The Isaiah Berlin Virtual Library
Logical Positivism Assessed
Review of Julius Weinberg, An Examination of logical positivism,
Criterion 17 (1937), 174–82
Dr Julius Weinberg’s survey of logical positivism is a methodical,
patient, cautiously-worded, uncommonly painstaking, piece of
work. The subject is of the first importance, and this is, so far as I
know, the first comprehensive account of it in any language; it
consists of an exposition, interrupted by occasional criticism, of
the views of several closely connected groups of contemporary
logicians whose agreement on fundamental issues is sufficiently
great to entitle them to be regarded as a single philosophical
movement. The rise, development and growing influence of logical
positivism is one of the most striking intellectual phenomena of
the present time: its doctrines are the subject of violent
controversy wherever philosophy is pursued seriously and for its
own sake: the immense passion which it has aroused, in both its
adherents and its opponents, is a guarantee of its genuinely
revolutionary character; consequently no one interested in the
history of thought, or the present condition of philosophy, can
afford to neglect this school, whatever he may think of its intrinsic
worth.
It is, as its name implies, a movement directed against all forms
of metaphysical or transcendental thought and expression. It is
rigorously empirical, and in this respect continues the central
tradition of English philosophy, counting amongst its ancestors
Berkeley, Hume and Mill, and their most famous German disciple,
Ernst Mach. It differs from its predecessors, in particular from
Hume, with whose main doctrines it possesses the closest affinity,
in rejecting the obsolete psychology, the loose and figurative
language, the frequent inconsistent intrusions of metaphysical
statements, above all the faulty treatment of the nature of
mathematics and logic, which mar the works of the fathers of
modern empiricism. The Humean doctrine which forms the
central tenet of all forms and brands of logical positivism is that
significant assertions can be of two kinds and only two: either they
1
LO G I CA L PO S I T I V I S M A S S E S S E D
are statements about matters of fact (the vérités du fait of Leibniz),
in which case they are empirical, contingent and verifiable, at any
rate in principle, by observation; or else they are necessary truths
(Leibniz’s vérités de la raison), in which case they are not indeed
verifiable, but only because they are, in a special sense of the word,
tautologies, which follow from definitions regulating the use of
symbols, and are true unconditionally, but on the other hand say
nothing about any matters of fact.
Any statement which claims to fall outside these two categories
will on examination, so it is alleged, be found to refer to nothing at
all, and so be literally meaningless. Into the first class fall the
majority of the statements of science and common sense. Into the
second the truths of mathematics, logic and all other deductive
systems. Into the third, i.e. into the ‘nonsense’ class, the majority
of the statements made by theologians and metaphysicians, which
have so long puzzled the layman, not because of their intrinsic
depth or complexity, but because, while in fact meaningless, they
are expressed in grammatically correct sentences which make it
difficult at first to expose their real nature, betray the reader into
believing that they possess literal significance, and so lead to a host
of verbal obscurities and pseudo-problems which have too long
passed for profound philosophical utterances.
Some languages, e.g. German, lend themselves better than
others to this kind of mystification: and the fact that many
metaphysical pseudo-statements are expressed in phrases which
are not only grammatically correct and beautiful in themselves, but
also emotionally expressive and evocative, and thus move and
communicate as poetry or rhetoric do, helps to conceal their total
lack of literal significance. Philosophers (the argument continues)
who study such sentences professionally have in the past too
frequently been trapped into enquiring into their ‘meaning’. Since,
in fact, they have none, this has led to various paradoxes and
confusions, the ‘elucidation’ of which, undertaken by persons
unable to free themselves from the magical properties of obscure
words, led in turn to fresh obscurities, at least as puzzling as the
problems which they claimed to solve. A systematically misleading
terminology grew out this, which, by implying the existence of real
entities corresponding to the new technical terms, and so creating
a mythology of its own, made the truth (as Berkeley pointed out
long ago) even more inaccessible; not until the bluff of
2
LO G I CA L PO S I T I V I S M A S S E S S E D
metaphysicians was called, and the smoke screen of words finally
lifted, would progress be made. And this propaedeutic function
could only be performed by the careful analysis of the literal
meanings of words and expressions, and their classification into
one of the two exhaustive compartments indicated above.
Armed with this programme, the anti-metaphysicians
proceeded to operate on two fronts simultaneously. On the one
hand they began to examine, classify and clearly define various
types of proposition used in the sciences and in ordinary life. Here
the chief instrument of research was to consider how such
propositions could in principle be verified: with the result that
while some, which began by appearing unverifiable, later proved
tractable because it was found possible to reduce them without
residue to others which could be verified, some remained
recalcitrant to the end and had to be rejected, because, while they
appeared to state something, no conceivable state of affairs
seemed indicated by them: from which it followed that they were
literally senseless. The province of the formal proposition of logic
and mathematics has been no less intensively cultivated: this was
largely made possible by the remarkable renaissance of
mathematical logic in the second half of the nineteenth century, an
exciting and brilliant development whose apex was marked by the
publication of Principia Mathematica by Russell and Whitehead. This
celebrated monument to the beauty and power of logical method
represents an attempt to demonstrate the reducibility of the
propositions of mathematics to those of pure logic, unreinforced
by any non-formal additions: it was clear that its success would
greatly weaken the position of those philosophers who regarded
the intuitive basis of arithmetic and geometry as the best refutation
of the thesis that if propositions are significant and not empirical
they are not necessarily for that reason purely formal.
The task of the elucidators of this period was not wholly unlike
that of the great figures of the French Enlightenment: both
believed themselves to be engaged on the task of dissipating a state
of darkness largely maintained by stubborn obscurantists with a
vested interest in the prevailing ignorance. There are modern
parallels to Hume and Voltaire, Diderot and Condillac; the wit, the
indignation, the passion for the truth, the belief in the immense
potency of the new method are to be found in full measure in the
works of modern positivists.
3
LO G I CA L PO S I T I V I S M A S S E S S E D
The forces which they had to fight were very formidable. The
dominant metaphysical systems of the Continent had had a
curiously intoxicating effect even upon this intellectually abstinent
and unadventurous country: the currency of language became
enormously inflated, a premium was placed on vagueness,
suggestiveness, ‘literariness’ of style; the relatively precise language
of the eighteenth century was replaced by a resonant and often
genuinely eloquent rhetoric, rich in imagery and in a particular kind
of allusive ambiguity which obscured such genuine, strictly philo-
sophical, arguments as could sometimes be perceived beneath its
broad and turbid stream. This atmosphere was conducive to the
production of great masterpieces which, as in the case of the great
Hegelians, or of Nietzsche or Schopenhauer, etc., expressed a
profoundly interesting and original attitude to life, with a depth of
spiritual and intellectual emotion which gives their work a place in
literature. To philosophy, interpreted as rigorous thought, the
result was ruinous: there is an exceedingly vivid and even lurid
description of the chaotic and degraded state of German
philosophy during the 1830s and 1840s in the introduction of Karl
Marx’s German Ideology. Something similar, although in a weaker
degree, was beginning, after the inevitable time-lag, to occur in
England.
It was as a natural reaction against this state of intellectual
dissipation that certain philosophers both in England and in
Germany introduced a new and stricter discipline of thought and
expression, and began to write with that meticulous and
undeviating accuracy which characterises the prose style of writers
so utterly different and even sharply opposed in every other
respect as Edmund Husserl and G. E. Moore, in the beginning of
the twentieth century. Moore’s influence is particularly notable: the
lucidity, sharpness and infinite intellectual honesty with which he
began to conduct his minute and exhaustive analyses of current
concepts and expressions introduced a new method and set a new
standard in English philosophy. He put forward the now
celebrated view that the main task of philosophy was not to
discover new facts about the world, which was done by the
sciences or ordinary experience, whose findings philosophers, as
such, could neither confirm nor refute; but to discriminate and
LOGICAL POSITIVISM ASSESSED 5
4
LO G I CA L PO S I T I V I S M A S S E S S E D
elucidate what exactly is meant by technical phrases and
ordinary expressions where obscure or imprecise use confounded
philosophers or ordinary men – an in a sense Cartesian activity
which by precept and many convincing examples he has made into
the most powerful instrument in use at the present time for the
discovery of the truth.
The corresponding movement on the Continent confined itself
more narrowly to the analysis of mathematics and of logic. It was
largely due to the writings of Russell and of Wittgenstein that the
two tendencies – the treatment as empirical of propositions of fact
and the formalisation of the rest – came together and were seen to
support each other. If parts are to be assigned, it is to Russell more
than to any other philosopher that the present movement owes its
existence. If Moore’s relentless and scrupulous integrity, his
concentration and his precision, have had a greater indirect
influence on what is most serious in contemporary philosophy, it is
Russell’s amazing virtuosity, the fertile stream of new and original
suggestions, which seemed to flow effortlessly to the advantage of
the new philosophy until it became deflected in other directions,
that provided the material and the stimulus for the later work of
others.
Dr Weinberg’s book begins after the end of what may be called
Russell’s early period, and conducts us to the present day. The
greatest and most original philosopher of this generation is Ludwig
Wittgenstein, whose Tractatus Logico-Philosophicus is a work which
has radically altered the prevailing views on the relation of
language and fact, which for him, and for logical positivism in
general, constitutes the subject-matter of philosophy. Dr
Weinberg’s chapter on the Tractatus is much the best in his whole
book: his exposition is careful and lucid, and the objections which
he has collected or discovered are extremely formidable.
Wittgenstein appears to maintain:
1 That the meaning of a proposition about the world is the
means of its verification.
2 That what cannot be expressed clearly cannot be expressed at
all: that only atomic propositions, or propositions analysable into
them (and tautologies which follow from nominal definitions) can,
in his sense, be expressed clearly.
To which Dr Weinberg (following others) objects (a) that this
entails that such expressions as are commonly described as general
5
LO G I CA L PO S I T I V I S M A S S E S S E D
propositions, e.g. the causal propositions of the sciences, fail on
both counts, since they are neither conclusively verifiable nor
reducible to a finite number of atomic propositions. They are,
nevertheless, bona fide propositions, since they can be conclusively
falsified (Dr Weinberg seems to me mistaken in thinking that all
general propositions can be falsified: but his point holds even if
only some are).
(b) That a proposition of logic – e.g. Wittgenstein’s own bold
and highly original doctrine that a relation between objects can
ultimately be indicated not by a symbol but only by a relation
between symbols, so that the symbolic structure of the expression
‘shows forth’ but does not in the strict sense ‘symbolise’ the
structure of the fact, the meaning being expressed by means of a
kind of ‘pictorial’ relation – cannot itself, being neither empirical
nor entailed by explicit definitions, be clearly stated at all, but only
‘shown forth’. Since logic largely consists of statements about
types of symbolisation, it would, if this were true, not be
expressible in the strict sense. This equally applies to the various
attempted analyses of general propositions – whatever the solution
may be, they cannot in the strict sense be stated; which creates a
philosophical puzzle of the first order. Dr Weinberg, who sees the
difficulty only too clearly in the case of Wittgenstein, appears to
think that it has, by means of appropriate symbolism, been at least
half removed by Carnap: his arguments for this are no more
convincing than those of Carnap himself.
(c) That the definition of meaning in terms of verifiability leads
to solipsism of a peculiar kind. Since all empirical propositions are
reducible to atomic propositions about my experience, all my
propositions refer ultimately to me, even when I appear to be
referring to physical objects or other selves. The ‘me’ they refer to,
not being contrastable with any ‘you’ or ‘it’, is very similar to
Kant’s transcendental ego. This is queer enough in any case in an
empirical philosophy; cannot be expressed; and leads to an
inevitably solipsistic universe, in which I am compelled to analyse
propositions about ‘your headache’ behaviouristically, into my
observations or my expectations in connection with the set of
sensible appearances which is all that is meant by ‘you’ or ‘your’.
This is the old egocentric predicament translated into verbal rather
than metaphysical terms. But, as Dr Weinberg points out, it is an
unplausible account of what either ordinary men or scientists mean
6
LO G I CA L PO S I T I V I S M A S S E S S E D
by physical objects or by communication, or by other selves. Dr
Weinberg agrees that Carnap has not finally solved this crux.
Rather diffidently, he appends a solution of his own at the end of
his volume; but I cannot see how, by recommending that we take
trans-subjective communication for granted, construct our
symbolism to reflect this, and then start from this higher level
taking the initial problem as solved, Dr Weinberg is doing anything
but avoiding the answer to the question altogether. If the paradox
is a genuine one, and Dr Weinberg seems to think so, since he
quotes it against Carnap, adoption of this or that special
symbolism cannot possibly remove it.
This section of the book is stimulating and valuable: after it
both author and reader begin to plod; steadily, and not without
profit. The discussion of probability is too discursive and rather
colourless; but anyone who has himself been concerned with that
difficult subject, whose present condition is more than
unsatisfactory, will appreciate the importance of Dr Weinberg’s
distinction between its three types: (a) mathematical or a priori
probability, a branch of mathematics or formal logic, and relevant
to logical positivism only in so far as such followers of
Wittgenstein and Waismann apply it to the propositional calculus,
by co-ordinating atomic facts with the series of natural numbers
and then describing compatibility, or the partial overlappings of
molecular propositions, in numerical terms; (b) scientific
probability, when an arithmetical schema is applied to empirical
data whose quantitative properties and modes of behaviour are
taken as given. How these are established, or rather what is meant
by saying that they are established, neither Dr Weinberg nor
anyone else has yet adequately explained. (c) Probability in sciences
which do not allow of quantitative treatment, e.g. psychology,
anthropology, history, detection etc., where one speaks of evidence
which makes this or that conclusion more or less probable, but not
of precise degrees of probability, because it seems nonsensical to
do so.
It is perhaps the inherent vagueness of such sciences that leads
to the popular view that they are in some special sense the concern
of philosophy, i.e. that philosophy takes charge of everything
rejected, because it is unquantifiable, by the exact sciences. If so, it
is important to destroy this illusion; and Dr Weinberg’s analysis of
probability is a means towards it; but in so far as he hints, without
7
LO G I CA L PO S I T I V I S M A S S E S S E D
developing his thesis, that these distinct senses of probability are
nevertheless species of one single general sense, he seems to
promise more than he can give, and fails to solve the problem.
His treatment of induction is far less satisfactory. He appears to
favour the pragmatist analysis of causal propositions, but
unaccountably fails to mention Ramsey, who gave the classical
account of it, and quotes instead the paler words of Dr Feigl.
Causal propositions, according to this view, are not propositions at
all, and so neither true nor false, but rules which apply to a
specified range of hypothetical situations. Dr Weinberg does not
mention the stock objections to this view, e.g. in what sense
scientific laws, if they are not true or false, can be rendered
probable, refuted by crucial experiments, or indeed logically differ
from ordinary hypotheticals (‘I am always unhappy when I am
hungry’), which seem either true or false, and are killed by a single
negative instance. Nor is there any reference to the objections to
phenomenalism which, analysing as it does singular-seeming
statements about physical objects into complexes containing causal
propositions, seems highly relevant to the issue.
After some valuable remarks on the distinction between the
older pragmatism, which equates truth and usefulness, and logical
positivism, which reserves truth for the verifiable and pragmatises
only the unverifiable, he deals with Carnap, for whom he reserves
his warmest admiration. Oddly enough, it is here that he is at his
weakest: he gives a lucid account of the celebrated distinction
drawn by Carnap between the formal and material mode of
speech, the first consisting of definitions and tautologies, the
second of empirical statements, which between them exhaust the
possibilities of language, and when confused, as in ordinary
speech, lead to the pseudo-paradoxes of metaphysics; but his
efforts to show that Carnap has done something towards
removing the difficulties inherent in Wittgenstein’s doctrine cannot
be judged successful; nor is his own solution, stated in outline and
very half-heartedly, of any help. The whole last section of the book
is written in a singularly unemphatic, lifeless style, which imparts
drabness to a subject whose proper nature is to be immensely taut
and sharp, moving drily but dramatically towards some decisive
climax: Ramsey’s essays are admittedly unlike this: his method was
to think aloud informally; where the thought is of such quality, this
is a more than adequate substitute for more formal writing.
8
LO G I CA L PO S I T I V I S M A S S E S S E D
There are, apart from numerous misprints, one or two textual
errors, for which Dr Weinberg is probably not responsible: e.g.
‘intentional’ for ‘intensional’, the omission to add ‘but aRc only
follows for aRb and bRc if R transitive’ (p. 63), and the repetition
of ‘~(∃ x )~φ x ’ for what? possibly ‘~(∃ x )φ x ’ (p. 131).
Having said so much in criticism, I should like to add that in
my opinion the merits of the book greatly outweigh its defects: the
subject is of paramount importance, the treatment is well
documented, highly informative and, broadly speaking, quite
accurate; its pace slackens as it progresses, and, apart from the
excellent chapter on Wittgenstein, it is not conspicuously original.
It is, in fact, a useful, exceptionally intelligent critical account of a
major subject, which is a rare phenomenon, not to be greeted with
light or grudging praise.
Copyright Isaiah Berlin 1937
Posted 5 January 2012
