Philosophy of the Social Sciences

41(2) 278­–294
Current Philosophy © The Author(s) 2011
Reprints and permission: http://www.
of Science sagepub.com/journalsPermissions.nav
DOI: 10.1177/0048393109352877
http://pos.sagepub.com

Joseph Agassi1

Abstract
This Companion to the philosophy of science reflects fairly well the gloomy
state of affairs in this subfield at its best—concerns, problems, prejudices,
and all. The field is still stuck with the problem of justification of science,
refusing to admit that there is neither need nor possibility to justify science
and forbid dissent from it.

Keywords
philosophy of science, induction, dogmatism, agreement, sociology of science,
Vienna Circle
Stathis Psillos and Martin Curd, eds.
The Routledge Companion to Philosophy of Science. London: Routledge, 2008. xxvii +
604 pp.

The Value of Accord

Why do we strive at agreement? Descartes, Leibniz, Laplace, and many other
occupants of our hall of fame said that agreement brings peace. Relativists go
for less, for peaceful coexistence. They nonetheless disturb the peace, as their
view precludes peaceful negotiations. Negotiation, not agreement, is what
brings democratic cooperation, said John Watkins (1957-1958). As the fourth
and final part of this volume handles the specific sciences, I hoped to find
there some discussion of the value of agreement and of negotiations. That part

Received 28 September 2008

1
Tel Aviv University, Israel, and York University, Toronto, Canada

Corresponding Author:
Joseph Agassi, 37, Levi Eshkol Street, Herzliya 46745 ISRAEL
Email: [email protected]

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
Agassi 279

includes eight chapters, four “hard” and four “human”: cognitive science, eco-
nomics, psychology, and social science.
Cognitive science is in part “hard”—computation theory, neurobiology,
and such; it also has some psychology and anthropology. The author Paul
Thagard centers on the controversy between reductionists and their oppo-
nents. He proves his opponents' view implausible and he dismisses it
(p. 535). He exaggerates, of course, but he has a point: for now, opponents
who wish to join the action may have to join. They do not have to agree: they
can use common sense (Agassi, 1977). Thagard declares the “logical” posi-
tivists his allies (pp. 539-41). They, however, viewed the opponents thesis as
confused, not implausible.
Economics is “a controversial discipline” (p. 543), says Uskali Mäki. He
offers a historical list of dissenters, overlooking Marx and Keynes. (The
latter appears two pages later.) He discusses the controversy between Popper
and Lakatos (p. 544). The latter offered a “modified version” of the former
that survived only 15 years (p. 545). Both were not clear, says Mäki, about
what theory was being tested “and what kind of performance it is tested for”
(p. 546). He does not explain. He refers to the tradition of Mill as one
that deemed the basic assumption of economics “more-or-less true” (p. 541).
The “Popperian-Laktosian episode . . . ignored” that tradition, but it “has
been upheld—but not in one choir—by Daniel Hausman (1992), Nancy
Cartwright (1989), and myself (1992), of whom the last two have also been
influenced by the Poznań school on idealization and the Aristotelian tradi-
tion” (p. 546). Again, no explanation.1 Mäki then snubs those who complain
that the models of economics are abstract, declaring that “the Greek economy
is a Walrasian system” true-or-false (p. 547). Of course: every claim that an
idealization has an instance is true or false. Now since idealizations have no
instances, how do they advance science that seeks true explanations? Mäki
does not say. Switching to applied science, he explains how false hypotheses
can be useful. This holds for Galileo’s hypothesis too, as Milton Friedman
stressed. Mäki criticizes Friedman briefly (loc. cit.), throwing no light on the
value of accord or discord.
Psychology may be a place for discussing the ambivalence about debates.
Richard Samuels chose to discuss briefly other items. Are we computers?
To what extent are we modular? (That is, do we have some self-contained
units of thought captured by inborn fixed neural structure?) And, what mental
structure is innate? What (rightly) troubles Samuels is that reductionism
precludes originality.

1
Musgrave notices (1995, 151) that Cartwright deems laws vacuously true, since they are
idealizations.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
280 Philosophy of the Social Sciences 41(2)

Discussing the social sciences, Harold Kinkaid comments on “the role of

idealized models, the place of individual behavior in social explanation, the
status of teleological and evolutionary explanations, and the role of values”
(p. 594). The role of models engages most authors of this Companion. Fol-
lowing Friedman, he presents idealizations as belonging to the instrumentalist
philosophy of science. As the paradigm case of idealization is that of Galileo,
the formidable enemy of instrumentalism, this invites explanation. Thus far,
no one has offered any, except for Larry Boland, whom only Mäki mentions
and merely as a follower of Popper. Kinkaid contrasts individualism not with
collectivism, as tradition has it, but with mechanism, whose great advocates,
beginning with Julien Offray de La Mettrie, were traditionally sworn indi-
vidualists. If any whiff of collectivism enters his essay, it is under the rubric
of evolutionism: “Marx thought that the state exists in order to defend the
interests of the ruling class. Durkheim claimed that the division of labor
exists in order to promote social solidarity” (p. 599). Under the same rubric
enter the functionalists and the school that favors replacing explanation in
social studies with understanding. He dismisses them all as they make “illicit
biological analogies” (p. 600). Kinkaid does not say what would legitimize
their views. Finally, he claims that science is value-free and dismisses the
view of Gunnar Myrdal that science is value laden, and Quine’s claim that
there is no sharp line dividing fact from value: they too have not legitimized
their views (p. 602).
This is strange. The Companion is divided interestingly into four parts:
the fourth presents specific sciences; the third discusses “concepts,” that is,
specific technical issues, including causation, evidence, idealization, mea-
surement, and probability; the first offers a general and historical background;
and the second is titled “Debates.” Proof kills debates, observed the great
Bacon. How then is disagreement possible? What role do debates play in
learning? No answer here, nor in the editors’ seven-page introduction to this
volume that is largely historical. They say of part 3 that the debates it dis-
cusses are important. Why? As debates come to eliminate error, why is here
so little about scientific error? One discussion here concerns scientific fail-
ure; it ascribes to Larry Laudan the idea that as we deem past science failures,
so will future commentators deem our science. He thus deems all error fail-
ure (p. 232). To show him in error, consider a set of suspects and an equal set
of investigators who shadow them; at most only one investigator will find the
suspicion correct, yet the success is of the group as a whole—if they are
lucky. Fallibilism finds error or at least possible one of the most valuable
products of the human mind. This Companion lamely reconciles fallibilism
with the theory of rational belief (pp. 92, 98, 479). Discussion of

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
Agassi 281

falsificationism should refer to scientific errors. Alas, here it does not. Its
author, Gürol Irzik, reports what seems to him its fatal philosophical error,
namely, its inability to claim certitude for its thesis that science approximates
the truth (p. 62). Why he finds certitude necessary, I cannot say. As fallibilism
rules out all claims for certitude, its adherents wait to be told what merit cer-
titude has today, after it has ceased to guarantee the truth.

The Expression of Accord Here

The present book seems defensive. This is a pity, as its task is not to defend
but to help potential users. Perhaps I am expressing here my bias. Perhaps
an outsider like me should not review handbooks, as these are bound to rep-
resent establishments that outsiders decline to join. If they do, they may feel
obliged to rehash familiar criticism of received views. Attempts at fair pre-
sentations facilitate this task. To my delight, this book presents both
established and fringe views, both critical realism (to which my output
belongs) and subjectivism-relativism (which is my pet aversion). Admit-
tedly, establishment presentations of fringe ideas may be inaccurate, but this
comes with the territory.
The least allowable distortion here appears in the few discussions of prob-
ability, especially the axiom system for it (p. 118). The famous system of
Kolmogorov (1933) is better. Alfred Rényi offered a mathematically better
system (1955). Popper’s system (1955) goes further (see below). He showed
repeatedly that the probability of a hypothesis does not follow the axioms of
probability.2 Thus, the degree of confirmation of any tautology is minimal
and its degree of probability is maximal. Moreover, proper probabilities are
positive; the probability of a hypothesis is all too often negative.3 Popper
called “corroboration” the confirmation that follows his theory. As he noted,
to formalize corroboration fully is impossible, as the data it considers are
results of tests, and tests are intensional. Moreover, confirmations in science
and in technology differ (Agassi 1985, 33-37).
The best comments on Popper’s formal axiom system for the calculus of
probability are due to Hughes Leblanc, here conspicuous by his absence.
However, its novelty is obvious even without commentary. It is the first pre-
sentation of probability theory that follows the presentation of group theory: a

2
Those who find this odd may notice that likewise the market does not abide by the laws of the
free market, that some acids are sweet, and that the English horn is no horn (but an oboe).
3
It is easy to transform a function whose values are between zero and one to a function whose
values are between minus one and plus one. The resultant function, however, does not follow the
calculus of probability.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
282 Philosophy of the Social Sciences 41(2)

formal system postpones discussing the question, What are the objects to
which the theory applies? This signifies philosophically as it allows for as
many answers to the postponed question as possible, namely, for as many
interpretations as possible of a group or of probability, this leading to easy
tests for the adequacy of each of them. The nearest to a reference to all this is
an opaque denial of it in an implied contrast that Maria Carla Galavotti makes
between Popper’s new propensity interpretation of probability that “gains
increasing popularity, but also elicited several objections” and “the pluralistic
approach . . . [that] is gaining ground” tout court (p. 423). Galavotti omits to
mention the name of the author of “the pluralistic approach”; it is Popper: his
fully formal axiom system for probability embodies it.4
The interpretation of the calculus of probability as the degree of rational
belief in hypotheses permeates the literature on the philosophy of science, in
disregard of Popper’s suggestion. Rational belief theory serves no purpose,
he said, yet the least objectionable version of it is popular among scientific
researchers, not among philosophers of science (Popper 1935/1959, §62,
note *1). It abides not by the axioms of probability but by Popper’s theory of
corroboration that ignores all information other than results of tests. These
are efforts at refutation.
This Companion esteems rational belief as rational agreement. By default,
writers here identify it with faith in science. Some philosophers will disagree.
Some of them it disregards as hostile to science. (Heidegger does make a
token appearance here, not his opinions.) Others belong to the commonsense
school of philosophy that takes common sense as prior to science.5 Theirs is
the realist commonsense version of the sense data theory. Russell advocated it
as the hypothesis that he repeatedly tried to render a logically adequate foun-
dation of scientific realism—adding that if it fails in this, we should relinquish
it (Schilpp 1944, 718). Regrettably, this Companion overlooks him. The paper
of Joanne Waugh and Roger Ariew about “The History of Philosophy and the
Philosophy of Science” has no reference to him. They thus insult a this great
light who was the initiator of the process that they describe—of rendering the
field both more rigorous and more historical. They ascribe it to Carnap whose
philosophy was eminently a-historical.

4
Popper’s enrichment of the logical interpretation of probability by deducing Boolean algebra
from its axioms is sorely missing here, as is Leblanc’s proof that Popper’s system has new models
(Leblanc, 1989).
5
This was the dominant view in England in the middle of the 20th century, when I was a student
there. It puzzled me very much, as I knew that in most places common sense is steeped in magic, and
even in places where science runs supreme magic is still rampant.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
Agassi 283

The Praise of Accord

The book is reasonably comprehensive—it includes over 600 pages of about 450
words per page (about 900 of more ordinary pages). How can it be useful? How
can practitioners use it? What can handbooks do better than encyclopedias? Prac-
titioners can see which way the wind is blowing. Those who have no wish to trim
their sails to the wind, then, will ignore handbooks. These are the familiar two
kinds of practitioners (the normal and the leading, to use Kuhn’s terminology). A
new breed is growing, the meta-philosophers (Kuhn himself?). They do not spec-
ulate; they observe. This is pleasing, as observation is less open to controversy
than speculation. (Even Popper, Kuhn’s most formidable adversary, admitted
Kuhn’s observation that the normal is ubiquitous.) What do they observe? This is
a bothersome question. Meta-philosophers may stave it off by limiting their
observations to philosophy departments. The answer will not work: not all phi-
losophers dwell there: Locke and Sartre were not academics, and Frege was a
mathematician. Meta-philosophers may then limit their observations to philo-
sophical matters like alienation or science. What is alienation? The question is
welcome—unlike the parallel question regarding science. This is an observed
fact. The reason for it is obvious: controversy is annoying and staving it off is
pleasing; and, of the two questions, only the latter invites controversy.
Why does controversy annoy? How can we limit it? Efforts to proscribe it
have failed, as have incentives for its avoidance such as, the promise of scientific
status on the ground that science spells unanimity. Valuing unanimity, we may
tolerate some received dogma, of course. Kuhn advocated this, knowing that
allegiance to science did not save him from the advocacy of a measure of dogma-
tism. He noted that scientists are often dogmatic, and that there are worse dogmas
than science. His point is unnerving: what if, as Popper said in response, science
as a whole is deteriorating and loses its edge? Kuhn answered this too. He used
Popper’s own contribution to the philosophy of science: as the problem of induc-
tion is insoluble and as science progresses nonetheless, clearly it can do without
induction. Hence, added Kuhn, research rests on some baseless suppositions.
These are dogmas; so Kuhn observed that science avoids stagnation by replacing
its dogmas under the pressure of accumulating experience. This Companion
treats him harshly, as the wind has ceased to blow his way. It addresses the
problem, What distinguishes science from dogma?
Tradition contrasts the dogmatic adherence to opinion with its rational
justification. By this means the problem of demarcation of science becomes
the problem of inductive justification: how does experience justify theory?
This is odd. We learn from experience; science is learning from experience;
we cannot say how. Meta-philosophers come to the rescue: rather than strug-
gle with the problem they observe philosophers struggling with it. Thus,
almost all of the 55 essays in this Companion touch upon this question, and

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
284 Philosophy of the Social Sciences 41(2)

more than half concentrate on it. Who should be a meta-philosopher then?

Compilers of handbooks and their authors. The editors of this Companion
testify that their contributors excel as philosophers too (p. xxvi).
Agreeably, meta-philosophers need not arbitrate controversies. Bayesians
and critical rationalists hardly appreciate each other’s output. The editors of
this Companion allow them both to have their say. It thus records the chief
ideas and standard clichés of different schools. The Companion also offers
major arguments of one school against the other, but not fully, thus minimiz-
ing controversy. This is well within tradition, of course.
Here is a one-paragraph report on five examples of this kind of treatment of
dissent here. (1) The editors admit, “Popper put forward a different conception
of scientific method” from “the logical positivists”—different, not conflict-
ing—yet he “shared” with them “hostility to psychologism”6 and the view that
“the philosophy of science is . . . normative” (p. xxiii). (2) In an essay on “logi-
cal” empiricism, Thomas Uebel notices (p. 79) that the “correct formulation”
of its meaning criterion “proved controversial and elusive.” “It is not clear
whether the entire logical empiricist project is derailed by this,” he adds. In
conclusion Uebel says (p. 87), “It is not easy to separate sharply the logical
empiricist philosophy of science from all approaches that dissent” from it.
(3) Collin Wilson’s essay on Bayesianism goes further and posits a general
“convergence of opinions” (p. 112): Bayesian philosophy allows for all diver-
gence of opinions, since empirical evidence leads to their quick convergence
(p. 111).7 (4) Alan Hájek and James M. Joyce discuss diverse theories of
confirmation. They assert (p. 115) that Popper was “unfriendly to confirmation
theory.”8 Hájek and Joyce conclude (p. 127) by siding with Carnap against
Hume and Popper in insisting that confirmation is important for science.9
(5) Robert Nola, on the social studies of science, tries to equate as much as he
can the views of Robert K. Merton and of Karl Marx (p. 262).

6
This is the received opinion. Carnap wrote a book (1928/1998) that employed a thoroughly psy-
chologistic idea the he called “methodological solipsism,” and he never gave it up (Creath and Fried-
man 2007, 161). It is erroneous. No text of any member of the Vienna Circle rejected psychologism
as definitely as Popper did in §2 of his The Logic of Scientific Discovery.
7
Divergent prejudices refuse to yield to evidence and thus preclude the convergence that Bayes-
ians promise.
8
Popper explained the great import of confirmation. He was unfriendly to theories of confirma-
tion that do not confine it to severe tests. Those who may object that the difference is too abstract to
matter should note that it was not too abstract for the U.S. Supreme Court to notice (Edmond and
Mercer 2004, 199).
9
The boot is on the other foot. Carnap’s book on induction ends with the admission of failure and
the promise to try again to overcome Hume’s criticism (Carnap 1950, 365); Popper’s theory strength-
ens Hume’s criticism and describes scientific progress as in accord with it.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
Agassi 285

The Editors’ Agenda

Agreement is valued everywhere. The peculiarity to Western philosophy is the
demand that it should be rational and thus transcend the parochial. The modern
version of this idea is due to Francis Bacon. He demanded the rejection of all
metaphysics as preconceived opinions that distort perceptions and disturb the
peace by destroying unanimity. He considered progress vital for science; he
praised humility and small contributions to science, promising that no matter
how small these are, they stay secure in escrow to yield great profit. He prom-
ised that (unlike theology) science would leave nothing hidden, so that the
neglect of metaphysics now will later on advance metaphysics that will rest on
science, thus achieving scientific status.10 He deemed rational only proven
belief and had no theory of proof. His influence today in the philosophy of
science survives as the lead item on its agenda. It is the question, What belief
is rational? The same holds for his categorical opposition to all daring think-
ing and to all controversy as rooted in personal ambition. After the debates
between Einstein and Bohr about quantum mechanics won so much acclaim,
this is hard to maintain. Thus, Bacon’s influence persists for no good reason.
Science does not command unanimity. It prevails because it is as easy to agree
as to disagree, and agreement sounds friendlier. Active agreement is still easy,
as it invites mere repetition, whereas active disagreement is criticism, and
criticism is creative. Thus, philosophers of science who today often praise
Einstein with little or no knowledge of his ideas. Thus, Laudan says (1983),
the problem of the demarcation of science is that of the credibility of the
scientists (cp. pp. 257, 282, 309). Is this true? This Companion dodges the
question, even in its meager discussion of the sociology of science. The ideas
that this Companion enlarges on boost ideas of Bacon.
Kant found scandalous the disagreement about the basis of agreement in
science. He claimed to have resolved the scandal by his view of science as
synthetic a priori knowledge. Einstein deemed this idea harmful. Under
Einstein’s influence, Schlick and Reichenbach relinquished certitude for a
while. They then assumed that scientific theories are synthetic a priori puta-
tive truths. They later switched to a Wittgenstein-style verifiability criterion of
meaning, soon to bump into the hard fact that no theory is open to verification.
They then refused to give up verification. Pity. Here we find a mere echo of
their older view unexplained (pp. 81-82).

10
All proselytizing religions share the idea that we should surrender what little we have in return
for a promise of great treasures in the near future. Comte viewed science as the religion that rejects
speculative metaphysics.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
286 Philosophy of the Social Sciences 41(2)

The major task of the philosophy of science, say the editors in their open-
ing kick (p. xix), is “to understand science as a cognitive activity that is
uniquely capable of yielding justified beliefs about the world.” This clashes
with the discussion here of critical rationalism that explicitly rejects all justi-
fication (p. 59). Science has its agenda, and only rational debates should
change it. How can we do that?11 By received opinion, rational belief is at the
top of the agenda of the philosophy of science. A proper vademecum should
discuss the agenda. The editors here have missed their call. They adjudicate
rather than explain.
Philosophers of science take rational belief to be all and only belief in (con-
temporary) science.12 What they look for is not an alternative to this idea or for
an improvement of it but the process that leads to it. They lack philosophical
justification, yet if they must have it, they should prefer it scientific.13
The problem of induction concerns the birth, confirmation, and employment
of scientific theory. Bacon said well-born theories require no confirmation.
Whewell disagreed: the discovery of a theory must precede its test and tests
seldom lead to confirmation. Duhem said that scientific theory has a limited
domain of application; kept within these limits it is true. The critical rational-
ist view of the situation is much simpler. The proverbial “all swans are white”
is a generalized observation that rests on the hypothesis14 that any subspecies
of birds are likely to share colors. The presence of black swans need not over-
turn that generalization, then, since they may belong to a different subspecies.

11
The current jargon term for the change of agenda is “paradigm shift.” Kuhn, its author, said the
leaders are in charge of the agenda. He did not say who these are and how they achieve their status.
He suggested that they gain it by exhibiting their intellectual abilities. This idea is empirically refuted.
12
Thus Quine, a confirmed nonjustificationist, committed himself faithfully to the ontology that
science advocates.
13
The idea that any justification is better scientific than metaphysical is Popper’s last word. See
the last appendices to the latest German edition of his The Logic of Scientific Discovery. Indeed: every
justification leads to infinite regress. The philosophical one is barren; the scientific one is fruitful.
14
This hypothesis is questionable: it behooves researchers to ask, of what subspecies it holds and
why, and test the answer to it. Thus, science is forever tentative. People who want scientists to be
infallible hate this. They cannot annul fallibility, however. This is Popper’s older and better version
of Putnam’s meta-induction cited here (p. 232) that dodges the conclusion that science is fallible
(p. 520). The dodge rests on a hypothesis regarding natural kinds. Discussion of natural kinds takes
much space here and is of no use as all its versions are hypothetical. The only interesting versions of
natural kinds are those that we find in given cultures. Their dependence on cultures makes them all
a priori unscientific.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
Agassi 287

This makes us more critically minded about generalizations. Inductivists want

the order reversed: they want claims about a subspecies to rest on observa-
tions. This does not work.15
Belief and disbelief 16 come in a context that specifies whether they may
stay or should undergo tests that may lead to their rejection or modification.17
Justifications of belief too depend on context; they are thus qualified. The
demand to have them unqualified is the hallmark of classical modern phi-
losophy that demands starting afresh.18 Hume’s criticism of this demand is
today uncontested: learning without presuppositions is impossible. Husserl
tried to outdo Descartes; even his disciples agree that he failed.19 And so
today discussion of justification often focuses on its context: it is at times
problematic, as legal test cases and appeals amply testify. Efforts to wriggle
out of this situation take a major portion of this Companion.
If belief should stay on the agenda of the philosophy of science, it should
appear together with disbelief. Bacon’s demand to justify every belief had a
rationale: it rested on his doctrine of prejudice that says that people never
correct their errors: they always dismiss criticism with some excuse. The
only viable policy for researchers, he said, is to avoid error. Whewell wit-
nessed the switch from the particle theory of light to wave theory and took
this as a refutation of Bacon’s doctrine.

15
The search for the universal is for freedom from constraints of contexts. The conditional with
the context of an observation as its antecedent and the observation as its consequent is context-free.
The tacit program of the Vienna Circle was to create such context-free observations. It was explicit
in the artificial intelligence program. Its advocates admitted failure: we do not know the context; we
much less observe it.
16
Bacon held the quaint idea that belief requires justification but disbelief does not. This was his
concession to skepticism. The law speaks differently: it requires that doubt be reasonable. Robert
Boyle and John Locke declared court procedures as the model for scientific testimony. They did not
refer to reasonable doubt, much less to laws against witchcraft. (They lived in the midst of a legal
witch-hunt.) See also next note.
17
Surprisingly the idea is often admitted without debate that scientists cling to refuted hypotheses
for want of better alternatives to them. Among its advocates were Lenin, Hempel, Kuhn, Feyerabend,
and Lakatos. It holds not in science but in courts, where dissenters from a default opinion may have
to defend their dissent.
18
The idea of starting afresh is radicalism, the idea of uprooting traditional ideas (radix = root). It
began in political thinking (Popper 1945, I, 9). Bacon (who was politically a conservative) introduced
it in science; his followers applied it to politics. That received philosophy of science is radical need
not make its practitioners radicals: they are simply unaware of their radicalism or else unable to free
themselves of it. Thus, some of them deny that all observation-reports are true, yet they use them as

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
288 Philosophy of the Social Sciences 41(2)

The original designers of the 20th-century theory of rational belief did not
ignore contexts. Sir Harold Jeffreys did not specify any context; J. M. Keynes
offered a single context for all induction, his principle of limited variation. So
did Russell, who reluctantly introduced the principle of induction as a syn-
thetic a priori valid truth. The most influential thinker is Bruno de Finetti, the
father of Bayesianism.20 He made a complex set of psychological hypotheses
that look untestable. This Companion seems to deny this (p. 105, first para-
graph; p. 115, final two lines; p. 119; p. 124, first two lines; pp. 134-35;
p. 250).21 Perhaps not: context-less justification, we learn, must exist, or else
skepticism is true (p. 115). This is a tacit transcendental proof.
On the first page of this vademecum, speaking on the philosophy of sci-
ence in general, the editors introduce four problems from that field:

1. What characteristic is specific to science?

2. Are scientific theories true descriptions of reality?
3. What is cause, explanation, confirmation, etc.?
4. What rules and what values (if any) govern theory-change?

Answers to questions 1, 2, and 4 enjoy a broad consensus, I daresay,

although the agreed answers leave much for further discussions that pertain
to possible answers to question 3. They are as follows.

1. Science comprises empirically testable explanations.

2. Scientific theories are true-or-false; they are series of approxima-
tions to the truth.
4. A refuted or an otherwise unsatisfactory theory is a challenge to
researchers to try to devise a better alternative to it.

given truths in their formulas of (mock) probability of hypotheses. Yet the situation is problematic:
is it wise to declare a theory false when it conflicts with an observation? This is Boyle’s problem. It
disturbed him particularly since he declared only repeatable observations scientific, and that makes
them all hypothetical. Russell has offered a lovely solution to this problem: theory and observations
serve to correct each other. But in this volume, the treatment of Russell is worse than that of Popper.
19
Husserl is absent from this volume; Heidegger appears here only as the target of some famous
criticism.
20
This Companion uses systematically the popular misnomer for his theory: Bayesian probability.
It is probability only metaphorically. Bayes’s law is provable. Were its use justified the way Bayes-
ians suggest, induction would be valid and Hume’s critique answered. The assertion (p. 105) that “its
applicability is quite general” is unwise.
21
The feminist essay similarly notes that traditional philosophy tacitly considers philosophizing
masculine (p. 183).

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
Agassi 289

Observing the tremendous growth that the philosophies of the specific

sciences have undergone over the last few decades, the editors insist: as sci-
ence is one, the general conception of it precedes its detailed manifestations.
They choose the Vienna Circle’s “conception of philosophy of science” as an
example. It “was strongly challenged by three important and influential
thinkers,” Quine, Sellars, and Kuhn, they report. This is inappropriate. Kuhn
explicitly expressed agreement with them. Quine repeatedly criticized Carnap,
never the Vienna Circle. And the most recent book on Sellars (Rosenberg
2007) mentions them once, apropos of an influential collection of essays of
his. Still, the editors are right in observing the ubiquity of debates. Should
we examine valuable historical cases? The editors end their introduction
with a historical three pages that introduce 20th-century ideas. They mention
Russell as a logician, overlook Wittgenstein, and ascribe his message—
metaphysics is meaningless—to Schlick and his Circle (p. xxii). They laud
it and hint (p. xxiii) that Popper shared it.
The editors’ report on Quine’s criticism of the views of the Vienna Circle
confuses the naturalism of the Vienna Circle that Popper refuted with Quine’s
naturalism that Popper had announced earlier.22 Let me quote the Stanford
Internet Encyclopedia,23 on this point:

Following Quine, naturalism is usually taken to be the philosophical

doctrine that there is no first philosophy and that the philosophical
enterprise is continuous with the scientific enterprise.  . . . [S]cience,
thus construed . . . , is . . . the complete story of the world.

(Popper’s naturalism goes further.24) On method, Quine said, “slogans aside,”

he was “in substantial agreement” with Popper (Hahn and Schilpp 1986, 621).
The editors declare the idea of the unity of science as “the dominant
dogma . . . favored by the logical empiricists” (p. xxiv), overlooking the ubiq-
uity of this idea. Moreover, as it is metaphysical, the logical empiricists

22
Popper (1935/1959, concluding section). Hume’s naturalism, incidentally, is still different
(Agassi 1986).
23
This quote is from the essay “Indispensability Arguments in the Philosophy of Mathematics.”
That argument says, since sets are indispensable for science, they are real. The majority deny this;
a quote from Putnam there (loc. cit.) accuses them of “intellectual dishonesty.” This is odd: it is not
clear whether Putnam admits the existence electrons (p. 232), let alone sets. The situation is thus a real
mess that this Companion does not clean up.
24
Popper (1963, chap. 5: “Back to the Presocratics”) goes further than Quine, as he adds myth to
the blend that includes science and metaphysics.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
290 Philosophy of the Social Sciences 41(2)

rejected it. They replaced it with physicalism,25 the claim that the language of
all science is the language of physics. No matter; the point of the editors is that
in the 80s the tide changed and the disunity of science became fashionable.
They ignore my “Unity and Diversity in Science” of 1969, where I say that
efforts at explanation, when successful, unify; criticism, when successful,
diversifies.26
In this Companion I found names of people who dropped out of the litera-
ture decades ago, and I missed names of active ones who undergo much
exposure to date. All my efforts to find a method here failed. I read with inter-
est reports on debates and wanted to see the current situation regarding them
and the recommendations authors make about them. Particularly debates on
the theory of confirmation here reported are murky. Carnap criticized Hempel
for his having “conflated” different concepts of confirmation (p. 116). No
explanation, no mention of the defense of Hempel, and no mention of
Popper’s concept of confirmation even though he proved that the same con-
flation in Carnap’s work leads to inconsistency. I would love to know how
things stand these days. This Companion should help me. It does not. A Com-
panion has an advantage over an anthology in that the editors can participate
in the production of its parts; it has an advantage over an encyclopedia in that
it can be more boldly future-oriented. The editors of this Companion did not
make sufficient use of these advantages. Perhaps the next Companion will be
more useful and herald a healthy return to Russell and accent on his follow-
ers, Popper, Quine, Bunge, Gellner, Lancelot Law Whyte, and others whom
I missed as I examined this Companion. Perhaps my slogan should be, back
to Russell’s Problems of Philosophy (1912).
After this cursory review of the editors’ preface, let me skip to the final
part and report on the four “hard” items on “individual sciences.”
Alexander Rosenberg discusses biology. Since Darwinism relies on random
variations, it defies natural laws, he observes. Given a terrain and random
motion of rocks, eventually they will end up situated in valleys: randomness
leads to stability. Since species are reasonably stable, biologists try to explain
them as points of stability. Efforts to aid research by analyzing the concepts
of “levels and units of selection” have opened new avenues of research
(p. 516), and this, it seems, Rosenberg welcomes.
Robin Findlay Hendry discusses chemistry. Its building blocks are ele-
ments, usually but erroneously identified with the chemical atoms: the
assumption behind it, that atoms are indivisible, is false. Hendry uses big

25
Neurath gave an example of a protocol sentence allegedly in the physicalist language. It is
grotesque.
26
See Agassi (1975, 404-68: “Unity and Diversity in Science”).

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
Agassi 291

cannons, Kripke and Putnam, needlessly. Consider the assumption behind

hydrodynamics, that fluids are continuous. It is worse than atomism, and the
big cannons will not help here. Discussing chemistry as a part of physics,
though an ancient idea, Hendry brings new cannons to back it up. He would
have been more helpful had he reported diverse successes as well as the
diverse extant problems, not to mention the surprising ability of the noble
gases to combine with other elements in refutation of Pauli’s beautiful expla-
nation of their nobility.
Peter Clark discusses mathematics. It is hard for me to comment on it,
partly for want of background knowledge. He says he wants his study to be
useful for the philosophy of science, and this eludes me. His discussion of
Dedekind’s pioneering study of numbers (p. 560) is so incomplete that it
falls between the stools of the popular and the accurate. He expresses dis-
satisfaction with the ad hoc proscription of paradoxical sets. This is a basic
problem in the foundations of mathematics, but it does not signify for the
philosophy of science. He says that classical abstract set theory “could hardly
be philosophically satisfying, for . . . we are left entirely in the dark as to
what sort of structures they are” (p. 561). This is surprising. The great devel-
opment of modern mathematics began with Lagrange’s discussion of
dimensions in the abstract and with the rise of group theory that discusses
some aspects of structures without asking what sort of structures they are.
Boole’s extensionalism then allowed for all sorts of crazy classes and thus
opened the road to formalization and thus to modern logic (Bar-Am 2008).
Hilbert found formalism great just because it leaves unsaid so much about the
objects under study. This has direct application to the philosophy of science,
not only because the techniques of Lagrange appear in physics and in eco-
nomics, but also because it is easier to make abstract testable hypotheses than
to speak concretely. Thus, discourse about democracy in the abstract may be
more testable than about some definite democratic structures (Agassi 1989). I
cannot guess why it seems to Clark so unpleasant that we have so many ways
of understanding natural numbers. But perhaps this is my error: Clark accepts
Hilbert’s view that this characteristic is inherent in mathematics (p. 562).
Also, he moves to the “Quine-Putnam indispensability arguments and the key
argument in defense of naturalism,” where he loses me. The question here is,
Do mathematical entities exist, and if so, in what sense? Quine told me he
gave up hope of ever receiving an intelligent comment on his publications
because commentators refused to believe that he meant what he said when he
said that in his view numbers exist just like tables and chairs. Is this the indis-
pensability thesis? I doubt it, and I cannot see how it matters to the philosophy
of science.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
292 Philosophy of the Social Sciences 41(2)

Simon Saunders discusses physics. He opens with a difficulty that he

ascribes to Heisenberg. Russell portrayed it well when he said (Russell
1940, 15),

Science seems to be at war with itself. . . . Naïve realism leads to phys-

ics, and physics, if true, shows that naïve realism is false.

Saunders ignores him. Eddington presents the matter as a discrepancy between

his two desks, one made of tactile and smooth wood and the other almost
empty and atomic (Eddington 1928, Preface). Saunders ignores him too. At
least he takes the wave-particle duality seriously. Unfortunately, he
misrepresents Einstein (p. 568). In 1905, presenting a new particle theory of
light, Einstein said, Maxwell’s wave theory approximates the new particle
theory for strong fields. In 1913 Bohr came up with a different idea: a particle
behaves according to Maxwell’s theory when far from nuclei and over long
distances. The two criteria for approximation clash for a quantum particles in
a weak field traveling along great distances: they should display characteristics
of quantum particle according to Einstein and wave characteristics according
to Bohr. The famous two-slit experiment complies with this setup. In it,
radiation displays wave characteristics along its path and quantum particle
characteristics at its target. For researchers in physics, the paper here is too
sketchy, inaccurate, and scarcely new. For what audience then is it written, and
to what end? And how does it fit in this vademecum?
How can one prepare a better guide for the philosophy of science? Back
to Russell’s Problems of Philosophy. Perhaps also his The Scientific Outlook
that presents science as a Promethean madness.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research,
authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publi-
cation of this article.

References
Agassi, Joseph. 1975. Science in flux. Dordrecht, the Netherlands: Reidel.
Agassi, Joseph. 1977. Towards a rational philosophical anthropology. The Hague,
the Netherlands: Nijhoff.
Agassi, Joseph. 1985. Technology: Philosophical and social aspects. Dordrecht, the
Netherlands: Kluwer.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
Agassi 293

Agassi, Joseph. 1986. A note on Smith’s term “Naturalism.” Hume Studies 12:92-98.
(Reprinted in Stanley Twyman, ed., David Hume: Critical Assessments, vol. 3,
London: Routledge, 1994)
Agassi, Joseph. 1989. The logic of consensus and of extremes. In Freedom and
rationality: Essays in honour of John Watkins, edited by F. D’Agostino and
I. C. Jarvie, 3-21. Boston Studies in the Philosophy of Science, vol. 117. Dordrecht,
the Netherlands: Reidel.
Bar-Am, Nimrod. 2008. Extensionalism: The revolution in logic. Berlin: Springer.
Carnap, Rudolf. 1928/1998. Die Logische Aufbau der Welt. Hamburg, Germany:
Felix Meiner Verlag.
Carnap, Rudolf, 1950. Logical Foundations of Probability, Chicago: Chicago university
Press.
Creath, Richard, and Michael Friedman. 2007. The Cambridge companion to Carnap.
Cambridge: Cambridge University Press.
Eddington, Arthur Stanley. 1928. The nature of the physical world. Cambridge:
Cambridge University Press.
Edmond, Gary, and David Mercer. 2004. The Invisible branch: The authority of sci-
ence studies in expert evidence jurisprudence. In Expertise in regulation and law,
edited by Gary Edmond, 197-242. Farnham, Surrey, UK: Ashgate.
Hahn, L. E., and P. A. Schilpp. 1986. The philosophy of W. V. Quine. Peru, IL: Open
Court.
Kolmogoroff, Andrey, 1933. Grundbegriffe der Wahrscheinlichkeitsrechnung. Berlin:
Julius Springer; 1956. Foundations of the Theory of Probability, 2nd ed. New York:
Chelsea.
Laudan, Larry. 1983. The demise of the demarcation problem. In Physics, philosophy
and psychoanalysis: Essays in honor of Adolf Grünbaum, edited by Robert S.
Cohen and Larry Laudan, 111-27. Boston Studies in the Philosophy of Science,
vol. 7. Dordrecht, the Netherlands: Reidel. (Reprinted in Michael Ruse, But Is
It Science? The Philosophical Question in the Creation/Evolution Controversy,
Buffalo, NY: Prometheus Books, 1988; and Larry Laudan, Beyond Positivism and
Relativism: Theory, Method, and Evidence, Boulder, CO: Westview, 1996)
Leblanc, Hugues 1989. The Autonomy of Probability Theory (Notes on Kolmogorov,
Rényi, and Popper). British Journal for the Philosophy of Science 40 (2):167-181.
Musgrave, Alan. 1995. Realism and idealization (metaphysical objections to scientific
realism). In The problem of rationality in science and its philosophy: on Popper
vs. Polanyi, the Polish conferences, 1988-89, edited by Józef Misiek, 143-66.
Boston Studies in the Philosophy of Science, vol. 160. Dordrecht, the Netherlands:
Reidel.
Popper, Karl R. 1935/1959. The logic of scientific discovery. London: Hutchinson.
Popper, Karl R. 1945. The open society and its enemies. London: Routledge.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014
294 Philosophy of the Social Sciences 41(2)

Popper, Karl R., 1955. Two Autonomous Axiom Systems for the Calculus of
Probabilities. British Journal for the Philosophy of Science 6 (21):51-57.
Popper, Karl R. 1963. Conjectures and refutations. London: Routledge.
Rényi, Alfréd, 1955. On a new axiomatic theory of probability. Acta Mathematica
Academiae Scientiarum Hungaricae 6, 285-335.
Rosenberg, Jay F. 2007. Wilfrid Sellars: Fusing the images. Oxford: Oxford University
Press.
Russell, Bertrand. 1912. The problems of philosophy. London: Williams and Norgate
and many later editions.
Russell, Bertrand. 1940. An Inquiry into meaning and truth. London: Allen & Unwin.
(Pelican ed., 1962)
Schilpp, Paul Arthur. 1944. The philosophy of Bertrand Russell. LaSalle, IL: Open
Court.
Watkins, John. 1957-1958. Epistemology and politics. Proceedings of the Aristotelian
Society 58:144-55.

Bio
Joseph Agassi, FRSC, is professor emeritus in Tel Aviv University and in York
University, Toronto. He is the author of about 20 books and editor of a few, as well
as of over 400 contributions to the learned press.

Downloaded from pos.sagepub.com at J. Robert Van Pelt and Opie Library on August 30, 2014