NORMATIVE

LOGIC AND ETHICS

PAULlL
- ~O
. RENZEN

PROBESSOR OF PHILOSOPHY AT T E E U N I V E R S I T Y E R L A N Q E X
A N D AT T H E U N I V E R S I T Y OR T E X A S I N AUSTIN

BIBLIOGRAPHISCHES INSTITUT . MANNHEIM/ZuRICH

 4; To KAETHE LORENZEN

A h Bechte vorbehalten. Nachdruck. mob auampmlse, verboten

.
O Blbfla~aphlwhesInatitut A 0 Mannhalrn 106V
 4; To KAETHE LORENZEN

A h Bechte vorbehalten. Nachdruck. mob auampmlse, verboten

.
O Blbfla~aphlwhesInatitut A 0 Mannhalrn 106V
 PREFACE

This book conhim the John-Locke-lecturesgiven during the

academic year 67-68 in Oxford.

Chapter VI (Modal Logic) has been added and Chapter VII

(Foundations of practical philosophy) has been considerably en-
larged.

I am grateful to Prof. Marjorie GRENE for many improvements

in the text.

Erlangen, September 1968 P a u l Lorenzen

 PREFACE

This book conhim the John-Locke-lecturesgiven during the

academic year 67-68 in Oxford.

Chapter VI (Modal Logic) has been added and Chapter VII

(Foundations of practical philosophy) has been considerably en-
larged.

I am grateful to Prof. Marjorie GRENE for many improvements

in the text.

Erlangen, September 1968 P a u l Lorenzen

 CONTENTS I
1 . Elementary Sentences ................ 11
2. Logical Particlea .................. 21
a1 Truths . . . . . . . . . . . . . . . . . . . 31
4 . Non-empirical Truths in the Formal Soiences . . . . . 40
5. Non-empirical Truths in the Material Sciences . . . . . 60
6 . Modal Logic . . . . . . . . . . . . . . . . . . . . 61
7. Foundations of Practical Philosophy . . . . . . . . . 73
 CONTENTS I
1 . Elementary Sentences ................ 11
2. Logical Particlea .................. 21
a1 Truths . . . . . . . . . . . . . . . . . . . 31
4 . Non-empirical Truths in the Formal Soiences . . . . . 40
5. Non-empirical Truths in the Material Sciences . . . . . 60
6 . Modal Logic . . . . . . . . . . . . . . . . . . . . 61
7. Foundations of Practical Philosophy . . . . . . . . . 73
 NORMATIVE LOGIC AND ETKICS

I . Elementary Bentences

I am grateful to have the privilege of speaking t o such a dimtin-

guished audience - the privilege of speaking as JOHN LOCKE
lecturer in Oxford-the privilege of speaking about philosophy
in this philosophical place war' QCopjv.
Exactly 300 years ago philosophy was quite different a t Oxford.
JOHN LOCKEhad just left Oxford-to be more precise, Christ
Church College. In 1667 he was living in London, where he wrote
hia first essay on toleration and began preparing his famous essay
concerning human understanding.
Philosophy was still taught in those days a t Oxford primarily
in the tradition of Occam, though there was a strong influence of
Bacon and Hobbes. Bolder spirits, like LOCKE,even studied
Descartes. This was the stream of thought which determined the
course of history. LOCKE was the English response to the Cartesian
challenge of making a philosophy out of the new exact science;
LEIBNIZw88 the German response. This dialogue between LOCKE
and LEIBNIZ was continued by HUME and KANT- but as YOU
know the dialogue deteriorated into monologues. I n the 19th and
20th century we have. for example, MILLand G. E. MOOREon the
English side, DILTHEYand HUSSERLon the German side. They
seem t o belong to different worlds. In the sciences the situation is
different. There is no one who can compare to NEWTON-yet the
English adopted the L E I B N I Zdifferentials,
~ and the fundamental
equations of electrodynamics are rightly called MAXWELL-HERTZ-
equations.
Similarly, the rise of modern formal logic is due t o the co-
operation of FREGE and RUSSELL.The so-called elementary pre-
dicate mlculus could rightly be called the FREGE-RUSSELL CBICUIUS.
I n these lectures I'am not going to talk about modern science.
I shall try to philosophize insteed. But what is that: "to philo-
sophize"?
 NORMATIVE LOGIC AND ETKICS

I . Elementary Bentences

I am grateful to have the privilege of speaking t o such a dimtin-

guished audience - the privilege of speaking as JOHN LOCKE
lecturer in Oxford-the privilege of speaking about philosophy
in this philosophical place war' QCopjv.
Exactly 300 years ago philosophy was quite different a t Oxford.
JOHN LOCKEhad just left Oxford-to be more precise, Christ
Church College. In 1667 he was living in London, where he wrote
hia first essay on toleration and began preparing his famous essay
concerning human understanding.
Philosophy was still taught in those days a t Oxford primarily
in the tradition of Occam, though there was a strong influence of
Bacon and Hobbes. Bolder spirits, like LOCKE,even studied
Descartes. This was the stream of thought which determined the
course of history. LOCKE was the English response to the Cartesian
challenge of making a philosophy out of the new exact science;
LEIBNIZw88 the German response. This dialogue between LOCKE
and LEIBNIZ was continued by HUME and KANT- but as YOU
know the dialogue deteriorated into monologues. I n the 19th and
20th century we have. for example, MILLand G. E. MOOREon the
English side, DILTHEYand HUSSERLon the German side. They
seem t o belong to different worlds. In the sciences the situation is
different. There is no one who can compare to NEWTON-yet the
English adopted the L E I B N I Zdifferentials,
~ and the fundamental
equations of electrodynamics are rightly called MAXWELL-HERTZ-
equations.
Similarly, the rise of modern formal logic is due t o the co-
operation of FREGE and RUSSELL.The so-called elementary pre-
dicate mlculus could rightly be called the FREGE-RUSSELL CBICUIUS.
I n these lectures I'am not going to talk about modern science.
I shall try to philosophize insteed. But what is that: "to philo-
sophize"?
12 Normative Logic and Ethiea I . Elementary S e n t e m 13

Let me begin with Locke's philosophy. He tried to understand Nevertheleas I have the hope that we can come to Bome common
modern science. As J. H. RANDALL formulates it: he "began with understanding of words and things, even assuming my rather poor
the rationalistic conception of Descastes and emerged with ob- knowledge of English.
servationalism". Of course, this result of his philosophizing may I t seem to me that the difference of opinion between MR.
be doubted but there is no historian of philosophy who seriously , MITCHELL and MR. STRAWSON may be described roughly - very
disagrees with the general appraise1 of RANDALL: Locm was 8 roughly indeed-aa the difference between the rationalist and
good, honest, plain-spoken Englishman. The standard German the empiricist opinion. Locke's endeavour to overcome Deaeertes'
history of philosophy, UBERUTE~, mys: "Er war ein wahrheita- rationalism was not a complete success. MR. STRAWSON we*
liebender, vorsichtiger, niichterner und frommer Charakter." that logical rules can be reduced to linguistic rules - and these sse
As far as I underatand Oxford philosophy, these values are- conceived of as empirical facts. MR. MITCHELL wants to know what
fortunetely - still in highest esteem, perhapa with one minor this "reduction" is, if not a logical deduction. As he does not get
deviation. The highest praise today would be: "He was a good, an amwer, he asserts that a t least aome logical rules are "absolute"
honest, plain-meta-spoken Englishman.': - a precondition for the acquisition of any empirical fact. This I
That is the whole problem. Why can we no longer speak in a should like to call rationalistic.
straightforward manner about, say, what we are doing, if we are
doing science? This question can be answered, it seems to me. Though this sounds rather hopeless, there remains the possibility
when we take into account that logic is an eeaential pert of all that some linguistic rules are logical rules. Then the disagreement
sciences. The acholaatics of the 16th and 18th century were in the between rationalism and empiricism could be formulated in terms
same situation: they could not speak about anything without of the following questions: 1. What are these logical linguistic
speaking about speaking. They knew, namely, that they should rules? 2. Are these rules to be accepted because they are factually
speak logically; and therefore, they tried,firat to i k logical rules. - accepted by the fluent speakers of some or all natural languages,
However this forced them into speaking about logic, that is, into or are they to be accepted because it is reasonable to speak accord-
metatak. ing to these rules, independently of their factual usage? We shall
Let us compare this with an argument of Mr. DAVIDMIWHELL still have the two opinions: of the empiricist, who states the
against the conventionalists (exemplified by Mr. STRAWSON). He acceptance of logical rules as a mtttter of fact. and of the rationalist,
attacks the claim that logical rules are based on linguistic rules. who accepts logical rules as a matter of reason.
"Any attempt to base logical principles on something more ulti- As the title of my lectures is: "Normative Logic and Ethics",
mate, whether it be our system of contingent rules for the use of you will gueea that I shall argue for the rationalist aide. But you
ianguage or anything elm, must be self-defeating. For the attempt may be aqsured that thia style of "arguing" will not consist in
consists of deducing conclusions from premisaes and for deduction "deducing". To argue in the way I propose, we shall have to look
to be poeaible the prior validity of logical lews is a prerequisite." ! first for a common basis, from which we can agree to begin. As
There ia an interesting note to this paaaage where MR. MITCHELL such a basis I p*pose the use of elementary sentences. Of course,
answers a "suggestion" to the effect that "to bese" one thing on ' we could even begin one step deeper, namely, with one-word-
another does not mean "to deduce" the first from the second. He phrases which are used as commands. Examples are: "Stop!",
answers that he does not understand what can be meant, if not "Faster!", "Silent!". Counter-examples are: "Yes?", "John!".
deduction. "Ouch!". Havinggiven these examples and counterexamples, I ask
Here now is the place where I should like to begin with apologiz- you, of course, to imagine appropriate situations. And in saying
ing for giving these lectures. How could I know if in thia context this, I expect that yo: understand the phrase "to imagine appro-
"to baee on" can mean something differentfrom "to deduce from" t ' priate situatiom". The phraeas which are used as commands I
In questions about the English languege I am- unfortunetely- propose to cell "imperative-phrasee". Now you will observe that
not a t ell competent. in talking about these one-word-phrases I have already reached,
i
12 Normative Logic and Ethiea I . Elementary S e n t e m 13

Let me begin with Locke's philosophy. He tried to understand Nevertheleas I have the hope that we can come to Bome common
modern science. As J. H. RANDALL formulates it: he "began with understanding of words and things, even assuming my rather poor
the rationalistic conception of Descastes and emerged with ob- knowledge of English.
servationalism". Of course, this result of his philosophizing may I t seem to me that the difference of opinion between MR.
be doubted but there is no historian of philosophy who seriously , MITCHELL and MR. STRAWSON may be described roughly - very
disagrees with the general appraise1 of RANDALL: Locm was 8 roughly indeed-aa the difference between the rationalist and
good, honest, plain-spoken Englishman. The standard German the empiricist opinion. Locke's endeavour to overcome Deaeertes'
history of philosophy, UBERUTE~, mys: "Er war ein wahrheita- rationalism was not a complete success. MR. STRAWSON we*
liebender, vorsichtiger, niichterner und frommer Charakter." that logical rules can be reduced to linguistic rules - and these sse
As far as I underatand Oxford philosophy, these values are- conceived of as empirical facts. MR. MITCHELL wants to know what
fortunetely - still in highest esteem, perhapa with one minor this "reduction" is, if not a logical deduction. As he does not get
deviation. The highest praise today would be: "He was a good, an amwer, he asserts that a t least aome logical rules are "absolute"
honest, plain-meta-spoken Englishman.': - a precondition for the acquisition of any empirical fact. This I
That is the whole problem. Why can we no longer speak in a should like to call rationalistic.
straightforward manner about, say, what we are doing, if we are
doing science? This question can be answered, it seems to me. Though this sounds rather hopeless, there remains the possibility
when we take into account that logic is an eeaential pert of all that some linguistic rules are logical rules. Then the disagreement
sciences. The acholaatics of the 16th and 18th century were in the between rationalism and empiricism could be formulated in terms
same situation: they could not speak about anything without of the following questions: 1. What are these logical linguistic
speaking about speaking. They knew, namely, that they should rules? 2. Are these rules to be accepted because they are factually
speak logically; and therefore, they tried,firat to i k logical rules. - accepted by the fluent speakers of some or all natural languages,
However this forced them into speaking about logic, that is, into or are they to be accepted because it is reasonable to speak accord-
metatak. ing to these rules, independently of their factual usage? We shall
Let us compare this with an argument of Mr. DAVIDMIWHELL still have the two opinions: of the empiricist, who states the
against the conventionalists (exemplified by Mr. STRAWSON). He acceptance of logical rules as a mtttter of fact. and of the rationalist,
attacks the claim that logical rules are based on linguistic rules. who accepts logical rules as a matter of reason.
"Any attempt to base logical principles on something more ulti- As the title of my lectures is: "Normative Logic and Ethics",
mate, whether it be our system of contingent rules for the use of you will gueea that I shall argue for the rationalist aide. But you
ianguage or anything elm, must be self-defeating. For the attempt may be aqsured that thia style of "arguing" will not consist in
consists of deducing conclusions from premisaes and for deduction "deducing". To argue in the way I propose, we shall have to look
to be poeaible the prior validity of logical lews is a prerequisite." ! first for a common basis, from which we can agree to begin. As
There ia an interesting note to this paaaage where MR. MITCHELL such a basis I p*pose the use of elementary sentences. Of course,
answers a "suggestion" to the effect that "to bese" one thing on ' we could even begin one step deeper, namely, with one-word-
another does not mean "to deduce" the first from the second. He phrases which are used as commands. Examples are: "Stop!",
answers that he does not understand what can be meant, if not "Faster!", "Silent!". Counter-examples are: "Yes?", "John!".
deduction. "Ouch!". Havinggiven these examples and counterexamples, I ask
Here now is the place where I should like to begin with apologiz- you, of course, to imagine appropriate situations. And in saying
ing for giving these lectures. How could I know if in thia context this, I expect that yo: understand the phrase "to imagine appro-
"to baee on" can mean something differentfrom "to deduce from" t ' priate situatiom". The phraeas which are used as commands I
In questions about the English languege I am- unfortunetely- propose to cell "imperative-phrasee". Now you will observe that
not a t ell competent. in talking about these one-word-phrases I have already reached,
i
14 Normative Logic and Elhicd I. Elementary Sentences 15

on the metalevel, the type of expression which I call 'elementary basis; we have all accepted a t least the use of elementary sentences.
sentences',, namely : Granted, it may be the case that some patterns of behaviour,
which all people of a certain group have accepted, are rather
"Silent!" is an imperative-phrase
stupid. For example, non-philosophers like to argue that philoso-
"John!" is not an imperative-phrase.
phizing is a stupid activity. B_ut no one can argue that the use of
The last sentence spoken was, of course, even more complicated alamantaqLsantaocas is stupid. Of course, it is easy to say: "Oh,
since it was on the metametalevel. This very word belongs to a how ridiculous that I am taIkingl", but then thia should he one's
metametametalevel. In this way we can easily obtain an infinite last utterance. +m&qumant deoenda on the use of e&mentery
progression- but this is not an infinite regression. If you prefer, sentences, one cannot argue that the use of elementary sentences
I need not introduce the word "imperative-phrase" - nevertheleas, lptq&
we could come to a common understanding of imperative-phrases. So we who have accepted the use of elementary sentences may
So let us not bother about poasible extensions on the metalevel: agree to call this usage "non-stupid" or "re_asonable". This does
let us work our way upwards with phrases applied to things, not not mean anything more than to confirm our decision, but we have
words: that is, with phrases on the so-called object-level. the consolation that those who silently, without speaking, look
Now suppose there is more than one person who could be silent. a t our behaviour cannot even think that we are stupid. For if
Then if you want John t o be silent, you may use "John! Silent!" we use the term "to think" in the sense of-Plato, thia would mean
as an imperative phrase. Let us distinguish the first word as a that they would speak t o their own eoul, using the elementary
Toper mme from the second word as a predicator. If he obeys, sentence "They are stupid".
it may be reported to you: "John is silent". It is such reporting What now about this playing with the phrases "to accept a
phrases, which I shall call "elementary sentences". It will be wage" and "a usage is stupid"1 Have we agreed to the rule that
sufficient to indicate that I am going t o use the linguistic exprea- we will not accept stupid wages? No. I did not yet propose this
,ion "elementary sentences" for phrases of the form: rule to you. But I should now like to do so, or let me rather propose
h affirmative rule: U E accepted 3 U E reasonable.
81,. ..,SnEp
(In moral philosophy this "subjective" notion of reason will be
81, ..., S , E ' ~ modified.)
with proper names 81, . . ., 8, and predicators p. & is used as the If we have accepted a usage, we call this usage "reasonable".
Englieh "is", E' as the English "is not". The word "reasonable" does not make any difference. The deeiaion
Obviously all of you havedecided to use suchelementrtry sentences always has t o be made; namely, whether to accept or not to accept
ratherfrequently. But what about a person who refuses t o use them O the usage, perhaps a new use.
If he refuses t o use words a t all, he will be brought t o a school In the case of elementary sentences nothing haa to he said. We
for mutes. If he has a kind family, they will try t o help him- but, do not persuade our children to begin to speak with the aaeertion
of course, they cannot argue with him. If he accepts the use of "To speak is reasonable", but simply by doing it, so that they may
imperativ,tive phrases, but refuses the use of elementary sentences, imitate us.
he will be treated as mentally retarded. Only after having accepted a certain number of linguistic
Why am I telling you this? The point is that this decision t o patterns, can we use these p a t t e r n "to argue". In the cese of
accept elementary ways of apeakin elementary sentences, we use proper names for naming objects
--
T;, dons
- not make sense to ask and we use predicators to assert or deny properties of the named
"reason". For t o "ssk" for such things demand8 a much more objects. But why do we do thia? "Bemuse there are objects and
compl~cateduse of language than the use of elementary sentencea because there are properties which the objects have or do not
'
~tself.If you ask such questlone, in other words, you have already have." This would be a fairly usual answer. But how do we know

-
accepted the more elemenGFpISge. Xo all or w do have a common thie? Well, it is eaay t o prove such existential eentencea, begin-
14 Normative Logic and Elhicd I. Elementary Sentences 15

on the metalevel, the type of expression which I call 'elementary basis; we have all accepted a t least the use of elementary sentences.
sentences',, namely : Granted, it may be the case that some patterns of behaviour,
which all people of a certain group have accepted, are rather
"Silent!" is an imperative-phrase
stupid. For example, non-philosophers like to argue that philoso-
"John!" is not an imperative-phrase.
phizing is a stupid activity. B_ut no one can argue that the use of
The last sentence spoken was, of course, even more complicated alamantaqLsantaocas is stupid. Of course, it is easy to say: "Oh,
since it was on the metametalevel. This very word belongs to a how ridiculous that I am taIkingl", but then thia should he one's
metametametalevel. In this way we can easily obtain an infinite last utterance. +m&qumant deoenda on the use of e&mentery
progression- but this is not an infinite regression. If you prefer, sentences, one cannot argue that the use of elementary sentences
I need not introduce the word "imperative-phrase" - nevertheleas, lptq&
we could come to a common understanding of imperative-phrases. So we who have accepted the use of elementary sentences may
So let us not bother about poasible extensions on the metalevel: agree to call this usage "non-stupid" or "re_asonable". This does
let us work our way upwards with phrases applied to things, not not mean anything more than to confirm our decision, but we have
words: that is, with phrases on the so-called object-level. the consolation that those who silently, without speaking, look
Now suppose there is more than one person who could be silent. a t our behaviour cannot even think that we are stupid. For if
Then if you want John t o be silent, you may use "John! Silent!" we use the term "to think" in the sense of-Plato, thia would mean
as an imperative phrase. Let us distinguish the first word as a that they would speak t o their own eoul, using the elementary
Toper mme from the second word as a predicator. If he obeys, sentence "They are stupid".
it may be reported to you: "John is silent". It is such reporting What now about this playing with the phrases "to accept a
phrases, which I shall call "elementary sentences". It will be wage" and "a usage is stupid"1 Have we agreed to the rule that
sufficient to indicate that I am going t o use the linguistic exprea- we will not accept stupid wages? No. I did not yet propose this
,ion "elementary sentences" for phrases of the form: rule to you. But I should now like to do so, or let me rather propose
h affirmative rule: U E accepted 3 U E reasonable.
81,. ..,SnEp
(In moral philosophy this "subjective" notion of reason will be
81, ..., S , E ' ~ modified.)
with proper names 81, . . ., 8, and predicators p. & is used as the If we have accepted a usage, we call this usage "reasonable".
Englieh "is", E' as the English "is not". The word "reasonable" does not make any difference. The deeiaion
Obviously all of you havedecided to use suchelementrtry sentences always has t o be made; namely, whether to accept or not to accept
ratherfrequently. But what about a person who refuses t o use them O the usage, perhaps a new use.
If he refuses t o use words a t all, he will be brought t o a school In the case of elementary sentences nothing haa to he said. We
for mutes. If he has a kind family, they will try t o help him- but, do not persuade our children to begin to speak with the aaeertion
of course, they cannot argue with him. If he accepts the use of "To speak is reasonable", but simply by doing it, so that they may
imperativ,tive phrases, but refuses the use of elementary sentences, imitate us.
he will be treated as mentally retarded. Only after having accepted a certain number of linguistic
Why am I telling you this? The point is that this decision t o patterns, can we use these p a t t e r n "to argue". In the cese of
accept elementary ways of apeakin elementary sentences, we use proper names for naming objects
--
T;, dons
- not make sense to ask and we use predicators to assert or deny properties of the named
"reason". For t o "ssk" for such things demand8 a much more objects. But why do we do thia? "Bemuse there are objects and
compl~cateduse of language than the use of elementary sentencea because there are properties which the objects have or do not
'
~tself.If you ask such questlone, in other words, you have already have." This would be a fairly usual answer. But how do we know

-
accepted the more elemenGFpISge. Xo all or w do have a common thie? Well, it is eaay t o prove such existential eentencea, begin-
16 Nonnative Logic and Ethied

ning with "there are objects", "there ere properties" by giving I sssume that you are in the same situation as I: namely, of
examples. " S O C R ~ T E is~ a philosopher" proves that there ere not knowing any ''Ersatz" or different behaviow to replace the
objects: namely, SOORATES, and t h h there are properties: namely, use of proper names and predicators. This means that we have
to be a philosopher. the common baais of having accepted the use of elementary
sentences as a reasonable usage. It does not mean that you have
It will he obvious by now that we prove the essertion by ming
agreed to the aasertion that there are objects which have proper.
an elementary sentence. Our first question is: "Why do we use
ties, hut i t is now easy to get to this aasertion. As we are already
elementm sentences 9"
using proper names, we now introduce the term "object". We may
We e t n -and we need no answer. We do not prescribe
then say that our proper names name objecta. Of course, we
that--st-t to our children the camot introduce the term "property" in this way because, if we
use of proper names. We do not prescribe that there are properties -: are using a predicator. we just have the objects of which we are
but we rec6mmend to our children the use of predicators. affirming or denying the predicator- but we do not have anything
To this distinguished audience I surely will not have to re. other than the objecta and the predicator. Where is a property
commend the use of elementery sentences. I can only remind you "represented" by the predicator? To the best of my knowledge:
that we all accepted this usage long ago. Once more, provisionally, Nowhere.
we may call tthi s g e “reasonable". So, why is it reasonable to say "8 has the property p" instead
With linguistic rules of a particular natural language the situa- of saying "8 is p"? I t does not look reasonable, because it seems
tion is dierent. Surely all those who have consented to speak to be merely longer. However there are intermediate forms of
English a t aU say, for example, "Socrates does not write poetry", sentences which are all true if and only if "8 is p" is true and it is
if they want to deny the predicator "writing poetry" of SO CRATE^. rewonable to use these intermediate forma. Let us begin with
It is 8 convention to use "to do" in such negative sentences,
"It is true, that S ia p"- I believe I need not repeat the situations
however, i t is not a convention to use negative sentences. Why in which this latter form makes sense. Then inatead of "It is true
not 8 convention? Instead of answering this question, let me that S is p", we may say "p applies to 8"- this is only a shift of
a m r the question: "Why don't I call it a conventionl" Well, emphasis to the predicator p. We may then say, a bit more expli-
I call a usage a convention if I know of another usage which I citly. "the predicator p applies to S'.
could accept instead. Insteed of using "to do" in negative sentences, Now comes the decisive step to "S has the property p". This step
there would be the possibility of affirming predicator6 in the form is justified if we are aware of the conventional choice of the predi-
S 8 p and of denying them in the form S E' p. cator. We could have chosen any other sound for the same purpose,
However I do not know of another behaviour which could namely, for the purpose of distinguishing objects by affirming or
replace the use of elementary sentences. If I did not accept proper denying the sound of them. If I am saying "S has the property p",
names and predicators. I would not know bow to speak a t all. , I am not only just asserting "S is p" or: "It is true that S is p".
Of course. I could happily be silent-for a while. But I know but I am in the same moment reflecting on my predicating.
this and only this way of beginning to speak-namely, beginning con6rming that p applies to S,and, in addition, expressing that
with elementary sentences in any of the conventional forms. Each I am not interested in the particular predicator p of the English
proper name is a convention (because I know many sounds I could language, but that I am interested in p only insofar as I may substi-
use instead), but to Use a proper name at all i~ not a convention: tute for p any other sound with the same use.
it is a unique pattern of linguistic behaviow. Therefore, I am going I am sorry that it is such a long story to explain what I propose
to call i t "logical". The same is the case with predicators. Each to be doing by saying "S has the property p" - hut this is quite
predicator is a convention. This is shown by the existence of more frrUlkly my proposal. In more usual terms, the step from "the
than one natural language. But a11 languagw d o u s e predicatom. pmhcator p applies to S" to "Shas the property p" is an nbshoetia.
This is a logical feature of our linguistic behaviour. In order to express the abstraction from the perticularities of the
16 Nonnative Logic and Ethied

ning with "there are objects", "there ere properties" by giving I sssume that you are in the same situation as I: namely, of
examples. " S O C R ~ T E is~ a philosopher" proves that there ere not knowing any ''Ersatz" or different behaviow to replace the
objects: namely, SOORATES, and t h h there are properties: namely, use of proper names and predicators. This means that we have
to be a philosopher. the common baais of having accepted the use of elementary
sentences as a reasonable usage. It does not mean that you have
It will he obvious by now that we prove the essertion by ming
agreed to the aasertion that there are objects which have proper.
an elementary sentence. Our first question is: "Why do we use
ties, hut i t is now easy to get to this aasertion. As we are already
elementm sentences 9"
using proper names, we now introduce the term "object". We may
We e t n -and we need no answer. We do not prescribe
then say that our proper names name objecta. Of course, we
that--st-t to our children the camot introduce the term "property" in this way because, if we
use of proper names. We do not prescribe that there are properties -: are using a predicator. we just have the objects of which we are
but we rec6mmend to our children the use of predicators. affirming or denying the predicator- but we do not have anything
To this distinguished audience I surely will not have to re. other than the objecta and the predicator. Where is a property
commend the use of elementery sentences. I can only remind you "represented" by the predicator? To the best of my knowledge:
that we all accepted this usage long ago. Once more, provisionally, Nowhere.
we may call tthi s g e “reasonable". So, why is it reasonable to say "8 has the property p" instead
With linguistic rules of a particular natural language the situa- of saying "8 is p"? I t does not look reasonable, because it seems
tion is dierent. Surely all those who have consented to speak to be merely longer. However there are intermediate forms of
English a t aU say, for example, "Socrates does not write poetry", sentences which are all true if and only if "8 is p" is true and it is
if they want to deny the predicator "writing poetry" of SO CRATE^. rewonable to use these intermediate forma. Let us begin with
It is 8 convention to use "to do" in such negative sentences,
"It is true, that S ia p"- I believe I need not repeat the situations
however, i t is not a convention to use negative sentences. Why in which this latter form makes sense. Then inatead of "It is true
not 8 convention? Instead of answering this question, let me that S is p", we may say "p applies to 8"- this is only a shift of
a m r the question: "Why don't I call it a conventionl" Well, emphasis to the predicator p. We may then say, a bit more expli-
I call a usage a convention if I know of another usage which I citly. "the predicator p applies to S'.
could accept instead. Insteed of using "to do" in negative sentences, Now comes the decisive step to "S has the property p". This step
there would be the possibility of affirming predicator6 in the form is justified if we are aware of the conventional choice of the predi-
S 8 p and of denying them in the form S E' p. cator. We could have chosen any other sound for the same purpose,
However I do not know of another behaviour which could namely, for the purpose of distinguishing objects by affirming or
replace the use of elementary sentences. If I did not accept proper denying the sound of them. If I am saying "S has the property p",
names and predicators. I would not know bow to speak a t all. , I am not only just asserting "S is p" or: "It is true that S is p".
Of course. I could happily be silent-for a while. But I know but I am in the same moment reflecting on my predicating.
this and only this way of beginning to speak-namely, beginning con6rming that p applies to S,and, in addition, expressing that
with elementary sentences in any of the conventional forms. Each I am not interested in the particular predicator p of the English
proper name is a convention (because I know many sounds I could language, but that I am interested in p only insofar as I may substi-
use instead), but to Use a proper name at all i~ not a convention: tute for p any other sound with the same use.
it is a unique pattern of linguistic behaviow. Therefore, I am going I am sorry that it is such a long story to explain what I propose
to call i t "logical". The same is the case with predicators. Each to be doing by saying "S has the property p" - hut this is quite
predicator is a convention. This is shown by the existence of more frrUlkly my proposal. In more usual terms, the step from "the
than one natural language. But a11 languagw d o u s e predicatom. pmhcator p applies to S" to "Shas the property p" is an nbshoetia.
This is a logical feature of our linguistic behaviour. In order to express the abstraction from the perticularities of the
18 Nonnative Logic and Ethics 1. Elementary Sentemu 19

chosen predicator, the interest shifts t o the use which might be The proposition that S ~p is then represented, e. g. by the
performed with any other sound. I n order to expreea this, I propose sentence "T E q".
t o say "S has the property p" instead of speaking about the This procedure can obviously be extended to two-place sentences
predicator p. and then obviously to any number of places. For example if we
I n general I propose to use a sentence of the form A (property come t o two-place sentences, S, T E p and T, S E q, they may be
.
0 .) if I want to assert A (predicator p) in abstraction from the called synonymous if .p and q aro explicitly introduced as converse
2-place predicators.
of the predicator p.
Let q be any predicator which I a m using synonymously with p, Still more complications arise if we extend our elementary
e. g. "rot" instead of "red", and let A (predicator p) be a sentence sentences in such a way that deaeriptiom are d o w e d instead
such that: of proper names. Let me call the new elementary sentences "semi-
A (predicator p) iff A (predicator q) elementnry". Elementary sentences have the form:
e. g. "red applies to S'iff "rot applies t o S ' t h e n I shall call, the
sentence A (predicator p) invariant, more precisely : invariant with
with c for E, r' and with proper names Sl, . . ., S and the pre-
respect to synonymity.
dicator p..
I n the case of invariant sentences only, I propose to say "A
By replacing any one of the "subjects" 51,. . ., S, by a variable
(property p)" instead of "A (predicator p)". My claim is that this
z we form the phrese
proposed way of speaking is fairly reasonable. I do not say that it
is necessary. You may not he interested in the invariance with
respect to synonymity a t all; hut if you are, it is convenient to use
(which we shall read the z with SI,. . ., x,. . .. S, c p).
this proposed technique of abstraction. This technique is applicable

-
whenever there is an equivalence relation "-", such as synonymity
between parts of our sentences. If x y implies A(x) iff A(y). A(x)
will be celled invariant with respect t o -. I invent a n abatmctor
Such phrases may he called potential deaeriptions. They have
the form 1% A(x), where A(x) is a sentence-form. These potential
descriptions are called proper descriptions if and only if there
exists exactly one object S with A(S).
,Iand write in the csse of invbiant sentences A(=) only:
I would like t o call the other potentid descriptions p e e d -
A(A z ) instead of A(x) . descriptions. For example: from "z, England E king" we get
I shall say that the object x represents the abstract object Ax. I, z. England E king
I n the case of synonymity I am using the word "property" as such
an abstroclm. Each property is, thereby, an abstrmt object, (read as: the king of England).
represented by synonymous predicators. Using the word "property" The problem is whether or not to introduce propositions repre-
in this way, you will admit that I have proven my case: namely, sented by such sentences as e. g. "the king of England is old".
that there are objects which have properties. If any one really desires such propositions, I would recommend
The term "proposition" may also be introduced as an abstractor. that we allow h ~ mto assert them. On the other hand we can
However this presupposes that we extend the equivalence relation obviously get along without them. Therefore, my proposal would
of synonymity from predicators to elementary sentences. This is be to use only proper descriptions in "meaningful" sentences.
no difficulty; different proper names are called synonyms if they i. e. in sentences for which we are going to introduce propositions
are used to name the same object. represented by them.
If 1, A(=) is a proper description- and if we let A(S) be true -
it is reesonable to use'this description as subject in another ele-
synonymous if and only if 9 and T are aynonyms and if p mentary sentence B(r, A(x)) by using this sentence aa synonymous
and q are sgnonymoua. with B (5).
18 Nonnative Logic and Ethics 1. Elementary Sentemu 19

chosen predicator, the interest shifts t o the use which might be The proposition that S ~p is then represented, e. g. by the
performed with any other sound. I n order to expreea this, I propose sentence "T E q".
t o say "S has the property p" instead of speaking about the This procedure can obviously be extended to two-place sentences
predicator p. and then obviously to any number of places. For example if we
I n general I propose to use a sentence of the form A (property come t o two-place sentences, S, T E p and T, S E q, they may be
.
0 .) if I want to assert A (predicator p) in abstraction from the called synonymous if .p and q aro explicitly introduced as converse
2-place predicators.
of the predicator p.
Let q be any predicator which I a m using synonymously with p, Still more complications arise if we extend our elementary
e. g. "rot" instead of "red", and let A (predicator p) be a sentence sentences in such a way that deaeriptiom are d o w e d instead
such that: of proper names. Let me call the new elementary sentences "semi-
A (predicator p) iff A (predicator q) elementnry". Elementary sentences have the form:
e. g. "red applies to S'iff "rot applies t o S ' t h e n I shall call, the
sentence A (predicator p) invariant, more precisely : invariant with
with c for E, r' and with proper names Sl, . . ., S and the pre-
respect to synonymity.
dicator p..
I n the case of invariant sentences only, I propose to say "A
By replacing any one of the "subjects" 51,. . ., S, by a variable
(property p)" instead of "A (predicator p)". My claim is that this
z we form the phrese
proposed way of speaking is fairly reasonable. I do not say that it
is necessary. You may not he interested in the invariance with
respect to synonymity a t all; hut if you are, it is convenient to use
(which we shall read the z with SI,. . ., x,. . .. S, c p).
this proposed technique of abstraction. This technique is applicable

-
whenever there is an equivalence relation "-", such as synonymity
between parts of our sentences. If x y implies A(x) iff A(y). A(x)
will be celled invariant with respect t o -. I invent a n abatmctor
Such phrases may he called potential deaeriptions. They have
the form 1% A(x), where A(x) is a sentence-form. These potential
descriptions are called proper descriptions if and only if there
exists exactly one object S with A(S).
,Iand write in the csse of invbiant sentences A(=) only:
I would like t o call the other potentid descriptions p e e d -
A(A z ) instead of A(x) . descriptions. For example: from "z, England E king" we get
I shall say that the object x represents the abstract object Ax. I, z. England E king
I n the case of synonymity I am using the word "property" as such
an abstroclm. Each property is, thereby, an abstrmt object, (read as: the king of England).
represented by synonymous predicators. Using the word "property" The problem is whether or not to introduce propositions repre-
in this way, you will admit that I have proven my case: namely, sented by such sentences as e. g. "the king of England is old".
that there are objects which have properties. If any one really desires such propositions, I would recommend
The term "proposition" may also be introduced as an abstractor. that we allow h ~ mto assert them. On the other hand we can
However this presupposes that we extend the equivalence relation obviously get along without them. Therefore, my proposal would
of synonymity from predicators to elementary sentences. This is be to use only proper descriptions in "meaningful" sentences.
no difficulty; different proper names are called synonyms if they i. e. in sentences for which we are going to introduce propositions
are used to name the same object. represented by them.
If 1, A(=) is a proper description- and if we let A(S) be true -
it is reesonable to use'this description as subject in another ele-
synonymous if and only if 9 and T are aynonyms and if p mentary sentence B(r, A(x)) by using this sentence aa synonymous
and q are sgnonymoua. with B (5).
20 Nonnative Logic and EWIied

This wage for proper descriptions is once more a linguistic "not" 88 junctom; "for all", "for some" as quantifie~a;but again
technique, which could be avoided, hut which is convenient. I n we will have to establish, to justify, norms for their uae.
contrast t o the technique of abstractors, the technique of descrip- , The instruction for the use of abstractors needed the statement
tion is rather unavoidable in practice. I n practical life we cannot of the condition of invariance
always introduce proper names. We use indicators such as "this",
"you" and so on, and we use indicator-descriptions such as "my
dog". Also, the simple sentence "the dog is barking" in its ordinary The instruction for the use of the description-operetor needed
usage does not presuppose that there exists exactly one dog, only the statement of the conditions of unique existence:
that there exists exactly one dog here and now in the situation in
VZ A @ )
which the sentence is wed. The definite article "the" in "the dog AW .A(4 A A ( Y )+ x = Y .
is harking" is not the logical operator r. It is rather a n abbreviated
indicator "this". As we 8ee here, a t least the quantifiers A , V and the junctors
But let me repeat that I am not competent a t all to assert any- A , -+ are needed.

thing about the English language. The purpose of talking about But why should we accept the techniques of composing sentences
elementary sentences is only t o perform a logical construction with such particles? What rules ahould we accept and why ahould
of these sentences. Obviously it is a logical reconstruction of what we accept just these rules? If in the English language certain
we have always done in natural languages, but this is merely a rules, for example, about substituting the simple A for not not A,
contingent fact. Only in the case of elemenbary sentences in the were generally accepted, - I do not know whether this is the case
strict sence we have no choice; either we accept the use of proper - b u t in any case, this would only be a contingent fact. This
names and predicators or we do not speak a t all. I n introducing alone would be no reason for me t o accept the usage. The problem
abstractors and descriptions we can more or less deviate from which we encounter here is the problem of a "foundation" for formal
traditional linguistic techniques. We will have to use our reason logic.
to decide which ways of speaking we will accept as logical ways of .
speaking. We will have t o establish norms for our speaking. We
may compare these with the factual usage in our natural language, 2. L o g i d Particles
but the factual will never give a justification. In the beginning we
can have no theoretical justification, i. e. n o justification by In these lectures I am trying to come with you to a common
arguments-we have t o take the risk of actually beginning. understanding of certain sentences. The difficulty seems not only
Then immediately after accepting elementary sentences, we have to lie in my difficulties to understand English- it seems to lie
a t once the choice of extending our linguistic meem, e. g. by ab-. more in my use of the term "understand". I am using it as syn-
stractors or descriptions, but thia extension can be accomplished onymous with the German "Verstehen". "Verstehen" is very
by speaking about our use of elementary sentences. We need not fashionable in German philosophy nowadays- though mostly
use abstractors or descriptions on the meta-level in order to : the Greek term "hermeneutics" is preferred. Hermeneutical
introduce them on the object-level; however, if we reflect on what philosophy does not look for theories, systems of true assertions,
we have done, we can see that we have used some linguistic meana but tries to understand-to understand man, t o understand the
in addition to elementary sentences in order to be able to my world, t o understand art, t o understand language, i. e. t o under-
what the new techniques are. This irreducible factor, which has stand everything.
to be supplied in addition t o elementary sentences, if we want to ; In the lest lecture y e have - I hope - agreed that we under-
be able to reduce abstractors and descriptions, consists- as you @d st least elementary sentences, e. g.,
will not be surprised to hem- of the logical particles, junctors and - .

InEnglish1 shall use the words "and", "or", "if- then", ' S o c a a ~ ~iaswise',
20 Nonnative Logic and EWIied

This wage for proper descriptions is once more a linguistic "not" 88 junctom; "for all", "for some" as quantifie~a;but again
technique, which could be avoided, hut which is convenient. I n we will have to establish, to justify, norms for their uae.
contrast t o the technique of abstractors, the technique of descrip- , The instruction for the use of abstractors needed the statement
tion is rather unavoidable in practice. I n practical life we cannot of the condition of invariance
always introduce proper names. We use indicators such as "this",
"you" and so on, and we use indicator-descriptions such as "my
dog". Also, the simple sentence "the dog is barking" in its ordinary The instruction for the use of the description-operetor needed
usage does not presuppose that there exists exactly one dog, only the statement of the conditions of unique existence:
that there exists exactly one dog here and now in the situation in
VZ A @ )
which the sentence is wed. The definite article "the" in "the dog AW .A(4 A A ( Y )+ x = Y .
is harking" is not the logical operator r. It is rather a n abbreviated
indicator "this". As we 8ee here, a t least the quantifiers A , V and the junctors
But let me repeat that I am not competent a t all to assert any- A , -+ are needed.

thing about the English language. The purpose of talking about But why should we accept the techniques of composing sentences
elementary sentences is only t o perform a logical construction with such particles? What rules ahould we accept and why ahould
of these sentences. Obviously it is a logical reconstruction of what we accept just these rules? If in the English language certain
we have always done in natural languages, but this is merely a rules, for example, about substituting the simple A for not not A,
contingent fact. Only in the case of elemenbary sentences in the were generally accepted, - I do not know whether this is the case
strict sence we have no choice; either we accept the use of proper - b u t in any case, this would only be a contingent fact. This
names and predicators or we do not speak a t all. I n introducing alone would be no reason for me t o accept the usage. The problem
abstractors and descriptions we can more or less deviate from which we encounter here is the problem of a "foundation" for formal
traditional linguistic techniques. We will have to use our reason logic.
to decide which ways of speaking we will accept as logical ways of .
speaking. We will have t o establish norms for our speaking. We
may compare these with the factual usage in our natural language, 2. L o g i d Particles
but the factual will never give a justification. In the beginning we
can have no theoretical justification, i. e. n o justification by In these lectures I am trying to come with you to a common
arguments-we have t o take the risk of actually beginning. understanding of certain sentences. The difficulty seems not only
Then immediately after accepting elementary sentences, we have to lie in my difficulties to understand English- it seems to lie
a t once the choice of extending our linguistic meem, e. g. by ab-. more in my use of the term "understand". I am using it as syn-
stractors or descriptions, but thia extension can be accomplished onymous with the German "Verstehen". "Verstehen" is very
by speaking about our use of elementary sentences. We need not fashionable in German philosophy nowadays- though mostly
use abstractors or descriptions on the meta-level in order to : the Greek term "hermeneutics" is preferred. Hermeneutical
introduce them on the object-level; however, if we reflect on what philosophy does not look for theories, systems of true assertions,
we have done, we can see that we have used some linguistic meana but tries to understand-to understand man, t o understand the
in addition to elementary sentences in order to be able to my world, t o understand art, t o understand language, i. e. t o under-
what the new techniques are. This irreducible factor, which has stand everything.
to be supplied in addition t o elementary sentences, if we want to ; In the lest lecture y e have - I hope - agreed that we under-
be able to reduce abstractors and descriptions, consists- as you @d st least elementary sentences, e. g.,
will not be surprised to hem- of the logical particles, junctors and - .

InEnglish1 shall use the words "and", "or", "if- then", ' S o c a a ~ ~iaswise',
 Nwmative and Ethics 2. Logicel Particles 23

or semi-elementary sentences such as: had an abstractor or description", how could I answer the question.
'The teacher of PLATO is WiBB.' "Why am I using "or" here?" without wing "or" on the metalevel?
'SOCRATES has the property wise.' The usual question is: What does "or" mean? Obviously i t is
eaay to learn the how-how of using "or"; all native speakers have
To understand a sentence very often means to know how to learned it. But what about understanding philosophically what
use it, that is, to have the "know-how" of its use. But this is not we me doing when we use such logical particles as "or"?
understanding as hermeneutical philosophy understands man and GODEL proved (1931) that in order to prove the consistency of
his world. To understand a asntence in this philosophical sense certain axiomatic theories you must have a meta-theory which is
of understanding means not the know-how of its use, but the in certain respects richer than the object theory.
know-why. Why do we use proper names? Why do we use pre- Since then it has become the mcepted style not t o be critical of
dicators? Only after having made up o w minds on these questions, the linguistic means on the meta-level. If one talks on the meta-
may we say that we understand philosophically what we are doing level, one normally says: "I am talking in a meta-languagew-
when we use element,ary sentences. though it is aU more or less deteriorated English.
I n this lecture I should like to prooeed to non-elementary Now, to the best of my knowledge, Godel's proof is correct,
sentences. I am going to extend the class of sentences which we but I have written a textbook on Metamathematics with the
understand. hidden purpose of showing that the so-called philosophical conse-
Let us take an example: "Logically true sentence8 are true in d l quences which are normally drawn from it have nothing to do
possible worlds." Do you understand this sentence? Perhaps you do with it.
-
not believe it but you understand it, do you not l Perhaps some If namely I begin-as I propose to do, in the tradition of
of you may even understand this sentence in the philosophical Cartesian doubt- to doubt whether I understand "or", then
s e w , but I have t o a b i t that I cannot understand this sentence it is of no help to teach me the use of "or" with the help of meta-
with the meam which I have developed so far. Strictly speaking, t a k which conteina "or". Thus, in o w case it is reasonable to
therefore, I cannot even know whether this sequence of words insist on a method of introducing "or" into that part of language
is a sentence a t all. Up t o now, I understand only elementary which we are going to use - without already using logical particles
sentences and semi-elementary sentences which contain abstrac- on a mete-level.
tors or descriptions. The usual truth-table method does not seem to satisfy this
Let us reflect on what I have just said: "Semi-elementary sen- requirement. The truth-table method begins with the assertion
tences contain abstractors or descriptions." Do you understand that all sentences ere true or false. We can avoid this by giving the
this? I do hope, of course, that you understand- because I would truth-table the form of rules. For "or" we have the following mles:
not have said it otherwise. However, now we have the rather odd
AcT,,BcT + A v BcT
situation that I ought not t o understand this, because i t is not &I
AeT.,BzF +A v BET
elementav sentence.
AEF,,BET 3 A V BET
If I had a i d about one of my earlier sentences, e. g.
AcF,.BcF +A v BEF
"My sentence contained an ahstractor."
or "My sentence contained a deacription." (with T = true, F = false).
Of course, we have to understand the . . . ,. . . . 3 . .. notation
then I would he entitled to understand, since both are semi- for rules, however, it is easy to learn whet we have to do if we are
elementary sentences with a description "my sentencew-and "follow" such rules by sufficient practice.
I will take the rest, "had a so-and-so", as a predicator. True, both I n the case of elementary sentences I mean by (Se p ) e F
sentences are on the metalevel, but they are =mi-elementary nothing more than (S E' p) E T and correspondingly (S e' p) e F
nevertheleea. The difficulty lies in the "or". If I say, "My sentence +(Scp) eT.
 Nwmative and Ethics 2. Logicel Particles 23

or semi-elementary sentences such as: had an abstractor or description", how could I answer the question.
'The teacher of PLATO is WiBB.' "Why am I using "or" here?" without wing "or" on the metalevel?
'SOCRATES has the property wise.' The usual question is: What does "or" mean? Obviously i t is
eaay to learn the how-how of using "or"; all native speakers have
To understand a sentence very often means to know how to learned it. But what about understanding philosophically what
use it, that is, to have the "know-how" of its use. But this is not we me doing when we use such logical particles as "or"?
understanding as hermeneutical philosophy understands man and GODEL proved (1931) that in order to prove the consistency of
his world. To understand a asntence in this philosophical sense certain axiomatic theories you must have a meta-theory which is
of understanding means not the know-how of its use, but the in certain respects richer than the object theory.
know-why. Why do we use proper names? Why do we use pre- Since then it has become the mcepted style not t o be critical of
dicators? Only after having made up o w minds on these questions, the linguistic means on the meta-level. If one talks on the meta-
may we say that we understand philosophically what we are doing level, one normally says: "I am talking in a meta-languagew-
when we use element,ary sentences. though it is aU more or less deteriorated English.
I n this lecture I should like to prooeed to non-elementary Now, to the best of my knowledge, Godel's proof is correct,
sentences. I am going to extend the class of sentences which we but I have written a textbook on Metamathematics with the
understand. hidden purpose of showing that the so-called philosophical conse-
Let us take an example: "Logically true sentence8 are true in d l quences which are normally drawn from it have nothing to do
possible worlds." Do you understand this sentence? Perhaps you do with it.
-
not believe it but you understand it, do you not l Perhaps some If namely I begin-as I propose to do, in the tradition of
of you may even understand this sentence in the philosophical Cartesian doubt- to doubt whether I understand "or", then
s e w , but I have t o a b i t that I cannot understand this sentence it is of no help to teach me the use of "or" with the help of meta-
with the meam which I have developed so far. Strictly speaking, t a k which conteina "or". Thus, in o w case it is reasonable to
therefore, I cannot even know whether this sequence of words insist on a method of introducing "or" into that part of language
is a sentence a t all. Up t o now, I understand only elementary which we are going to use - without already using logical particles
sentences and semi-elementary sentences which contain abstrac- on a mete-level.
tors or descriptions. The usual truth-table method does not seem to satisfy this
Let us reflect on what I have just said: "Semi-elementary sen- requirement. The truth-table method begins with the assertion
tences contain abstractors or descriptions." Do you understand that all sentences ere true or false. We can avoid this by giving the
this? I do hope, of course, that you understand- because I would truth-table the form of rules. For "or" we have the following mles:
not have said it otherwise. However, now we have the rather odd
AcT,,BcT + A v BcT
situation that I ought not t o understand this, because i t is not &I
AeT.,BzF +A v BET
elementav sentence.
AEF,,BET 3 A V BET
If I had a i d about one of my earlier sentences, e. g.
AcF,.BcF +A v BEF
"My sentence contained an ahstractor."
or "My sentence contained a deacription." (with T = true, F = false).
Of course, we have to understand the . . . ,. . . . 3 . .. notation
then I would he entitled to understand, since both are semi- for rules, however, it is easy to learn whet we have to do if we are
elementary sentences with a description "my sentencew-and "follow" such rules by sufficient practice.
I will take the rest, "had a so-and-so", as a predicator. True, both I n the case of elementary sentences I mean by (Se p ) e F
sentences are on the metalevel, but they are =mi-elementary nothing more than (S E' p) E T and correspondingly (S e' p) e F
nevertheleea. The difficulty lies in the "or". If I say, "My sentence +(Scp) eT.
24 Normative Logic and Ethiur 2. Logical Particlea 25

it. This meta-dialogical aasertion is, of course, rather difficult to

Let us call a sentence (if you prefer: the proposition represented
by the sentence) definite if it is true or fabe. We can formulate defend. Strictly speaking, it has up to now no agreed upon meaning.
This shows that in dealing with quantified sentences the classi-
this proposal without using "or" by proposing the following rules:
cal disjunction "either true or false" no longer applies. It is, how-
A E T + A E definite ever, posaible t o introduce a negation --r by fixing its diagolical use:
A E F 3 A E definite

I am not going t o bother you with the question whether all ele-
mentary sentences are definite. It would be a reasonable move to with no response other than to counterattack A.
restrict our attention to definite sentences -and it would follow - For the some-quantifier we have already fixed the following attack-
that, if we start with defmite elementary sentences, all composite defense-rule:
sentences are defmite. For the composition of the latter we can use
the iunctors: A , V , -, according to rules which correspond to
the truth-tables.
This two-valued-approach to logic is called classical (although For the all-quantifier we may now propose the following attack-
it,
.. is-~ verv anti-Aristotelian). The difficulty with the classical defense-rule :
approach lies - as BROUWERdiscovered 60 years ago -
in the
quantifiers. We may introduce the some-quantifiek (the so-called
"existential quintifier") by the rule: Here the opponent chooses an object S. If an all-sentence has
been defended against an attack, this does not justify calling the
thesis "true". I propose to call a composite sentence true if and
only if i t can be defended as a thesis against all "possible" opposition;
From this rule no one up to now has been able to prove that e. g., but this proposal does not make sense as long as the rules of the
the sentence "some odd numbers am perfect" is true. For what dialogical game, the game of attacking and defending sentences,
we have here is no longer an elementery sentence. The symbol T has not been determined in full detail. In order to do this, I begin
is not a predicator; it only looks like one. We do not understand by introducing junctora into this game. But we cannot use the
an elementary sentence S s p, unless we underatand S E' p also. '
junctora as they are determined by the truth-tables, because the
Predication is the decision between affirming or denying the predi- use of quantifiers may yield sentences which have no truth-value.
cator p of the object S. However, t o assert a some-sentence ia For the adjunction v (I use this term for the non-exclusive
something different, because we have to defend such an aaaertion disjunction) we have in analogy to the some-quantifier two rules:

.
according to the given rule: we have to name an object S. There-
fore, the sequence of symbols V % A (x) &' T (or V A (x) E F) has
no meaning up to now; it has no use that has been explicitly
agreed upon.
--,n let,
R .-- us introduce the assertion of - V z A(x) by the follow-
ing proposal: If I m e r t this, you may aaaert the a f h a t i v e part,
The proponent of A v B may chose which part he will defend
after the opponent has attacked by "?". For the conjunction A the
attacking opponent shall have the choice of the left or right part,
i. e. v A(x) and then I shall ask you to defend it. If you succeed, ' which the proponent has to defend. We indicate this choice by
I will have loat my assertion; otherwise, I shall have won. In ease
: L 7, or R? respectively; so we get 'two rules in analogy to the
not to challenge the original thesis -
you do not know how to defend V z A(x), i t is Bdvisable for you
V. A(x).
I do not propose to call -. V. A ( z ) true if you munot refute
; dl-quantifier:
. < ..,
>-,
... .
\?.,,
A A B
the aa&ion of it: I would cell it "true" only if no one could refute .* A A B
,, I . .
24 Normative Logic and Ethiur 2. Logical Particlea 25

it. This meta-dialogical aasertion is, of course, rather difficult to

Let us call a sentence (if you prefer: the proposition represented
by the sentence) definite if it is true or fabe. We can formulate defend. Strictly speaking, it has up to now no agreed upon meaning.
This shows that in dealing with quantified sentences the classi-
this proposal without using "or" by proposing the following rules:
cal disjunction "either true or false" no longer applies. It is, how-
A E T + A E definite ever, posaible t o introduce a negation --r by fixing its diagolical use:
A E F 3 A E definite

I am not going t o bother you with the question whether all ele-
mentary sentences are definite. It would be a reasonable move to with no response other than to counterattack A.
restrict our attention to definite sentences -and it would follow - For the some-quantifier we have already fixed the following attack-
that, if we start with defmite elementary sentences, all composite defense-rule:
sentences are defmite. For the composition of the latter we can use
the iunctors: A , V , -, according to rules which correspond to
the truth-tables.
This two-valued-approach to logic is called classical (although For the all-quantifier we may now propose the following attack-
it,
.. is-~ verv anti-Aristotelian). The difficulty with the classical defense-rule :
approach lies - as BROUWERdiscovered 60 years ago -
in the
quantifiers. We may introduce the some-quantifiek (the so-called
"existential quintifier") by the rule: Here the opponent chooses an object S. If an all-sentence has
been defended against an attack, this does not justify calling the
thesis "true". I propose to call a composite sentence true if and
only if i t can be defended as a thesis against all "possible" opposition;
From this rule no one up to now has been able to prove that e. g., but this proposal does not make sense as long as the rules of the
the sentence "some odd numbers am perfect" is true. For what dialogical game, the game of attacking and defending sentences,
we have here is no longer an elementery sentence. The symbol T has not been determined in full detail. In order to do this, I begin
is not a predicator; it only looks like one. We do not understand by introducing junctora into this game. But we cannot use the
an elementary sentence S s p, unless we underatand S E' p also. '
junctora as they are determined by the truth-tables, because the
Predication is the decision between affirming or denying the predi- use of quantifiers may yield sentences which have no truth-value.
cator p of the object S. However, t o assert a some-sentence ia For the adjunction v (I use this term for the non-exclusive
something different, because we have to defend such an aaaertion disjunction) we have in analogy to the some-quantifier two rules:

.
according to the given rule: we have to name an object S. There-
fore, the sequence of symbols V % A (x) &' T (or V A (x) E F) has
no meaning up to now; it has no use that has been explicitly
agreed upon.
--,n let,
R .-- us introduce the assertion of - V z A(x) by the follow-
ing proposal: If I m e r t this, you may aaaert the a f h a t i v e part,
The proponent of A v B may chose which part he will defend
after the opponent has attacked by "?". For the conjunction A the
attacking opponent shall have the choice of the left or right part,
i. e. v A(x) and then I shall ask you to defend it. If you succeed, ' which the proponent has to defend. We indicate this choice by
I will have loat my assertion; otherwise, I shall have won. In ease
: L 7, or R? respectively; so we get 'two rules in analogy to the
not to challenge the original thesis -
you do not know how to defend V z A(x), i t is Bdvisable for you
V. A(x).
I do not propose to call -. V. A ( z ) true if you munot refute
; dl-quantifier:
. < ..,
>-,
... .
\?.,,
A A B
the aa&ion of it: I would cell it "true" only if no one could refute .* A A B
,, I . .
26 Normative Logic and Ethk 2. L q i c a l Particles 27

Up to now I have said nothing about the junctor "if - then", (divided by a double line(().Then the opponent attacks the thesis
and the dialogue continues by means of alternating moves of the
and I shall not assert anything about the English usage of these
wods. On the other hand, I have already used these words rather players.
frequently. In classical logic it is usual to define a junctor .- by Az'X E B + X E8 V Z C W
Russell 1 R ~ a - + R e av REW

Let me call the sentences A - B "conditionaLs". The term "impli-

cation" I should like to reserve for a relation between sentences.
Rea
( R c a)
?

I
?
R E B VR E W
REW
?
For the dialogical game I introduce a junctor + by the following
attack-defense rule: An elementary sentence in parentheses indicates a verification of
the enclosed sentence. If the asserted elementary sentence R E a
A+B ) A ? B.
had been falsified, the opponent would have lost the game.
In wo&: If A + B has been asserted, it may he attacked by In the last position of the game the proponent has to show
asserting A ; as defense B may be-asserted. that Russell is wicked. But let us wsume that he cannot do this:
The converse conditional t may be defined by: then the proponent has lost the game. His thesis is not true.
In order to fix the rules of this game, I propose in a first attempt
the following :
It has the following attack-defense rule: Glarting Rule: The proponent begins by assertmng a thesis. The
players make their moves alternatingly.
Oenerol Rule: Each player may either attack a mntence w e r t e d
The introduction of these 4 junctors A , v , +, c will look
It
by his partner or defend himaelf against a n attack
rather arbitrary, but I must confeas that I do not know of anything by his partner.
less arbitrary. Our 4 junctors exhaust all "simple" junctors. By
a "simple7' junctor I mean such a junctor that its attack-defense .'. I hope that you agree that this is a reasonable convention to
rules contain in the attacks, as well as in the defenses, each partial follow in order to find out whether or not a thesis is true. Actually
sentence exactly once. We require also that there is always a I have to define "true" as the defensibility of a the& in this
defense because otherwise we would come to dialogical negation. game against any opponent. As long as the thesis is composed
It is juat our 4 junctors. which satisfy thew requirements. All .. without quantifiers, there are always only finitely many strategies
other junctors are definable in t e r n of the simple ones, perhaps for the players. We can decide if there is 8 winning strategy for
including negation. E. g., the proponent; only then the thesis shall be called "true". If the ele-
mentary sentences that occur are definite, this notion of truth
A*B%.A+BAA+B coincides with the classical one. The defensibility (i. e. the existence
A u B + A A - B \i - A A B . of a winning strategy) may not be decidable if quantifiers occur-
I still have not shown how a dialogue runs. Let me first give an but "truth" is to mean defensibility in all oases.
example. The thesis may be Before I ask you to accept this "dialogical" notion of truth, I
. must unfortunately complicate the situation a bit more. The
'AU atheists are stupid or wioked.' "general rule" hes to be modified for cases which are more compli-
insymhols: A . . ~ E ~ - + X VE XBE W . cated than o w first example.
Let UE conaider a conditional of the 3rd degree:
The dialogue begins with the sasertion of this thesis, let ua say.
mitten down by the proponent on the right side of the bleckboard
26 Normative Logic and Ethk 2. L q i c a l Particles 27

Up to now I have said nothing about the junctor "if - then", (divided by a double line(().Then the opponent attacks the thesis
and the dialogue continues by means of alternating moves of the
and I shall not assert anything about the English usage of these
wods. On the other hand, I have already used these words rather players.
frequently. In classical logic it is usual to define a junctor .- by Az'X E B + X E8 V Z C W
Russell 1 R ~ a - + R e av REW

Let me call the sentences A - B "conditionaLs". The term "impli-

cation" I should like to reserve for a relation between sentences.
Rea
( R c a)
?

I
?
R E B VR E W
REW
?
For the dialogical game I introduce a junctor + by the following
attack-defense rule: An elementary sentence in parentheses indicates a verification of
the enclosed sentence. If the asserted elementary sentence R E a
A+B ) A ? B.
had been falsified, the opponent would have lost the game.
In wo&: If A + B has been asserted, it may he attacked by In the last position of the game the proponent has to show
asserting A ; as defense B may be-asserted. that Russell is wicked. But let us wsume that he cannot do this:
The converse conditional t may be defined by: then the proponent has lost the game. His thesis is not true.
In order to fix the rules of this game, I propose in a first attempt
the following :
It has the following attack-defense rule: Glarting Rule: The proponent begins by assertmng a thesis. The
players make their moves alternatingly.
Oenerol Rule: Each player may either attack a mntence w e r t e d
The introduction of these 4 junctors A , v , +, c will look
It
by his partner or defend himaelf against a n attack
rather arbitrary, but I must confeas that I do not know of anything by his partner.
less arbitrary. Our 4 junctors exhaust all "simple" junctors. By
a "simple7' junctor I mean such a junctor that its attack-defense .'. I hope that you agree that this is a reasonable convention to
rules contain in the attacks, as well as in the defenses, each partial follow in order to find out whether or not a thesis is true. Actually
sentence exactly once. We require also that there is always a I have to define "true" as the defensibility of a the& in this
defense because otherwise we would come to dialogical negation. game against any opponent. As long as the thesis is composed
It is juat our 4 junctors. which satisfy thew requirements. All .. without quantifiers, there are always only finitely many strategies
other junctors are definable in t e r n of the simple ones, perhaps for the players. We can decide if there is 8 winning strategy for
including negation. E. g., the proponent; only then the thesis shall be called "true". If the ele-
mentary sentences that occur are definite, this notion of truth
A*B%.A+BAA+B coincides with the classical one. The defensibility (i. e. the existence
A u B + A A - B \i - A A B . of a winning strategy) may not be decidable if quantifiers occur-
I still have not shown how a dialogue runs. Let me first give an but "truth" is to mean defensibility in all oases.
example. The thesis may be Before I ask you to accept this "dialogical" notion of truth, I
. must unfortunately complicate the situation a bit more. The
'AU atheists are stupid or wioked.' "general rule" hes to be modified for cases which are more compli-
insymhols: A . . ~ E ~ - + X VE XBE W . cated than o w first example.
Let UE conaider a conditional of the 3rd degree:
The dialogue begins with the sasertion of this thesis, let ua say.
mitten down by the proponent on the right side of the bleckboard
28 Normative Logic and Ethiw 2. Logical Partides 29

The dialogue may run as follows: agree to a reasonable use. The traditional use in natural languages
is a rather dubious authority. For example, the double use of a
negation sign sometimes means a strong negation, sometimes a n
affirmation. We have t o decide for ourselves how to take it.
Admittedly, my proposal, with its attack-defense-rules, general
rule and winning rule, may look a bit too complicated to be rea-
I n this situation the proponent may think of defending his sonable. I n particular, the choice of a number as a limit of repeti-
thesis by easerting A , hut the opponent may-rightly I think- tions seems too complicated for a fundamental rule of logic Ae
out that he aaeerted A only in the attack against A + B; a matter of fact the use of such a number can he avoided. This can
therefore, he may insist that the proponent first defend himself he done by a modification of the general rule which simplifies the
game without BBecting the defenaibility of any thesis. This will
against this attack, i. e. that he aeeerts B.
This example shows that the following restriction of the general be the finel statement of the General Rule. For the proponent i t
reads :
rule is justified:
Each player may either attack a sentence aeeerted by his partner 1. The proponent may either attack 8 sentence asserted by the
or he may defend himself against the last attack against which opponent or he may defend himself against the last attack of
he has not already defended himself. the opponent.
This formulation is nearly final. It still has the disadvantage of But for the opponent it contains, inatead of a limit of repetition,
allowing the opponent indefinite repetition of attacks. Thus the a much stronger restriction:
proponent could not win even in the simplest cases. This requires
2. The opponent may either attack the sentence asserted by the
us to impose a limit on repetitions by the opponent. Still we have
proponent in the preceding move or he may defend himself
to avoid arbitrarineas. This leaves only the possibility of letting
against the attack of the proponent in the preceding move.
the opponent choose such a limit himself. So let the opponent
choose a number m from 0, 1, 2 , . .. The g e n e d rule then yields That this simplification of the general rule does not affect the
the further restriction that the opponent may attack a sentence defensibility of any thesis is a logically composite metedialogical
a t most m + 1 times. assertion. Since the meta-dialogue may be played with the unmodi-
With this restriction of the opponent we may formulate a rule fied geneml rule, there is no circularity here; but this meta-dia-
for the end of the game in a simple manner by the following l o p e is too complicated $0 he deelt with in these lectures. The
modification of the general rule serve8 only t o simplify the dia-
Winning Rule: If the opponent cannot make any further move, logue; therefore I will continue to use it, although it could be
the proponent has won.
dispensed with.
If the proponent has succeded in reaching such a h a 1 position, With the rules of the dialogical game we now have sentences at
the opponent cannot defend himaelf against a n attack and he our disposal which are compounded by means of logical particles,
cannot attack any assertion of the proponent (because the pro- using elementary sentences as their park. For these new sentencea
ponent haa already defended himself successfullyagainst all attacks). we have defined "truth" as defensibility against every opposition.
If such a fins1 position can be reached, against the ingenuity The dialogical game leads us in a natural way to generalize this
of any opponent, I propose to call the thesis "true". I have simply notion of truth to "implication". In order to do this, consider the
to ask YOU to accept this dialogical notion of truth. I n the case of development of a dialogue. It s t a r t s with asserting a thesis Ao.
elementary sentences we start with a common understand~ngof If Ao is a conjunctivn or adjunction, an all- or some-sentence.
these aentences, and therefore with their truth. But for non-ele- the attacks will be made by writing question marks, perhaps with
mentary sentences, composed with logical perticlea. we heve to L, R or with the choice of a n a c e for a variable. If a negation
28 Normative Logic and Ethiw 2. Logical Partides 29

The dialogue may run as follows: agree to a reasonable use. The traditional use in natural languages
is a rather dubious authority. For example, the double use of a
negation sign sometimes means a strong negation, sometimes a n
affirmation. We have t o decide for ourselves how to take it.
Admittedly, my proposal, with its attack-defense-rules, general
rule and winning rule, may look a bit too complicated to be rea-
I n this situation the proponent may think of defending his sonable. I n particular, the choice of a number as a limit of repeti-
thesis by easerting A , hut the opponent may-rightly I think- tions seems too complicated for a fundamental rule of logic Ae
out that he aaeerted A only in the attack against A + B; a matter of fact the use of such a number can he avoided. This can
therefore, he may insist that the proponent first defend himself he done by a modification of the general rule which simplifies the
game without BBecting the defenaibility of any thesis. This will
against this attack, i. e. that he aeeerts B.
This example shows that the following restriction of the general be the finel statement of the General Rule. For the proponent i t
reads :
rule is justified:
Each player may either attack a sentence aeeerted by his partner 1. The proponent may either attack 8 sentence asserted by the
or he may defend himself against the last attack against which opponent or he may defend himself against the last attack of
he has not already defended himself. the opponent.
This formulation is nearly final. It still has the disadvantage of But for the opponent it contains, inatead of a limit of repetition,
allowing the opponent indefinite repetition of attacks. Thus the a much stronger restriction:
proponent could not win even in the simplest cases. This requires
2. The opponent may either attack the sentence asserted by the
us to impose a limit on repetitions by the opponent. Still we have
proponent in the preceding move or he may defend himself
to avoid arbitrarineas. This leaves only the possibility of letting
against the attack of the proponent in the preceding move.
the opponent choose such a limit himself. So let the opponent
choose a number m from 0, 1, 2 , . .. The g e n e d rule then yields That this simplification of the general rule does not affect the
the further restriction that the opponent may attack a sentence defensibility of any thesis is a logically composite metedialogical
a t most m + 1 times. assertion. Since the meta-dialogue may be played with the unmodi-
With this restriction of the opponent we may formulate a rule fied geneml rule, there is no circularity here; but this meta-dia-
for the end of the game in a simple manner by the following l o p e is too complicated $0 he deelt with in these lectures. The
modification of the general rule serve8 only t o simplify the dia-
Winning Rule: If the opponent cannot make any further move, logue; therefore I will continue to use it, although it could be
the proponent has won.
dispensed with.
If the proponent has succeded in reaching such a h a 1 position, With the rules of the dialogical game we now have sentences at
the opponent cannot defend himaelf against a n attack and he our disposal which are compounded by means of logical particles,
cannot attack any assertion of the proponent (because the pro- using elementary sentences as their park. For these new sentencea
ponent haa already defended himself successfullyagainst all attacks). we have defined "truth" as defensibility against every opposition.
If such a fins1 position can be reached, against the ingenuity The dialogical game leads us in a natural way to generalize this
of any opponent, I propose to call the thesis "true". I have simply notion of truth to "implication". In order to do this, consider the
to ask YOU to accept this dialogical notion of truth. I n the case of development of a dialogue. It s t a r t s with asserting a thesis Ao.
elementary sentences we start with a common understand~ngof If Ao is a conjunctivn or adjunction, an all- or some-sentence.
these aentences, and therefore with their truth. But for non-ele- the attacks will be made by writing question marks, perhaps with
mentary sentences, composed with logical perticlea. we heve to L, R or with the choice of a n a c e for a variable. If a negation
30 Normative Lcgic and Ethics 3. Lcgical Truth 31

or a conditional has to be attacked, the opponent has to assert a will be logical truths, but this does not mean that we m y not
sentence himself. In the course of the dialogue the proponent will intmduce later on still different sentences, and thereby come to
have to defend himself, and unless he defend8 an elementary truths which are neither empirical nor logical.
sentence - and thereby wins the game - he will have to assert a
new sentence A. Let us assume that for some reason the players
want to have a break. After the break the opponent will have to
attack A, and the proponent may attack all sentences asserted 3. Logical Truths
by the opponent before the break. We may say that after the break
a new game begins. The starting position now is determined not By meam of the didogieal introduction of the logical particles
merely by the thesis A, but by the opponent's assertions, too. We . ' we are now in a position to understand composite sentences. This
may use the following notation does not mean that we already understand what logic - in the sense
of "formal logic" - is. If we want to know the truth of a composite
sentence, we have to look a t the defensibility of that sen-
tence as a thesis; but we do not need to know any so-called
"logical rules" or "logical truths": Indeed, we have still to discover
a. distinction which allows us to interpret these term.
Let us consider some examplos which will exhibit such a distic-
This is a generalization of the game which starts with a thesis
tion. As the &st example I will take:
only. I would like to call the sentences given in advance by the
opponent hypotheses. If the thesis A can be defended with the "In all Bavarian lakes there me f%hesW.
hypotheses A l , . . . , A n given, I propose to say that the hypotheses
imply the thesis. I take this aa an ordinary version of the fouowing sentence:
This is obviously the case if and only if the thesis :
Az V v yinz
Bar. tlsh
bke
is true.
I n the Platonic Academy, during the time of pre-Aristotelian A dialogue may run as follows:
logic, this game wes played in a rather tricky way. The proponent
A. V u Y inz
began by asking a q u ~ t i o n A
: or not A t H e persuaded the oppo- Bav. flah
nent to make a choice. Let us my the opponent aaserted A. Then lake
Tegernaea ? V y in Tegernsee
the proponent took as his thesis - A , but he did not aay so. Bah
Instead he began to ask the opponent seemingly harmless ques- 1 S in Tegemsee
..
tions: What about A l ? What about A s ? . He tried to get a ?
sufficient system of hypotheses. If he attained such a system, he
performod the fkal stroke by saying: Well, my dear friend, now This dialogue concludes with an elementary sentence (of the form
..
you have admitted A1 , . ,An ; therefore, you must admit - A , S, T E p, with S as a proper name for a fish, T for Tegernsee and
the contrary of your original sr*sertion. The problem which Aristotle the predicator p for "to be in").
discovered here was the problem of justifying this therefore. This Depending on the truth of the elementary sentence, this dialogue
led him to the discovery of formd logic. will be won or lost. Of course. if this dialogue has been won. we
For our lectures this means that in the class of logically composite still do not know whether or not the the& is true. There are quite
sentenoes we can now arrive a t the distinction between empli- a few lakes in Bavaria, m y of which could be chosen by the
cally and non-empirioelly true sentanoes. The non-empirid truth8 opponent in his first attack.
30 Normative Lcgic and Ethics 3. Lcgical Truth 31

or a conditional has to be attacked, the opponent has to assert a will be logical truths, but this does not mean that we m y not
sentence himself. In the course of the dialogue the proponent will intmduce later on still different sentences, and thereby come to
have to defend himself, and unless he defend8 an elementary truths which are neither empirical nor logical.
sentence - and thereby wins the game - he will have to assert a
new sentence A. Let us assume that for some reason the players
want to have a break. After the break the opponent will have to
attack A, and the proponent may attack all sentences asserted 3. Logical Truths
by the opponent before the break. We may say that after the break
a new game begins. The starting position now is determined not By meam of the didogieal introduction of the logical particles
merely by the thesis A, but by the opponent's assertions, too. We . ' we are now in a position to understand composite sentences. This
may use the following notation does not mean that we already understand what logic - in the sense
of "formal logic" - is. If we want to know the truth of a composite
sentence, we have to look a t the defensibility of that sen-
tence as a thesis; but we do not need to know any so-called
"logical rules" or "logical truths": Indeed, we have still to discover
a. distinction which allows us to interpret these term.
Let us consider some examplos which will exhibit such a distic-
This is a generalization of the game which starts with a thesis
tion. As the &st example I will take:
only. I would like to call the sentences given in advance by the
opponent hypotheses. If the thesis A can be defended with the "In all Bavarian lakes there me f%hesW.
hypotheses A l , . . . , A n given, I propose to say that the hypotheses
imply the thesis. I take this aa an ordinary version of the fouowing sentence:
This is obviously the case if and only if the thesis :
Az V v yinz
Bar. tlsh
bke
is true.
I n the Platonic Academy, during the time of pre-Aristotelian A dialogue may run as follows:
logic, this game wes played in a rather tricky way. The proponent
A. V u Y inz
began by asking a q u ~ t i o n A
: or not A t H e persuaded the oppo- Bav. flah
nent to make a choice. Let us my the opponent aaserted A. Then lake
Tegernaea ? V y in Tegernsee
the proponent took as his thesis - A , but he did not aay so. Bah
Instead he began to ask the opponent seemingly harmless ques- 1 S in Tegemsee
..
tions: What about A l ? What about A s ? . He tried to get a ?
sufficient system of hypotheses. If he attained such a system, he
performod the fkal stroke by saying: Well, my dear friend, now This dialogue concludes with an elementary sentence (of the form
..
you have admitted A1 , . ,An ; therefore, you must admit - A , S, T E p, with S as a proper name for a fish, T for Tegernsee and
the contrary of your original sr*sertion. The problem which Aristotle the predicator p for "to be in").
discovered here was the problem of justifying this therefore. This Depending on the truth of the elementary sentence, this dialogue
led him to the discovery of formd logic. will be won or lost. Of course. if this dialogue has been won. we
For our lectures this means that in the class of logically composite still do not know whether or not the the& is true. There are quite
sentenoes we can now arrive a t the distinction between empli- a few lakes in Bavaria, m y of which could be chosen by the
cally and non-empirioelly true sentanoes. The non-empirid truth8 opponent in his first attack.
32 Nonnative Logic and Ethica

A s the truth of the thesis depends on the truth of a t least one

I 3. Logical Trutha

(The number following the question mark indicates the line which
33

of its elementary parts, I shall call the thesis "empirically" true. is attacked). In this situation the opponent has to attack the
This term obviously is merely traditional. We my that the truth elementary sentence a, which he himself has asserted in line 3.
of elementary sentences is "empirically known", if it is known a
It is only fair, therefore, to give up - instead of finding out whether
a t all. We say that we come to know the truth of an elementary a is true or false (assuming that a is definite a t all). This leads us
sentence by experience, if we come to know the truth a t all. to separate the following composite sentences from the rest:
The Greek word tpxerp[a and the Latin word experientia have namely, those sentences whose truth can be defended in such a
the same radical.' per, which is also in the German word "Erfah- way that the proponent finally has to defend merely an elementary
rung", hut we need not speculate about the activities or passivi- sentence which has been asserted by the opponent.
ties by which we come to the truth of elementary sentences. If I All sentences of the forms a -t a and -. a A --.a. satisfy this
should propose to call the true elementary sentences "empirically condition. Aristotle discovered the existence of sentences which
true", you would rightly object that there is no need for such a are true in this special way by such examples as the famous:
term. Those elementary sentences which are true are true and that "If all Greeks are men and if all men are mortal, then a11 Greeks
is all there is to it. In the case of composite sentences the situation is
are mortal".
different: I propose to introduce a term for the truth of those
composite sentences whose truth is independent of the truth of We need not interpret this as a conditional. I think it is nearer to
their elementary parts. Though it may appear a t first that the Aristotle to interpret his syllogisms as implications. Two hypo-
truth of a composite sentence always depends on the truth of ita theses are given and the thesis has to be defended in such a way
elementary parts, this is not the case if we understand "to depend that the proponent finally has to defend merely an elementary
on" correctly. sentence which has been asserted by the opponent.
In order to see this point, I shall first take the simple exemple: Such a strategy exists if we interpret a sentence of the form
.
"all p are q" as A,. x r p + x r q. I shall show that the follow-
"If Oxford is a river, then Oxford is a river". ing dialogue can he won against any opponent:
The dialogue rum as follows : A s . a(.) + b(x) .

Oer ? I/ Orr-Orr
1
2 Az.b(x)+c(x).
The dialogue may run as follows:
I/ A = . a ( 4 +~ ( 2 ) .

and the opponent, who has carelesely challenged the thesis, will 3 Y t a(y) + C(Y)
have lost the game because he cannot defend the elementary 4 a(y) t Y 71
sentence. And even if he could defend the elementary sentence 5 a(y)+b(y) a(y) ?5
he would lose the game; since the proponent could then defend 6 b(y) Y t2
himself with just the =me elementary sentence. Let "a" be any 7 WY) -+ C(Y) WY) 17
elementary sentence. The thesis a + a is defensible independently 8

-
C(Y) C(Y)
of the defensibility of a.
The aame is the case with many other theses, e. g. a A o;-. and the opponent gives up.
Here the proponent can use the following strategy: It sounds rather implausible that the first "logical implications"
should have been such complicated cases; nevertheless we need
1 -.a A -a. not assume that Aristotle had in mind exactly our dialogical game.
2 a A-a? L ?2 He may have had in mind such rules as "p + q" (from p go to q)
3 a Rt 2 when he a i d "q belongs to all p". I shell deal with rules of this
4 . -a a 7 4 , form in a later lecture. Meantime, let us introduce officially the
32 Nonnative Logic and Ethica

A s the truth of the thesis depends on the truth of a t least one

I 3. Logical Trutha

(The number following the question mark indicates the line which
33

of its elementary parts, I shall call the thesis "empirically" true. is attacked). In this situation the opponent has to attack the
This term obviously is merely traditional. We my that the truth elementary sentence a, which he himself has asserted in line 3.
of elementary sentences is "empirically known", if it is known a
It is only fair, therefore, to give up - instead of finding out whether
a t all. We say that we come to know the truth of an elementary a is true or false (assuming that a is definite a t all). This leads us
sentence by experience, if we come to know the truth a t all. to separate the following composite sentences from the rest:
The Greek word tpxerp[a and the Latin word experientia have namely, those sentences whose truth can be defended in such a
the same radical.' per, which is also in the German word "Erfah- way that the proponent finally has to defend merely an elementary
rung", hut we need not speculate about the activities or passivi- sentence which has been asserted by the opponent.
ties by which we come to the truth of elementary sentences. If I All sentences of the forms a -t a and -. a A --.a. satisfy this
should propose to call the true elementary sentences "empirically condition. Aristotle discovered the existence of sentences which
true", you would rightly object that there is no need for such a are true in this special way by such examples as the famous:
term. Those elementary sentences which are true are true and that "If all Greeks are men and if all men are mortal, then a11 Greeks
is all there is to it. In the case of composite sentences the situation is
are mortal".
different: I propose to introduce a term for the truth of those
composite sentences whose truth is independent of the truth of We need not interpret this as a conditional. I think it is nearer to
their elementary parts. Though it may appear a t first that the Aristotle to interpret his syllogisms as implications. Two hypo-
truth of a composite sentence always depends on the truth of ita theses are given and the thesis has to be defended in such a way
elementary parts, this is not the case if we understand "to depend that the proponent finally has to defend merely an elementary
on" correctly. sentence which has been asserted by the opponent.
In order to see this point, I shall first take the simple exemple: Such a strategy exists if we interpret a sentence of the form
.
"all p are q" as A,. x r p + x r q. I shall show that the follow-
"If Oxford is a river, then Oxford is a river". ing dialogue can he won against any opponent:
The dialogue rum as follows : A s . a(.) + b(x) .

Oer ? I/ Orr-Orr
1
2 Az.b(x)+c(x).
The dialogue may run as follows:
I/ A = . a ( 4 +~ ( 2 ) .

and the opponent, who has carelesely challenged the thesis, will 3 Y t a(y) + C(Y)
have lost the game because he cannot defend the elementary 4 a(y) t Y 71
sentence. And even if he could defend the elementary sentence 5 a(y)+b(y) a(y) ?5
he would lose the game; since the proponent could then defend 6 b(y) Y t2
himself with just the =me elementary sentence. Let "a" be any 7 WY) -+ C(Y) WY) 17
elementary sentence. The thesis a + a is defensible independently 8

-
C(Y) C(Y)
of the defensibility of a.
The aame is the case with many other theses, e. g. a A o;-. and the opponent gives up.
Here the proponent can use the following strategy: It sounds rather implausible that the first "logical implications"
should have been such complicated cases; nevertheless we need
1 -.a A -a. not assume that Aristotle had in mind exactly our dialogical game.
2 a A-a? L ?2 He may have had in mind such rules as "p + q" (from p go to q)
3 a Rt 2 when he a i d "q belongs to all p". I shell deal with rules of this
4 . -a a 7 4 , form in a later lecture. Meantime, let us introduce officially the
34 Nonnative Logic and Ethica 3. Logical T r u t h 35
term "logically true" for those theses which can be defended in the formal game, we need not understand what sentences are, what
this special way, without bothering a t the moment about Aristotle it means to asaert a sentence, or what logical particles are. We may
or about the question whether "logically true" ahould by synony- forget about all this completely. However, in order to understand
mous with "non-empirically true". I would like t o propoee only the formal game, that is, in order to anewer the question, why it is
the following rule : reasonable t o spend our time with this game, we will have to
A E logically true 3 A E' empirically true remember that the formal game is a formalization of the material
If a thesis can he defend@ in this special way with a system of game. The material game has to be understood first. then it has
hypotheses given, I shall say that the hypotheses logically imply to be formalized. The result is the formal game: With the formal
the thesis. game we are simulating material dialogues.
I n order t o study logical truth and logical implications, i t is Once we have established the formal game aa a reasonable.
convenient t o introduce a formd variant of the dialogical game. though auxiliary instrument for investigating material dialogues.
Insteed of beginning with elementary sentences, we now begin with we may investigate the formal game ae if i t were an end in itaelf.
We will call formulas which can be defended against a U possible
prime-formulas :
a, b, . .. opposition l o g i c d y t w formulae. If a formula B is defensible as
.
a thesis when some formulee A l , . . A , are given as hypotheses.
QW,WY). .. . we will say that A1, . . .,A. l o g i d l y imply B. I shall write this in
a(., y), ... . .
the following way:
...
We form compound formulas by using the symbols:
t . A l , . . ., A,, < B .
In this notation < represents a two-place relation between a
system of formulae and a formula. Of course, the system could be
replaced by the formula A1 A . . . A A . .
The attack-defense-rules are the same as before; they have merely Formal logic is a rather ancient discipline. since it began with
to be completed by a rule for prlme formulae p Aristotle, flourished with the Stoics, flourished again with the later
PITI Soholastics, and has come t o bear fruit in our century. Aristotle
investigated material dialogues in the Topics, but since his Andy-
w ~ t h9 as attack and no defense; but in the formal game, however, tics we find the logician concerned mainly with logical implications,
the proponent is never to be allowed to attack a prime-formula. especially with rules which yield new logical implicationa from
The Gene~alRule, therefore, reads: given ones. Aristotle uses, for example, the rule of contraposition:
1. The proponent may either attack a composite formula of the
A,B<C *A.-C<-B
opponent or he may defend himself against the last attack of
the opponent. in order to derive further syllogism from those which he eslled
"perfect". The Stoica (or perhaps earlier the Megaream) added
2. The opponent may either attack the formula put by the pro-
rules of transitivity such as:
ponent in the preceding move or he may defend himself egainat
the attack of the proponent in the preceding move.
The Winning Rule in the formal game is the following: They did not start with syllogisms, but with logical implications
If the proponent ha8 to defend a prime-formula which of the logic of junctors, e. g.
has been put by the opponent, the proponent has won.
This game with formulae I shall call the f o d game; the game These a m the begmrnga of loglcal calculi. After the m d e v a l
with sentences may be called the moleriaE gome. I n order to play .
t,
p o d LEIBNIZwas the first logman to look for a cornwe lo-
34 Nonnative Logic and Ethica 3. Logical T r u t h 35
term "logically true" for those theses which can be defended in the formal game, we need not understand what sentences are, what
this special way, without bothering a t the moment about Aristotle it means to asaert a sentence, or what logical particles are. We may
or about the question whether "logically true" ahould by synony- forget about all this completely. However, in order to understand
mous with "non-empirically true". I would like t o propoee only the formal game, that is, in order to anewer the question, why it is
the following rule : reasonable t o spend our time with this game, we will have to
A E logically true 3 A E' empirically true remember that the formal game is a formalization of the material
If a thesis can he defend@ in this special way with a system of game. The material game has to be understood first. then it has
hypotheses given, I shall say that the hypotheses logically imply to be formalized. The result is the formal game: With the formal
the thesis. game we are simulating material dialogues.
I n order t o study logical truth and logical implications, i t is Once we have established the formal game aa a reasonable.
convenient t o introduce a formd variant of the dialogical game. though auxiliary instrument for investigating material dialogues.
Insteed of beginning with elementary sentences, we now begin with we may investigate the formal game ae if i t were an end in itaelf.
We will call formulas which can be defended against a U possible
prime-formulas :
a, b, . .. opposition l o g i c d y t w formulae. If a formula B is defensible as
.
a thesis when some formulee A l , . . A , are given as hypotheses.
QW,WY). .. . we will say that A1, . . .,A. l o g i d l y imply B. I shall write this in
a(., y), ... . .
the following way:
...
We form compound formulas by using the symbols:
t . A l , . . ., A,, < B .
In this notation < represents a two-place relation between a
system of formulae and a formula. Of course, the system could be
replaced by the formula A1 A . . . A A . .
The attack-defense-rules are the same as before; they have merely Formal logic is a rather ancient discipline. since it began with
to be completed by a rule for prlme formulae p Aristotle, flourished with the Stoics, flourished again with the later
PITI Soholastics, and has come t o bear fruit in our century. Aristotle
investigated material dialogues in the Topics, but since his Andy-
w ~ t h9 as attack and no defense; but in the formal game, however, tics we find the logician concerned mainly with logical implications,
the proponent is never to be allowed to attack a prime-formula. especially with rules which yield new logical implicationa from
The Gene~alRule, therefore, reads: given ones. Aristotle uses, for example, the rule of contraposition:
1. The proponent may either attack a composite formula of the
A,B<C *A.-C<-B
opponent or he may defend himself against the last attack of
the opponent. in order to derive further syllogism from those which he eslled
"perfect". The Stoica (or perhaps earlier the Megaream) added
2. The opponent may either attack the formula put by the pro-
rules of transitivity such as:
ponent in the preceding move or he may defend himself egainat
the attack of the proponent in the preceding move.
The Winning Rule in the formal game is the following: They did not start with syllogisms, but with logical implications
If the proponent ha8 to defend a prime-formula which of the logic of junctors, e. g.
has been put by the opponent, the proponent has won.
This game with formulae I shall call the f o d game; the game These a m the begmrnga of loglcal calculi. After the m d e v a l
with sentences may be called the moleriaE gome. I n order to play .
t,
p o d LEIBNIZwas the first logman to look for a cornwe lo-
 Normative Logic and Ethica

,
with --. A. a(z) given alone, but only with the additions1 cleaaioal
gical calculus. Completeness mmns that dl logical implications
should be derivable from the initial implications of the calculus
by the rules of the ca~culus.BOOLE was the first to give, though in
an artificid algebraic form, a solution for the logic of junctors.
FBEGE gave the first calculus for the classical logic of junctors and
hypotheaea:

and
- V Z-a(%) v - V. -a(x)

quantifiers. A completeness proof is due to GODEL. Az.a(x) v - a ( % ) .

In our century, however, difficulties began with BROUWER'S The dialogue runs as follows :
critique of the classicel tertium w n datur used for quantificational
logic. HEY TIN^ gave a logical calculus for nonclassical logic, nor-
mally called "intuitionistic". I would like to use the term "con-
structive logic" instead, since intuitionism is only one variant
within the broader field of constructive logic and mathematics.
The problem of the completeness of the HEYTINO calcu1us has not
been settled along the lines of the traditional approach. The dif-
ficulty is simply this: one has to establish a notion of constructive
logical implication (or constructive logical truth) without using
the derivability of a calcuhls.
The notion of dialogical defensibility yields exactly what is
wanted.
Let us consider the problem of double negation; does -
imply a9 The claasiciat says "yes", the construotivist "no". The
-a
I t is easy to see that the thesis could not be defended without
the first hypothesis.
We have now to show that the dialogically defined logical implica-
dialogical game says "no", also. tions are identical with those of the constructive logical calculi as
a proposed by HEYTINO, GENTZEN and othera. This is rather tedious,
2 but not difficult in principle. It is eaay to arrive a t a complete
3 a ? logical calculus, if we ask ourselves which positions

The proponent has no move besides repeating his attack of --

but as this does not change the situation, he cannot win. Never-
a;

affirmat if we add the hypothesis a v -

theless, we can understand the classical rule: duplex negatio
a. This clessical hypothe-
sis is never mentioned by the claasicista; that is the whole trick of
"claesical" logic.
of the formal game are defensible. We now write these positions in
one line as:

We then ask for a calculus which yields all the winning-positions.

We denote a 8ptBm of hypotheses by Z, and a system of hypath-
eses in which a formula A occurs by Z(A). Then we have a t leaat
The opponent has to give up.

-
In cleasicd quantiiicational logic the formula
implies V I a(%).In the formal game V
-
A z a(z) '
a(z) is not defensible
for dl prime-formulae p the following winning positions:
 Normative Logic and Ethica

,
with --. A. a(z) given alone, but only with the additions1 cleaaioal
gical calculus. Completeness mmns that dl logical implications
should be derivable from the initial implications of the calculus
by the rules of the ca~culus.BOOLE was the first to give, though in
an artificid algebraic form, a solution for the logic of junctors.
FBEGE gave the first calculus for the classical logic of junctors and
hypotheaea:

and
- V Z-a(%) v - V. -a(x)

quantifiers. A completeness proof is due to GODEL. Az.a(x) v - a ( % ) .

In our century, however, difficulties began with BROUWER'S The dialogue runs as follows :
critique of the classicel tertium w n datur used for quantificational
logic. HEY TIN^ gave a logical calculus for nonclassical logic, nor-
mally called "intuitionistic". I would like to use the term "con-
structive logic" instead, since intuitionism is only one variant
within the broader field of constructive logic and mathematics.
The problem of the completeness of the HEYTINO calcu1us has not
been settled along the lines of the traditional approach. The dif-
ficulty is simply this: one has to establish a notion of constructive
logical implication (or constructive logical truth) without using
the derivability of a calcuhls.
The notion of dialogical defensibility yields exactly what is
wanted.
Let us consider the problem of double negation; does -
imply a9 The claasiciat says "yes", the construotivist "no". The
-a
I t is easy to see that the thesis could not be defended without
the first hypothesis.
We have now to show that the dialogically defined logical implica-
dialogical game says "no", also. tions are identical with those of the constructive logical calculi as
a proposed by HEYTINO, GENTZEN and othera. This is rather tedious,
2 but not difficult in principle. It is eaay to arrive a t a complete
3 a ? logical calculus, if we ask ourselves which positions

The proponent has no move besides repeating his attack of --

but as this does not change the situation, he cannot win. Never-
a;

affirmat if we add the hypothesis a v -

theless, we can understand the classical rule: duplex negatio
a. This clessical hypothe-
sis is never mentioned by the claasicista; that is the whole trick of
"claesical" logic.
of the formal game are defensible. We now write these positions in
one line as:

We then ask for a calculus which yields all the winning-positions.

We denote a 8ptBm of hypotheses by Z, and a system of hypath-
eses in which a formula A occurs by Z(A). Then we have a t leaat
The opponent has to give up.

-
In cleasicd quantiiicational logic the formula
implies V I a(%).In the formal game V
-
A z a(z) '
a(z) is not defensible
for dl prime-formulae p the following winning positions:
38 Normatiue Logic and Ethice 3. Logical Trutha 39

These will be the initial positions of our calculus. We now ask for Each of the logical particles +, -, A,, V, gives us in this way
rules which lead from winning-positionsto further winning-positions. two "admissible" rules. The junctora A , v give ua three rules emh,
Them rules may be called "admisaible" rulq. We get admimible because either the opponent or the proponent has a choice of two
rules if we ask e. g. how a position attack-defense-rulea. This e v e s us 14 = 4 . 2 + 2 . 3 admimible
rules altogether. With the initial positions Z ( p )Ilp and these 14
z A+B 11 admissible rules, we have a complete logical calculus: The system
with a conditional aa thesis can be defended. The opponent hse .
A I , . . . , A , logically implies B if and only if the position Ax, . .,
A . (1 B is derivable in this calculus. We can formulate this'reault
as a completeness theorem

and there is this defense.

with t for derivability in the above calculus.
A 7 The logical calculus which can be justified in this way is k n o w
This yields the following: I f the position Z, AI/B is a winning . in the literature as a GENTZENcalculus: namely, a s the "intui-
tionistic f33" in KLEENE'S textbook on metamathematics. KLEENE
position, then ZIJA+ B is a winning position also. proves that this calculus is equivalent to the H E Y ~ NC Q~ C U ~ U ~ .
We have, therefore, derived the following admissible rule:
Philosophically there is no reason t o start with the historical
fact that HEYTINGpublished a certain calculus or to look for an
I: -
The situation is more complicated, if we ask h o q t o defend a interpretation of that calculus. It is, however. reasonable t o start
position in which a conditional W u r s a s hypothesis with material dialogues, to formalize this game, to look for admis-
sible rules for winning-positions; this procedure leads us directly
Z( A + B) C 11 . to a n interpretation of the GENTZEN calculus and then indirectly
The opponent has to attack C. Let us consider the case where the to a n interpretation of the HEVTINQ~ a h h s I. would claim,
proponent attacks A + B as response therefore, that the dialogical approach justifies the logical intuitions
of BROUWER and HEYTING.Independently of this relation to the
recent history of formal logic, moreover, we have justified the
... GENTZENcalculus U 3 as a convenient tool for investigating the
The opponent may only continue with formal dialogical game.
The justification of classical logical calculi is more difficult,
... because we have to justify the admimion of tertium.non-datur.
... t .4,
. hypotheses. There are, of course, many cases in material dialogues
in which these c l a ~ i c a hypotheses
l are true, and then we need no
or with . , ,
justification. However, in arithmetic we do not know whether or not
... they are true; therefore, we need consistency proofs, if we are going
to use classical logical implications in arithmetic. By a consistency
B
proof I mean here a proof which tells us which 8entences are con-
If both these positions are winning-positions, Z ( A -+ B ) 1C will structively true if we have pkven aentences to be "true" with
be a winning-position also. We have, ther?foh, mother adiniaSibE
I ,: ,,!! , :
" the help of classical logic. The simplest result. fimt attained by
GENTZEN, ia that all implications between prime sentences which
;>den 'Ixi o h i d l y proven to hold, hold also constmctively.
I: '
38 Normatiue Logic and Ethice 3. Logical Trutha 39

These will be the initial positions of our calculus. We now ask for Each of the logical particles +, -, A,, V, gives us in this way
rules which lead from winning-positionsto further winning-positions. two "admissible" rules. The junctora A , v give ua three rules emh,
Them rules may be called "admisaible" rulq. We get admimible because either the opponent or the proponent has a choice of two
rules if we ask e. g. how a position attack-defense-rulea. This e v e s us 14 = 4 . 2 + 2 . 3 admimible
rules altogether. With the initial positions Z ( p )Ilp and these 14
z A+B 11 admissible rules, we have a complete logical calculus: The system
with a conditional aa thesis can be defended. The opponent hse .
A I , . . . , A , logically implies B if and only if the position Ax, . .,
A . (1 B is derivable in this calculus. We can formulate this'reault
as a completeness theorem

and there is this defense.

with t for derivability in the above calculus.
A 7 The logical calculus which can be justified in this way is k n o w
This yields the following: I f the position Z, AI/B is a winning . in the literature as a GENTZENcalculus: namely, a s the "intui-
tionistic f33" in KLEENE'S textbook on metamathematics. KLEENE
position, then ZIJA+ B is a winning position also. proves that this calculus is equivalent to the H E Y ~ NC Q~ C U ~ U ~ .
We have, therefore, derived the following admissible rule:
Philosophically there is no reason t o start with the historical
fact that HEYTINGpublished a certain calculus or to look for an
I: -
The situation is more complicated, if we ask h o q t o defend a interpretation of that calculus. It is, however. reasonable t o start
position in which a conditional W u r s a s hypothesis with material dialogues, to formalize this game, to look for admis-
sible rules for winning-positions; this procedure leads us directly
Z( A + B) C 11 . to a n interpretation of the GENTZEN calculus and then indirectly
The opponent has to attack C. Let us consider the case where the to a n interpretation of the HEVTINQ~ a h h s I. would claim,
proponent attacks A + B as response therefore, that the dialogical approach justifies the logical intuitions
of BROUWER and HEYTING.Independently of this relation to the
recent history of formal logic, moreover, we have justified the
... GENTZENcalculus U 3 as a convenient tool for investigating the
The opponent may only continue with formal dialogical game.
The justification of classical logical calculi is more difficult,
... because we have to justify the admimion of tertium.non-datur.
... t .4,
. hypotheses. There are, of course, many cases in material dialogues
in which these c l a ~ i c a hypotheses
l are true, and then we need no
or with . , ,
justification. However, in arithmetic we do not know whether or not
... they are true; therefore, we need consistency proofs, if we are going
to use classical logical implications in arithmetic. By a consistency
B
proof I mean here a proof which tells us which 8entences are con-
If both these positions are winning-positions, Z ( A -+ B ) 1C will structively true if we have pkven aentences to be "true" with
be a winning-position also. We have, ther?foh, mother adiniaSibE
I ,: ,,!! , :
" the help of classical logic. The simplest result. fimt attained by
GENTZEN, ia that all implications between prime sentences which
;>den 'Ixi o h i d l y proven to hold, hold also constmctively.
I: '
40 Nonnative Logio and E t h k 4. Non-empirical Trulha tn the Formal Sciencee 41

I shall not go into such proofs in any detail, but I would like to tradition. Neverthelass it is necessary to keep in mind that, in
remark that such metamathematical resulta have to be proven term8 of the dialogical approach, these formal sciences are material
without using the doubtful tertium-non-datur. The metamnthe- games. The formal game uses the symbolism of prime-formulae
matical results are, therefore, constructively true. But in order to a, b(z),c(z, y), . . . and the logical particles only. If a terminological
clarify the notion of a constructive metamathematics, i t is neees- distinction is necewry, I shall call the formulae of the formal game
sary to clarify f h t the notion of constructive arithmetic. Though "logical formulae."
I have occasionally already used arithmetical sentences. I shall I should not mind calling logic a formal science; but our interest
have to introduce this new kind of sentence methodically without is now in the extra-logical formal sciences. I avoid calling these
presupposing any prior knowledge of arithmetic. Since F B ~ and E extrdogical formal sciences "mathematics" because, while traditio-
PEANO it is usual to represent arithmetic as an axiomatic theory, nally "mathematics" always includes geometry, geometry will
but we will have t o justify the axioms aa true sentences. They turn out not t o be a formal science. Thus I must attempt t o get
will turn out to be our first examples of true sentences which are along without using the term "mathematics". Nevertheless, our
neither logically nor empirically true. notion of the formal sciences will coincide roughly with the tradi-
tional notion of logic and mathematics, if geometry is excluded.
I n order to introduce arithmetic, I begin - aa mankind did at
least 10.000 years ago - with counting. To count means to invant
4. Non-empirical truths in the formal sciences a sequence of symbols called numerals and to use them in the well-
known way which we call "counting" objectsl. The simplest sequence
Thus far I have used two pairs of traditional terms: empirical- of numerals is mede up with one symbol only, say
logical and materid-formal. The first distinction is justified be- .. .
I, 11, 111, . (one, one-one, me-one-one, . .).
cause in the dialogical game with sentences composed of elementary
The introduction of these numerals is independent of the existence
sentences some truths are defensible independently of the defen-
of numeral words in one's natural language. Nor does the use of the
sibility of their elementary parts. This special defensibility I
stroke.numerals need to be called a n extension of one's language;
designate aa "logically true". The interest in logical truth then
but surely it is a new use of symbols.
leads to a game with non-sentences; namely, with newly introduced
I m u m e that you accept. as I do, the use of stroke-numerals
symbols, called tbrmulaa, This name "formula" suggests the use
for counting purposes as a reasonable activity. The sequence of
of the Aristotelian terms material-fonnal to distinguish the two
stroke numerals is COn0t~Ctedby the following rules:
games.
However, this does not exclude the search for senlencee which a m
neither elementary nor composed of elementary sentences. I f wo
find such sentences, we may logically compoae them; t h a t is, we
may compose them by using the logical particles. We will thus with n as a n "eigenvariable": that is, as a variable for strings of
arrive a t a new dialogical game which is not the formal game; so symbols constructed by meam of these rules themselves.
we will have to call it a material game - but its non-compounded $ % i s becomes rather inconvenient when we come to longer
sentences, its prime-sentences, will not be ebment&ry sentences. If strings of symbols, but we need not consider the devices for abbre-
true, they will not be empiricdy true. viationsat this stage of our discussion. In particular, Ineednot recall
Arithmetic will be our first example of such a new dialogical ..
our usual figures 1, 2, 3, . of Indian origin, i. e. -, a,r . . .
game. Analysis, i. e. the theory of real and complex functions, For our philosophical purpose of understanding how and why we
and Metamathematics will be other examples. It is customary t o use arithmetical sentendes, i t is sufficient to consider the simple
call these sciences "formal" because their eantences are u s d y stroke-notation; but as we know that there are different notations
called "formulae". I should like t o follow this tenninologicel for numerals. we will use the tam "number" aa a n abstractor.
40 Normtise Logic and Ethics 4. Non-empirical T~lUhsin the Formal Sciences 41

I shall not go into such proofs in any detail, hut I would like to tradition. Nevertheless i t is necessary to keep in mind that. in
remark that such metamathematical resulta have t o be proven terms of the dialogical approach, these formal sciences are material
without using the doubtful tertium-non-datur. The metamathe- games. The formal game usee the synlbolism of prime-formulae
matical results are, therefore, constructively true. But in order to a, b(x), c(x, y), . . . and the logical particles only. If a terminological
clarify the notion of a constructive metamathematics. i t is neces- distinction is necessary, I shall cell the formulae of the formal game
sary t o clarify f i s t the notion of constructive arithmetic. Though "logical formulae."
I have occasionally already used arithmetical sentences. I shall I should not mind calling logic a. formal science: but our interest
have to introduce this new kind of sentence methodically without is now in the ext.ra-logical formal sciences. I avoid calling these
presupposing any prior knowledge of arithmetic. Since FREQEand extralogical formal sciences "mathematics" because, while traditio-
PEANO i t is usual to represent arithmet,ic as an axiomatic theory, nally "mat,hematicsn always includes geometry, geometry will
but we will have to j ~ ~ s t i fthe
y axioms as true sontenccs. They turn out not t,o he a formal science. Thus I must attempt to get
will turn out t o he our first examples of true sentences which are along without using the term "mathematics". Nevertheless, our
neither logically nor empirically true. notion of the formal sciences will coincide roughly with the tradi-
tional notion of logic and mathematics, if geometry is excluded.
I n order to introduce arithmetic, I begin - as mankind did a t
least 10.000 years ago - with counting. To count means t,o invent
4. Nan-empirical trt~thsi n the formal srioncea a sequence of symbols called numerals and to use them in the well-
known way which we call "counting" objects. The simplest sequence
Thus far I have used two pair8 of traditional terms: empirical- of numerals is made up with one symbol only, say
logical and material-formal. The first distinction is justified be- I, 11, 111, . . . (one, one-one, one-one~ono,. . .).
cause in the dialogical game with sentences composed of elementary
sentences some truths are defensible independently of the defen- The introduction of these numerals is independent of the existence
sibility of their elementary parts. This special defensibility I of numeral words in one's natural language. Nor does the use of the
designate a s "logically true". The interest in logical truth then stroke-numerals need to be called an extension of one's language;
leads t o a game w ~ t hnon-sentences; namely, with newly introduced but surely it is a new use of symbols.
symbols, called formulae, This name "formula" suggests the use I assume that you accept, as I do, the use of stroke-numerals
of the Aristotelian terms material-formal to distinguish the two for counting purposes as a reasonable activity. The sequence of
games. stroke numerals is constructed by the following rules:
However, this does not exclude the search for sentences which are
neither elementary nor composed of elementary sentences. If we
find such sentences, we may logically compose them; that is, we
may compose them by using the logical particles. We will thus with n as an "eigenvariable": that is, as a variable for strings of
arrive at a new dialogical game which is not the formal game; so symbols constructed by means of these rules themselves.
we will have to call i t a material game - but its non-compounded $his becomes raLher inconvenient when we come to longer
sentences, its prime-sentences, will not be elementary sentences. If strings of symbols, but we need not consider the devices for ahhre-
true, they will not be empirically true. viationsat this stage of our discussion. In particular, I need not recall
Arit.hmntic will be our first example of such a new dialogical our usual figures 1, 2, 3, . . . of Indian origin, i. e. -, L , s.. .
pwrto. Analynin. i. o. tho theory of real and complex functions. For our philosophical purpose of understanding how and why we
~ ~ r i cMl o t . t r t , , t b l . l ~ o ~ ~ ~will i ~ ~otl~or
~ ~ t ,In) n oxarnplns. I t is customary to use arithmetical sentences, it is sufficient to ronsider the simple
call tltt,so sciunwe "I'orrrml'' hocmmn t,heir sentancos are usually stroke-notation; but as we know that there are different notations
rallod "formulae". I should like to follow this terminological for numerals, we will use the term "number" a s an abstractor.
42 Normative Logic and Ethics 4. Non-empirical Truths in the Formal Sciences 43

The ~lrrnlberI I I can be said to be represented, e. g., by the fignre 3. the sense that they can be defended as theses in a dialogical game.
We do not count objects as a mere pastime but in order t o We have now constructibility sentences t m, n as prime-sentences.
know wbethcr the objects counted will be sufficient, say for a fair If they are attacked, they have to be defended by performing the
distribution. This means that we want t o compare different groups construction.
of objects by counting them. Instead of comparing the groups, There are infinitely many sentences which are true in this sense,
we count them a n d compare the numbers. The order of numbers i. e. defensible. And it is on this foundation that the problem of
is not determined merely by examples. Of course, we could give axiomatization is possible; that is, of finding a convenient system
examplos of pairs of numbers m, n such that wc want t o say of true sentences such that all true sentences are logically implied
m < . . g. 1. I 1 1 a n I I. I I I I I would be such pain. Instead by them.
however, we give rules for all pairs m, n such that we wish to say
m < n . The rules I propose are:
=> l , n l
sider the false prime-sentences, i. e. the true sentences
All these sentences are logically implied by:
-
One simple answer t o this question is the following. We con-
m <n.

m, n => m 1 , n 1 . (T3) - m i l
If a pair m, n is constructible according t o these - as I am going
: and
j
t o say for short - < -rules, we put m < n :
Of course, we have first, t o assure ourselves that these new axioms
1 are true. Why will no one be able to construct nr, 1 according t o
I f we reflect on what we have done and ask ourselves: "How do the i -rules? Because only pairs 1, n / and m I , n I are constructible.

i
we know t h a t I I I < I I I 1 1 ?", tho answer has t o he "because These pairs have n I as their second member, but m. I has I as its
I I I, I I I I I is a constructible pair according t o the <-rules". The second member and we have I < n I .
next question, retrospectively, is: "Why do we accept the <-rules?" In order to "prove" the second assertion it issufficient t,o defend
This is no longer a question of trnth, but a practical question. No
(T4) t n I < n I + m < n .
one is forced t o accept these roles, but they are t o be recommended
if we wish t o open a new field of symbolic activity which is some- The strategy is simply this: I f the opponent has constructed
times useful in dealing with groups of objects. Only practice can 1 m 1, n 1, he will have used the second <-rule (the pair m 1, n I
is not un initial pair 1. n I because for the firat members we have
teach tho value of such a tochnique. Terminologically, I would like
t o say that the <-rules have a pragmatic justification. 1 I ~ n l ) .
Let us compare t,his answer with the axiomatic approach. Now we add t o ( T I ) - ( T 4) the principle of induction, i. e. for
II
Instead of the <.rules one states <-axioms: each formula A ( n ) we take the formula:
I
(T5 ) A(I) A A m . A(m) A(ml). -L A , A ( n )
l<n l
+
(T 1)
(T2) ? n < n + m I < n I as a n axiom.
or, if you prefer, with universal quantifiers in front: These formulae are indeed true, i. e. defensible. A dialogue may
run as follows:
1 A([)
i 2 A m . A ( m )+ A ( m l ) . A A(n)
All true prime-sentences m < n are now logically implied by these i 3 I I I ? A ( II I)
axioms. The question: "How do we know the truth of the axiomal"
I
4 ? I ?2
is not permissible; one refers vaguely t o empirical verification or I 5 4)+ A ( I I ) A(!) 7 5
confirmations. Starting with the pragmatically justified <-rules, I 6 A(I I ) I I ?2
o n the other hand we see that the <-axioms are indeed true, in I 7 4 1 I ) + A ( ( I I) A(I1) ? 7
42 Normative Logic and Ethics 4. Non-empirical Truths in the Formal Sciences 43

The ~lrrnlberI I I can be said to be represented, e. g., by the fignre 3. the sense that they can be defended as theses in a dialogical game.
We do not count objects as a mere pastime but in order t o We have now constructibility sentences t m, n as prime-sentences.
know wbethcr the objects counted will be sufficient, say for a fair If they are attacked, they have to be defended by performing the
distribution. This means that we want t o compare different groups construction.
of objects by counting them. Instead of comparing the groups, There are infinitely many sentences which are true in this sense,
we count them a n d compare the numbers. The order of numbers i. e. defensible. And it is on this foundation that the problem of
is not determined merely by examples. Of course, we could give axiomatization is possible; that is, of finding a convenient system
examplos of pairs of numbers m, n such that wc want t o say of true sentences such that all true sentences are logically implied
m < . . g. 1. I 1 1 a n I I. I I I I I would be such pain. Instead by them.
however, we give rules for all pairs m, n such that we wish to say
m < n . The rules I propose are:
=> l , n l
sider the false prime-sentences, i. e. the true sentences
All these sentences are logically implied by:
-
One simple answer t o this question is the following. We con-
m <n.

m, n => m 1 , n 1 . (T3) - m i l
If a pair m, n is constructible according t o these - as I am going
: and
j
t o say for short - < -rules, we put m < n :
Of course, we have first, t o assure ourselves that these new axioms
1 are true. Why will no one be able to construct nr, 1 according t o
I f we reflect on what we have done and ask ourselves: "How do the i -rules? Because only pairs 1, n / and m I , n I are constructible.

i
we know t h a t I I I < I I I 1 1 ?", tho answer has t o he "because These pairs have n I as their second member, but m. I has I as its
I I I, I I I I I is a constructible pair according t o the <-rules". The second member and we have I < n I .
next question, retrospectively, is: "Why do we accept the <-rules?" In order to "prove" the second assertion it issufficient t,o defend
This is no longer a question of trnth, but a practical question. No
(T4) t n I < n I + m < n .
one is forced t o accept these roles, but they are t o be recommended
if we wish t o open a new field of symbolic activity which is some- The strategy is simply this: I f the opponent has constructed
times useful in dealing with groups of objects. Only practice can 1 m 1, n 1, he will have used the second <-rule (the pair m 1, n I
is not un initial pair 1. n I because for the firat members we have
teach tho value of such a tochnique. Terminologically, I would like
t o say that the <-rules have a pragmatic justification. 1 I ~ n l ) .
Let us compare t,his answer with the axiomatic approach. Now we add t o ( T I ) - ( T 4) the principle of induction, i. e. for
II
Instead of the <.rules one states <-axioms: each formula A ( n ) we take the formula:
I
(T5 ) A(I) A A m . A(m) A(ml). -L A , A ( n )
l<n l
+
(T 1)
(T2) ? n < n + m I < n I as a n axiom.
or, if you prefer, with universal quantifiers in front: These formulae are indeed true, i. e. defensible. A dialogue may
run as follows:
1 A([)
i 2 A m . A ( m )+ A ( m l ) . A A(n)
All true prime-sentences m < n are now logically implied by these i 3 I I I ? A ( II I)
axioms. The question: "How do we know the truth of the axiomal"
I
4 ? I ?2
is not permissible; one refers vaguely t o empirical verification or I 5 4)+ A ( I I ) A(!) 7 5
confirmations. Starting with the pragmatically justified <-rules, I 6 A(I I ) I I ?2
o n the other hand we see that the <-axioms are indeed true, in I 7 4 1 I ) + A ( ( I I) A(I1) ? 7
44 Nownative Logic and Ethics 4. Non-empirical Truth in the Formal Sciences 45

The opponent gives up, because he has to assert A(I I I ) which the For multiplication the procedure is the same. We begin with
proponent has t o defend. Against other choices than I I 1 3 in line pragmatically justified construction roles for triplets and define
3 the defense is similar. the prime-sentences m . n = p by means of constructibility.
( T I ) - (T 4) together wit,h the principle of induction (T 5) are a To defend a prime-sentence means to perform a construction.
complete axiomatization of the <-arithmetic. Even the numbers (or numerals, if you prefer) have to be con-
As GODEL has proved, if we admit addition and multiplication structed before arithmetic can begin. So "construction" marks the
in the prime-sentences, a complete axiomatization is no longer difference between arithmetical prime-sentences and elementary
possible. There are always formulae A(n) such that A(m) is logically sentences with proper names and predicators.
implied for each m, but A, A(n) is not logically implied by the I n order to come closer to the traditional K A N T ~ ~ I~ shall
terms,
axiom system. Nevertheless, A n (An) will be true: t o defend it use "synthetical" instead of "constructive". At least, according
means nothing more than to defend A(m) with m chosen by the to the dictionaries, both the Greek word "synthesis" and the
opponent. Latin word "constrnctio" ran be translated as "Zusarnrnenset-
To prove such a result is rather difficult, but in these lectures zung", "putting together".
only the philosophical questions are relevant. How and why do we To defend an arithmetical truth, we need not use the "crnpir-
get true arithmetical sentences? We have only to ask for the ical" truth of elementary sentences; we need only use "synthesis"
prime-sentences. We have seen that the prime-sentences m < n according to pragmatically justified rules. Therefore, I propose to
have their origin in rules for constructing pairs of numbers. We call arithmetical truths synthetic truths. They are non-empirical
can define equality and inequality in the following way: truths. If we follow the K n ~ ~ i ausage
n and call all non-empirical
truths "a priori", we have with KANTthe arithmetical truths as
synthetic a priori truths.
Let me add that there is also an important similarity between
T h e usual axioms for equality: the arit,hmetical truths and logical truths. Logical truths are
defensible in the formal game. Nothing has to be known about the
world except a game with uninterpreted symbols; therefore, logical
truths are called formal truths. I n arithmetic we have the same
are logically implied by our axioms. situation. Once the pragmatic justification of the construction
I n order t o introduce addition, I shall give rules for construct- rules by which we introduce the numerals and the prime-sentences
ing triplets of numbers; namely: has been understood, we may forget it. We then have a game with
3m, I, ml
symbols before us. Arithmetical truths are true sentences in this
a m , nl, pl . game, beginning with constructible sentences. Therefore, I should
m, n, p
like to qualify the "synthetic a priori" as "fomally synthetic a
T h e justification for proposing (and accepting) these rules is again priori".
pragmatic. As all of us have practiced addition since elementary This terminological proposal may not be quite satisfactory, but.
school, I shall take it for granted. it suggests the distinction between four kinds of a priori truths:
Once the rules are accepted, everything has to be proven. For formally analytic - formally synthetic
addition we first have to show (which I shall not do here) that the materially analytic - materially synthetic
third member is uniquely determined:
As I offer in these lectures just four kinds of a priori troths, thp
terminology serves my purposes satisfactorily.
Then we may define: The material a priori will be dealt with in the next lecture.
The term "formally analytic" I propose to use for such truths as
44 Nownative Logic and Ethics 4. Non-empirical Truth in the Formal Sciences 45

The opponent gives up, because he has to assert A(I I I ) which the For multiplication the procedure is the same. We begin with
proponent has t o defend. Against other choices than I I 1 3 in line pragmatically justified construction roles for triplets and define
3 the defense is similar. the prime-sentences m . n = p by means of constructibility.
( T I ) - (T 4) together wit,h the principle of induction (T 5) are a To defend a prime-sentence means to perform a construction.
complete axiomatization of the <-arithmetic. Even the numbers (or numerals, if you prefer) have to be con-
As GODEL has proved, if we admit addition and multiplication structed before arithmetic can begin. So "construction" marks the
in the prime-sentences, a complete axiomatization is no longer difference between arithmetical prime-sentences and elementary
possible. There are always formulae A(n) such that A(m) is logically sentences with proper names and predicators.
implied for each m, but A, A(n) is not logically implied by the I n order to come closer to the traditional K A N T ~ ~ I~ shall
terms,
axiom system. Nevertheless, A n (An) will be true: t o defend it use "synthetical" instead of "constructive". At least, according
means nothing more than to defend A(m) with m chosen by the to the dictionaries, both the Greek word "synthesis" and the
opponent. Latin word "constrnctio" ran be translated as "Zusarnrnenset-
To prove such a result is rather difficult, but in these lectures zung", "putting together".
only the philosophical questions are relevant. How and why do we To defend an arithmetical truth, we need not use the "crnpir-
get true arithmetical sentences? We have only to ask for the ical" truth of elementary sentences; we need only use "synthesis"
prime-sentences. We have seen that the prime-sentences m < n according to pragmatically justified rules. Therefore, I propose to
have their origin in rules for constructing pairs of numbers. We call arithmetical truths synthetic truths. They are non-empirical
can define equality and inequality in the following way: truths. If we follow the K n ~ ~ i ausage
n and call all non-empirical
truths "a priori", we have with KANTthe arithmetical truths as
synthetic a priori truths.
Let me add that there is also an important similarity between
T h e usual axioms for equality: the arit,hmetical truths and logical truths. Logical truths are
defensible in the formal game. Nothing has to be known about the
world except a game with uninterpreted symbols; therefore, logical
truths are called formal truths. I n arithmetic we have the same
are logically implied by our axioms. situation. Once the pragmatic justification of the construction
I n order t o introduce addition, I shall give rules for construct- rules by which we introduce the numerals and the prime-sentences
ing triplets of numbers; namely: has been understood, we may forget it. We then have a game with
3m, I, ml
symbols before us. Arithmetical truths are true sentences in this
a m , nl, pl . game, beginning with constructible sentences. Therefore, I should
m, n, p
like to qualify the "synthetic a priori" as "fomally synthetic a
T h e justification for proposing (and accepting) these rules is again priori".
pragmatic. As all of us have practiced addition since elementary This terminological proposal may not be quite satisfactory, but.
school, I shall take it for granted. it suggests the distinction between four kinds of a priori truths:
Once the rules are accepted, everything has to be proven. For formally analytic - formally synthetic
addition we first have to show (which I shall not do here) that the materially analytic - materially synthetic
third member is uniquely determined:
As I offer in these lectures just four kinds of a priori troths, thp
terminology serves my purposes satisfactorily.
Then we may define: The material a priori will be dealt with in the next lecture.
The term "formally analytic" I propose to use for such truths as
46 Normative Logic and Ethics 4 . Non-empirical Trutha in the Fwmal Sciences 47

need for their defense not merely the logical rules, but also &fi- A(x) defines implicitly a n object y that satisfies the condition.
nitions. Definitions occur, of course, also in connection with ele- Here I would prefer to d e h e explicitly:
mentary sentences. If we start with the two-place predicator
"married", such that "2. y c married" is the standard version
of "x is married to y"-and if we introduce for legal purposes The so-called implicit definition of a function F by a formula
the predicator "bachelor" by defining:

r c bachelor % z c male A - V x, y e married,

A(x, y) is also reducible to an explicit definition with a description:

we get the t r u e sentenre: There remains only one interesting case, the case of so-called induc-
. . . . of predi-
tive definitions. I n arithmetic sequences p I . p ~p3,
x. y r married + x c' bachelor
cators are often introduced by conditions:
(i. e.: If x is married t.u y, then x is not a bachelor).
An example in arit.hmetic would be:

Since DEDEKIND i t has .been customary t o "prove" that there

exists for each n exactly one set S, such that
A logical implication of thin definition would be:

I n order t o defend the truth of these sentences, one has to know By this "proof" we could define 8 , explicitly with a description,
merely some definitions. One does not have to know whether the bnt this "proof" uses a naive or formalized version of Cantorian
prime-sentences are elementary or arithmetical prime-sentences. set theory. I propose instead to introduce sets as abstracts repre-
a n d one need know nothingabout theirdefensibility. When wesubsti- sented by sentence-forms; I shall give some details later.
t u t e the definiem for the definiendum, the sentences become logically For "inductive definitions" the abstraction-theory of sets has
true. Such sentences I propose to call "fornrally analytic truths". the consequence that we have to establish the sentence-formsr c p n
Logical truths a r e a proper part of them. Forrually analytic truths as meaningful, independently of set-theory. The sin~plestway I
which are not logically true I would like to call "formally analytic know is t o introduce the predicators by means of construction
t r ~ ~ t hins the strict sense". "Formally analytically true s. s." rules for true sentences.
t,trrrefnre means that the sentence is defensible in virtue of its
logical form and a t least one definition. The term "definition" haa
a long complex history. I shall use i t here only in the narrow sense
of esplicit definition :
With these rules, to defend x c p l means to defend A(x), and to
defend x c p,+l means to defend B(p,, x ) . So by an application of
arithmetical induction we get the result: that for each n tho thesis
The predicator p is introdnced by the rule that the loft side has to n c p, has a meaning; it is determined how to defend it in a dia-
be replaced b y the right side. logue. Thus it is obvious that these predicators satisfy the initial
This narrow sense excludes so-called "implicit definitions" (not conditions of so-called inductive definition. Moreover according
t o mention "real definitions"). Very often, when i t is said that a n t o the terminology proposed, we now have. in all caaes where a n
axiom system "implicitly defines" its objects, this is only a myth. "inductive definition" has t o be used, not analytic truths, but
B u t there are more serious cases, e. g. if it is said, that a formula formally synthetic truths.
46 Normative Logic and Ethics 4 . Non-empirical Trutha in the Fwmal Sciences 47

need for their defense not merely the logical rules, but also &fi- A(x) defines implicitly a n object y that satisfies the condition.
nitions. Definitions occur, of course, also in connection with ele- Here I would prefer to d e h e explicitly:
mentary sentences. If we start with the two-place predicator
"married", such that "2. y c married" is the standard version
of "x is married to y"-and if we introduce for legal purposes The so-called implicit definition of a function F by a formula
the predicator "bachelor" by defining:

r c bachelor % z c male A - V x, y e married,

A(x, y) is also reducible to an explicit definition with a description:

we get the t r u e sentenre: There remains only one interesting case, the case of so-called induc-
. . . . of predi-
tive definitions. I n arithmetic sequences p I . p ~p3,
x. y r married + x c' bachelor
cators are often introduced by conditions:
(i. e.: If x is married t.u y, then x is not a bachelor).
An example in arit.hmetic would be:

Since DEDEKIND i t has .been customary t o "prove" that there

exists for each n exactly one set S, such that
A logical implication of thin definition would be:

I n order t o defend the truth of these sentences, one has to know By this "proof" we could define 8 , explicitly with a description,
merely some definitions. One does not have to know whether the bnt this "proof" uses a naive or formalized version of Cantorian
prime-sentences are elementary or arithmetical prime-sentences. set theory. I propose instead to introduce sets as abstracts repre-
a n d one need know nothingabout theirdefensibility. When wesubsti- sented by sentence-forms; I shall give some details later.
t u t e the definiem for the definiendum, the sentences become logically For "inductive definitions" the abstraction-theory of sets has
true. Such sentences I propose to call "fornrally analytic truths". the consequence that we have to establish the sentence-formsr c p n
Logical truths a r e a proper part of them. Forrually analytic truths as meaningful, independently of set-theory. The sin~plestway I
which are not logically true I would like to call "formally analytic know is t o introduce the predicators by means of construction
t r ~ ~ t hins the strict sense". "Formally analytically true s. s." rules for true sentences.
t,trrrefnre means that the sentence is defensible in virtue of its
logical form and a t least one definition. The term "definition" haa
a long complex history. I shall use i t here only in the narrow sense
of esplicit definition :
With these rules, to defend x c p l means to defend A(x), and to
defend x c p,+l means to defend B(p,, x ) . So by an application of
arithmetical induction we get the result: that for each n tho thesis
The predicator p is introdnced by the rule that the loft side has to n c p, has a meaning; it is determined how to defend it in a dia-
be replaced b y the right side. logue. Thus it is obvious that these predicators satisfy the initial
This narrow sense excludes so-called "implicit definitions" (not conditions of so-called inductive definition. Moreover according
t o mention "real definitions"). Very often, when i t is said that a n t o the terminology proposed, we now have. in all caaes where a n
axiom system "implicitly defines" its objects, this is only a myth. "inductive definition" has t o be used, not analytic truths, but
B u t there are more serious cases, e. g. if it is said, that a formula formally synthetic truths.
48 Normative Logic and Ethica 4 . Nan-empirical T r u t h in the Formal Sciences 49

I want t o conclude my remarks about the formal sciences by may be chosen for representing a set. Whatever we decide, it
indicating how arithmetic can be generalized a n d extended without should be clear that the formula A ( n ) cannot be admitted as
leaving the limits of formal truth. The generalizations are easy. representing a set. A ( n ) cannot be admitted, since we are attempt-
Instead of beginning with the simplest construct.ion possible, ing to fix the range of the variable S. Only after we have fixed the
namely, the construction of the stroke-numerals, we may investi- range, A ( n ) has been defined. Of course, it may be that we can
gate constructions of arbitrary strings of symbols. I n this way we define a formula A' ( n ) without using quantified set-variables--and
get the theory of calculi or rnetan~athematics.This last name indi- that we can prove:
cates the main application of the general theory, in which we take
as strings of symbols the sentences of particular formal theories.
Methodologically, the theory of calculi does not. raise problems
But to use set-variables without fixing their range and then to
which we do n o t face already in arithmetic. Bnt something new
use formulae with quantified set-variables in order to define sets
seems t o happen if we extend arithmetic, from its elementary level,
is k vicious circle. Although POINCAR;:had discovered this mistake
the level we have been considering so far, t o higher levels, i. e. t o
of CANTOR^^^ set theory, RUSSELL comitted the same mistake by
the theory of real numbers, customarily referred to as analysis.
adding the axiom of reducibility t o his ramified theory of typos.
The new phenomenon here is the use of sets of numbers as objects
of the theory. Since then axiomatic set theories have been en vogue. One
Without going into the details of real numbers, we can formulate simply postulates the existence of sets in such a way that the
the essential problem in the following way. Wo start with arithme- comprehension axiom :
tical sentence-forms. We define a n equivalence relation between
such formulae, A ( n ) ,R ( n ) ,by A. . A ( n ) ++ B ( n ) . By a n abstrac-
tion with respect to this equivalence we proceed from the formidae holds for all formulae A ( x ) , including those with quantified set-
t o the sets represented. This is the abstraction-theory of sets variables. This is called "impredicative" comprehension. Instcad
mentioned above. This is the way all mathematicians use sets in of postulating the existence of sets, HILBERT proposed that we
practice, although their philosophy normally forbids them t o admit consider axiomatic set theories as a fornlal game; bnt he added
this. We denote the set represented by A ( n ) as c , A ( n ) . that this game makes sense only if its formal consistency can be
The c-rolation between elomcnts and sets is defined by proven as a metamathematical theorem. Such a proof would give
a constructive interpretation of axiomatic theory.

As a11 8et.s have t,o be represented by a formula, all questions about So there are two possibilities of giving a foundation to traditional
t h e existence of sets of numbers are questions about the existence analysis: either t o restrict comprehension to predicati\.e compre-
of formulae in t h c arithmetical language. For the purposes of ana- hension or to prove the consistency of impretlicati\.e axiomatic
lysis this language need not he specified; on the contrary it may theories. I am in favour of a predicrttive analysis and have carried
always be held open for the introduction of new prime-sentences. out this prograni in my book "Differential und Integral", but I
Now let US have a sentence-form referring t o numbers a.ld sets would not mind if impredicative theories could constructively be
of nunhers: B ( n , 8 ) .If we quantify the set-variable S, we get new proven t o be consistent.
formulae, for instance The so-called working mathematicians of our time prefer t o
work with impredicative theories without a consistency proof. To
understand this unjustified preference is a matter of contemporary
Is this a definition? I n order t o answer this question we have t o history. I shall not go into such an "explanation", because it is not
fix a dialogical use of the definiens. To defend V s B(m, S), we relevant t o the philosophical attempt t o understand how formal
have t o name a set. We have therefore to decide which formulae sciences a r e "possible".
48 Normative Logic and Ethica 4 . Nan-empirical T r u t h in the Formal Sciences 49

I want t o conclude my remarks about the formal sciences by may be chosen for representing a set. Whatever we decide, it
indicating how arithmetic can be generalized a n d extended without should be clear that the formula A ( n ) cannot be admitted as
leaving the limits of formal truth. The generalizations are easy. representing a set. A ( n ) cannot be admitted, since we are attempt-
Instead of beginning with the simplest construct.ion possible, ing to fix the range of the variable S. Only after we have fixed the
namely, the construction of the stroke-numerals, we may investi- range, A ( n ) has been defined. Of course, it may be that we can
gate constructions of arbitrary strings of symbols. I n this way we define a formula A' ( n ) without using quantified set-variables--and
get the theory of calculi or rnetan~athematics.This last name indi- that we can prove:
cates the main application of the general theory, in which we take
as strings of symbols the sentences of particular formal theories.
Methodologically, the theory of calculi does not. raise problems
But to use set-variables without fixing their range and then to
which we do n o t face already in arithmetic. Bnt something new
use formulae with quantified set-variables in order to define sets
seems t o happen if we extend arithmetic, from its elementary level,
is k vicious circle. Although POINCAR;:had discovered this mistake
the level we have been considering so far, t o higher levels, i. e. t o
of CANTOR^^^ set theory, RUSSELL comitted the same mistake by
the theory of real numbers, customarily referred to as analysis.
adding the axiom of reducibility t o his ramified theory of typos.
The new phenomenon here is the use of sets of numbers as objects
of the theory. Since then axiomatic set theories have been en vogue. One
Without going into the details of real numbers, we can formulate simply postulates the existence of sets in such a way that the
the essential problem in the following way. Wo start with arithme- comprehension axiom :
tical sentence-forms. We define a n equivalence relation between
such formulae, A ( n ) ,R ( n ) ,by A. . A ( n ) ++ B ( n ) . By a n abstrac-
tion with respect to this equivalence we proceed from the formidae holds for all formulae A ( x ) , including those with quantified set-
t o the sets represented. This is the abstraction-theory of sets variables. This is called "impredicative" comprehension. Instcad
mentioned above. This is the way all mathematicians use sets in of postulating the existence of sets, HILBERT proposed that we
practice, although their philosophy normally forbids them t o admit consider axiomatic set theories as a fornlal game; bnt he added
this. We denote the set represented by A ( n ) as c , A ( n ) . that this game makes sense only if its formal consistency can be
The c-rolation between elomcnts and sets is defined by proven as a metamathematical theorem. Such a proof would give
a constructive interpretation of axiomatic theory.

As a11 8et.s have t,o be represented by a formula, all questions about So there are two possibilities of giving a foundation to traditional
t h e existence of sets of numbers are questions about the existence analysis: either t o restrict comprehension to predicati\.e compre-
of formulae in t h c arithmetical language. For the purposes of ana- hension or to prove the consistency of impretlicati\.e axiomatic
lysis this language need not he specified; on the contrary it may theories. I am in favour of a predicrttive analysis and have carried
always be held open for the introduction of new prime-sentences. out this prograni in my book "Differential und Integral", but I
Now let US have a sentence-form referring t o numbers a.ld sets would not mind if impredicative theories could constructively be
of nunhers: B ( n , 8 ) .If we quantify the set-variable S, we get new proven t o be consistent.
formulae, for instance The so-called working mathematicians of our time prefer t o
work with impredicative theories without a consistency proof. To
understand this unjustified preference is a matter of contemporary
Is this a definition? I n order t o answer this question we have t o history. I shall not go into such an "explanation", because it is not
fix a dialogical use of the definiens. To defend V s B(m, S), we relevant t o the philosophical attempt t o understand how formal
have t o name a set. We have therefore to decide which formulae sciences a r e "possible".
 50 Normative Logic and Ethics 5 . Non-empirical Truths in the Material Scienced 51

5. Non-empirical truths in the material sciences I n what follows I shall deal with linguist,^^ norms which arc
proposed for acceptance only because they deal wit.h special objects.
I n the last lectures I indicated how formal sciences can be We are now looking more closely a t the world, not merely realizing
reconstructed. The aim was not to reconstruct uncritically the that there are objects which may be treated equally or differently
contemporary fashions of science, but to present a critical recon- by means of predicators. The result of speaking a b o ~what
~ t we see
struction. The result was a reconstruction of constructive logical through this closer examination I shall call material sciences.
calculi and of classical logical calculi with the help of a construc- The first thing to do is to refine the met,hod of speaking with
tive consistency proof. I have indicated the beginnings of con- elementary sentences only. I f we have a t our disposal nothing
structive arithmetic, of constructive metamathematics and of con- other than elementary sentences, we can determine the use of a
structive analysis. Here the usual terminology is "predicative" predicator by examples and counterexamples only. I restrict myself
analysis. The reconstruction did not allow the justification of t o a one-place predicator p. The examples may be:
impredicative analysis or, more generally, CANTORIAN set theory.
These theories are to be considered as fashionable games only. Of and
course, a ~onstructiveinterpretation may be found in the future,
hut t,his is a t present merely wishful thinking. the counterexamples.
I n a refined version of KANTIAN terminology the true sentences If we now come to a new object R, how are we going to decide
of the formal sciences may be divided into three classes as shown by whether R c p or R c' p ? It is an easy answer to say: "by irnme-
the following system : diate comparison", or to say: "in virtue of the similarity or dissi-

formal truths
.
milarity with 81, . . , S , or T I ,. . . , T,". Only if I am already
able to use a t least some predicators in new cases, am I able to
\
/
/ formally a priori a. a.
/ \
understand such predicators a s "to compare" or "similar".
Let me call the method of examples and counterexamples the
exemplary determination of a predicator. This is approximately
the same as what is usually called "ostensive definition", but it
logical formally analytic formally synthetic
is not a definition in my strict sense and the examples may not be
T h e criterion for the distinction is the use of certain linguistic physical objects, as is normally assumed in so-called ostensive
norms for verification: whether the rules of logic alone, logic and definitions. With examples and counterexamples we may also
definitions alone, or logic, definitions and constructive rules are determine the use of predicators for our own activities, such as
sufficient. e. g. "naming" (sc. an object with a proper name) in contrast to
I n all cases of formal truths we need not know anything "affirming or denying" (sc. a predicator of an object). To learn
about particular objects. Logic, definitions and symbolic construc- these predicators, it is useless merely t,o point to the physical
tions are to be justified pragmatically. This means that we have to events of uttering words.
understand our human situation in the world; we have to under- Exemplary detcrmination has an indirect variant: namely,
s t a n d that the acceptance of linguistic norms is good for us; that telling stories. In St. Luke 10: 29 Jesus is asked "Who is niy
we cannot become truly human without them. All this has to be neighbour?". St. Luke continues: "And Jesus ansneriny said, a
understood even if we have no words to formulate it. certain man went down from Jerusalem etc. etc." Jesus tells the
Yet, I repeat, we need know nothing about special objects. We story of the good Samaritan as an example of the phrase "to be
use elementary sentences to speak about all objects whatever they neighbour unto someone". Though thin is accomplished solely
m a y be. Then we form composite sentences out of elementary with words, no definition of the phrase is given. The words of the
sentences by t h e use of logical particles. We may consider objects, story are, of course, assumed to be already understood. 1 should Like
i. e. things given by proper names, in whatever way we choose. to call this method of exen~plificationby stories "indirect exem-
 50 Normative Logic and Ethics 5 . Non-empirical Truths in the Material Scienced 51

5. Non-empirical truths in the material sciences I n what follows I shall deal with linguist,^^ norms which arc
proposed for acceptance only because they deal wit.h special objects.
I n the last lectures I indicated how formal sciences can be We are now looking more closely a t the world, not merely realizing
reconstructed. The aim was not to reconstruct uncritically the that there are objects which may be treated equally or differently
contemporary fashions of science, but to present a critical recon- by means of predicators. The result of speaking a b o ~what
~ t we see
struction. The result was a reconstruction of constructive logical through this closer examination I shall call material sciences.
calculi and of classical logical calculi with the help of a construc- The first thing to do is to refine the met,hod of speaking with
tive consistency proof. I have indicated the beginnings of con- elementary sentences only. I f we have a t our disposal nothing
structive arithmetic, of constructive metamathematics and of con- other than elementary sentences, we can determine the use of a
structive analysis. Here the usual terminology is "predicative" predicator by examples and counterexamples only. I restrict myself
analysis. The reconstruction did not allow the justification of t o a one-place predicator p. The examples may be:
impredicative analysis or, more generally, CANTORIAN set theory.
These theories are to be considered as fashionable games only. Of and
course, a ~onstructiveinterpretation may be found in the future,
hut t,his is a t present merely wishful thinking. the counterexamples.
I n a refined version of KANTIAN terminology the true sentences If we now come to a new object R, how are we going to decide
of the formal sciences may be divided into three classes as shown by whether R c p or R c' p ? It is an easy answer to say: "by irnme-
the following system : diate comparison", or to say: "in virtue of the similarity or dissi-

formal truths
.
milarity with 81, . . , S , or T I ,. . . , T,". Only if I am already
able to use a t least some predicators in new cases, am I able to
\
/
/ formally a priori a. a.
/ \
understand such predicators a s "to compare" or "similar".
Let me call the method of examples and counterexamples the
exemplary determination of a predicator. This is approximately
the same as what is usually called "ostensive definition", but it
logical formally analytic formally synthetic
is not a definition in my strict sense and the examples may not be
T h e criterion for the distinction is the use of certain linguistic physical objects, as is normally assumed in so-called ostensive
norms for verification: whether the rules of logic alone, logic and definitions. With examples and counterexamples we may also
definitions alone, or logic, definitions and constructive rules are determine the use of predicators for our own activities, such as
sufficient. e. g. "naming" (sc. an object with a proper name) in contrast to
I n all cases of formal truths we need not know anything "affirming or denying" (sc. a predicator of an object). To learn
about particular objects. Logic, definitions and symbolic construc- these predicators, it is useless merely t,o point to the physical
tions are to be justified pragmatically. This means that we have to events of uttering words.
understand our human situation in the world; we have to under- Exemplary detcrmination has an indirect variant: namely,
s t a n d that the acceptance of linguistic norms is good for us; that telling stories. In St. Luke 10: 29 Jesus is asked "Who is niy
we cannot become truly human without them. All this has to be neighbour?". St. Luke continues: "And Jesus ansneriny said, a
understood even if we have no words to formulate it. certain man went down from Jerusalem etc. etc." Jesus tells the
Yet, I repeat, we need know nothing about special objects. We story of the good Samaritan as an example of the phrase "to be
use elementary sentences to speak about all objects whatever they neighbour unto someone". Though thin is accomplished solely
m a y be. Then we form composite sentences out of elementary with words, no definition of the phrase is given. The words of the
sentences by t h e use of logical particles. We may consider objects, story are, of course, assumed to be already understood. 1 should Like
i. e. things given by proper names, in whatever way we choose. to call this method of exen~plificationby stories "indirect exem-
 52 Normative Logic and Ethics 5 . Nan-empirical Trutha in the Material Sciencea 53

plary determination". If we avoid circularities it is a very powerful

method of extending our common vocabulary. A % . x E proper narne-tx c' predicator.
Nevertheless, if we used only exemplary determinations, direct Tommy ? Tommy c proper name +
or indirect, for all our predicators, there would be too much Tommy c' predicator
disagreement, too much uncertainty. The use of some predicators Tommy E proper name ? Tommy c' predicator
may be made relatively stable by merely exemplary determination, ? (...)
but we have t o look for another method of stabilizing the use of
most of our predicators. I propose to use rules which prescribe a The defense of the elementary sentence "Tommy c' predicator"
connection bet,ween different predicators. The simplest rules of can be performed by deriving it, beginning with sentences asserted
this sort were dealt with by ARISTOTLE. His universal categorical by the opponent and using the accepted rule (R).
sontcnces may be int,orprct,ed a s the following rules for one-place Let me call rnles for predicators which are accepted for pragma-
predicators p and q : tical reasons, namely, in order to improve the use of the predicators
affirmative: z e p => x e q involved, materially analytic rules. If a sentence can be defended
negative: zc p => r c' q . in virtue of its logical form, perhaps with the help of definitions
and with the help of a t least one materially analytic rule, this
These rules prescribe the transition from an elementary sentence sentence will be called, a "materially analytic truth". Whether or
of the form x c p to the sentence x c q or z e' q, as the case may be. not a sentence is materially-analytically true always depend8 on
The question is: why accept such rules? the rules accepted. The acceptance of the rules depends on our
I n order t o answer this question, let me begin with an example practice of using the predicators. We may say that the accept.ance
of my own terminology. When introdncing elementary sentences, of the rules depends on "experience", but then "experienre" is
I used two kinds of words, proper names and predicators. These something different from our knowledge of empirically true sen-
terms are determined exemplarily. Thus we must imagine that tences. We do not accept materially analytic rules because we
up t o now the nse of these terms has been determined merely by experience the truth or falsity of elementary sentences.
giving some examples and counterexamples. This may lead to
I should like to illustrate this with the famous case of colour-words.
. difficulties, e. g . someone may assert that the word "sun" is a Whereas the bull becomes nervous by the colour red, the philoso-
proper name a n d also a predicator. A similar difficulty normally
pher since antiquity already becomes nervous by the word 'colour'.
arises with the word "God". Now I hope that along with me you
Let us assume that-independently of English usage-we have
accept it a s reasonable to fix the use of our terms in such a way
introduced "red" by examples of fresh blood, and "greenish" by
t h a t a proper name shall never be a predicator. Nothing other than
examples of different leaves. If we have a violet spot S brforc us,
this is proposed by the nogative rule:
why not call i t "a greenish red", i. e. S cred A S c greenish? This
( R ) x e proper name => x E' predicator. would be possible. However, if we have introduced a predicator
"blueish" with suitable examples and counterexamples, i t would be
We do not have a vocabulary and a set of definitions such that better to speak of "blueish red" a n d to accept a rule which makes
t h e universal sentence: No proper name is a predicator (I take "red" and "greenish" contraries,
this a s a n ordinary version of
z E red => x E' greenish
Ar . x c proper name -t x E' predicator.)
One could oppose accepting this rule by pointing to an orange
is logically implied. This universal sentence is not a formally ana- spot, but, once more, it is possible to propose first the introduction
lytical truth. However, after having accepted the proposed rule. of a new predicator "yellowish", and then to insist on the negative
i t is easy t o defend this sentence: rule.
 52 Normative Logic and Ethics 5 . Nan-empirical Trutha in the Material Sciencea 53

plary determination". If we avoid circularities it is a very powerful

method of extending our common vocabulary. A % . x E proper narne-tx c' predicator.
Nevertheless, if we used only exemplary determinations, direct Tommy ? Tommy c proper name +
or indirect, for all our predicators, there would be too much Tommy c' predicator
disagreement, too much uncertainty. The use of some predicators Tommy E proper name ? Tommy c' predicator
may be made relatively stable by merely exemplary determination, ? (...)
but we have t o look for another method of stabilizing the use of
most of our predicators. I propose to use rules which prescribe a The defense of the elementary sentence "Tommy c' predicator"
connection bet,ween different predicators. The simplest rules of can be performed by deriving it, beginning with sentences asserted
this sort were dealt with by ARISTOTLE. His universal categorical by the opponent and using the accepted rule (R).
sontcnces may be int,orprct,ed a s the following rules for one-place Let me call rnles for predicators which are accepted for pragma-
predicators p and q : tical reasons, namely, in order to improve the use of the predicators
affirmative: z e p => x e q involved, materially analytic rules. If a sentence can be defended
negative: zc p => r c' q . in virtue of its logical form, perhaps with the help of definitions
and with the help of a t least one materially analytic rule, this
These rules prescribe the transition from an elementary sentence sentence will be called, a "materially analytic truth". Whether or
of the form x c p to the sentence x c q or z e' q, as the case may be. not a sentence is materially-analytically true always depend8 on
The question is: why accept such rules? the rules accepted. The acceptance of the rules depends on our
I n order t o answer this question, let me begin with an example practice of using the predicators. We may say that the accept.ance
of my own terminology. When introdncing elementary sentences, of the rules depends on "experience", but then "experienre" is
I used two kinds of words, proper names and predicators. These something different from our knowledge of empirically true sen-
terms are determined exemplarily. Thus we must imagine that tences. We do not accept materially analytic rules because we
up t o now the nse of these terms has been determined merely by experience the truth or falsity of elementary sentences.
giving some examples and counterexamples. This may lead to
I should like to illustrate this with the famous case of colour-words.
. difficulties, e. g . someone may assert that the word "sun" is a Whereas the bull becomes nervous by the colour red, the philoso-
proper name a n d also a predicator. A similar difficulty normally
pher since antiquity already becomes nervous by the word 'colour'.
arises with the word "God". Now I hope that along with me you
Let us assume that-independently of English usage-we have
accept it a s reasonable to fix the use of our terms in such a way
introduced "red" by examples of fresh blood, and "greenish" by
t h a t a proper name shall never be a predicator. Nothing other than
examples of different leaves. If we have a violet spot S brforc us,
this is proposed by the nogative rule:
why not call i t "a greenish red", i. e. S cred A S c greenish? This
( R ) x e proper name => x E' predicator. would be possible. However, if we have introduced a predicator
"blueish" with suitable examples and counterexamples, i t would be
We do not have a vocabulary and a set of definitions such that better to speak of "blueish red" a n d to accept a rule which makes
t h e universal sentence: No proper name is a predicator (I take "red" and "greenish" contraries,
this a s a n ordinary version of
z E red => x E' greenish
Ar . x c proper name -t x E' predicator.)
One could oppose accepting this rule by pointing to an orange
is logically implied. This universal sentence is not a formally ana- spot, but, once more, it is possible to propose first the introduction
lytical truth. However, after having accepted the proposed rule. of a new predicator "yellowish", and then to insist on the negative
i t is easy t o defend this sentence: rule.
54 Nwmatice Logic and Ethics 5 . Non-empirical Truth., in the Material 8cieneea 55

As an admirer of G O E T K E Farbenlehre
~ 1 should like to follow his
proposals, i. e. t o work with two basic colours and with red as the
top-colour arranged in a triangle:

red
Let 81 be vert~calto 11, t z and let s? br vrrt~(.aIto t l . What then
about s z , t z . I n Eucldean geometry we have, of course, the theo-
red-yellow rem :

blue (-) green ( f ) yellow But how do we justify this theorem-how do we justify teaching
this to our children all over the world?
Within the pure colours blue is the darkest colour, yellow the The empiricist says: "because in all instances of this general
lightest, and red is "the colour of the colonrs". Natural languages sentence, if the premisses were true, the conclusion has been found
cannot decide anything about such questions. The physicist with to be a true elementary sentence." Well, all red spots have been
his electrodynamic waves may be of some help, but he does not found t o be not-green; all mammals have been found to have
know anything about the "sinnlich-sittliche Wirkung", the sen- lungs. I n the lat,ter case, no one seriol~nlyproposrs a linquislie rnlr
nuous-rnoral effccts of the coloum.
The different linguistic approaches t o colonrs aro of no philoso- x E rnammals => x E having lungs
phical interest in themselves. They are merely a convenient examp-
le of.those linguistic conventions which I have called "materially We have nothing more than an empirically true general sentence.
analytic rules". For o m philosophical purposes i t is sufficient to In the case of the colours i t is a t least arguable to accept certain
know t h a t sometimes it may be reasonable t o propose and to linguistic rules for colour words, so that, by this acceptance the
accept such rules. For my own terminology I have already made use general rules become materially analytic troths. But the rase of
of this posibility more than once. geometry is different.
I n our terminology we now have logical truths, analytic truths Nor is i t of any help t o say that our theorem is logically implied
s. s. a n d formally synthetic truths. They all are non-empirical by axioms a s stated by EUCLIDor HILBERT.We should then have
troths. I n t h e following I wish t o show that geometry provides us t o ask about the truth of the axioms. It is only accidentnl that
with a n example of a priori truths of still another kind. I shall EUCLID did not choose our theorem as a n axiom.
propose for t h e m the term "materially synthetic", though the term How, then, can our theorem be defended, if i t is neither empiri-
does not matter a t the moment. cally true, nor formally true, nor materially-analytically true?
"Geometry" is the traditional word for a theory which originated Perhaps, and modern physicists seem to say so, it is not true a t all.
with THALES. EUCLIDwrote its first classical text-book. However I n order to come t o a reasonable opinion about this question, I
there is a long prehist,ory of geometry, because there is a prescien- should like to begin with PLATO.He made i t clear that the points,
tific language for our working with rigid bodies. This pregeometry lines and planes of geometry are something different from the
is the basis of geometry for us. We begin with the exemplary corners, edges and sides of a body, of a real body, a s we say, in
determination of a pregeometrical vocabulary. Let us take body, order t o distinguish i t from ideal bodies. Nevertheless, t,he difficulty
side, edge, corner -and, e. g., with the help of a brick, the predi- remains of justifying this Platonic talk of ideal bodies. I n my
cators: plane, straight, incident and vertical. Let us consider, in answer I follow the line of KANTIAN philosophy, especially the
<' pragmatic idealism" of Huoo DINOLER,a German philosopher
this pregeon~etricalway of speaking, a plane side with stra<gU
edges who lived from 1881-1954. "Idealism" in this sense starts with the
54 Nwmatice Logic and Ethics 5 . Non-empirical Truth., in the Material 8cieneea 55

As an admirer of G O E T K E Farbenlehre
~ 1 should like to follow his
proposals, i. e. t o work with two basic colours and with red as the
top-colour arranged in a triangle:

red
Let 81 be vert~calto 11, t z and let s? br vrrt~(.aIto t l . What then
about s z , t z . I n Eucldean geometry we have, of course, the theo-
red-yellow rem :

blue (-) green ( f ) yellow But how do we justify this theorem-how do we justify teaching
this to our children all over the world?
Within the pure colours blue is the darkest colour, yellow the The empiricist says: "because in all instances of this general
lightest, and red is "the colour of the colonrs". Natural languages sentence, if the premisses were true, the conclusion has been found
cannot decide anything about such questions. The physicist with to be a true elementary sentence." Well, all red spots have been
his electrodynamic waves may be of some help, but he does not found t o be not-green; all mammals have been found to have
know anything about the "sinnlich-sittliche Wirkung", the sen- lungs. I n the lat,ter case, no one seriol~nlyproposrs a linquislie rnlr
nuous-rnoral effccts of the coloum.
The different linguistic approaches t o colonrs aro of no philoso- x E rnammals => x E having lungs
phical interest in themselves. They are merely a convenient examp-
le of.those linguistic conventions which I have called "materially We have nothing more than an empirically true general sentence.
analytic rules". For o m philosophical purposes i t is sufficient to In the case of the colours i t is a t least arguable to accept certain
know t h a t sometimes it may be reasonable t o propose and to linguistic rules for colour words, so that, by this acceptance the
accept such rules. For my own terminology I have already made use general rules become materially analytic troths. But the rase of
of this posibility more than once. geometry is different.
I n our terminology we now have logical truths, analytic truths Nor is i t of any help t o say that our theorem is logically implied
s. s. a n d formally synthetic truths. They all are non-empirical by axioms a s stated by EUCLIDor HILBERT.We should then have
troths. I n t h e following I wish t o show that geometry provides us t o ask about the truth of the axioms. It is only accidentnl that
with a n example of a priori truths of still another kind. I shall EUCLID did not choose our theorem as a n axiom.
propose for t h e m the term "materially synthetic", though the term How, then, can our theorem be defended, if i t is neither empiri-
does not matter a t the moment. cally true, nor formally true, nor materially-analytically true?
"Geometry" is the traditional word for a theory which originated Perhaps, and modern physicists seem to say so, it is not true a t all.
with THALES. EUCLIDwrote its first classical text-book. However I n order to come t o a reasonable opinion about this question, I
there is a long prehist,ory of geometry, because there is a prescien- should like to begin with PLATO.He made i t clear that the points,
tific language for our working with rigid bodies. This pregeometry lines and planes of geometry are something different from the
is the basis of geometry for us. We begin with the exemplary corners, edges and sides of a body, of a real body, a s we say, in
determination of a pregeometrical vocabulary. Let us take body, order t o distinguish i t from ideal bodies. Nevertheless, t,he difficulty
side, edge, corner -and, e. g., with the help of a brick, the predi- remains of justifying this Platonic talk of ideal bodies. I n my
cators: plane, straight, incident and vertical. Let us consider, in answer I follow the line of KANTIAN philosophy, especially the
<' pragmatic idealism" of Huoo DINOLER,a German philosopher
this pregeon~etricalway of speaking, a plane side with stra<gU
edges who lived from 1881-1954. "Idealism" in this sense starts with the
 56 Normati~ieLogic and Ethics 5. Non-empirical Truth in the Material Sciences 57

simple truth t h a t a brick is not a natural object but an artefact. Here, me use the definition:
We have to make its side plane and we have to make it vertical. If
we start inquiring how plane sides are made, we can easily go on
from there to the a r t of telescope-making. Spherical lenses are Spheres merely have to frilfill the first principle. Lens grinders
produced by grinding two blocks of glans against one another. If try to fulfill the ideal norm of outer homogeneity by grinding three
you have sufficient patience, you can do i t yourself without any sides mutually against each other in order to realize planes.
machinery. Yet you will not get an ideal sphere, merely a reali- What now about joining the lens grinders? No one has to join.
zation of a n ideal sphere. How can we understand these queer But then he should not join the geometcrs either. 'This is a t least
phrases of the lens grinders? Are they bad metaphysicians? Obvi- my proposal for the term "geometry": it is the theory of ideal
ously they are not, though normally -with the exception of SPINOZA points and planes as they are determined by ideal norms.
-lens grinders are not good nretaphysicians either. They are not Cutting two planes we get straight lines k, 1, . . . .
speaking merely about what they have made; they are also speak- Then we have to determine verticality by an ideal norm. We
ing about what they want to make. They have a norm which they write S t 2 for 1 being vertical to the plane S . EIJCLID "defines"
try to fulfill. This norm in an ideal norm, which moans that they the right angle as being equal to its adjacent angle. We can gene-
will never fulfill the norm. So they seem to be very unreasonable. ralize this idea and get the following principle of homogenotiy:
This is just the question. Let me explain their ideal norm so
t h a t you may decide for yourself. The grinding-process shall make
t h e sides of the lenses "homogeneous". This means that all corners I do not want to call these principles "definitions"; they are ideal
-in geometrical language, we say "points" -shall be undistin- norms which we have to accept in order to establish interpersonal
guishable. The homogeneity, i. e. the undistinguishability of all cooperation, e. g. in architecture, fortification, and generally
points, means according to the usage of LEIBNIZ,that a11 sentence- speaking in the BioS noiqrixd<. Of cokse, these principles of
forms which are true for one point of the sphere are also true for homogeneity may at first appear arbitrary, but a closer examina-
every other point. tion will show you that each proposal of producing an inhomoge-
Let us first agree on our geometrical language. We have-at neity would be arbitrary. The principles of homogeneity are simply
least -two types of variables: P I , P z . . . as point,-variables and the only possible way to avoid arbitrariness.
.
8 1 , Sz, . . as side-variables. As prime-sentence forms we have This last statement may be misleading. By the phrase "the
I' ( S (P incident with S) and Sl l S z (81 vertical to S z ) All only possible way" I do not wish to refer vaguely to a system of
other sentences are logically compounded, with quantification true sentences, such that all norms which are different from the
allowed for both types of variables. principles of homogeneity are logically inconsistent with these
Now in this language the ideal norm for spheres can be formula- sentences. The case is rather that the principles of homogeneity
t e d as the following principle of inner homogeneity: are one proposal which makes geometry possible; I simply do not
know of any other proposal. And I am fairly sure that you do not
know of any either. Thus I take the risk of beginning geometry
Here A (S, P) is a sentence form which has no other free variables with homogeneity because otherwise I should not know how to
than S and P , b u t it may have other quantified variables of both begin a t all.
types. Taking the principles of homogeneity as axiom-schemata of a
I call this geometrical axiom-schema the principle of inner formal theory, we may now look for implied theorems. There is no
homogeneity, since for plane instead of spherical sides we have a reason to restrict "implications" to pure logic; we may also freely
second norm, the principle of outer homogeneity: use arithmetic. However, we have first to show holv numbers enter
the scene. This is done by defining "measurement", beginning with
the length of segments (of straight lines), Length is defined as 8
 56 Normati~ieLogic and Ethics 5. Non-empirical Truth in the Material Sciences 57

simple truth t h a t a brick is not a natural object but an artefact. Here, me use the definition:
We have to make its side plane and we have to make it vertical. If
we start inquiring how plane sides are made, we can easily go on
from there to the a r t of telescope-making. Spherical lenses are Spheres merely have to frilfill the first principle. Lens grinders
produced by grinding two blocks of glans against one another. If try to fulfill the ideal norm of outer homogeneity by grinding three
you have sufficient patience, you can do i t yourself without any sides mutually against each other in order to realize planes.
machinery. Yet you will not get an ideal sphere, merely a reali- What now about joining the lens grinders? No one has to join.
zation of a n ideal sphere. How can we understand these queer But then he should not join the geometcrs either. 'This is a t least
phrases of the lens grinders? Are they bad metaphysicians? Obvi- my proposal for the term "geometry": it is the theory of ideal
ously they are not, though normally -with the exception of SPINOZA points and planes as they are determined by ideal norms.
-lens grinders are not good nretaphysicians either. They are not Cutting two planes we get straight lines k, 1, . . . .
speaking merely about what they have made; they are also speak- Then we have to determine verticality by an ideal norm. We
ing about what they want to make. They have a norm which they write S t 2 for 1 being vertical to the plane S . EIJCLID "defines"
try to fulfill. This norm in an ideal norm, which moans that they the right angle as being equal to its adjacent angle. We can gene-
will never fulfill the norm. So they seem to be very unreasonable. ralize this idea and get the following principle of homogenotiy:
This is just the question. Let me explain their ideal norm so
t h a t you may decide for yourself. The grinding-process shall make
t h e sides of the lenses "homogeneous". This means that all corners I do not want to call these principles "definitions"; they are ideal
-in geometrical language, we say "points" -shall be undistin- norms which we have to accept in order to establish interpersonal
guishable. The homogeneity, i. e. the undistinguishability of all cooperation, e. g. in architecture, fortification, and generally
points, means according to the usage of LEIBNIZ,that a11 sentence- speaking in the BioS noiqrixd<. Of cokse, these principles of
forms which are true for one point of the sphere are also true for homogeneity may at first appear arbitrary, but a closer examina-
every other point. tion will show you that each proposal of producing an inhomoge-
Let us first agree on our geometrical language. We have-at neity would be arbitrary. The principles of homogeneity are simply
least -two types of variables: P I , P z . . . as point,-variables and the only possible way to avoid arbitrariness.
.
8 1 , Sz, . . as side-variables. As prime-sentence forms we have This last statement may be misleading. By the phrase "the
I' ( S (P incident with S) and Sl l S z (81 vertical to S z ) All only possible way" I do not wish to refer vaguely to a system of
other sentences are logically compounded, with quantification true sentences, such that all norms which are different from the
allowed for both types of variables. principles of homogeneity are logically inconsistent with these
Now in this language the ideal norm for spheres can be formula- sentences. The case is rather that the principles of homogeneity
t e d as the following principle of inner homogeneity: are one proposal which makes geometry possible; I simply do not
know of any other proposal. And I am fairly sure that you do not
know of any either. Thus I take the risk of beginning geometry
Here A (S, P) is a sentence form which has no other free variables with homogeneity because otherwise I should not know how to
than S and P , b u t it may have other quantified variables of both begin a t all.
types. Taking the principles of homogeneity as axiom-schemata of a
I call this geometrical axiom-schema the principle of inner formal theory, we may now look for implied theorems. There is no
homogeneity, since for plane instead of spherical sides we have a reason to restrict "implications" to pure logic; we may also freely
second norm, the principle of outer homogeneity: use arithmetic. However, we have first to show holv numbers enter
the scene. This is done by defining "measurement", beginning with
the length of segments (of straight lines), Length is defined as 8
 58 Normative Logic and Ethies 5 . Non-empirical T r u t h in the Material Sciemtces 59

I real number which represents a class of congruent segments.

Therefore we have first to define congruence and to show that
All the discussion about "non-euclidean" geometries does not
I apply here, because they have "congruence" as a primitive.
congruence is an cquivalence relation, such that the abstract After geometry I should like to sketch very briefly a program for
objects represented by congruent segments correspond uniquely chronomet~yas an extension of the program for geometry. Proced-
to real numbers. That is the program of Descartes' analytical ing from geometry to chronometry, we have to go from lens-grin-
geometry. The problem is that for EUCLID(and for HILBERT) ders t o the watch-maken. In order to get a criterion for an ideal
"
congruence" is taken as a primitive term. I n our geometrical clock, we have to establish the idea of uniform motion. A uniform
language "congruence" has to be defined with the help of verticality. motion has to be indistinguishable in all its points. The specific
We may first of all define "parallel", either as EUCLID did, difficulty of chronometry is that we can compare a t first only
simultaneous motions. It is easy to define geometrically a ratio
of velocities of two moving bodies for simultaneous positions. We
or more simply as call two motions "repeatable relative to each other", if they have
for all repetitions the same ratio in every position. The ideal norm
for uniform motion, then requires that the ratio of velocities of
repeatable motions is constant and always the same in~lep~mdently
Then we can "compare" two given segments a, b by the following of a temporal displacement (cf. P. Janich: Die Protophysik der
construction of auxiliary segments 81. . . ., 8 7 Zeit, to be published by Bibl. Inst. Mamheim, 1969). In ot,her
A. 8. B. words, the "similarity" of two motions, defined by a constant ratio
of velocities, has to be preserved in every repetition and in every
displacement of the beginning or ending of the motions. Each of
two motions fulfilling these conditions is, therefore, indistinguish-
able in all parts, the criterion of distinction being the ratio of
D velocity a t a point relative to a repeatable motion. Such a motion
These segments are in connnon geometrical notation determined is uniform - and on the basis of this notion "time" becomes nrea-
as follows : surable.
81 = A ~ B I , A z I s z ) ) S ~ I83IIa,
,B~ The next step after chronometry is to make materia measurable.
C = s 2 U 83, B2 1 8 4 1 8 3 9 C 1 8 5 1 b, Following rather closely the historical origin of the notion of the
D = 8 4 n s ~S~. = BZC , s7 = BI D inert maas, I would propose to begin with the collision of bodies.
I n the case of inelastic collisions we can measure the ratio of the
I f and only if sg 1 8 7 , the segments a, b are defined as congruent. velocities wl, vz of the colliding bodies. (The vdocitics have to he
The required properties of congruence so delined have to be measured relative to the bodies after the collision.)
proved from t h e principles of homogeneity. But obviously, this We have to determine what an ideal (undisturbed) collision
is impossible; we must a t least add existence-axioms, e. g. that shall be, but, as we want to measure the ratio of the velocities
for each triplet of points there exists a plane incident with the and as we cannot measure the velocities themselves without intro-
point. I n EUCLID such an axiom is called an a i q p a , in contrast ducing arbitrary units there is once more only one proposal which
t o a xoiwj &vvoia. The principles of homogeneity are hinted a t in is reasonable, i. e. not arbitrary, that is, when the collision is repeat-
t h e deiapoi. ed, the ratio shall always be the same.
IJnSort~mnLolytho appnm-h which I have indicated has not been The rest is mathematics. We may define the ratio of "masscs"
published in dotail. It is a program. But for onr philosophical by
task of understanding how geometry is possible the principles of - VZ.
3L-
homogeneity already give, i t seems to me, a sufficient answer. mz v1
 58 Normative Logic and Ethies 5 . Non-empirical T r u t h in the Material Sciemtces 59

I real number which represents a class of congruent segments.

Therefore we have first to define congruence and to show that
All the discussion about "non-euclidean" geometries does not
I apply here, because they have "congruence" as a primitive.
congruence is an cquivalence relation, such that the abstract After geometry I should like to sketch very briefly a program for
objects represented by congruent segments correspond uniquely chronomet~yas an extension of the program for geometry. Proced-
to real numbers. That is the program of Descartes' analytical ing from geometry to chronometry, we have to go from lens-grin-
geometry. The problem is that for EUCLID(and for HILBERT) ders t o the watch-maken. In order to get a criterion for an ideal
"
congruence" is taken as a primitive term. I n our geometrical clock, we have to establish the idea of uniform motion. A uniform
language "congruence" has to be defined with the help of verticality. motion has to be indistinguishable in all its points. The specific
We may first of all define "parallel", either as EUCLID did, difficulty of chronometry is that we can compare a t first only
simultaneous motions. It is easy to define geometrically a ratio
of velocities of two moving bodies for simultaneous positions. We
or more simply as call two motions "repeatable relative to each other", if they have
for all repetitions the same ratio in every position. The ideal norm
for uniform motion, then requires that the ratio of velocities of
repeatable motions is constant and always the same in~lep~mdently
Then we can "compare" two given segments a, b by the following of a temporal displacement (cf. P. Janich: Die Protophysik der
construction of auxiliary segments 81. . . ., 8 7 Zeit, to be published by Bibl. Inst. Mamheim, 1969). In ot,her
A. 8. B. words, the "similarity" of two motions, defined by a constant ratio
of velocities, has to be preserved in every repetition and in every
displacement of the beginning or ending of the motions. Each of
two motions fulfilling these conditions is, therefore, indistinguish-
able in all parts, the criterion of distinction being the ratio of
D velocity a t a point relative to a repeatable motion. Such a motion
These segments are in connnon geometrical notation determined is uniform - and on the basis of this notion "time" becomes nrea-
as follows : surable.
81 = A ~ B I , A z I s z ) ) S ~ I83IIa,
,B~ The next step after chronometry is to make materia measurable.
C = s 2 U 83, B2 1 8 4 1 8 3 9 C 1 8 5 1 b, Following rather closely the historical origin of the notion of the
D = 8 4 n s ~S~. = BZC , s7 = BI D inert maas, I would propose to begin with the collision of bodies.
I n the case of inelastic collisions we can measure the ratio of the
I f and only if sg 1 8 7 , the segments a, b are defined as congruent. velocities wl, vz of the colliding bodies. (The vdocitics have to he
The required properties of congruence so delined have to be measured relative to the bodies after the collision.)
proved from t h e principles of homogeneity. But obviously, this We have to determine what an ideal (undisturbed) collision
is impossible; we must a t least add existence-axioms, e. g. that shall be, but, as we want to measure the ratio of the velocities
for each triplet of points there exists a plane incident with the and as we cannot measure the velocities themselves without intro-
point. I n EUCLID such an axiom is called an a i q p a , in contrast ducing arbitrary units there is once more only one proposal which
t o a xoiwj &vvoia. The principles of homogeneity are hinted a t in is reasonable, i. e. not arbitrary, that is, when the collision is repeat-
t h e deiapoi. ed, the ratio shall always be the same.
IJnSort~mnLolytho appnm-h which I have indicated has not been The rest is mathematics. We may define the ratio of "masscs"
published in dotail. It is a program. But for onr philosophical by
task of understanding how geometry is possible the principles of - VZ.
3L-
homogeneity already give, i t seems to me, a sufficient answer. mz v1
 No~rnativeLogic and Ethics

d(rn.9) truths
We may define "impulse" a s m.v, "force" as
dt
, etc. / \
analvtic svnthetic
By analogy t o geometry and chronometry I should like t o call

\/ \/
this part of mechanics, which is concerned with the ideal norms a priori s. s.
for measuring materia, "hylomelry" from the Greek word for
materia. Hylometry states the ideal norms for collisions and
thereby gives t h e first a priori causal relations ("ca~~sallynecessary
truths"). Hylornetry coincides with clafisical mechanics without analytic s. s. synthetic s. s.
gravitation. This part is sometimes called "rational mechanics",
but a s it is usually presented it is difficult to understand the ratio-
nale of this discipline. logical form. anal. mat. anal. form. synth. mat. synth. empirical
Geometry, chronometry and hylometry are a-priori theories
which make empirical measurement of space, time and materia
"possible". They have to be established before physics in the 6. Modal Logic
modern sense of an empirical science, with its hypothetical fields
of forces, can begin. Therefore, I should like to call these three
The final purpose of philosophy, as I understand it, is t o make
disciplines by t h e common name: protophysics. The true sentences
the art of practical reasoning teachable. But no one can know
of protophysics are those sentences which are defendable on the
whether there is such a thing as "the a r t of practical reasoning"
basis of logic, arithmetic and analysis, definitions a n d the ideal
or, if there is, whether it can be made teachable, unless he tries
norms which make measurements possible.
both to practice it and to teach it. As a propaedeutical step toward
These ideal norms are clearly different from the construction finding practical uses of reason, however, it is useful to investigate
rules of arithmetic, nor are they definitions. They are not formal theoretical reason first.
determinations, because we are no longer dealing merely with Since ~ I S T O T L E , the father of logic, there has been a tradition
strings of symbols. We are dealing with. materia, grinding its sides, of using so-called "modclities" in the language of the sciences*.
regulating its movements and producing collisions. We prescribe For example, the worda "necessary" and "possible" are usually
b y norms how the materia shall "behave", if I may use this bio- used in English a s modalities. It is the task of modal logic t o
logical metaphor. In contrast to materially analytic determina- establish a reasonable use for these modalities. In the Language of
tions, where we prescribe rules for our predicators in order t o fit morals, law and politics we have the corresponding modalities
t h e world, we now force the materia t o fit our ideal norms. I n "obligatory" and "permitted".
protophysics our relation t o the world is no longer passive (with I am not going to investigate the actual usages of these English
our activity merely on the linguistic side), we are now actively words. Instead, I will propose a use in such a way that it can be
changing the world. I n order to avoid a new term for expressing claimed t h a t the proposed use reconstructs the intentions of the
this difference, I wish to give the task t o the traditional term authors of texts on modal logic. I do not claim to reconstruct every
"synthetic". B y this terminological convention we get the result such intention, since some traditions are inconsistent with one
t h a t these protophysical true sentences which are dependent on another; in such cases we have t o come t o a reasonable decision
t h e ideal norms are to be called "materially synthetic" truths. about which intention to reconstruct. Thus the teaching of THEO-
Assuming t,hat a t least some of you will look benevolently on PHRASTUS, ARISTOTLE'S successor, on modal logic already was incon-
my attempts t o revise traditional terms, I should like t o offer
Modal logic was Aristotle's favorite child, as we can pee fmm the proportion of hia
finally the following system of all the truths reconstructed in these text devoted to modal loglo In comparison with thst dealing wlth the aasertoric
lectures thus far: syllogistic.
 No~rnativeLogic and Ethics

d(rn.9) truths
We may define "impulse" a s m.v, "force" as
dt
, etc. / \
analvtic svnthetic
By analogy t o geometry and chronometry I should like t o call

\/ \/
this part of mechanics, which is concerned with the ideal norms a priori s. s.
for measuring materia, "hylomelry" from the Greek word for
materia. Hylometry states the ideal norms for collisions and
thereby gives t h e first a priori causal relations ("ca~~sallynecessary
truths"). Hylornetry coincides with clafisical mechanics without analytic s. s. synthetic s. s.
gravitation. This part is sometimes called "rational mechanics",
but a s it is usually presented it is difficult to understand the ratio-
nale of this discipline. logical form. anal. mat. anal. form. synth. mat. synth. empirical
Geometry, chronometry and hylometry are a-priori theories
which make empirical measurement of space, time and materia
"possible". They have to be established before physics in the 6. Modal Logic
modern sense of an empirical science, with its hypothetical fields
of forces, can begin. Therefore, I should like to call these three
The final purpose of philosophy, as I understand it, is t o make
disciplines by t h e common name: protophysics. The true sentences
the art of practical reasoning teachable. But no one can know
of protophysics are those sentences which are defendable on the
whether there is such a thing as "the a r t of practical reasoning"
basis of logic, arithmetic and analysis, definitions a n d the ideal
or, if there is, whether it can be made teachable, unless he tries
norms which make measurements possible.
both to practice it and to teach it. As a propaedeutical step toward
These ideal norms are clearly different from the construction finding practical uses of reason, however, it is useful to investigate
rules of arithmetic, nor are they definitions. They are not formal theoretical reason first.
determinations, because we are no longer dealing merely with Since ~ I S T O T L E , the father of logic, there has been a tradition
strings of symbols. We are dealing with. materia, grinding its sides, of using so-called "modclities" in the language of the sciences*.
regulating its movements and producing collisions. We prescribe For example, the worda "necessary" and "possible" are usually
b y norms how the materia shall "behave", if I may use this bio- used in English a s modalities. It is the task of modal logic t o
logical metaphor. In contrast to materially analytic determina- establish a reasonable use for these modalities. In the Language of
tions, where we prescribe rules for our predicators in order t o fit morals, law and politics we have the corresponding modalities
t h e world, we now force the materia t o fit our ideal norms. I n "obligatory" and "permitted".
protophysics our relation t o the world is no longer passive (with I am not going to investigate the actual usages of these English
our activity merely on the linguistic side), we are now actively words. Instead, I will propose a use in such a way that it can be
changing the world. I n order to avoid a new term for expressing claimed t h a t the proposed use reconstructs the intentions of the
this difference, I wish to give the task t o the traditional term authors of texts on modal logic. I do not claim to reconstruct every
"synthetic". B y this terminological convention we get the result such intention, since some traditions are inconsistent with one
t h a t these protophysical true sentences which are dependent on another; in such cases we have t o come t o a reasonable decision
t h e ideal norms are to be called "materially synthetic" truths. about which intention to reconstruct. Thus the teaching of THEO-
Assuming t,hat a t least some of you will look benevolently on PHRASTUS, ARISTOTLE'S successor, on modal logic already was incon-
my attempts t o revise traditional terms, I should like t o offer
Modal logic was Aristotle's favorite child, as we can pee fmm the proportion of hia
finally the following system of all the truths reconstructed in these text devoted to modal loglo In comparison with thst dealing wlth the aasertoric
lectures thus far: syllogistic.
62 Normative Logic a d Ethics 6. Modal Logic 63

sistent with t h a t of ARISTOTLE. These inconsistent teachings have conclusion. Using the same symbol < as before (but now for modal-
been carried through all our history, a n d even modern commen- logical implication, too) we get
tators (or historians of logic) such as B O C H E ~ ~andK I KNEALE fail
t o come t o a decision on the matter. AA A AB<A(A A B).
However, in our century a new initiat,ive has been taken by LEWIS
The task of modal logic is to determine the class of modal-
with his invention of a system of calculi SI- 8 s . These calculi were
Analytica priors, book 1,
logical implications. Already in ARISTOTLE.
intended t o be formalizations of a n "intuitive" notion of necessity.
chapter 8, we find (if we are looking for it) the following rule:
Starting from the uninterpreted calculi, mathematical logicians
such as BETH and KRIPKEhave managed t o "interpret" these
calculi within a metalanguage, using only classical quantificational
logic. But these "interpretations" are justified only insofar a s The validit,y of this "Rule of ARISTOTLE" is obvious under the pro-
they int.erpret the LEWIS-calculi.We see here the order in which posed interpretation. Moreover, the rule is convertible, because,
formalistic logicians typically proceed: first invent any calculus
(it is sufficient t o refer vaguely to "reason" or "intuition" t o make
the invention acceptable), and then, using all one's ingenuity,
ciall5- for Z'-
if A,Al A '1,Az. . . < ill-B holds for all nytems Z i t holds espe-
A1 A Az A . . . and this gives immediately A2B,
i . e . A l A A Z A . . . <l3.
look for a n interpretation within modern mathematical theories. The theorem mentioned above: AA A AD < A ( A A B ) is, of
However, there is another way of doing modal logic, the way which course, a consequence of ARISTOTLE'S Rule.
is t o be followed here. Besides this rule we tind also in ARISTOTLE the trivial implication
Imagine a situation in which a system 2 of sentences has been AA < A (nccrssity in~pliestruth). This follows from the assumption
generally accepted as true by a certain group of people. I n such a that Z is a basis of true sentences. I n modal logic we have to in-
situation, all sentences logically implied by Z have a special vestigate, which implications hold between arbitrarily compound
importance, namely, they should all be generally accepted in the sentences. Besides the junctors and quantifiers the modality A
group. This importance justifies a special notation, "<", by which may be used. Under the given interpretation A z A + 2 < A the
I will denote logical implication. (This is a two-place relation be- iteration of does not make sense, because A2 A z A would be
tween sentences, not t o be confused with the junctor "+" in con- Z<(Z'<A).
ditionals.) The symbol A2 (read a s "necessary relative to Z") is Only as a formal game, the iteration of A may be admitted and
introduced b y t h e following definition: then e. g . the arbitrary rule, t o treat AAA as equivalent with AA
will rasily owur to evcry mathematically trained mind. Also the
false analogy with Z -t (Z' + A ) (which is equivalent to Z -+ A )
may sugyrst to "postulate" the equivalence of Ad with A .
Here "A" is a variable for modal-free sentences. Instead of playfully suggesting such postulates I'll restrict the
By itself, this definition would not justify our setting u p an use of A ~ I itsI nou-iterated w e : only if A does not already occur in
extra theory, called "modal logic". But there are theomms about a forrnda A, AA shall be a formula.
statements of t h e form AzA which are true independently of the But the logical particles may be applied without restriction.
particular Z. Because of this indepedence I will write A 86 EII We have then to investigate how a compound sentence can be
abbreviation for A=. Thus the following theorem is obviously true defended as a thesis, if some hypotheses are given. The admission
(as a theorem about logical implication): of A leads to two new situations:
1. I n the course of the dialogue the opponent may have put a
formula AA. If we would not havo the restriction that Z has
As this conditional is true for all Z, II'I1 call it modal-logically t o be tmatcd as unknown, the proponent would surely ask,
true-and I'll say, that the premissos modal-logically imply the relative t o which Z the "necessity" is assnrted. But in modal
62 Normative Logic a d Ethics 6. Modal Logic 63

sistent with t h a t of ARISTOTLE. These inconsistent teachings have conclusion. Using the same symbol < as before (but now for modal-
been carried through all our history, a n d even modern commen- logical implication, too) we get
tators (or historians of logic) such as B O C H E ~ ~andK I KNEALE fail
t o come t o a decision on the matter. AA A AB<A(A A B).
However, in our century a new initiat,ive has been taken by LEWIS
The task of modal logic is to determine the class of modal-
with his invention of a system of calculi SI- 8 s . These calculi were
Analytica priors, book 1,
logical implications. Already in ARISTOTLE.
intended t o be formalizations of a n "intuitive" notion of necessity.
chapter 8, we find (if we are looking for it) the following rule:
Starting from the uninterpreted calculi, mathematical logicians
such as BETH and KRIPKEhave managed t o "interpret" these
calculi within a metalanguage, using only classical quantificational
logic. But these "interpretations" are justified only insofar a s The validit,y of this "Rule of ARISTOTLE" is obvious under the pro-
they int.erpret the LEWIS-calculi.We see here the order in which posed interpretation. Moreover, the rule is convertible, because,
formalistic logicians typically proceed: first invent any calculus
(it is sufficient t o refer vaguely to "reason" or "intuition" t o make
the invention acceptable), and then, using all one's ingenuity,
ciall5- for Z'-
if A,Al A '1,Az. . . < ill-B holds for all nytems Z i t holds espe-
A1 A Az A . . . and this gives immediately A2B,
i . e . A l A A Z A . . . <l3.
look for a n interpretation within modern mathematical theories. The theorem mentioned above: AA A AD < A ( A A B ) is, of
However, there is another way of doing modal logic, the way which course, a consequence of ARISTOTLE'S Rule.
is t o be followed here. Besides this rule we tind also in ARISTOTLE the trivial implication
Imagine a situation in which a system 2 of sentences has been AA < A (nccrssity in~pliestruth). This follows from the assumption
generally accepted as true by a certain group of people. I n such a that Z is a basis of true sentences. I n modal logic we have to in-
situation, all sentences logically implied by Z have a special vestigate, which implications hold between arbitrarily compound
importance, namely, they should all be generally accepted in the sentences. Besides the junctors and quantifiers the modality A
group. This importance justifies a special notation, "<", by which may be used. Under the given interpretation A z A + 2 < A the
I will denote logical implication. (This is a two-place relation be- iteration of does not make sense, because A2 A z A would be
tween sentences, not t o be confused with the junctor "+" in con- Z<(Z'<A).
ditionals.) The symbol A2 (read a s "necessary relative to Z") is Only as a formal game, the iteration of A may be admitted and
introduced b y t h e following definition: then e. g . the arbitrary rule, t o treat AAA as equivalent with AA
will rasily owur to evcry mathematically trained mind. Also the
false analogy with Z -t (Z' + A ) (which is equivalent to Z -+ A )
may sugyrst to "postulate" the equivalence of Ad with A .
Here "A" is a variable for modal-free sentences. Instead of playfully suggesting such postulates I'll restrict the
By itself, this definition would not justify our setting u p an use of A ~ I itsI nou-iterated w e : only if A does not already occur in
extra theory, called "modal logic". But there are theomms about a forrnda A, AA shall be a formula.
statements of t h e form AzA which are true independently of the But the logical particles may be applied without restriction.
particular Z. Because of this indepedence I will write A 86 EII We have then to investigate how a compound sentence can be
abbreviation for A=. Thus the following theorem is obviously true defended as a thesis, if some hypotheses are given. The admission
(as a theorem about logical implication): of A leads to two new situations:
1. I n the course of the dialogue the opponent may have put a
formula AA. If we would not havo the restriction that Z has
As this conditional is true for all Z, II'I1 call it modal-logically t o be tmatcd as unknown, the proponent would surely ask,
true-and I'll say, that the premissos modal-logically imply the relative t o which Z the "necessity" is assnrted. But in modal
 i
Normative Logic and Ethics I 6. M c d d Logic

logic, this is not allowed. So the proponent may only force the jugded by its own reasonableness. In order to improve this rea-
opponent to admit A, if he has admitted AA. sonableness, I would like to sketch, how the interpretation can be
This gives us the following attack-defense-rule for A-formulae applied to the traditional problems of modal syllogistics.
(i. e. formulae beginning with A): First of all, we have to decide how we are to interpret necessary
universal affirmative propositions. In ordinary English we can
AA I ? I A
say "All 8 are necessarily P" (e. g. "All men are necessarily
2. I f the proponent has put a A-formula d B and if the opponent mortal").
attacks this (by"?"), the proponent has to defend AB as logically What is the form of such sentences in modal logic? Superficially
implied by all formulae put by the opponent beforehand. Only there seem to be two different forms appropriate:
the "Rule of ARISTOTLE" is available for defending dB.
This gives the following additional
A-defense-rule: If the proponent defends a A-formula he may
attack only the A-formulae (the beginning A ARISTOTLE
Uses in his modal syllogistics very oft,en the first form,
deleted) put by the opponent heforehand. but sometimes (e. g. Anal. I, 1, ch. 9) he uses the second. We know
It is easy to extend the logical GENTZEN-calculus established in that this was the main objection of THEOPHRAST: he insisted on
using only the first form.
the 3. lecture to yield exactly all the winning-positions of this
dialogical game with A. With the help of our interpretation, it is easy to see that only
In order to defend a thesis B, if a system S(AA) of hypotheses THEOPERAST'S point of view leads to an adequate solution of the
is given which contains a A-formula AA, the proponent may problems of modal syllogistics. Already the necessary universal
negative proposition is convertible only in the fint form (cf. Anal.
attack AA. The opponent has to defend himse:f by putting A.
This gives the new position S(AA), A I/ B. Therefore, the follow- I, 1, ch. 3)
ing rule (corresponding to AA < A) leads from winning-positions
to winning-positions:
S(AA),A 11 B +S(AA) 11 B . The second form, whether read as A, . x E S -t A x E'P. or as
(0, .
A x E S + -I A x E P. is not convertible.
I n order to defend a thesis dB, if a system S(AA1,. . . , AA,) ARLRISTOTLE'Sreasoning with the second form is mostly sound, e.g.
of hypotheses is given, the proponent may -if AB is attacked-try if he infers S a A P from S a M and M a A P, but on the inter-
to defend B with the system A1, . . . , A, as hypotheses. pretation here proposed, the second form is obviously not intended
This gives a new version of the Rule of ARISTOTLE: by ARISTOTLE: If ARISTOTLE asserts that a11 men are necessarily
living beings (Anal. I , 1, ch. 9) he does not refer to a system Z
( P ~ ) A1,. . . ,A, /I B +-S ( d A 1 , . . . , AA,) (1 d B . such that for all proper names "z" of men: Z implies x E living.
This G E N T Z E N - C for
~ ~ modal
~ U ~ Ulogic
~ is consistent and complete But he refers to a system Z such that Z implies "all men are
with respect to the given interpretation of A. I t is known from the living beings".
literature (e. g. K. SCHUTTE, Vollstiindige Systerne modaler und I n arithmetic we may have systems Z (e. g. the system of
intuitionistisoher Logik, 1967) that this G~NTze~-ca~culus is equi- Peano-Axioms) such that 2 implies the formulae A(n) for all
valent to VON WRI~HT'S calculns M', if the restriction not to iterate numerala n without implying the formula A, A (2) -that is the
A is omitted. famous case of wincompleteness proved by GODEL-but
That our interpretation leads to a calculus already well-known in ARISTOTLEnever deala in his syllogistics with systems in which dl
the modern formalistic literature on modal logic will-I hope- proper names of his objects occur. He bears in mind systems of
give no additional authority to the interpretation. It has to be predicator-rules given as "necessary" (in my terminology they
 i
Normative Logic and Ethics I 6. M c d d Logic

logic, this is not allowed. So the proponent may only force the jugded by its own reasonableness. In order to improve this rea-
opponent to admit A, if he has admitted AA. sonableness, I would like to sketch, how the interpretation can be
This gives us the following attack-defense-rule for A-formulae applied to the traditional problems of modal syllogistics.
(i. e. formulae beginning with A): First of all, we have to decide how we are to interpret necessary
universal affirmative propositions. In ordinary English we can
AA I ? I A
say "All 8 are necessarily P" (e. g. "All men are necessarily
2. I f the proponent has put a A-formula d B and if the opponent mortal").
attacks this (by"?"), the proponent has to defend AB as logically What is the form of such sentences in modal logic? Superficially
implied by all formulae put by the opponent beforehand. Only there seem to be two different forms appropriate:
the "Rule of ARISTOTLE" is available for defending dB.
This gives the following additional
A-defense-rule: If the proponent defends a A-formula he may
attack only the A-formulae (the beginning A ARISTOTLE
Uses in his modal syllogistics very oft,en the first form,
deleted) put by the opponent heforehand. but sometimes (e. g. Anal. I, 1, ch. 9) he uses the second. We know
It is easy to extend the logical GENTZEN-calculus established in that this was the main objection of THEOPHRAST: he insisted on
using only the first form.
the 3. lecture to yield exactly all the winning-positions of this
dialogical game with A. With the help of our interpretation, it is easy to see that only
In order to defend a thesis B, if a system S(AA) of hypotheses THEOPERAST'S point of view leads to an adequate solution of the
is given which contains a A-formula AA, the proponent may problems of modal syllogistics. Already the necessary universal
negative proposition is convertible only in the fint form (cf. Anal.
attack AA. The opponent has to defend himse:f by putting A.
This gives the new position S(AA), A I/ B. Therefore, the follow- I, 1, ch. 3)
ing rule (corresponding to AA < A) leads from winning-positions
to winning-positions:
S(AA),A 11 B +S(AA) 11 B . The second form, whether read as A, . x E S -t A x E'P. or as
(0, .
A x E S + -I A x E P. is not convertible.
I n order to defend a thesis dB, if a system S(AA1,. . . , AA,) ARLRISTOTLE'Sreasoning with the second form is mostly sound, e.g.
of hypotheses is given, the proponent may -if AB is attacked-try if he infers S a A P from S a M and M a A P, but on the inter-
to defend B with the system A1, . . . , A, as hypotheses. pretation here proposed, the second form is obviously not intended
This gives a new version of the Rule of ARISTOTLE: by ARISTOTLE: If ARISTOTLE asserts that a11 men are necessarily
living beings (Anal. I , 1, ch. 9) he does not refer to a system Z
( P ~ ) A1,. . . ,A, /I B +-S ( d A 1 , . . . , AA,) (1 d B . such that for all proper names "z" of men: Z implies x E living.
This G E N T Z E N - C for
~ ~ modal
~ U ~ Ulogic
~ is consistent and complete But he refers to a system Z such that Z implies "all men are
with respect to the given interpretation of A. I t is known from the living beings".
literature (e. g. K. SCHUTTE, Vollstiindige Systerne modaler und I n arithmetic we may have systems Z (e. g. the system of
intuitionistisoher Logik, 1967) that this G~NTze~-ca~culus is equi- Peano-Axioms) such that 2 implies the formulae A(n) for all
valent to VON WRI~HT'S calculns M', if the restriction not to iterate numerala n without implying the formula A, A (2) -that is the
A is omitted. famous case of wincompleteness proved by GODEL-but
That our interpretation leads to a calculus already well-known in ARISTOTLEnever deala in his syllogistics with systems in which dl
the modern formalistic literature on modal logic will-I hope- proper names of his objects occur. He bears in mind systems of
give no additional authority to the interpretation. It has to be predicator-rules given as "necessary" (in my terminology they
66 Normative Logic and Ethics

would have t o be called "analytic"). With respect to such systems

2 only t h e first form A S a P makes sense.
The second main difference between Aristotelian and Theophra-
stean modal syllogistics is a verbal one. It is concerned with
modalities definable in terms of A. With the negation --. only, we
may define A', V, V' as follows:

We omit the roir~plicationsintrorluce(1 by drrlll~lencgation in non-

classical logic. I n classical logic we have besides the well-known
tertium non datur X A v X' A also the following valid adjunctions:
These four modalities form a modal square quartwn non datur A A v X A v A' A
qiiintrirn non datur A A v X A v Y' A v '1' A .
The task of modal syllogistics is a generalization of the task of
assertoric syllogistics. There we have to investigate the implica-
tions
A and A' a r e contraries, A , V' and A', V are contradictories. SQIV ~ M o p < S r l '
Therefore we have the two implications
with Q , u and r as variahles for the relations n, e, i , o. We add the
A <V (i.e.AA<V A ) converse relations 6 and z, so t h a t we can restrict o~irselvesto
and A ' < V' (i. e. A'A < V' A ) the above standard form (in cont,rast to the traditional 4 figures).
We denote the standard syllogisms for short as
If we construct conjunt.tlons of Lilesc four m ~ u ~ r l ~ t lwo
o s ,get

XA+VA A V'A (e. g. a I a < a for the modus barbara)

I n modal syllogistics we have the standard form
a s a new modality.
ARISTOTLEuses the Greek term 2v6dX~rnlnormally in the sense @ S Q MA Y M o I ' < R S r P
of X, only occasionally in the sense of V. T H E ~ P H R A Srestricts
T
E V ~ L ~ tEo ~t hJ e~ sense of V. I n the following I'll translate with @, Y , R as variables for the affirmative modalities A.9,V, X,X.
We abbreviate the standard syllogisms as
V as possible
X as contingout. @elYa<Rr
(e. g. A a I A a < A a for the modus barbara of the class A A A ).
I f we add (as A R I S T ~ T Land
E THEOPERAST did) the "simple" truth Altogether we have 303= 27000 standard syllogisms. I n spite
X a n d falsity X' t o the modalities we get by conjrinction of this large number it turns out rather easily that there are exactly
507 valid standard syllogisms.
X A % X A A V' A (contingently true)
On the basis of ARISTOTLE'S Rule each valid assertoric syllogism
X' A + X' A A V A (contingently false).
@lo< r - and there are 21, as well-known, in standard form -
This gives nine modalities with implications a s indicated in the immediately yields the validity of
following figure AQ~Ao<AT.
66 Normative Logic and Ethics

would have t o be called "analytic"). With respect to such systems

2 only t h e first form A S a P makes sense.
The second main difference between Aristotelian and Theophra-
stean modal syllogistics is a verbal one. It is concerned with
modalities definable in terms of A. With the negation --. only, we
may define A', V, V' as follows:

We omit the roir~plicationsintrorluce(1 by drrlll~lencgation in non-

classical logic. I n classical logic we have besides the well-known
tertium non datur X A v X' A also the following valid adjunctions:
These four modalities form a modal square quartwn non datur A A v X A v A' A
qiiintrirn non datur A A v X A v Y' A v '1' A .
The task of modal syllogistics is a generalization of the task of
assertoric syllogistics. There we have to investigate the implica-
tions
A and A' a r e contraries, A , V' and A', V are contradictories. SQIV ~ M o p < S r l '
Therefore we have the two implications
with Q , u and r as variahles for the relations n, e, i , o. We add the
A <V (i.e.AA<V A ) converse relations 6 and z, so t h a t we can restrict o~irselvesto
and A ' < V' (i. e. A'A < V' A ) the above standard form (in cont,rast to the traditional 4 figures).
We denote the standard syllogisms for short as
If we construct conjunt.tlons of Lilesc four m ~ u ~ r l ~ t lwo
o s ,get

XA+VA A V'A (e. g. a I a < a for the modus barbara)

I n modal syllogistics we have the standard form
a s a new modality.
ARISTOTLEuses the Greek term 2v6dX~rnlnormally in the sense @ S Q MA Y M o I ' < R S r P
of X, only occasionally in the sense of V. T H E ~ P H R A Srestricts
T
E V ~ L ~ tEo ~t hJ e~ sense of V. I n the following I'll translate with @, Y , R as variables for the affirmative modalities A.9,V, X,X.
We abbreviate the standard syllogisms as
V as possible
X as contingout. @elYa<Rr
(e. g. A a I A a < A a for the modus barbara of the class A A A ).
I f we add (as A R I S T ~ T Land
E THEOPERAST did) the "simple" truth Altogether we have 303= 27000 standard syllogisms. I n spite
X a n d falsity X' t o the modalities we get by conjrinction of this large number it turns out rather easily that there are exactly
507 valid standard syllogisms.
X A % X A A V' A (contingently true)
On the basis of ARISTOTLE'S Rule each valid assertoric syllogism
X' A + X' A A V A (contingently false).
@lo< r - and there are 21, as well-known, in standard form -
This gives nine modalities with implications a s indicated in the immediately yields the validity of
following figure AQ~Ao<AT.
68 Normalitre Logic and Ethics 6 . Modal Logic 69

Modal syllogisms of this form will be called "of the class A A d " . "ontological" modalities) which occur in the context of descriptive
By contraposition one gets 21 valid syllogisms for each of the (indicative) sentences.
classes A V V and V A V . The situation changes, however, when we consider "ethical"
Addtng X we have the assertoric class X X X and we get the modalities, those called "deontic" or "deontological". (For the
following "multiplication table" for modalities: rest of the paper, t,he physical modalities will be denoted with a
subscript "0" e. g. A., X , , V,.) Even the most enlightened society
depends on generally accepted universal norms which prescribe
actions.
I will render unconditional imperatives (such as "Go away!"
and "John, get some water!" in ordinary English) in the standard
This table of 6 classes satisfies the "regula peiorern" which was form of an indicative sentence with! prefixed, e. g. "! John gets
upheld by THE~PHRAST against ARISTOTLE. sonle water." ("!" may be read as "please.") Let Z now be a
In this table we may strengthen the premisses S by X (this system of such imperative sentences! B l , . . . , ! B,. If the system
givcs 5 classes) and the premisses V by X, X' or X' (6 classes) BI , . . . , B , of indicative sentences logically implies a sentence B ,
we may weaken the conclusion A by X or V ( 2 ?lasses) and the 1 propose to say "Ac B", where A now is an ethical modality (in
conclusion X b y V (8 classes). Altogether we get 27 classes, i. e. English, e. g., "obligatory" or "rt.quired"). We especially get
27.21 = 567 valid syllogisms in standard form. .
A= B1, Az B2, . . . A x Bn Modal logic comes in if we ornit Z
To show t h a t 567 is the exact number requires ruling out all in these sentences and rewrite them as A B l , . . . , A B , . The
other claaees. For example the classes V X V and X V V are ruled ahove sentence Az B , restated as A B, now in logically implied (in
out aa they are reducible-by contraposition-to A X A and the sense of modal logic). The system of modal sentences A B1, . . , .
X A A. E. g. n o implication A B , therefore may replace the original imperatives. I propose
to call them "norms".
Legal authority need not issue imperatives, but may instead,
as has been done since the beginning of law, nse modal sentences,
is valid, as S a P is not implied by M o P alone.
There is no philosophical importance attached to the number i. e., norms. This has the advantage that conditional norms such
567. But with the help of a complete account of modal syllogistics as A1 + A B I , Az -t A B z . . . . , etc., may be used instead of
archaic unconditional ones without causing logical difficulties (if
the history of logic (and metaphysics) can be looked a t in a new
way. I n particular, all the traditional talk about "absolute neces- we ignore the difficulties of modern logicians). Modal logic tells
us which further conditional norms are logically implied.
sity" where some modal logic is used (and a t the same time the
ahove definition of relative necessity is rejected) becomes, to say The modalities V and X may be mtrodnced in ethical modal
the least, still more suspect. But, of course, i t is easy to interpret logic as before :
"absolute" necessity as relative to a system Z of a priori truths. A'A%A,A (forbidden)
The inventions of modal calculi in the style of LEWISon the
V'AA=,AA (negatively permitted)
other hand, can no longer be claimed t o be of any help in under- V A+,A'A (permitted)
standing ARISTOTLE(or his followers). Modalities are of no use a t
X A A= V A A V' A (discretionary)
all in mathematics; if they occur in texts of mathematics, they
occur only a s a sloppy use of language. So it seems as though Ordinary nroral talk usually is restricted to A ("required"), A'
there is no place for modalities in a critical society in which no ("forbidden") and X ("discretionary" in the sense of "neither
system of sentences is accepted without question. But this applies forbidden nor required"), because no logical implication holds
only t o "physioal" modalities (the so-called "ontic" or even between them.
68 Normalitre Logic and Ethics 6 . Modal Logic 69

Modal syllogisms of this form will be called "of the class A A d " . "ontological" modalities) which occur in the context of descriptive
By contraposition one gets 21 valid syllogisms for each of the (indicative) sentences.
classes A V V and V A V . The situation changes, however, when we consider "ethical"
Addtng X we have the assertoric class X X X and we get the modalities, those called "deontic" or "deontological". (For the
following "multiplication table" for modalities: rest of the paper, t,he physical modalities will be denoted with a
subscript "0" e. g. A., X , , V,.) Even the most enlightened society
depends on generally accepted universal norms which prescribe
actions.
I will render unconditional imperatives (such as "Go away!"
and "John, get some water!" in ordinary English) in the standard
This table of 6 classes satisfies the "regula peiorern" which was form of an indicative sentence with! prefixed, e. g. "! John gets
upheld by THE~PHRAST against ARISTOTLE. sonle water." ("!" may be read as "please.") Let Z now be a
In this table we may strengthen the premisses S by X (this system of such imperative sentences! B l , . . . , ! B,. If the system
givcs 5 classes) and the premisses V by X, X' or X' (6 classes) BI , . . . , B , of indicative sentences logically implies a sentence B ,
we may weaken the conclusion A by X or V ( 2 ?lasses) and the 1 propose to say "Ac B", where A now is an ethical modality (in
conclusion X b y V (8 classes). Altogether we get 27 classes, i. e. English, e. g., "obligatory" or "rt.quired"). We especially get
27.21 = 567 valid syllogisms in standard form. .
A= B1, Az B2, . . . A x Bn Modal logic comes in if we ornit Z
To show t h a t 567 is the exact number requires ruling out all in these sentences and rewrite them as A B l , . . . , A B , . The
other claaees. For example the classes V X V and X V V are ruled ahove sentence Az B , restated as A B, now in logically implied (in
out aa they are reducible-by contraposition-to A X A and the sense of modal logic). The system of modal sentences A B1, . . , .
X A A. E. g. n o implication A B , therefore may replace the original imperatives. I propose
to call them "norms".
Legal authority need not issue imperatives, but may instead,
as has been done since the beginning of law, nse modal sentences,
is valid, as S a P is not implied by M o P alone.
There is no philosophical importance attached to the number i. e., norms. This has the advantage that conditional norms such
567. But with the help of a complete account of modal syllogistics as A1 + A B I , Az -t A B z . . . . , etc., may be used instead of
archaic unconditional ones without causing logical difficulties (if
the history of logic (and metaphysics) can be looked a t in a new
way. I n particular, all the traditional talk about "absolute neces- we ignore the difficulties of modern logicians). Modal logic tells
us which further conditional norms are logically implied.
sity" where some modal logic is used (and a t the same time the
ahove definition of relative necessity is rejected) becomes, to say The modalities V and X may be mtrodnced in ethical modal
the least, still more suspect. But, of course, i t is easy to interpret logic as before :
"absolute" necessity as relative to a system Z of a priori truths. A'A%A,A (forbidden)
The inventions of modal calculi in the style of LEWISon the
V'AA=,AA (negatively permitted)
other hand, can no longer be claimed t o be of any help in under- V A+,A'A (permitted)
standing ARISTOTLE(or his followers). Modalities are of no use a t
X A A= V A A V' A (discretionary)
all in mathematics; if they occur in texts of mathematics, they
occur only a s a sloppy use of language. So it seems as though Ordinary nroral talk usually is restricted to A ("required"), A'
there is no place for modalities in a critical society in which no ("forbidden") and X ("discretionary" in the sense of "neither
system of sentences is accepted without question. But this applies forbidden nor required"), because no logical implication holds
only t o "physioal" modalities (the so-called "ontic" or even between them.
70 Normatiae Logic and Ethics 6. Malal Logic 71

The distinctive featnres of ethical modal logic, in contrast to I - A h

physicalmodal logic, are the following: 1. there isnoethicalmodality 2 AzA P X
corresponding to the simple truth X,, 2. the fact that basic norms 3 A-Azpx ?
have to be issued by lawgivers makes i t essential, to establish 4 ?
norms for norm-giving, i. e. to introduce iterated modalities. 5 A PYl
1'11 first treat simple truth X, in ethical modal logic. If some G
- A PYZ
action should be done, it does not follow that this action will be 7 ?
done. 8 y3 ?
While in physical modal logic, necessity implies truth: 9 ?
The proponent has to give up-and, obviously, there is no essenti-
ally different strategy available.
we have to omit this implication in ethical modal logic. For de- (In metamathematics this situation is well-known: there are
fending a modal thesis (some modal hypotheses given) we may
still use the attack-defense rule derivable for all x and -
consistent axiom-systems which are o-inconsist,mt, i. e. p x is
A. p x is derivable, too).
Though it is trivial that there is no logical implication from A
(obligatory) to X, (true)-this is nothing else than HUBIE'Sgap
between "ought" and "is" in the other direction-it is more
but, if the opponent has put a A-formula A A, he is not obliged
difficult to argue t2hatthere is no ethical modality X correspondi~~g
to admit A . This means, that the dialogue is restricted by the
to X,.
following additional rule
In the context of physical modalities X, is characterised by the
A-attack-rule: The proponent may not attack any A-formula. following implication
The A-defense-rule will be used for ethical modalities with the -X,A<X,-A
same wording as for physical modalities. The Gentzen-calculus for (we have A, - . A < A , A in this direction only).
winning-positions, therefore, will only contain the "Rule of
ARISTOTLE"( I > ) , but no longer the rule (0.9).
As a formal system ethical modal logic is a proper subsystem of
ethical modality X such that -
The problem, therefore, is to investigate the possibility of an
X A iX --. A . Ordinary language
very often uses the words "good" or "right" such that for all
physical modal logic as long as modalities are not iterated. actions A, if it is not good (or right) to do A , then it is good (or
An example of an implication between modal sentences, valid right) to omit A. Also the phrase "the right thing to do" indicates,
in physical modal logic, but invalid in ethical modal logic is the that for carh action A either to do A or not to do A has tn hc
following : right.
But the proposed interpretation of ethical modalities starting
with a system Z of basic norms, does not yield such an ethical
The proof in physical modal logic is obvious: from tertium non datur (even if classical logic is applied).
If we start with a system Z I , it may follow
n,, A A,, ,A
for a certain class of actions A, but this tertium non datnr will
which I d s with the help of A < V, A to the above theorem. never hold for all A (because ZI has to be finite). We may enrich
I n ethical modal logic - even if we add as a hypothesis that no 21 in order to get a refined system Zz, in which for more A's
contradiction (i. e. n o logically false sentence A) is obligatory-
V /\, p x cannot he defended on the basis of A, A p z: A,, A v A=, - A
70 Normatiae Logic and Ethics 6. Malal Logic 71

The distinctive featnres of ethical modal logic, in contrast to I - A h

physicalmodal logic, are the following: 1. there isnoethicalmodality 2 AzA P X
corresponding to the simple truth X,, 2. the fact that basic norms 3 A-Azpx ?
have to be issued by lawgivers makes i t essential, to establish 4 ?
norms for norm-giving, i. e. to introduce iterated modalities. 5 A PYl
1'11 first treat simple truth X, in ethical modal logic. If some G
- A PYZ
action should be done, it does not follow that this action will be 7 ?
done. 8 y3 ?
While in physical modal logic, necessity implies truth: 9 ?
The proponent has to give up-and, obviously, there is no essenti-
ally different strategy available.
we have to omit this implication in ethical modal logic. For de- (In metamathematics this situation is well-known: there are
fending a modal thesis (some modal hypotheses given) we may
still use the attack-defense rule derivable for all x and -
consistent axiom-systems which are o-inconsist,mt, i. e. p x is
A. p x is derivable, too).
Though it is trivial that there is no logical implication from A
(obligatory) to X, (true)-this is nothing else than HUBIE'Sgap
between "ought" and "is" in the other direction-it is more
but, if the opponent has put a A-formula A A, he is not obliged
difficult to argue t2hatthere is no ethical modality X correspondi~~g
to admit A . This means, that the dialogue is restricted by the
to X,.
following additional rule
In the context of physical modalities X, is characterised by the
A-attack-rule: The proponent may not attack any A-formula. following implication
The A-defense-rule will be used for ethical modalities with the -X,A<X,-A
same wording as for physical modalities. The Gentzen-calculus for (we have A, - . A < A , A in this direction only).
winning-positions, therefore, will only contain the "Rule of
ARISTOTLE"( I > ) , but no longer the rule (0.9).
As a formal system ethical modal logic is a proper subsystem of
ethical modality X such that -
The problem, therefore, is to investigate the possibility of an
X A iX --. A . Ordinary language
very often uses the words "good" or "right" such that for all
physical modal logic as long as modalities are not iterated. actions A, if it is not good (or right) to do A , then it is good (or
An example of an implication between modal sentences, valid right) to omit A. Also the phrase "the right thing to do" indicates,
in physical modal logic, but invalid in ethical modal logic is the that for carh action A either to do A or not to do A has tn hc
following : right.
But the proposed interpretation of ethical modalities starting
with a system Z of basic norms, does not yield such an ethical
The proof in physical modal logic is obvious: from tertium non datur (even if classical logic is applied).
If we start with a system Z I , it may follow
n,, A A,, ,A
for a certain class of actions A, but this tertium non datnr will
which I d s with the help of A < V, A to the above theorem. never hold for all A (because ZI has to be finite). We may enrich
I n ethical modal logic - even if we add as a hypothesis that no 21 in order to get a refined system Zz, in which for more A's
contradiction (i. e. n o logically false sentence A) is obligatory-
V /\, p x cannot he defended on the basis of A, A p z: A,, A v A=, - A
72 A'ormalive Logic and Ethics 7. Foundations of Partical Philosophy 73

is valid-but still Zz will not be complete: some actions will is (classically) valid. The mistake to omit Sz is the mistake of
remain "discretionary", i. e. neither required nor forbidden. moral rigorism in which all actions are either "pood" or "bad".
The most important case of ethical modalities with respect to Modal logic with reepect to more than one basis hss to be distin-
two different bases Z1, and Zz is the case of a moral code Z z , part guished from modal logic with iterated modalities. The most
of which is made (by institutional enforcement) a legal code 21. important use is a "constitution" which restricts the lawgiver
If we write for short A1 instead of A=, and Az instead of A=, (e. g. the English king). Such a constitution forbids certain laws,
(and using V I , XI, . . . correspondingly) the assumption Z I c Zz which are called "unconstitutional" acrording to the formula:
yields the following implication A1 A < Az A . Other implications "Congress shall make no law . . . ".
follow. I n the administration there are similar situations in which a
We may easily select 9 modalities with the same implications as higher authority restricts the norms of a lower authority. There
in physical modal logic: may be a hierarchy of such authorities. In order to interpret a
modal sentence with wtimes iterated A , we have to assume a
A: chain of n authorities (cf. 0. BECKER, Untersuchungen iiber den
Modalkalkiil, Meisenheim 1952).
The dialogical rules for ethical modal logic with iterated moda-
lities are the same aa without iteration.
Under the assumption that all authorities are logically con-
sistent we have A < V on each level and therefore, e. g.

Here Azl has t o be defined by But e. g. the implication A A < 4 is equivalent with the aasump-
tion that the lower authority actually requires what it is required
to require (by the higher authority).
The similarity with the physical modalities is only a seeming one,
because the whole picture (with Xal A + V z A A V i A) looks as
follows : 7 . Foundations of Practical Philosophy

I n the lectures up to now, i t seems to me, everything has been

"plain sailing". No one has any serious difficulties with elemen-
tary sentences, with descriptions or abstractors, with the logical
particles or modal rules, with rules for predicaton or constructive
rules for arithmetical symbols, or even with ideal norms in proto-
physics. Of course, most people still insist on doubting every step.
but this is only a kind of philosophic amusement and the practice
of the scientists is not disturbed by such verbal games.
I n this sense, then, everything has been plain sailing. But now,
in this lecture. we are going to enter upon the rough open sea.
For I a m going to talk about "practical philosophy", and this
Only a sixtum non datur term, in my usage, stands for moral philosophy in a broad sense.
d l A v Azl A v Xz A v A i l A v A; A including ethics, legal philosophy and politics.
72 A'ormalive Logic and Ethics 7. Foundations of Partical Philosophy 73

is valid-but still Zz will not be complete: some actions will is (classically) valid. The mistake to omit Sz is the mistake of
remain "discretionary", i. e. neither required nor forbidden. moral rigorism in which all actions are either "pood" or "bad".
The most important case of ethical modalities with respect to Modal logic with reepect to more than one basis hss to be distin-
two different bases Z1, and Zz is the case of a moral code Z z , part guished from modal logic with iterated modalities. The most
of which is made (by institutional enforcement) a legal code 21. important use is a "constitution" which restricts the lawgiver
If we write for short A1 instead of A=, and Az instead of A=, (e. g. the English king). Such a constitution forbids certain laws,
(and using V I , XI, . . . correspondingly) the assumption Z I c Zz which are called "unconstitutional" acrording to the formula:
yields the following implication A1 A < Az A . Other implications "Congress shall make no law . . . ".
follow. I n the administration there are similar situations in which a
We may easily select 9 modalities with the same implications as higher authority restricts the norms of a lower authority. There
in physical modal logic: may be a hierarchy of such authorities. In order to interpret a
modal sentence with wtimes iterated A , we have to assume a
A: chain of n authorities (cf. 0. BECKER, Untersuchungen iiber den
Modalkalkiil, Meisenheim 1952).
The dialogical rules for ethical modal logic with iterated moda-
lities are the same aa without iteration.
Under the assumption that all authorities are logically con-
sistent we have A < V on each level and therefore, e. g.

Here Azl has t o be defined by But e. g. the implication A A < 4 is equivalent with the aasump-
tion that the lower authority actually requires what it is required
to require (by the higher authority).
The similarity with the physical modalities is only a seeming one,
because the whole picture (with Xal A + V z A A V i A) looks as
follows : 7 . Foundations of Practical Philosophy

I n the lectures up to now, i t seems to me, everything has been

"plain sailing". No one has any serious difficulties with elemen-
tary sentences, with descriptions or abstractors, with the logical
particles or modal rules, with rules for predicaton or constructive
rules for arithmetical symbols, or even with ideal norms in proto-
physics. Of course, most people still insist on doubting every step.
but this is only a kind of philosophic amusement and the practice
of the scientists is not disturbed by such verbal games.
I n this sense, then, everything has been plain sailing. But now,
in this lecture. we are going to enter upon the rough open sea.
For I a m going to talk about "practical philosophy", and this
Only a sixtum non datur term, in my usage, stands for moral philosophy in a broad sense.
d l A v Azl A v Xz A v A i l A v A; A including ethics, legal philosophy and politics.
 74 Xormalive Logic and Ethics 7. Foundations of Partical Philosophy 55

The very moment I utter the word "morals" I am aware t h a t Moreover, the norl-religious norms which are part of the Decalo-
the direction I have chosen t o go will lead me into a barrage of gue-Do not kill (other humans)! Do not eornn~itadultery! Do not
criticism from those who wish t o go in another direction. If 1 were steal! Do not bear false witness!-are still the core of penal codes
t o call attention t o the vanity of all our lives, if I were to say that all over the world today. Theoretically you may consider this t o
every €!fort is in vain, that only death is certain and that life be merely an historical fart, but since, in practice, you are sitting
itself is the vanity of vanities (you would also have to recognize in this room and not in a prison, I have to assume that yo11 are
my existentialist tone of voice), then many of you would accuse following these norms (with the possible exception of adultery,
me of platitudes. Yo11 would criticize me because everyone already which, interestingly, has not been considered a crime in England
knows the truth of sentences like ,,All men are mortal". since CROMWELL).
If I went on t o ask yon whether i t might he the case that you I take legal norms t,o he a species of rrroral n u r r n s t h e i r speritir
have only knowledge but not underfitanding of your mortality, difference hcing their nto<leof enfor~cnrnent.1 ~mderstandthat the
rather than answering me yo11 would probably ask yournelf: elrmcnt of pr~hlirforce in the law j~~stifios qrtite a lot of theorizing
"What is he trying t o d o ? I s he going t o try t o change our lives? in jurisprudence, especially since the political revolutions of modern
What feeling is he attempting to express? Why docs he misuse a tilnns no longer p n m i t 11ssimply t o rely on the will of the king.
lecture in thin wthy? i f he wants t o talk about moral philosophy, But t h i ~does not just.ify ollr forpett.ing that legal nnr~ns,i. e., en-
then he ought t o aim a t exactness, objectivity and truth, a s forced moral norms, provide us with a rich set of examples of
everyone who wishes t o do decent philosophy !" moral norms. So the term ,,moral norm" is already well determined,
B u t is "Attempt truth-and nothing else!" a justifiable norm though only by examples.
for the moral philosopher? To ask this question is already t o deny Thus we find ourselves in the situation of having already accepted
the imperative, for the question asks for the justification of the certain moral norms. The quest,ion now is: "Why do I accept such
norm-not its truth. I n this lecture I hope t o show that there norms?" It is, therefore, with this question that moral philosophy
are non-empirical truths in practical philosophy. But my main begins; and i t begins, therefore, as norn~ativcethics. To say that,
purpose will be t o show that practical philosophy deals not only however, is not t o say that I will use words uncritically. But my
with truth, but also with the justifmtion of norms. immediate concern is not with a meta-ethical examination of the
I n order t o do this, I will have to formulate two "super-norms". moral talk of other people. Rather, 1 will be very much concerned
called "principles", t o be u ~ e din justifying norms. Obviously, I with my own ways of talking with you.
will not be able t o "justify" these principles, sincn the term It is now time t o get down t o business.
4'.
j ~ ~ t i f i c a t i o nmakes
'' sense only a f a r one has accepted such prin- The first task of practical philosophy is the reconstruction of a
ciples. So, if one defines "faith" in a negative sense, a s "the minimunl of vocabulary so that we can argne for or against the
acceptance of something which is not justified", the acceptance of acceptnnre of norms. As we will see later, this constr~~ction has
these principles m a y be called a n act of faith. But thin act of faith already been introduced "implicitly" into our natural language,
will be different from acts of faith in traditional religions, if we under the influence of philosophical doct,rines; and this is tht'
take "religious faith" a~ a yielding of one's critical judgment reason for calling a new const,uction a "reconstruction". I n Illy
before a (religious) tradition. earlier lectures I hare already krgued for some norms, but so far I
Let me make a more harmless confession: I cannot find any have limited my concern t o norms for using words; now we will 1)c
justification for t h a t playing with the term "moral" which is concerned mit,h norms for actions.
commonplace in contemporary philosophy. Even the briefest One can introduce terms by exemplary determination alone,
history of the t e r m "moral philosophy" shows that a t least u p t o i. e., by using appropriate sit,uations and exercises. But when I
THOMASIUS,a younger contemporary of JOHN LOCKE,moral argued for Lhe rules of tho dialogical gamo, I hail to use nolne
philosophy always included attempts t o justify the law, i. e., to language in which to formulate the rules. I n such cases we need
justify legal nonns. t o distinguish between the language used for teachers only and
 74 Xormalive Logic and Ethics 7. Foundations of Partical Philosophy 55

The very moment I utter the word "morals" I am aware t h a t Moreover, the norl-religious norms which are part of the Decalo-
the direction I have chosen t o go will lead me into a barrage of gue-Do not kill (other humans)! Do not eornn~itadultery! Do not
criticism from those who wish t o go in another direction. If 1 were steal! Do not bear false witness!-are still the core of penal codes
t o call attention t o the vanity of all our lives, if I were to say that all over the world today. Theoretically you may consider this t o
every €!fort is in vain, that only death is certain and that life be merely an historical fart, but since, in practice, you are sitting
itself is the vanity of vanities (you would also have to recognize in this room and not in a prison, I have to assume that yo11 are
my existentialist tone of voice), then many of you would accuse following these norms (with the possible exception of adultery,
me of platitudes. Yo11 would criticize me because everyone already which, interestingly, has not been considered a crime in England
knows the truth of sentences like ,,All men are mortal". since CROMWELL).
If I went on t o ask yon whether i t might he the case that you I take legal norms t,o he a species of rrroral n u r r n s t h e i r speritir
have only knowledge but not underfitanding of your mortality, difference hcing their nto<leof enfor~cnrnent.1 ~mderstandthat the
rather than answering me yo11 would probably ask yournelf: elrmcnt of pr~hlirforce in the law j~~stifios qrtite a lot of theorizing
"What is he trying t o d o ? I s he going t o try t o change our lives? in jurisprudence, especially since the political revolutions of modern
What feeling is he attempting to express? Why docs he misuse a tilnns no longer p n m i t 11ssimply t o rely on the will of the king.
lecture in thin wthy? i f he wants t o talk about moral philosophy, But t h i ~does not just.ify ollr forpett.ing that legal nnr~ns,i. e., en-
then he ought t o aim a t exactness, objectivity and truth, a s forced moral norms, provide us with a rich set of examples of
everyone who wishes t o do decent philosophy !" moral norms. So the term ,,moral norm" is already well determined,
B u t is "Attempt truth-and nothing else!" a justifiable norm though only by examples.
for the moral philosopher? To ask this question is already t o deny Thus we find ourselves in the situation of having already accepted
the imperative, for the question asks for the justification of the certain moral norms. The quest,ion now is: "Why do I accept such
norm-not its truth. I n this lecture I hope t o show that there norms?" It is, therefore, with this question that moral philosophy
are non-empirical truths in practical philosophy. But my main begins; and i t begins, therefore, as norn~ativcethics. To say that,
purpose will be t o show that practical philosophy deals not only however, is not t o say that I will use words uncritically. But my
with truth, but also with the justifmtion of norms. immediate concern is not with a meta-ethical examination of the
I n order t o do this, I will have to formulate two "super-norms". moral talk of other people. Rather, 1 will be very much concerned
called "principles", t o be u ~ e din justifying norms. Obviously, I with my own ways of talking with you.
will not be able t o "justify" these principles, sincn the term It is now time t o get down t o business.
4'.
j ~ ~ t i f i c a t i o nmakes
'' sense only a f a r one has accepted such prin- The first task of practical philosophy is the reconstruction of a
ciples. So, if one defines "faith" in a negative sense, a s "the minimunl of vocabulary so that we can argne for or against the
acceptance of something which is not justified", the acceptance of acceptnnre of norms. As we will see later, this constr~~ction has
these principles m a y be called a n act of faith. But thin act of faith already been introduced "implicitly" into our natural language,
will be different from acts of faith in traditional religions, if we under the influence of philosophical doct,rines; and this is tht'
take "religious faith" a~ a yielding of one's critical judgment reason for calling a new const,uction a "reconstruction". I n Illy
before a (religious) tradition. earlier lectures I hare already krgued for some norms, but so far I
Let me make a more harmless confession: I cannot find any have limited my concern t o norms for using words; now we will 1)c
justification for t h a t playing with the term "moral" which is concerned mit,h norms for actions.
commonplace in contemporary philosophy. Even the briefest One can introduce terms by exemplary determination alone,
history of the t e r m "moral philosophy" shows that a t least u p t o i. e., by using appropriate sit,uations and exercises. But when I
THOMASIUS,a younger contemporary of JOHN LOCKE,moral argued for Lhe rules of tho dialogical gamo, I hail to use nolne
philosophy always included attempts t o justify the law, i. e., to language in which to formulate the rules. I n such cases we need
justify legal nonns. t o distinguish between the language used for teachers only and
 7. Follndationa of Partieel Philosophy 77

the language to be taught, e. g., the logical particles A , v , +, Since imperatives are one of the simplest kinds of talk, the
A , V, -. I would like to introduce the term "ortholanguage"
teacher may almady have introduced his student to such express-
for that language which is to be taught methodically and
" ions as "Get some water!" Imperatives refer directly to practice
paralanguage" for that language which may be used in ex-
(experience and action), either in the present or in the future. In
plaining how to teach the ortholanguage. Pedagogically a para-
a situation in which no action has yet been settled on the teacher
Ianguage will precede the ortholanguage and will be used primarily
can begin to develop the art of discussing plans for future actions.
to describe a situation in which ortho-linguistic terms can be
Such a discussion does, of course, center upon proposed action,
appropriately introduced. Since such descriptions must not
but any contemplated action must wait upon the outcome of the
presuppose a n understanding of those terms which need to be
discussion which therefore may be called "epipractical talk". Plans
taught, the paralanguage used must not contain any synonyms
can be talked over without the use of mental terms.
for these terms.
I n this final lecture, therefore, we have to construct an ortho- In a discussion the proposal, "Let us do A", may have been
language which can be used in practical arguments. If practical offered and agreed upon. Then the student will do A. However,
philosophy is to equip us with a vocabulary with which we can there may be circumstances such that it is not possible to do A
argue for or against norms, it must first provide us with an ortho- before the next day, and then those who are discussing plans may
linguistic "mental" terminology. In my preceding lectures I used have agreed to the proposal, "Let us do A tomorrow".
such terms as "want", "know", etc., rather freely. For example, I The student has to learn that, before doing A. he may describe
used paralanguage t o say that lens grinders want their surfaces to himself (i. e., his "mental state") by saying, "I b A". If the nest
be homogeneous a n d that I did not know of any other approach to day comes and he does not do A, he has to loam t,o say, "I b ' A"
geometry which is not arbitra~y.Now we must dismiss all such (with ' for negation). He has "changed his mind". If the student
,,mental terms" from tho paralanguage so that we can reintroduce
participates scnsibly in thc discussion of plans, ho will gradually
them methodically into an ortholanguage. learn how to use the term " b". If, for example, in the courst: of
Let me begin with the term "mind", i. e., with giving instruc- a discussion he changes his acceptance into a rejection, one could
tions for teaching t h e use of the phrase, "S hasamind". I will take this tell him that a t some moment before the linguistic act of rejecting
sentence as an expression which summarizes three more specific a plan, he must have "changed his mind", i. e., must have changed
"states of mind", namely, "S has the volition t o do A", " S wants from b to b'.
that X",and "S has the opinion that X". Here "S" stands for The student needs to learn that the phrases "b A" and " b' A"
proper names of persons, "A" stands for action-predicates, and do not refer to the linguistic acts of agreeing or disagreeing to a
"X" stands for (descriptive) sentences. I n giving these English proposed plan A. "I b A" does not stand for the sentence. "I have
"translations" of t h e three mental terms which are to he introduced, agreed to do A", and the instruction for " b " is not reducible to
I a m only giving a hint to those who know English. Officially, so a definition of the kind: "S b A + If asked, S would agree to do
to speak, I must introduce three phrases, namely, S b A , S D X, A". I take such a counterfactual as saying that S has the volition
and S D X , without giving either English synonyms for b, D, to manifest externally his aesent to do A under the condition that
or D or German synonyms as "Wollen", "Begehren", "Meinen". he is asked. The counterfactual, thus interpreted, is of the form:
I will now give instructions for the use of b, P, and D in a " S b (if B, then 6)" and is thereby reducible to " b ", but " b "
paralanguage without using synonyms. The instructions have to is not reducible t o the counterfactual. I am introducing "b A"
be given in concrete situations in which the use of the terms is and " b ' A" as referring to the student's "mental" activity, which
appropriate. Here, in a lecture, I can only describe to you how a he can perform without saying anything, perhaps without oven
teacher should instruct his student. The teacher himself should having been asked anything.
use ortholanguage only, but, in order to describe his teaching,
I propose to use " S has the volition to do A" in philosophical
I will have t o use paralanguage.
English, instead of "S b A".
 7. Follndationa of Partieel Philosophy 77

the language to be taught, e. g., the logical particles A , v , +, Since imperatives are one of the simplest kinds of talk, the
A , V, -. I would like to introduce the term "ortholanguage"
teacher may almady have introduced his student to such express-
for that language which is to be taught methodically and
" ions as "Get some water!" Imperatives refer directly to practice
paralanguage" for that language which may be used in ex-
(experience and action), either in the present or in the future. In
plaining how to teach the ortholanguage. Pedagogically a para-
a situation in which no action has yet been settled on the teacher
Ianguage will precede the ortholanguage and will be used primarily
can begin to develop the art of discussing plans for future actions.
to describe a situation in which ortho-linguistic terms can be
Such a discussion does, of course, center upon proposed action,
appropriately introduced. Since such descriptions must not
but any contemplated action must wait upon the outcome of the
presuppose a n understanding of those terms which need to be
discussion which therefore may be called "epipractical talk". Plans
taught, the paralanguage used must not contain any synonyms
can be talked over without the use of mental terms.
for these terms.
I n this final lecture, therefore, we have to construct an ortho- In a discussion the proposal, "Let us do A", may have been
language which can be used in practical arguments. If practical offered and agreed upon. Then the student will do A. However,
philosophy is to equip us with a vocabulary with which we can there may be circumstances such that it is not possible to do A
argue for or against norms, it must first provide us with an ortho- before the next day, and then those who are discussing plans may
linguistic "mental" terminology. In my preceding lectures I used have agreed to the proposal, "Let us do A tomorrow".
such terms as "want", "know", etc., rather freely. For example, I The student has to learn that, before doing A. he may describe
used paralanguage t o say that lens grinders want their surfaces to himself (i. e., his "mental state") by saying, "I b A". If the nest
be homogeneous a n d that I did not know of any other approach to day comes and he does not do A, he has to loam t,o say, "I b ' A"
geometry which is not arbitra~y.Now we must dismiss all such (with ' for negation). He has "changed his mind". If the student
,,mental terms" from tho paralanguage so that we can reintroduce
participates scnsibly in thc discussion of plans, ho will gradually
them methodically into an ortholanguage. learn how to use the term " b". If, for example, in the courst: of
Let me begin with the term "mind", i. e., with giving instruc- a discussion he changes his acceptance into a rejection, one could
tions for teaching t h e use of the phrase, "S hasamind". I will take this tell him that a t some moment before the linguistic act of rejecting
sentence as an expression which summarizes three more specific a plan, he must have "changed his mind", i. e., must have changed
"states of mind", namely, "S has the volition t o do A", " S wants from b to b'.
that X",and "S has the opinion that X". Here "S" stands for The student needs to learn that the phrases "b A" and " b' A"
proper names of persons, "A" stands for action-predicates, and do not refer to the linguistic acts of agreeing or disagreeing to a
"X" stands for (descriptive) sentences. I n giving these English proposed plan A. "I b A" does not stand for the sentence. "I have
"translations" of t h e three mental terms which are to he introduced, agreed to do A", and the instruction for " b " is not reducible to
I a m only giving a hint to those who know English. Officially, so a definition of the kind: "S b A + If asked, S would agree to do
to speak, I must introduce three phrases, namely, S b A , S D X, A". I take such a counterfactual as saying that S has the volition
and S D X , without giving either English synonyms for b, D, to manifest externally his aesent to do A under the condition that
or D or German synonyms as "Wollen", "Begehren", "Meinen". he is asked. The counterfactual, thus interpreted, is of the form:
I will now give instructions for the use of b, P, and D in a " S b (if B, then 6)" and is thereby reducible to " b ", but " b "
paralanguage without using synonyms. The instructions have to is not reducible t o the counterfactual. I am introducing "b A"
be given in concrete situations in which the use of the terms is and " b ' A" as referring to the student's "mental" activity, which
appropriate. Here, in a lecture, I can only describe to you how a he can perform without saying anything, perhaps without oven
teacher should instruct his student. The teacher himself should having been asked anything.
use ortholanguage only, but, in order to describe his teaching,
I propose to use " S has the volition to do A" in philosophical
I will have t o use paralanguage.
English, instead of "S b A".
78 No~maliveLogic nnd Elhics

Thc term "intention" may be used only in thosc special situa- which is minimally sufficient for formulating the "principles" of
tions in which someone has the volition to do A in orrler to bring pract,ical philosophy, however, this particnlar extension is not reb-
about (to cause) a cert,ain st,at,e of affairs, X. If it is agreed that vant. It is enongh that we now can say that the considerations
certain actions will cause tho state X , then "to intend that .Y" which precede a decision take different wantings into account and
can be defined as "to have t,he volition to rause t h a t X". finally end in a decision in favour of one over the others.
'I'hwc art: snvornl ut,hor t,crrns uhioh r:ari ht: ~ m r f ' ~ ~~nlrnduccd.
lly Ren~arkably, all such terms as "pain" and "pleasure", "to
t 1 prnl)osr: t~conjparc t.ho
txcause t,hey fncilit,atr! talking a b ~ n ~b. enjoy" and "to suffern-I'll call these "hedonistic" terms-are
"nrental ~ t a t e "of having a volitlion with the (ordinary) state of snparfluous for our purpose! If someone wants to do A and has
having a n object, esperially a n artifact (a statue, for example). tried t,o argue for i t by saying, "I enjoy to do A", he might just
This analogy allows us t o speak of the act of forming a volition as: as well have simply repeated t h a t he wants to do A. If, instead, he
deciding + forming a volition says that everyone in a similar situation finds to do A enjoyable,
then, in effect, he isasserting that everyone in such a situation should
a n d of preparing a decision (i. c., the act of deciding) as: want to do A or a t least nray want t o do A. But then he is asserting a
considering + preparing a decision. modal sentence, and fnrt.her discussion aould have to concern this
norm and its possible justification.
Thoso who participate sensibly in a discussion have t o consider Instead of asking whether sorneonc wants sornothing or not., it
varinns alternatives before deciding what t o do. The stndent has is common prartire t o ask hirn how he "feels" about it or whether
to learn such "considerations" hy participrhng in disc~~ssions. or not he "feels happy" ahout it. B I I in ~ the reconstruction I
Different proposals are formulated in sentences like "Let us do A," propose, these are only different ways of saying the same thing.
and "No, let ns d o A2". Discnssions also often involvc statements If someone asserts that he would be "happy" with X , the questions
t o tho effect "If we do A , we will callso that X", and "Let us canse still can be a s l d : Slr~~uldlrr bc "happy" with i t ? Should 110 not
t h a t XI", and "No, let us cawe t h a t Xz". want somet.hing else?
I will now doscribe how the teacher can instruct his student in One may roughly classify as "hedonistic" all those moral
t h e use of another basic ment,al term, " D m ,in connection with theories which 'take the wantings of people as merely given. Such
such considerations. "8 D X" shall be introduced in such a way theories elaborate on ways of harmonizing, reconciling, optimizing,
t,hat i t partially reconstructs tho English phrase, " S wants that etc., the given wantings (or "interests" as they often are called)
X", so t h a t if S has the volition to cause that X, we will say, "S in common volitions. But moral philosophy has &hetask of formulat-
D A"'. But "8 D X" should not be construed its being syno- ing principle8 which allow us to work on merely subjectively "gillen"
n y m s with "S int,onds t h a t X". Rather, if in the course of a u~antingsand so to discipline them that our deeisiov~sahoul how to
discussion S has proposed a plan and after having unsuccessfully wl are justifiable.
tried t o defend it, he finally adopted another plan, then S has to Quite a lot of this work is done by working on the "facts"
learn t h a t his mental stat,e before t h a t derision is described as involved in a decision. Practical disonssions include not only
"want,ing t h a t X", though definitely not as "intending that X". proposals for action bnt also tl~eoretical-especiw f~bctual-
I n the coulse of a cliscussion, therefore, we can loam what it questions. We have to look for a descripti~~n of our sitnation.
is t h a t we want. Our wantings become "manifest"; and other including the "facts" about ourselvw, in contrnst to inorcly irnagi-
people can infer from our talking what those wantings are (only, nary "opinions" that we may happen t o hold about oursclve~.In
of course, if the possibility of lying, pretending, etc., is reasonably order t o reconstruct the mental term "opinion", the t.cacher of
excluded). the ortholanguaar has to instruct his student about t h e IISC of
I n order t o explain our dreams (or actions which do not make ''8 D X", i. e., " S has the opinion that X". If "X" refers t o a
sense o n the basis of our manifest wantings), we may talk of future state, I propose t o use the term "belief" in philosophical
"
latent" wantings. For t h e purpose of constructing a terminology English, instead of "opinion".
78 No~maliveLogic nnd Elhics

Thc term "intention" may be used only in thosc special situa- which is minimally sufficient for formulating the "principles" of
tions in which someone has the volition to do A in orrler to bring pract,ical philosophy, however, this particnlar extension is not reb-
about (to cause) a cert,ain st,at,e of affairs, X. If it is agreed that vant. It is enongh that we now can say that the considerations
certain actions will cause tho state X , then "to intend that .Y" which precede a decision take different wantings into account and
can be defined as "to have t,he volition to rause t h a t X". finally end in a decision in favour of one over the others.
'I'hwc art: snvornl ut,hor t,crrns uhioh r:ari ht: ~ m r f ' ~ ~~nlrnduccd.
lly Ren~arkably, all such terms as "pain" and "pleasure", "to
t 1 prnl)osr: t~conjparc t.ho
txcause t,hey fncilit,atr! talking a b ~ n ~b. enjoy" and "to suffern-I'll call these "hedonistic" terms-are
"nrental ~ t a t e "of having a volitlion with the (ordinary) state of snparfluous for our purpose! If someone wants to do A and has
having a n object, esperially a n artifact (a statue, for example). tried t,o argue for i t by saying, "I enjoy to do A", he might just
This analogy allows us t o speak of the act of forming a volition as: as well have simply repeated t h a t he wants to do A. If, instead, he
deciding + forming a volition says that everyone in a similar situation finds to do A enjoyable,
then, in effect, he isasserting that everyone in such a situation should
a n d of preparing a decision (i. c., the act of deciding) as: want to do A or a t least nray want t o do A. But then he is asserting a
considering + preparing a decision. modal sentence, and fnrt.her discussion aould have to concern this
norm and its possible justification.
Thoso who participate sensibly in a discussion have t o consider Instead of asking whether sorneonc wants sornothing or not., it
varinns alternatives before deciding what t o do. The stndent has is common prartire t o ask hirn how he "feels" about it or whether
to learn such "considerations" hy participrhng in disc~~ssions. or not he "feels happy" ahout it. B I I in ~ the reconstruction I
Different proposals are formulated in sentences like "Let us do A," propose, these are only different ways of saying the same thing.
and "No, let ns d o A2". Discnssions also often involvc statements If someone asserts that he would be "happy" with X , the questions
t o tho effect "If we do A , we will callso that X", and "Let us canse still can be a s l d : Slr~~uldlrr bc "happy" with i t ? Should 110 not
t h a t XI", and "No, let us cawe t h a t Xz". want somet.hing else?
I will now doscribe how the teacher can instruct his student in One may roughly classify as "hedonistic" all those moral
t h e use of another basic ment,al term, " D m ,in connection with theories which 'take the wantings of people as merely given. Such
such considerations. "8 D X" shall be introduced in such a way theories elaborate on ways of harmonizing, reconciling, optimizing,
t,hat i t partially reconstructs tho English phrase, " S wants that etc., the given wantings (or "interests" as they often are called)
X", so t h a t if S has the volition to cause that X, we will say, "S in common volitions. But moral philosophy has &hetask of formulat-
D A"'. But "8 D X" should not be construed its being syno- ing principle8 which allow us to work on merely subjectively "gillen"
n y m s with "S int,onds t h a t X". Rather, if in the course of a u~antingsand so to discipline them that our deeisiov~sahoul how to
discussion S has proposed a plan and after having unsuccessfully wl are justifiable.
tried t o defend it, he finally adopted another plan, then S has to Quite a lot of this work is done by working on the "facts"
learn t h a t his mental stat,e before t h a t derision is described as involved in a decision. Practical disonssions include not only
"want,ing t h a t X", though definitely not as "intending that X". proposals for action bnt also tl~eoretical-especiw f~bctual-
I n the coulse of a cliscussion, therefore, we can loam what it questions. We have to look for a descripti~~n of our sitnation.
is t h a t we want. Our wantings become "manifest"; and other including the "facts" about ourselvw, in contrnst to inorcly irnagi-
people can infer from our talking what those wantings are (only, nary "opinions" that we may happen t o hold about oursclve~.In
of course, if the possibility of lying, pretending, etc., is reasonably order t o reconstruct the mental term "opinion", the t.cacher of
excluded). the ortholanguaar has to instruct his student about t h e IISC of
I n order t o explain our dreams (or actions which do not make ''8 D X", i. e., " S has the opinion that X". If "X" refers t o a
sense o n the basis of our manifest wantings), we may talk of future state, I propose t o use the term "belief" in philosophical
"
latent" wantings. For t h e purpose of constructing a terminology English, instead of "opinion".
 80 A'ornmtiue Logic and Ethics

The term " D " can be introduced as a special case of b . We have punuing the "philosophy of mind" only to get a minimum voca-
already seen t h a t we can distinguish between the linguistic act of bulary with which to formulate principles of practical philosophy.
assenting t o a proposal for action and the mental decision to do so. I t is not necessary for us to reconstrnct further "mental terms"
Here, instead of a volition to do A, we have the special case of a now, because we can learn to argue about the truth of practical
volition to assent to a (descriptive) sentence, X. So we may in- norms without using mental terms a t all.
troduce the phrase, "S [> X", as an abbreviation of "8 b assenting What i s of the utmost relevance for practical philosophy is
t o X". But of course tho student first has to learn how to parti- theoretical reasoning. Theoretical reasoning is concerned with the
cipate sensibly in a discussion about the truth of X. The phrase truth of sentences rather than with tho ethical modalities of
"8 D X" has t o be taught and learned within the context of actions. Practical philosophy has the task of finding principles
honest discussions. All variants of asserting-such as lying, pre- which allow us to argue for or against an action or, as we now
tending, etc.-will, unfortunately, be learned quickly enough can say, for or against a decision. During such practical reasoning
afterwards. we have to take different wantings into consideration, and we
If we use the same analogy with preparing to form a statue, have to modify the "given" ones-those which simply happen
then forming a statue and finally having it, we now can introduce to occur to us-until we end up with a "justifiable" decision. This
judging + forming an opinon means that we have to reject all those wantings which are not
thinking + preparing a judgment. "justifiable"; we have to discipline our wantings.
Here the relevance of theoretical reasoning becomes clear: it
The choice of these English terms may seem somewhat arbitrary, provides 11s with a paradigmatic case of how we can and should
but I do not want to lose time discussing their appropriateness. discipline our wantings. In theoretical matters we are accnstomed
Especially this terminology does not exclude that we often "judge" to restricting our opinons in such a way that we do not insist on
without any explicit preparation. opinions which simply happen to occur to us. On the contrary, we
I n the special case of judging (what to say), we have much try to overcome the imperfections of the opinions with which we
better training in doing "the right thing" (namely, in saying the
begin. Our wantings are more easily disciplined in theoretioal
truth) than we have in the general case of deciding (what to do). matters than in practical matters, precisely because theoretical
Very often "the right thing to do" is nothing more than a pseudo- reasoning is one step removed from a h o n .
description. But for elementary sentences of the form S E p, there Such a program is possible in theoretical matters, because we
are already rules for the use of the predicators p ; at least there can critically reconstruct the scientific-hnmanistic disciplines
are standard examples and counter-examples. I n cases such as color
(Wissenschaften), including a critical reconstruction of thcir
blindness-when someone fails to "feel" the same way as other terminology. (To say this is to anticipate a special application of
people do about the difference in colour between, e. g., blood and the "cultural principle" which will be formulated later. There
grass, -no common decision procedure is available. Then a quarrel we will see that practical reasoning also involves the more difficult
is fruitless, and somo predicators, e. g. "red" and "green" are job of reconstructing a genesis of concrete situations.) I n the
simply dropped from the discussion. scientific-humanistic disciplines we appropriate to ourselves the
I n our ordinary language we have such "mental" terms as
<' perception" (i. e., "seeing", "hearing", etc.) and "imagination" reasonable use of terminology, and by this method we learn to
apply the language in new situations. All mere fancy is methodi-
(i. e., "memory", "phantasy", etc.). These are used when the cally excluded; we learn to surrender ourselves to reason.
truth of a sentence is in question, e.g., "Is this red?" "Yes, I see I n philosophical language, the term "subjectivity" is often
t h a t it is red." I n the case of a past event, memory has to be
used to refer to a general unwillingness to surrender one's own
substituted for perception. I do not deny that the extension of opinions. I wish to adopt this term. But an opinion is not to be
t h e ortholanguage in the direction of such terms can be useful, construed as "merely subjective" only if it is idiosyncratic with
b u t for our purposes here such a n extension is irrelevant. We are the person who holds it. An opinion still may be "merely subjec-
 80 A'ornmtiue Logic and Ethics

The term " D " can be introduced as a special case of b . We have punuing the "philosophy of mind" only to get a minimum voca-
already seen t h a t we can distinguish between the linguistic act of bulary with which to formulate principles of practical philosophy.
assenting t o a proposal for action and the mental decision to do so. I t is not necessary for us to reconstrnct further "mental terms"
Here, instead of a volition to do A, we have the special case of a now, because we can learn to argue about the truth of practical
volition to assent to a (descriptive) sentence, X. So we may in- norms without using mental terms a t all.
troduce the phrase, "S [> X", as an abbreviation of "8 b assenting What i s of the utmost relevance for practical philosophy is
t o X". But of course tho student first has to learn how to parti- theoretical reasoning. Theoretical reasoning is concerned with the
cipate sensibly in a discussion about the truth of X. The phrase truth of sentences rather than with tho ethical modalities of
"8 D X" has t o be taught and learned within the context of actions. Practical philosophy has the task of finding principles
honest discussions. All variants of asserting-such as lying, pre- which allow us to argue for or against an action or, as we now
tending, etc.-will, unfortunately, be learned quickly enough can say, for or against a decision. During such practical reasoning
afterwards. we have to take different wantings into consideration, and we
If we use the same analogy with preparing to form a statue, have to modify the "given" ones-those which simply happen
then forming a statue and finally having it, we now can introduce to occur to us-until we end up with a "justifiable" decision. This
judging + forming an opinon means that we have to reject all those wantings which are not
thinking + preparing a judgment. "justifiable"; we have to discipline our wantings.
Here the relevance of theoretical reasoning becomes clear: it
The choice of these English terms may seem somewhat arbitrary, provides 11s with a paradigmatic case of how we can and should
but I do not want to lose time discussing their appropriateness. discipline our wantings. In theoretical matters we are accnstomed
Especially this terminology does not exclude that we often "judge" to restricting our opinons in such a way that we do not insist on
without any explicit preparation. opinions which simply happen to occur to us. On the contrary, we
I n the special case of judging (what to say), we have much try to overcome the imperfections of the opinions with which we
better training in doing "the right thing" (namely, in saying the
begin. Our wantings are more easily disciplined in theoretioal
truth) than we have in the general case of deciding (what to do). matters than in practical matters, precisely because theoretical
Very often "the right thing to do" is nothing more than a pseudo- reasoning is one step removed from a h o n .
description. But for elementary sentences of the form S E p, there Such a program is possible in theoretical matters, because we
are already rules for the use of the predicators p ; at least there can critically reconstruct the scientific-hnmanistic disciplines
are standard examples and counter-examples. I n cases such as color
(Wissenschaften), including a critical reconstruction of thcir
blindness-when someone fails to "feel" the same way as other terminology. (To say this is to anticipate a special application of
people do about the difference in colour between, e. g., blood and the "cultural principle" which will be formulated later. There
grass, -no common decision procedure is available. Then a quarrel we will see that practical reasoning also involves the more difficult
is fruitless, and somo predicators, e. g. "red" and "green" are job of reconstructing a genesis of concrete situations.) I n the
simply dropped from the discussion. scientific-humanistic disciplines we appropriate to ourselves the
I n our ordinary language we have such "mental" terms as
<' perception" (i. e., "seeing", "hearing", etc.) and "imagination" reasonable use of terminology, and by this method we learn to
apply the language in new situations. All mere fancy is methodi-
(i. e., "memory", "phantasy", etc.). These are used when the cally excluded; we learn to surrender ourselves to reason.
truth of a sentence is in question, e.g., "Is this red?" "Yes, I see I n philosophical language, the term "subjectivity" is often
t h a t it is red." I n the case of a past event, memory has to be
used to refer to a general unwillingness to surrender one's own
substituted for perception. I do not deny that the extension of opinions. I wish to adopt this term. But an opinion is not to be
t h e ortholanguage in the direction of such terms can be useful, construed as "merely subjective" only if it is idiosyncratic with
b u t for our purposes here such a n extension is irrelevant. We are the person who holds it. An opinion still may be "merely subjec-
82 Normative Logic a d Ethics 7. Foundations of Partical Philoaophy 83

t i v ~ "even if it, is held by many people. I n the sciences it is taken by examining how we can come t o a justifiable practiral decision
for grant,erl (and in t h r hr~manitieni t should bc taken for granted) in a concrete situation.
that the s~ibjert.ivityof <,pinion is imleprndent of the number of I n order t o reach a decision in a particular situation. we have
pcnplc who lidti an opinion. A scnlrnw rloes not her:onie true by t o apply "material" norms, and we have to justify the material
the mere fact t.1la.t il is asswted by many prople. The many, norms which we apply. According to the principle of transsuhjecti-
oYnoAAoi, ~ r s ~ ~ anael l y rt~.st.ricip~l
1.0 s~~hjertivit,y, w h ~ t h e rby their vity, each wanting wtiich is taken into consideration han to be
irliosynrracies or 1))- ~mwit~cally n . c r y ~ t t dfashi~msor traditions. formulated as a (universal) norln: "If the siauation is such ancl
including thc ira.rlitions o f ' n a t ~ ~ rItmg~~age.
al such. then the following action is obligatory (or permissible etc.). ..".
It, is usual t o i ~ n ~ t r n .thr s t I ~ P Ws~lhj~'ctivitj
of opinion with t,he But there will be neither permission nor obligation. unless some
i t y "tr11t11.h11t. as w v have scen i l l n l l ottr investi-
L ' o l ~ j c r ~ t , i ~ ,>I' wantings are distinguished from others in such a way that theic
gations into t,lleorrticnl philosophy, the trut,h of sentr:nrm always satisfaction is required to be taken into consideration (also by
h i ~ st o h r a h~lrnanacc:omplish~nr~nt a n achievcn~mtof persons. every other person concerned). I t may turn out t,hat some elements
No prrson can d o more t.han try to rlvercornr his s~ihjectivity, of the situation finally rule out the satisfaction-but "prima
and t,his is t h r niln of Inpiral cliscipline. facie" (or a s a "rule of thumb") i t has to be considered. Thr
principle of transsubjectivit,,y demands that all n~errlysubjoctivr
It nmy I r r hc,lpf~~I t o h a w a trrrn which A h in wit11 our philoso-
wantings are t o be rnled ont of consideration. Rut we still have the
phi(.al tradifions: so 1 prrqxnse (1 aln, for the rl~onl,mt,still restrict-
problem of distinguishing between those wantings which are and
ing o11r < ~ m c < wtot (.IN, t , t ' ~ ~ 01' t , l ~S , Y ! ~ O I I I , I ~ S )t o ( d l I,IIP rc?({~lired
those wantings which are not "merely subjective".
ov<.ru~rv~ing cnl' s ~ ~ l ) , i ~ ~ ~ "L~JLI~s<.(-II,~,~II<~<!
~livily of ~ ~ ~ J ~ j o ~ ~ L i v i l y " - o r
I n order t o have simpler terms than "n~erelvsubjective" and
" t r i ~ ~ ~ s ~ ~ ~ l , j ~ ~ ~ I: t . i sv hi tLy ." .'I-Ilis is st-ill st~hj,:,.tiviLy, tnlt a $<
transsnbjoctive", I propose to revive a rather oldfashioncd term.
s l j t i t I I is I I 'i s I I i ~ ~ i s - a ntriw ~ l ~ , I Io \ w -
Let me call the merely subjective wantings "inessantial". The*
cotrw thc!~~t. ' I ' r i x r ~ s s ~ ~ l , , i ~ ~ ~is~ n<,t
t i v i t;Ly f w t , i n ~ itL id not n p o s t ~ ~ l a t o
wantings which are left, even when one tries to be as transsrthjec-
either. Trr~r~ssulrjt:cti\~ity in si111p1ya ttmn churnoterizing that
tive as possible, are then to be called "essential" wantings. To
activity in rvhi?t) wr* rlrr rll\vil,vs rxlroa<ly involvr~lif \re hegin t o accept a wanting a s essential is, thereby, defined as a c c q h n g the
r.lvLsr,n 1st all.
obligation t o take it into consideration. Further, since it may bo
If we f o r r n n l a t ~the itrrperativr smtonce, "Let ria transcend the case t h a t someone doesnot actually want what would be an essen-
our subjertivit,y !", we have 11c1.ca "rzcrrn" which rnakes i t possible tial wanting if only he did want it, there is rcaqon for introducing
fbr 11s to justify all ot,ht:r not.rns (norrns in the ordinary sense). another traditionally well-known term, "need" (..Bedurfnis"). If,
T r a n s s ~ ~ b j w t i v iis, t , ~t,liet'ofi~re.a "supr:rnorrn"; it in n "first norm" under certain conditions, a wanting t o do A would be a n essentinl
or, as the Latin tern, goes, 11 "[winciplo". We may cornparc the wanting if it were wanted a t all, I proposc to say that there is a
principle of transsubjectivity t o a cnnstitntion which makes laws "need" t o do A .
possible. But since this principle is the basis not only of all law Of course, this added terminology does not solve the problem
but of all moralit,y. I propose t,o call i t for s t , "tlre moral of how we are t o decide whether there is a need for a "given"
principle" ( I h s Moralprinzip). wanting. But we now have a terminology which is adeqnate for
Looked at, histor,ically, thc moral 1n.ir1ciplc of trnnssubject,ivity formulating what can be said "in the abstract" (i. e., without
is, of collrse, nothing ot,hrr t,l~ana r c f o r ~ n ~ ~ l a t of i o nKANT'S"cute- dealing with any particular "concrete" situation) abont the
gorical imperat,ive". Further, the s a n ~ ec n ~ c i a lrj~lwt,ionwhich distinction between essent,ial and inessential wantings. (This claim
~ r i s o swith KANT'S"fonnnl" ethios also arises hero: r w n if n person will have t o be justified in the following pages.)
rncognizos t,hat transsuhjert.ivity is a nel.essnry colditim of moral Again we begin with practice. This means, as we have scrn, that
reasoning, he can still ask whether this one moral principle is also we are always already involved in the discussion of plans. \Vo arc
sufficient. for moral masoning. We can answer this q ~ i ~ s t i oonly n not only "all along" ( H ~ D E G Q E,,immer
R: schon") talking with
82 Normative Logic a d Ethics 7. Foundations of Partical Philoaophy 83

t i v ~ "even if it, is held by many people. I n the sciences it is taken by examining how we can come t o a justifiable practiral decision
for grant,erl (and in t h r hr~manitieni t should bc taken for granted) in a concrete situation.
that the s~ibjert.ivityof <,pinion is imleprndent of the number of I n order t o reach a decision in a particular situation. we have
pcnplc who lidti an opinion. A scnlrnw rloes not her:onie true by t o apply "material" norms, and we have to justify the material
the mere fact t.1la.t il is asswted by many prople. The many, norms which we apply. According to the principle of transsuhjecti-
oYnoAAoi, ~ r s ~ ~ anael l y rt~.st.ricip~l
1.0 s~~hjertivit,y, w h ~ t h e rby their vity, each wanting wtiich is taken into consideration han to be
irliosynrracies or 1))- ~mwit~cally n . c r y ~ t t dfashi~msor traditions. formulated as a (universal) norln: "If the siauation is such ancl
including thc ira.rlitions o f ' n a t ~ ~ rItmg~~age.
al such. then the following action is obligatory (or permissible etc.). ..".
It, is usual t o i ~ n ~ t r n .thr s t I ~ P Ws~lhj~'ctivitj
of opinion with t,he But there will be neither permission nor obligation. unless some
i t y "tr11t11.h11t. as w v have scen i l l n l l ottr investi-
L ' o l ~ j c r ~ t , i ~ ,>I' wantings are distinguished from others in such a way that theic
gations into t,lleorrticnl philosophy, the trut,h of sentr:nrm always satisfaction is required to be taken into consideration (also by
h i ~ st o h r a h~lrnanacc:omplish~nr~nt a n achievcn~mtof persons. every other person concerned). I t may turn out t,hat some elements
No prrson can d o more t.han try to rlvercornr his s~ihjectivity, of the situation finally rule out the satisfaction-but "prima
and t,his is t h r niln of Inpiral cliscipline. facie" (or a s a "rule of thumb") i t has to be considered. Thr
principle of transsubjectivit,,y demands that all n~errlysubjoctivr
It nmy I r r hc,lpf~~I t o h a w a trrrn which A h in wit11 our philoso-
wantings are t o be rnled ont of consideration. Rut we still have the
phi(.al tradifions: so 1 prrqxnse (1 aln, for the rl~onl,mt,still restrict-
problem of distinguishing between those wantings which are and
ing o11r < ~ m c < wtot (.IN, t , t ' ~ ~ 01' t , l ~S , Y ! ~ O I I I , I ~ S )t o ( d l I,IIP rc?({~lired
those wantings which are not "merely subjective".
ov<.ru~rv~ing cnl' s ~ ~ l ) , i ~ ~ ~ "L~JLI~s<.(-II,~,~II<~<!
~livily of ~ ~ ~ J ~ j o ~ ~ L i v i l y " - o r
I n order t o have simpler terms than "n~erelvsubjective" and
" t r i ~ ~ ~ s ~ ~ ~ l , j ~ ~ ~ I: t . i sv hi tLy ." .'I-Ilis is st-ill st~hj,:,.tiviLy, tnlt a $<
transsnbjoctive", I propose to revive a rather oldfashioncd term.
s l j t i t I I is I I 'i s I I i ~ ~ i s - a ntriw ~ l ~ , I Io \ w -
Let me call the merely subjective wantings "inessantial". The*
cotrw thc!~~t. ' I ' r i x r ~ s s ~ ~ l , , i ~ ~ ~is~ n<,t
t i v i t;Ly f w t , i n ~ itL id not n p o s t ~ ~ l a t o
wantings which are left, even when one tries to be as transsrthjec-
either. Trr~r~ssulrjt:cti\~ity in si111p1ya ttmn churnoterizing that
tive as possible, are then to be called "essential" wantings. To
activity in rvhi?t) wr* rlrr rll\vil,vs rxlroa<ly involvr~lif \re hegin t o accept a wanting a s essential is, thereby, defined as a c c q h n g the
r.lvLsr,n 1st all.
obligation t o take it into consideration. Further, since it may bo
If we f o r r n n l a t ~the itrrperativr smtonce, "Let ria transcend the case t h a t someone doesnot actually want what would be an essen-
our subjertivit,y !", we have 11c1.ca "rzcrrn" which rnakes i t possible tial wanting if only he did want it, there is rcaqon for introducing
fbr 11s to justify all ot,ht:r not.rns (norrns in the ordinary sense). another traditionally well-known term, "need" (..Bedurfnis"). If,
T r a n s s ~ ~ b j w t i v iis, t , ~t,liet'ofi~re.a "supr:rnorrn"; it in n "first norm" under certain conditions, a wanting t o do A would be a n essentinl
or, as the Latin tern, goes, 11 "[winciplo". We may cornparc the wanting if it were wanted a t all, I proposc to say that there is a
principle of transsubjectivity t o a cnnstitntion which makes laws "need" t o do A .
possible. But since this principle is the basis not only of all law Of course, this added terminology does not solve the problem
but of all moralit,y. I propose t,o call i t for s t , "tlre moral of how we are t o decide whether there is a need for a "given"
principle" ( I h s Moralprinzip). wanting. But we now have a terminology which is adeqnate for
Looked at, histor,ically, thc moral 1n.ir1ciplc of trnnssubject,ivity formulating what can be said "in the abstract" (i. e., without
is, of collrse, nothing ot,hrr t,l~ana r c f o r ~ n ~ ~ l a t of i o nKANT'S"cute- dealing with any particular "concrete" situation) abont the
gorical imperat,ive". Further, the s a n ~ ec n ~ c i a lrj~lwt,ionwhich distinction between essent,ial and inessential wantings. (This claim
~ r i s o swith KANT'S"fonnnl" ethios also arises hero: r w n if n person will have t o be justified in the following pages.)
rncognizos t,hat transsuhjert.ivity is a nel.essnry colditim of moral Again we begin with practice. This means, as we have scrn, that
reasoning, he can still ask whether this one moral principle is also we are always already involved in the discussion of plans. \Vo arc
sufficient. for moral masoning. We can answer this q ~ i ~ s t i oonly n not only "all along" ( H ~ D E G Q E,,immer
R: schon") talking with
84 Normatioe Logic and Ethics 7. Foundations of Partical Philosophg 85

one another, but also we are "all along" acting together, cooperat- The tern "genesis" can mean the "history" of a situation, i. e..
ing with one another. Some wantings are "all along" accepted as a detailed answer to the question asking how the situation factu-
needs. For example, a s children, all of us could grow up only ally came into being. We can call the answer to this question, the
because adults took care of our needs. All of us, therefore, know "factual genesis" of a situation. Part of the factual genesis is a
t h a t we are very often in need.
,'causal explanation" of the situation, i. e., an account which
Moreover, we human beings, unlike animals, anticipate our offers hypothetical universal laws by virtue of which the given
future needs, because we know them via language. Therefore, if situation can be derived from some-hypothetically assumed-
we try t o satisfy our present wantings just as they occur t o us. earlier situation.
we find ourselves worrying that we might be jeopardizing the Historical and causal questions are fully appropriate and suffi-
satisfaction of future needs. cient if we are dealing only with natural phenomena. Rut a con-
The needs which we share with animals may he called provisio- crete situation involving human beings includcs not only natural
nally "natural" needs; they are essential wantings for us insofar wantings but also cultural wantings. And when cult,ure is relevant.
as we are natural beings. But there are also other needs. Now then another kind of genetic question becomes relevant: Why was
t h e task of practical philosophy is t o establish principles which this brought about? What aims were sought? What was achieved
give us a method of deciding practical questions in our present and where did the attempts fail? What is asked for here will be
situation. And t h e very existence of language-and therefore the called a "normative genesis".
exintence of norms for our linguistic activities-shows that we I n order t o justify "normatively genetic" questioning (or, t o
cannot consider ourselves as merely natural beings. I therefore be more precise, the imperative to ask such questions") one need
propose-once more provisionally-to use the term "cultural" only realize that "culture" finally has t o be distinguished from
t o refer t o all essential wantings which cannot be called "natural". "nature" just by being "brought about": it is a human achieve-
So we may speak both of "cultural needs" and of "natural needs". ment. We can become clearer ahout what kind of genesis this is
B u t it still has t o be shown how, in a concrete situation, any if we look more critically at the ways in which cultural needs are
wantings-natural or cultural-are t o be distinguished as either discussed. Then we will be able t o arrive a t a more detailed pre-
essential or inessential. One cannot say, i n abstracto, which norms scription of an appropriate method for dealing with cultural needs.
will be justified in a concrete situation; too much depends upon Before we can discuss actions in a concrete situation, we have
t h e particularities of the situation. But we can say, i n abstracto, first t o be able t o "describe" the situation. I will call any such
t h a t if any norms are uncritically accepted in a concrete situation. description an "abstract situation". The "abstract situation"
there is always the risk of deciding against our needs. The only serves as a model for the concrete situation. But the choice of a
way open to us b y which we can overcome the limitations of our model already presupposes an anticipated evaluation of the concrete
own subjeotivity, i. e., of the uncritical acceptance of norms, is situation, since the construction of a model involves decisions
t h e attempt t o become more transsubjective. about whether certain elements in a possible description are
.< relevant" or "irrelevant". Therefore. I propose that the term
B u t how do we know that all this is not mere verbiage? I s it
n o t merely "formal',l" to repeat the moral principle of transsubjec- "model" be understood as a description of which one claims that
tivity over and over again? i t contains only "relevant" elements of the concrete situation.
Following the history of practical philosophy which leads from Here once again it becomes necessary to apply the moral
KANT'S"categorical imperative" t o M s x ' s "dialectics", I would principle of transsuhjectivity, for one must be guided by this
like t o suggest a 'further principle' which provides, still i n principle if he is t o overcome his own subjective illusions about
abstracto, the "materia" for the "forma" of the moral principle. a situation. Such a guide is especially important if one is himself
This second principle, fist of all, requires us t o look a t the a relevant part of a situation.
genesis of a concrete situation before deciding what t o do or not The only way we can avoid arbitrariness in our choice of a model
t o do. is t o begin with the explicit admission that no element of a con-
84 Normatioe Logic and Ethics 7. Foundations of Partical Philosophg 85

one another, but also we are "all along" acting together, cooperat- The tern "genesis" can mean the "history" of a situation, i. e..
ing with one another. Some wantings are "all along" accepted as a detailed answer to the question asking how the situation factu-
needs. For example, a s children, all of us could grow up only ally came into being. We can call the answer to this question, the
because adults took care of our needs. All of us, therefore, know "factual genesis" of a situation. Part of the factual genesis is a
t h a t we are very often in need.
,'causal explanation" of the situation, i. e., an account which
Moreover, we human beings, unlike animals, anticipate our offers hypothetical universal laws by virtue of which the given
future needs, because we know them via language. Therefore, if situation can be derived from some-hypothetically assumed-
we try t o satisfy our present wantings just as they occur t o us. earlier situation.
we find ourselves worrying that we might be jeopardizing the Historical and causal questions are fully appropriate and suffi-
satisfaction of future needs. cient if we are dealing only with natural phenomena. Rut a con-
The needs which we share with animals may he called provisio- crete situation involving human beings includcs not only natural
nally "natural" needs; they are essential wantings for us insofar wantings but also cultural wantings. And when cult,ure is relevant.
as we are natural beings. But there are also other needs. Now then another kind of genetic question becomes relevant: Why was
t h e task of practical philosophy is t o establish principles which this brought about? What aims were sought? What was achieved
give us a method of deciding practical questions in our present and where did the attempts fail? What is asked for here will be
situation. And t h e very existence of language-and therefore the called a "normative genesis".
exintence of norms for our linguistic activities-shows that we I n order t o justify "normatively genetic" questioning (or, t o
cannot consider ourselves as merely natural beings. I therefore be more precise, the imperative to ask such questions") one need
propose-once more provisionally-to use the term "cultural" only realize that "culture" finally has t o be distinguished from
t o refer t o all essential wantings which cannot be called "natural". "nature" just by being "brought about": it is a human achieve-
So we may speak both of "cultural needs" and of "natural needs". ment. We can become clearer ahout what kind of genesis this is
B u t it still has t o be shown how, in a concrete situation, any if we look more critically at the ways in which cultural needs are
wantings-natural or cultural-are t o be distinguished as either discussed. Then we will be able t o arrive a t a more detailed pre-
essential or inessential. One cannot say, i n abstracto, which norms scription of an appropriate method for dealing with cultural needs.
will be justified in a concrete situation; too much depends upon Before we can discuss actions in a concrete situation, we have
t h e particularities of the situation. But we can say, i n abstracto, first t o be able t o "describe" the situation. I will call any such
t h a t if any norms are uncritically accepted in a concrete situation. description an "abstract situation". The "abstract situation"
there is always the risk of deciding against our needs. The only serves as a model for the concrete situation. But the choice of a
way open to us b y which we can overcome the limitations of our model already presupposes an anticipated evaluation of the concrete
own subjeotivity, i. e., of the uncritical acceptance of norms, is situation, since the construction of a model involves decisions
t h e attempt t o become more transsubjective. about whether certain elements in a possible description are
.< relevant" or "irrelevant". Therefore. I propose that the term
B u t how do we know that all this is not mere verbiage? I s it
n o t merely "formal',l" to repeat the moral principle of transsubjec- "model" be understood as a description of which one claims that
tivity over and over again? i t contains only "relevant" elements of the concrete situation.
Following the history of practical philosophy which leads from Here once again it becomes necessary to apply the moral
KANT'S"categorical imperative" t o M s x ' s "dialectics", I would principle of transsuhjectivity, for one must be guided by this
like t o suggest a 'further principle' which provides, still i n principle if he is t o overcome his own subjective illusions about
abstracto, the "materia" for the "forma" of the moral principle. a situation. Such a guide is especially important if one is himself
This second principle, fist of all, requires us t o look a t the a relevant part of a situation.
genesis of a concrete situation before deciding what t o do or not The only way we can avoid arbitrariness in our choice of a model
t o do. is t o begin with the explicit admission that no element of a con-
 86 ATormaliueLogic and Ethics

Crete situation shall be ruled out a t the beginning. This means

which, since FICHTE, has been known as a "dialectical" triad,
t h a t we have t o look a t each situation a s a n open complex. It is
leading from a "thesis" C; via a n "antithesis" Cz t o a "synthesis"
assumed to be a complex, because more than one element may
C = (Cl, Cz).
be relevant. It is a n open complex, because the number of relevant
The term "dialectical" has been used specifically for that kind
elements is initially unrestricted. Of course, in a n y particular
of genetic account which is appropriate for the consideration of
dialogue one can admit only a finite number of elements which
practical problems. We therefore could use the term "dialectical',l"
are claimed t o be relevant, and this means, in particular that
for the imperative to look a t every concrete situation in just this
only a h i t c number of wantings can be considered essential.
nornlatively genetic fashion. The specific charactcr of this super-
But t o say t h a t a concrete situation is t o be looked a t as an norm is that we t ~ ; ~ vtoe jualily the genesis from ( C ; , . . . , C,_l)
open complex only serves to remind us that we a l r e d y have to be via (C;, . . . , C;..,), C; to (GI, . . . , C,). This justification is
transsubjective in our description of a conrrete situation. For our neither simply a n historical report nor a causal explanation;
purposes now, it is even more important that we realize that we it is a justification gnided by the moral principle of transsub-
can work with a model only if we look for a "normative genesis" jectivit,y.
of the concret,e situation as it is described by t,he model of it which As the term "dialectical" nowadays accnrs very often ss a
we construct. There will be some wantings which will have to be merely fashionable word without a reasonably fixed nsage, I wollld
accepted as needs without our going into the qliestion of a' nor- like t o mention, that the term is irrelevant here. It would 110
mative genesis, a n d these will he what I have termed "natural" sufficient, t,hough more cumbersome, to speak of' nor~nativeganesis
needs. (If some needs were not accepted as nat,ural, there would only. As normative genesis distinguishc!~crrlt~zral from natural
be no ending of a n argument.) So, t,here will be mainly the prob- needs, it seems to be appropriate t o use the phrase "the cultural
lem of deciding which wantings in the complex nnder conside- principle" (Das Kulturprinzip) for the super.norm to give a nor-
ration arc t o be accepted as cultural needs. mative genesis for norms.
Srhemat~icdly,we will have a complex (CI, . . . G , ) of wantings More important,, it seems t o me, is not Lo become too rigid in
which are claimed to be cultural needs. If we assnme that the prescribing the schemata of this prin?iple. They may ha a p p l i d
complex (CI. . . . C,-1) is already justified, a n attempt to justify in many variations.
C , could lead t o nnending circles; for one might have t,o rely on Instead of working out furtller schemata. I'll give two examples,
C , a s alrcady accepted in onler t o justify any one of C1. . . . where t,he c ~ d t l ~ r aprinciple
l is applie<l. Thr first exarr~ple is
Rut a "normative genesis" provides a non-circdar method of adapt.ed f r o ~ nCONDORCET.
justificat,ion (and the only one so far proposed and practiced). The Since t h r prehistory of mankind we find nomsxls forrerl to be
complex (C, , . . . C,) is considered -once more rather schemati- always on their way, because they have to find food for their
cally -as a n abstract situat,ion which has developed from a simplet animals (C;). On the other hand the tradition of gathe~,ingfruits
abst,ract sitrlntion ( C ; , . . . CL_l). Hrre C;, . . . (;lr;-, musk not be may have developed-say in a group of people who nlostl?
identical with Cj . . . . C,-l b111~ t ~ ; ~ , - b"earlir:rstngo"ofthclatt.~.~..
ran lived by fishing-into a primitive stage of agriculture (C;). If
Then a new want'ing, C;, is added t o the complex (6;,.... C;-,) now a t some place a fishing group was joined by a group of nomacis,
which is thereby transformed into a new situation (C1, . . . C,). . there was a chance of a synthesis: cattle-breeding together with
Schematically, we have a developed agriculture (such that enough food for tho aninlaln
was produced) resulted in a new type of economy: a p i c r ~ l t , ~(~Cr)e
; .. . . 9 - 1 ) = C . ..,C , C, => (Cl, . . . , C") as we know it since the neolithicum.
Such an example does not belnng t,o philosophy -so the ade-
The simplest case of this scheme IS quacy of this normative genesis is irrelevant f r ~ our
r pnrpones. Ihlt
the next example will belong to philosophy itself: the normative
genesis of dialognes. Let us start a t a stage where there arc only
 86 ATormaliueLogic and Ethics

Crete situation shall be ruled out a t the beginning. This means

which, since FICHTE, has been known as a "dialectical" triad,
t h a t we have t o look a t each situation a s a n open complex. It is
leading from a "thesis" C; via a n "antithesis" Cz t o a "synthesis"
assumed to be a complex, because more than one element may
C = (Cl, Cz).
be relevant. It is a n open complex, because the number of relevant
The term "dialectical" has been used specifically for that kind
elements is initially unrestricted. Of course, in a n y particular
of genetic account which is appropriate for the consideration of
dialogue one can admit only a finite number of elements which
practical problems. We therefore could use the term "dialectical',l"
are claimed t o be relevant, and this means, in particular that
for the imperative to look a t every concrete situation in just this
only a h i t c number of wantings can be considered essential.
nornlatively genetic fashion. The specific charactcr of this super-
But t o say t h a t a concrete situation is t o be looked a t as an norm is that we t ~ ; ~ vtoe jualily the genesis from ( C ; , . . . , C,_l)
open complex only serves to remind us that we a l r e d y have to be via (C;, . . . , C;..,), C; to (GI, . . . , C,). This justification is
transsubjective in our description of a conrrete situation. For our neither simply a n historical report nor a causal explanation;
purposes now, it is even more important that we realize that we it is a justification gnided by the moral principle of transsub-
can work with a model only if we look for a "normative genesis" jectivit,y.
of the concret,e situation as it is described by t,he model of it which As the term "dialectical" nowadays accnrs very often ss a
we construct. There will be some wantings which will have to be merely fashionable word without a reasonably fixed nsage, I wollld
accepted as needs without our going into the qliestion of a' nor- like t o mention, that the term is irrelevant here. It would 110
mative genesis, a n d these will he what I have termed "natural" sufficient, t,hough more cumbersome, to speak of' nor~nativeganesis
needs. (If some needs were not accepted as nat,ural, there would only. As normative genesis distinguishc!~crrlt~zral from natural
be no ending of a n argument.) So, t,here will be mainly the prob- needs, it seems to be appropriate t o use the phrase "the cultural
lem of deciding which wantings in the complex nnder conside- principle" (Das Kulturprinzip) for the super.norm to give a nor-
ration arc t o be accepted as cultural needs. mative genesis for norms.
Srhemat~icdly,we will have a complex (CI, . . . G , ) of wantings More important,, it seems t o me, is not Lo become too rigid in
which are claimed to be cultural needs. If we assnme that the prescribing the schemata of this prin?iple. They may ha a p p l i d
complex (CI. . . . C,-1) is already justified, a n attempt to justify in many variations.
C , could lead t o nnending circles; for one might have t,o rely on Instead of working out furtller schemata. I'll give two examples,
C , a s alrcady accepted in onler t o justify any one of C1. . . . where t,he c ~ d t l ~ r aprinciple
l is applie<l. Thr first exarr~ple is
Rut a "normative genesis" provides a non-circdar method of adapt.ed f r o ~ nCONDORCET.
justificat,ion (and the only one so far proposed and practiced). The Since t h r prehistory of mankind we find nomsxls forrerl to be
complex (C, , . . . C,) is considered -once more rather schemati- always on their way, because they have to find food for their
cally -as a n abstract situat,ion which has developed from a simplet animals (C;). On the other hand the tradition of gathe~,ingfruits
abst,ract sitrlntion ( C ; , . . . CL_l). Hrre C;, . . . (;lr;-, musk not be may have developed-say in a group of people who nlostl?
identical with Cj . . . . C,-l b111~ t ~ ; ~ , - b"earlir:rstngo"ofthclatt.~.~..
ran lived by fishing-into a primitive stage of agriculture (C;). If
Then a new want'ing, C;, is added t o the complex (6;,.... C;-,) now a t some place a fishing group was joined by a group of nomacis,
which is thereby transformed into a new situation (C1, . . . C,). . there was a chance of a synthesis: cattle-breeding together with
Schematically, we have a developed agriculture (such that enough food for tho aninlaln
was produced) resulted in a new type of economy: a p i c r ~ l t , ~(~Cr)e
; .. . . 9 - 1 ) = C . ..,C , C, => (Cl, . . . , C") as we know it since the neolithicum.
Such an example does not belnng t,o philosophy -so the ade-
The simplest case of this scheme IS quacy of this normative genesis is irrelevant f r ~ our
r pnrpones. Ihlt
the next example will belong to philosophy itself: the normative
genesis of dialognes. Let us start a t a stage where there arc only
 I
88 Normalive Logic and Ethim 7. Foundations of Partieal Philwophy 89
i
material dialogical games without using logical particles, i. e. dia-
logical gamcs using only elementary sentences and rules for the
( famous "hermeneutic circle", too-should be a spiral. As no other
term seems to be a t hand I would like to call this spiral movement
occurring prcdicatol.n (C1'). As a next step we may imagine a / for constructing normatively genetic accounts the "dialectical"
primitive use of logical particles (Cz'), separat,ed from the dialo- spiral. "The dialectical spiral" is not a descriptive term for a way
gical games. The synthesis (C) is a stage in which material dialogues people think, but i t is required to think this way in order to satis-
with logical partit:les (61)and formal dialogues (Cz) are f d y fy the cultural principle.
developed. The principle of constructing a normative gencsis allows us to
The culti~ral principle requires that, before we can come to justify "cultural" needs. At the same time, claims that some want-
a practical decision in a concrate situation, we must construct a ings are "natural" needs can he refuted-namely by constructinga
norn~ativegenrsis of a nwdel of the concrete s i t ~ ~ a t i oAnd
n . here, normative genesis which "explains" those wanlings (which are
a normative genesis is a genesis such that each step of it can be claimed to be "natural") as merely resulting from factual failures
'
justified transsubjectively. to understand the normativegenesis. In this way it is the principle of
Tho only principle guiding our practical judgment still will be .,normative genesis" which allows us to distinguish between cultural
transsubjectivity. but now this criterion is set to work on all needs and wantings, which are merely claimed to be (either natural
the material provided by a concrete situat,ion. In order to apply or cultural) needs.
the principle, however, we have to work with abstract situations, Together, the moral and the cultural principle prescribe in
i. e., with models, and we have to construct a normatively genetic abstract terms how one can and should prepare practical decisions
sequence of such models, stretching from a merely "natural" in a concrete situation. They are not arbitrary principles; rather,
state up to the cultural stages of the concrete nituation being they are the only principles which make "reasonable" decisions pos-
considered. sible. The term "reason" can be introduced into t,he ortholanguage
As HI(:EL haS ~tressed,it would be a rationalist~cillusion for only as its keystone, i. e., at the very end, after the moral and the
us to believe that the construction of a normative genesis should cultural principle have already been formulated. Insofar as it con-
be carried out without a thorough investigation of the actual cerns practical problems, "reason" is called "pract,ical reason"
history of the concrete situation. Very often only historical know- and its meaning can be understood only after the moral and the
ledge provides 11s with an adequate "understanding", i. e. a trans- cnltural principle first have been practiced and then have been
subjective justification for cultural wantings which may seem a t formulated.
first to be unjustifiable. Hist.orical knowledge is unique in enabling The ortholanguage I have proposed, toget,her with the two
us to become more transsubjective. principles of practical reason, are all a critical reconstruction of tra-
The construction of a normative genesis thereby turns out, ditions of practical philosophy, which I have carried out as trans-
normally, to be a partial reconstruction of the history of the subjectively as I c o ~ ~ lId have
. proposed a normative genesis of the
concrete situation. Not all historical details are relevant for a abstract doctrine called "practical philosophy ", so that t.he method
normative genesis, but it is a reasonable guess that the essential 1 have proposed-how to teach the terminology and the prin-
steps which led to our present situation actually were taken (in ciples of practical philosophy -is nothing other than an applica-
a more or less obscure way) during its history. tion of just t,hese principles: the method is identical with its own
The construction of a normative genesis has to be done stepwise: result.
we have to start with some norms taken from immediate practice Normative logic and ethics is, thereby, a "closed system" of pure
a n d have t o apply these norms for a critical understanding of some philosophy -but open for all applications.
parts of history. From history we have to go back t o our genetic
account of norms, which now-hopefully-can be "improved".
Then we have to go to a critical understanding of history again etc.
This movement should not be a circle, but-as in thecase of the
 I
88 Normalive Logic and Ethim 7. Foundations of Partieal Philwophy 89
i
material dialogical games without using logical particles, i. e. dia-
logical gamcs using only elementary sentences and rules for the
( famous "hermeneutic circle", too-should be a spiral. As no other
term seems to be a t hand I would like to call this spiral movement
occurring prcdicatol.n (C1'). As a next step we may imagine a / for constructing normatively genetic accounts the "dialectical"
primitive use of logical particles (Cz'), separat,ed from the dialo- spiral. "The dialectical spiral" is not a descriptive term for a way
gical games. The synthesis (C) is a stage in which material dialogues people think, but i t is required to think this way in order to satis-
with logical partit:les (61)and formal dialogues (Cz) are f d y fy the cultural principle.
developed. The principle of constructing a normative gencsis allows us to
The culti~ral principle requires that, before we can come to justify "cultural" needs. At the same time, claims that some want-
a practical decision in a concrate situation, we must construct a ings are "natural" needs can he refuted-namely by constructinga
norn~ativegenrsis of a nwdel of the concrete s i t ~ ~ a t i oAnd
n . here, normative genesis which "explains" those wanlings (which are
a normative genesis is a genesis such that each step of it can be claimed to be "natural") as merely resulting from factual failures
'
justified transsubjectively. to understand the normativegenesis. In this way it is the principle of
Tho only principle guiding our practical judgment still will be .,normative genesis" which allows us to distinguish between cultural
transsubjectivity. but now this criterion is set to work on all needs and wantings, which are merely claimed to be (either natural
the material provided by a concrete situat,ion. In order to apply or cultural) needs.
the principle, however, we have to work with abstract situations, Together, the moral and the cultural principle prescribe in
i. e., with models, and we have to construct a normatively genetic abstract terms how one can and should prepare practical decisions
sequence of such models, stretching from a merely "natural" in a concrete situation. They are not arbitrary principles; rather,
state up to the cultural stages of the concrete nituation being they are the only principles which make "reasonable" decisions pos-
considered. sible. The term "reason" can be introduced into t,he ortholanguage
As HI(:EL haS ~tressed,it would be a rationalist~cillusion for only as its keystone, i. e., at the very end, after the moral and the
us to believe that the construction of a normative genesis should cultural principle have already been formulated. Insofar as it con-
be carried out without a thorough investigation of the actual cerns practical problems, "reason" is called "pract,ical reason"
history of the concrete situation. Very often only historical know- and its meaning can be understood only after the moral and the
ledge provides 11s with an adequate "understanding", i. e. a trans- cnltural principle first have been practiced and then have been
subjective justification for cultural wantings which may seem a t formulated.
first to be unjustifiable. Hist.orical knowledge is unique in enabling The ortholanguage I have proposed, toget,her with the two
us to become more transsubjective. principles of practical reason, are all a critical reconstruction of tra-
The construction of a normative genesis thereby turns out, ditions of practical philosophy, which I have carried out as trans-
normally, to be a partial reconstruction of the history of the subjectively as I c o ~ ~ lId have
. proposed a normative genesis of the
concrete situation. Not all historical details are relevant for a abstract doctrine called "practical philosophy ", so that t.he method
normative genesis, but it is a reasonable guess that the essential 1 have proposed-how to teach the terminology and the prin-
steps which led to our present situation actually were taken (in ciples of practical philosophy -is nothing other than an applica-
a more or less obscure way) during its history. tion of just t,hese principles: the method is identical with its own
The construction of a normative genesis has to be done stepwise: result.
we have to start with some norms taken from immediate practice Normative logic and ethics is, thereby, a "closed system" of pure
a n d have t o apply these norms for a critical understanding of some philosophy -but open for all applications.
parts of history. From history we have to go back t o our genetic
account of norms, which now-hopefully-can be "improved".
Then we have to go to a critical understanding of history again etc.
This movement should not be a circle, but-as in thecase of the