, 241, 245, 260, 262, fuzzy set, 39, 40, 42
266, 271–273
exact, 38, 57, 78, 81, 83, 90, 183, Galilei G , 73, 150
200, 202, 210–213, 231, Galilei transformation, 91, 92
252–256, 263 Gardner M , 132, 134
gauge theory, 118, 119, 192, 261,
fallacy of misplaced concreteness, 264, 265
12, 202, 206, 229, 237, 273 Gauss C F , 19, 40, 76, 95, 151,
fallacy of simple location, 201, 217 184, 186, 255
false, 28, 29, 36, 40, 41, 43 Gauss principle of least constraint,
falsification, 20, 110, 139, 140, 145, 75, 165
150, 151, 155, 215, 216, Gaussian curvature, 95, 96
233, 254, 255, 257, 261, Gaussian error curve, 184, 254
262 Gell–Mann M , 107, 117, 119, 171,
FBI, 193 172, 192, 267
feedback, 5, 159–163, 180, 250, 251 gene, 169, 176, 177, 180
Ferguson K , 271 general covariance, 46, 150, 157
Fermat P de , 35, 208 general relativity, 46, 55, 94–99,
Fermat’s principle, 164, 165, 246 123, 129, 140, 150, 157,
fermion, 114, 115 201, 231, 237, 241, 252,
Feynman R P , 33, 107, 132, 165 253, 257, 260, 261, 264
Fichte J G , 11, 45, 51, 55, 57, 220– generalized inverse, 144
222, 224, 232 genetic information, 175, 176
final causation, 79, 246 genome, 176, 179, 180
finalism, 79, 164, 165 Gentzen G , 35
finalistic, 79, 161, 165 geodesic, 75, 76
Findley J N , 44, 50, 205, 212, 232, geodesy, 6, 39, 41, 87, 149, 150,
240, 270 255, 269
Fischer H , 274 geophysics, 135, 149, 150
Folse H J , 51, 59 German idealism, 156, 225
formal logic, 29, 36, 43–45, 50, 51, Gibbs J W , 80, 89
53, 54, 57, 58 Gilson E , 14, 239
formalized thinking, 44 Glashow S , 118, 264
fractals, 83 Glass L , 83
free will, freedom of the will, 121, Gleick J , 82, 261
191, 212, 228, 230, 241, Globus G G , 13, 53, 164, 219, 237
245–251, 266 Gnedenko B V , 90
Frege G , 27, 34, 156 God, 11, 14, 60, 158, 189, 202, 204,
frequency theory of probability, 87 206, 210, 214, 215, 218,
future, 120–123, 130, 132, 142, 226–230, 240, 241, 261,
170, 241, 242, 266 269–274
fuzzy, 38–43, 50, 57, 58, 152, 210– Gödel K , 34, 59, 128, 208, 233,
213, 254, 259 241, 242
290 INDEX

Gödel’s theorem, 4, 33, 34, 36, 44, Hilbert space, 62, 69, 87, 100, 105,
56, 58, 128, 158, 193, 212, 200
222, 248, 254 Hipparchus , 257
Gödelian uncertainty, 36, 266 Hittmayr O , 234
Goethe J W von , 134, 228, 230, Hofmann–Wellenhof B , 94, 99,
273 101, 257, 269
Gold T , 241 Hofstadter D R , 8, 14, 36, 58, 59,
Götschl J , 181 183, 189, 192, 238
grandmother cells, 16, 17 holism, 30, 43, 46, 108, 189, 190,
gravitation, 74, 75, 97–99, 115, 192
118, 119, 256, 264, 265 holistic, 108, 109, 183, 203, 222
Gribbin J , 110 Holzmüller W , 49, 177
group, 33, 116–118 homeostasis, 5
Gulyga A , 45, 225 Hubel D H , 17
Gutmann V , 235 human perception, 14, 138–140,
204, 214
Hadamard J , 30, 149, 242 Hume D , 146, 233
Haken H , 164, 167–169, 171, 183 Huxley A , 270
Haldane J B S , 13, 49, 56 Huygens C , 166
Halmos P R , 26 hypercycle, 171, 179, 180, 182
Hamilton W , 76, 77 hypothalamus, 5, 159
Hamilton’s equations, 76, 77, 82, hypothesis, 20, 21, 260
100 hypothetic realism, 21
hardware, 8, 9, 138, 140, 159, 164,
187, 212, 238, 246 idealism, 10, 13, 45, 55, 158, 197,
Hartmann N , 21, 45, 217, 225, 231 198, 200, 201, 204, 220–
Hartshorne Ch , 270 222, 232, 237, 239, 240
Havemann R , 57, 107, 108, 122 identity theory, 10
Hawking S , 107, 112, 132, 134, ill–posed problem, 137, 143, 144,
233, 234, 248, 265–267, 146, 149, 202, 203
269 image processing, 15
Hegel G W F , 11, 45, 50, 51, 55, immanence, 269, 274
58, 69, 110, 204, 205, 221, immortality, 9, 12, 121, 230, 238,
222, 224, 226, 229, 232 273
Heidegger M , 274 improperly posed problem, 137,
Heintel E , 234 149
Heisenberg W , 100, 105, 106, 110, indeterminism, 242, 247, 249, 251
226, 234, 263, 264 induction, 90, 139, 140, 145–155,
Heisenberg uncertainty relation, 216, 233, 261, 262
37, 103, 152, 193, 223, inertial force, 96–98
254, 266, 267 inertial navigation, 99
hemisphere (brain), 5, 15 inexact concepts, 38, 39, 184
Heraclitus , 226 infinity, 28, 37, 38, 59, 62, 270
Hilbert D , 32, 34, 37, 59, 209 informal reasoning, 43, 44, 212
INDEX 291

information, 109, 132, 145, 159, Küng H , 271

171–179, 188, 266, 273 Küppers B O , 182, 188, 189
initial condition, 74, 78, 79, 81, 83, Kuznecov B G , 45, 225
164, 187, 188, 191, 245,
266 Lagrange J L , 76
instability, 81, 242, 249–251 Lamarck J , 156
integer, 26, 27, 35, 37, 38, 183, 213 Lanczos C , 99
interactionism, 13, 203, 237 language, 5, 30, 31, 44, 50, 59, 207,
introspection, 31, 220, 223 233, 234
intuition, 5, 32, 37, 90, 139 Laplace P S de , 78, 84, 156, 272
intuitionism, 37, 38 Laplace’s demon, 78, 80, 82, 83,
inverse problem, 134–146, 149, 137, 242, 243, 263, 266
190, 191 law, 43, 49, 54, 88, 139, 145–148,
inversion of perspective, 54 150, 154, 155, 161, 164,
irreversible, 80, 130 172–175, 177, 178, 181–
183, 186–191, 193, 247,
Jantsch E , 165, 170, 180 248, 251–263, 265, 267,
Jaspers K , 225, 235 273
Jeffreys H , 42, 90, 148, 149, 152, exact, 183, 254, 255, 263
153, 155 objective, 19, 262
Jesus , 272 physical, 11, 13, 164, 188–191,
Johanson D C , 183 193, 248, 253–255, 257,
Jordan P , 247 258, 262, 263, 267
Joyce J , 114 law of contradiction, 29, 39, 57
law of logic, 29, 39
Kaku M , 56, 119, 265, 273 law of nature, 19, 154, 231, 251–
Kant I , 11, 18, 20, 21, 45, 55, 110, 263, 273
156, 206, 215, 224, 230, law of the excluded middle, 29, 39,
234, 262 68, 69
Kant’s first antinomy, 126 learning, 141, 142
Kauffman S A , 180, 182, 183 least action, 78, 165
Kendrew J , 220, 222, 223 least–squares adjustment, 41, 42,
Kepler J , 156, 226, 252, 258 47, 63, 66, 87, 186
Kepler’s laws, 55, 139, 150, 155, least–squares collocation, 87
251, 252, 254, 258–260 Leff H S , 175
Khinchin A Ya , 90 Legendre A M , 40, 156, 186
Knox R , 11 Leibniz G W von , 10, 62, 204, 217,
Koch K R, 149 230
Kohonen T , 7, 142 Lem S , 129
Kolakowski L , 271 Lenin V I , 164, 205, 232
Kolmogorov A N , 81, 82, 87, 88 lepton, 114
Koltermann R , 274 Lewin R , 171, 183
Kosko B , 43 light cone, 120, 121, 123, 241
Kreisgang, 51, 224 light velocity, 91, 92, 109, 113, 120,
Kuhn T S , 155, 257, 261 127, 140, 152, 153, 184,
292 INDEX

185, 257 269, 273

light wave, 104, 167, 226 dialectic, 10, 110, 199, 204–
likelihood, 153 206, 232, 237, 239, 269
limbic system, 4, 5, 8, 31, 238 materialistic dialectics, 108
Lindsay R B , 76, 79, 83, 89, 90, mathematical model, 136, 143,
99 157, 158, 170, 172, 259,
linear, 62, 63, 82, 100, 101, 136, 260
143, 144 matter, 8–14, 49, 55, 56, 96, 97,
linear operator, 100, 101, 136, 219 104, 110, 113, 164, 165,
linguistics, 30 188, 189, 199–201, 203–
Linné C von , 150, 156 206, 218–220, 223, 224,
Lobačevsky N I , 19 226, 227, 229, 237–240,
Locke J , 233 248, 266, 269
Lockwood M , 14, 111, 164, 248 matter tensor, 96
logic, 18, 20, 21, 23–46, 50, 51, matter wave, 104, 226, 239
53–55, 57–59, 67–69, 85, Maturana H R , 21, 165, 225, 262
145, 151, 193, 194, 204, Maupertuis P L de , 78, 165
205, 210, 222, 227, 230– Maxwell’s demon, 84, 174, 175,
233, 267 178
logical atomism, 30, 38, 43, 46, Maxwell’s equation, 92, 118, 154
203, 218, 233 Mayer–Kuckuk T , 119, 169, 170,
logical positivism, 21, 222, 225, 178, 183
228, 231–233, 259 measuring error, 47, 55, 103, 139,
Lorentz H A , 92 140, 151, 152, 186, 216,
Lorentz transformation, 92, 93, 254, 255
122, 150, 154, 157 Melchior P , 158
Lorenz E N , 30, 81–83, 170, 180, mental, 8, 10–12, 142, 190, 197–
242 199, 203, 204, 207, 208,
Lorenz K , 19, 21, 140 218, 220, 237, 242, 250,
Lowell P , 259 266
Lucas J R , 43, 44, 212, 228, 248, meson, 112, 114–116
263, 267 metabolism, 178, 181
metaphysics, 13, 14, 227, 230, 231,
Mach E , 232 233, 234, 262
Mackey M C , 83 Michelson–Morley experiment, 91,
Margenau H , 76, 79, 83, 89, 90, 140, 152–154
99, 164, 248 Miller D , 88, 107, 149, 155
Marx K , 45, 164, 205, 232 Mills R L , 118, 264
mass, 73, 75, 76, 98, 111, 113–115, Milne E A , 126
150, 200, 237 mind, 8–14, 17, 19, 20, 43, 49,
material, 8, 10–12, 20, 201, 203, 55, 56, 93, 105, 110, 122,
204, 218, 220, 237 143, 159, 161, 164, 165,
materialism, 10, 13, 45, 55, 197, 172, 188, 190, 198–201,
199, 201, 204–206, 220, 203–206, 210–212, 218–
232, 237, 239, 240, 248, 220, 222, 223, 226, 229,
INDEX 293

231, 237–239, 242, 245– non–equilibrium thermodynamics,

250, 256, 260 178, 182
Minkowski H , 93, 157, 241 nonlinear, 81, 82, 136, 178–180,
Mises R von , 87 182
Misner C W , 124, 165, 204, 235 nonlocality, 109, 262
Møller P M , 51, 220 nonverbal thinking, 30, 32
monad, 204, 217, 218, 230, 242, normal science, 156
250, 269 nuclear force, 112
monadic structure, 203, 217–219 nucleotide, 176
monism, 10, 201, 203–206, 210 number
neutral, 201, 203, 210 integer, 26, 27, 35, 37, 38, 183,
Monod J , 178, 183 213
Moritz H , 87, 94, 99, 101, 145, irrational, 27, 55, 221
257, 269 rational, 27, 55, 221
morphogenesis, 168, 170, 171, 177 number theory, 33, 37
Moser F , 108, 157, 183, 191, 222
Moses, 272 object, 30, 38, 49–52, 55, 58, 214–
Müller G E , 45 225, 231, 232, 239, 268
observer, 104–106, 121, 190, 200,
Nagel E , 35, 271 219, 220, 238, 240, 241,
Nahin P J , 130 262, 268
Napoleon , 272 Occam, William of , 229
Ne’eman Yu , 117 Occam’s razor, 107, 229
negation of negation, 54, 205 Oeser E , 8
negentropy, 174, 175, 178 ontological argument, 270
nervous system, 3, 138, 213, 254, ontological status, 9, 206, 210, 237
255 ontology, 10, 13, 14, 206, 209, 233
neural networks, 7, 8, 17, 141, 142 open universe, 122, 123, 240–242,
neuron, 6–8, 17, 43, 177, 192, 209, 245, 249
213 order out of chaos, 82, 84, 88, 192,
neurotransmitter, 6, 192 254
neutrino, 112, 114, 264 order out of order, 82, 182
neutron, 111–114, 117, 264 Ornstein R , 7
Newman J R , 35, 271 overdetermined problem, 142–144
Newton I , 55, 62, 73, 80, 150, 156,
230 panlogism, 204
Newton’s law of gravitation, 74, panpsychism, 10, 204, 218, 219
75, 98, 139, 252, 258–260, pantheism, 158, 203, 229, 234, 269
265 paradigm, 112, 150, 155–157, 215,
Newton’s law of motion, 74, 98, 257, 261, 262
199 paradox, 34, 36–38, 52, 58, 88,
Nicolis G , 183 102, 105, 109, 128, 188,
Nobel Prize, 49, 117, 119, 132, 193, 199, 211, 220, 222–
171, 179, 226, 264 224, 254, 270, 271
noise, 14, 99, 140 parallel–processing, 7, 31, 213
294 INDEX

Parmenides , 203, 226, 241 275

Pascal B , 84, 90 positron, 112, 113, 133, 134
passage of quantity into quality, 56 posterior probability, 147–149,
past, 120–123, 170, 242 152, 153
Paul, St. , 36 potentiality, 88, 108, 110, 210, 227
Pauli W , 234, 253 predetermination, 267
Peano G , 33, 156 present, 120–123, 132, 242
Penrose R , 7, 8, 14, 31, 103, 111, Prigogine I , 171, 178, 179, 181–
154, 164, 175, 182, 206, 183
208, 212, 213, 242, 243, Primas H , 191
249, 251, 253, 257, 260 prime mover, 227, 269
perception, 1, 3, 14–16, 19, 20, principle of covariance, 97
136, 138–141, 198, 204, principle of least squares, 75, 144
214, 215, 245, 253 prior probability, 147, 148, 152–
Perry J , 225, 238 154
perspective, 54, 207, 217, 231, 234, probability, 40–42, 83–90, 102–
256, 262, 274, 275 105, 107, 108, 110, 130,
Peter, St. , 272 131, 144, 146–149, 152–
Petsche H , 7 154, 184, 244
phase transition, 56, 172 a posteriori, 147, 148
photogrammetry, 15 a priori, 147–149, 152–154
photon, 46, 91, 100, 103, 104, 111– posterior, 147–149, 152, 153
115, 132, 133 subjective, 41, 87
Piaget J , 21 probability wave, 104
Planck M , 100, 158 problem
Planck’s constant, 101, 103, 140, ill–posed, 137, 143, 144, 146,
165 149, 202, 203
Plantinga A , 271 improperly posed, 137, 149
plate tectonics, 150, 156, 157, 167 overdetermined, 142–144
Plato , 11, 45, 54, 60, 110, 155, 207, properly posed, 137
226, 227 stable, 137, 203
Platonic solids, 118, 156 underdetermined, 142–145
Platonism, 207, 208 unstable, 137, 203
Plotinus , 58, 203, 222, 227, 274 well–posed, 137, 143, 146, 149,
pluralism, 206, 226, 274, 275 203
Poincaré H , 80–82, 92, 140, 150, profound truth (Bohr), 50, 53, 54,
157, 242, 257 190, 232, 267, 270, 274
Popper K R , 13, 14, 20, 21, 44, 88, propensity, 88, 89, 107, 108, 240
107, 108, 110, 139, 149, properly posed problem, 137
150, 155, 164, 192, 203, property, 23–27
207, 209, 213, 218, 226, proposition, 28–30, 89, 204
233, 241, 242, 249, 250, propositional calculus, 29
257 protein, 176, 177, 179, 180
positivism, 14, 21, 222, 225, 228, proton, 111–117, 199, 264
231–233, 257, 259, 262, provable, 35
INDEX 295

Ptolemy , 156, 257 Reichenbach H , 129, 134, 233

Pythagoras , 59, 226 relation, 26, 274
Pythagoras theorem, 60, 61, 93 relative standard error, 185
relativity, 19, 47, 55, 90–99, 109–
quantum fluctuations, 110, 248, 111, 119, 120, 122, 129,
254 130, 146, 150, 152–157,
quantum logic, 69, 194 200, 201, 206, 215, 219,
quantum of energy, 99 231, 235, 240, 243, 252–
quantum theory, 13, 14, 38, 46, 254, 256, 257, 263, 264,
49, 62, 69, 78, 79, 88, 92, 273, 274
93, 99–114, 119, 130, 140, religion, 271, 273
142, 146, 150, 152, 154, reverse engineering, 141
156–158, 164, 165, 172, reversible, 80, 130, 131, 134
178, 186, 190, 192, 194, Rex A F , 175
200, 206, 210, 216, 219, Riemann B , 96, 209, 273
220, 222, 226, 227, 230, RNA, 176, 177, 179, 180
231, 234, 237, 238, 240, root mean square error, 151, 184
243, 244, 247, 248, 251– Rosenfeld L , 51
253, 260–264, 267, 273, Rosenthal D M , 14, 238
274 rotation group, 116
quark, 111, 114–117, 119, 124, 259, Rucker R , 38, 130, 134
264 Ruhla C , 90
quasar, 124, 127 Russel’s chicken, 147, 154
Quine W V O , 21, 193, 215 Russell B , 8, 10, 11, 18, 20, 22, 26,
27, 30, 34, 90, 147, 155,
Rabbi type, vii, 108, 110, 111, 158, 156, 199, 201, 204, 208,
221, 225, 232, 265 210, 225, 230, 232, 233,
Rahner K , 271 237, 246
random, 80, 84–87, 103, 105, 110, Rutherford E , 100, 111, 226
154, 155, 167, 171, 172,
175, 183, 209, 213, 246, Salam A , 118, 264
248, 254, 267 Santa Fe Institute, 172, 183
realism Schelling F W J von , 45, 55
hypothetic, 21, 197, 198, 220 Schilpp P A , 128, 129, 241
scientific, 198, 200, 215, 220, Schröder W , 82, 150
221 Schrödinger E , 100, 101, 105, 108,
red shift, 127, 128 111, 158, 169, 178, 181,
reduction of the wave packet, 104– 183, 210, 222, 226, 234,
105 264
reductionism, 10, 156, 186–194, Schrödinger cat, 105–106, 154
263 Schrödinger equation, 103, 186,
moderate, 189 200, 248
strong, 189, 190 scientific revolution, 155–157
reflexive logic, 58, 249 searchlight, 20, 21, 82, 136, 139
reflexive structure, 58 Second Law of Thermodynamics,
296 INDEX

173–175, 177, 178, 181, 206, 226, 229, 262

182, 254 spiritualism, 273
Seitelberger F , 8 Squires E , 248
self–consciousness, 271 stability, 81, 82, 137, 149, 169, 178,
self–organization, 166–172, 177, 180, 242
180, 182, 188, 266 stable problem, 137, 203
self–reference, 36, 37, 58, 191, 193, standard error, 39, 151, 184, 185
220, 224, 249, 267, 268 standard model, 119, 192, 261, 264
sense data, 17, 19–21, 136–138, standard theory, 118
198, 203 Stapp H P , 14, 111, 164, 230, 247,
separability of gravitation and in- 248
ertia, 97 state function, 101, 102, 219, 239,
servomechanism, 161, 251 248
set, 23–27, 30, 34, 36–40, 42, 67, statistic, 86
68, 85, 86, 88, 89 statistical mechanics, 80, 83, 88,
set of possible solutions, 145, 206 89, 112, 130, 132, 192,
set theory, 23–27, 29, 34, 37, 38, 251, 254
67, 85 statistical thermodynamics, 80
Shaw C D , 83, 170, 180 Stegmüller W , 199, 232, 233
Sheldrake A R , 183 Stengers I , 178, 183
Singer W , 17 stereoscopic vision, 15
skepticism, 199 Stewart I , 7, 31, 56, 82, 172, 183,
Smart J J C , 134 192, 263
Smullyan R M , 36, 45, 59 stochastic, 55, 86, 87
software, 8, 9, 138, 140, 159, 164, strange attractor, 83, 180
187, 212, 238, 246 string theory, 33, 119
software law, 79, 164, 182, 187, strong force, 115, 116, 264
188, 190, 191, 246 subconscious, 31
solipsism, 198, 214, 245 subject, 214–225, 230–232, 239,
space–time, 90–99, 109, 120–134, 268
157, 158, 201, 218, 228, subjective probability, 41, 87
230, 231, 241, 256 substance, 9, 10, 12, 201–203, 206,
curved, 95–97, 157, 256 218, 229, 237, 238
flat, 96 Supek I , 20, 202, 210
special relativity, 90, 92, 93, 96, SUPERB, 154, 206, 260, 261
97, 110, 113, 120, 123, supergravity, 112, 119
140, 150, 152, 154, 157, superstring theory, 112, 119, 158,
219, 253, 260 192, 261, 265, 267
speed of light, 184, 257 supersymmetry, 33, 119, 158, 265
Speiser A , 51, 53, 59, 60, 220–222, survival of the fittest, 20, 54, 165,
232, 267 172, 175
Sperry R W , 13, 237 symbolic logic, 23–30, 59, 85, 145,
Spies M , 43, 142 230
spin, 113–117 symmetry, 84, 85, 87, 89, 113, 116–
Spinoza B de , 10, 59, 158, 203, 119, 169, 170, 253, 262
INDEX 297

symmetry breaking, 113, 169, 170 time, 74, 93, 94, 102, 120–124,
synapse, 7 126–134, 228, 240–242
synergetics, 156, 166, 167, 171, 179 time loop, 129, 130, 266
Synge J L , 98 time travel, 128, 130, 242
synthesis, 45, 47–49, 51, 53–55, 57, time–reversible, 80, 130
58, 98, 110, 123, 171, 198, Tipler F J , 126, 271
200, 220–222, 232, 241, TOE (Theory of Everything), 58,
252, 271, 274 146, 192, 261, 263–269
synthetic, 5, 18, 215, 230, 231, 266 tomography, 135, 136, 138
system identification, 141, 142 transcendence, 269
Szent–Györgyi A , 193 Treder H J , 99, 134, 221
Tresmontant C , 238, 273
triad
tautology, 30, 230 dialectic, 45, 51, 53, 98, 232,
Teilhard de Chardin P , 126, 183, 252, 274
204, 219, 271, 272 truth, 28–30, 32, 215, 262, 270,
TENTATIVE, 260, 261 271, 274, 275
thalamus, 5, 15 Turing A M , 169, 170
Thales , 225
theology, 126, 229, 271, 274 uncertainty relation, 37, 103, 152,
theorem of Pythagoras, 60, 61, 93 193, 223, 254, 266, 267
theoria, 155, 231, 256, 274 underdetermined problem, 142–
theory, 149–152, 154–158, 251–263 145
Theory of Everything, 58, 146, unified theory, 112, 113, 118, 119,
192, 261, 263–269 216, 249, 264, 265
theory of ideas, 226 union, 24, 25, 66, 67, 85
theory of knowledge, 3, 19, 137, unitary transformation, 103, 116
214, 217 unity of contraries, 53
evolutionary, 17, 20, 138 Universal Metaphysical Problem
thermodynamics, 171, 173, 175, Solver, 13, 202, 215, 229
177–179, 181, 182, 192, universe, 30, 78, 121–129, 223,
254, 260 224, 240–243, 267, 268,
thermostat, 5, 79, 159, 162, 165, 272, 273
246, 247, 250, 251 as an object, 223, 224
thermostat analogy, 250 creative, 123, 241, 242, 245
thesis, 45–49, 53, 54, 98, 123, 126, open, 123, 240–242, 245, 249
198, 232, 252, 274 unprovable, 35, 36, 44, 55, 212
unreasonable effectiveness of
thinking
mathematics (Wigner),
nonverbal, 30, 32
150, 154, 157, 255, 262
thinking about thinking, 50, 51, unstable, 81, 83, 137, 138, 167,
58, 220, 222, 268, 271, 274 169, 203, 242–244, 250
Thom R , 79, 170, 183 Urs von Balthasar H , 275
Thompson R F , 7 USEFUL, 260, 261
Thomson J J , 111
three–world model, 207 Varela F J , 21, 165, 225, 262
 298 INDEX

verification, 139, 140, 145, 150, working hypothesis, 20, 156, 198,
151, 155, 215, 233, 261, 206, 215, 260
262 world–line, 120, 121, 129, 130, 218
vicious circle, 26, 36, 52, 204 World 1, 207–213, 242, 245, 250
vital force, 161, 182, 186, 188, 190 World 2, 207–213, 242, 245, 250
vitalism, 157, 186, 192 World 3, 207–212, 220, 222, 226,
Vollmer G , 21 242, 250

Waldrop M M , 171, 172, 178, 182, Yang C N , 118, 264

183 Young J Z , 17, 21
weak force, 115, 264 Yukawa H , 112, 115
weather prediction, 40, 81, 87, 102,
Zadeh L A , 42
137, 242
Zurek W H , 172, 183
Wegener A , 150, 156
Weinberg S , 107, 118, 119, 132,
264, 267
Weizsäcker C F von , 51, 69, 88,
90, 106, 110, 158, 172,
182, 188, 194, 199, 205,
222, 224, 234, 235, 240,
241, 271, 272
well–posed problem, 137, 143, 146,
149, 203
Wells H G , 130
Weston–Smith M , 183, 220, 223,
224, 241
Wettstein H , 150
Weyl H , 28, 38, 118, 119, 169, 194,
241, 264
Wheeler J A , 132–134, 158, 180,
204, 205, 224, 225, 255,
262, 267
Whitehead A N , 10, 12, 20, 21, 27,
54, 59, 107, 122, 133, 142,
150, 190, 201, 202, 204,
207, 215, 217–219, 226,
230, 231, 233, 237, 238,
241, 242, 250, 252, 256,
259, 267, 270, 271, 274
Whitrow G J , 132, 134, 241
Wiener N , 161, 162, 170
Wigner E , 105, 154, 191, 257
Will C M , 99
Winkler R , 183
Wittgenstein L , 50, 232, 234, 262

View publication stats