Philosophy and Philosophers - an Introduction to Western Philosophy - Chapter 10
back to ...
or ...
CHAPTER TEN
Logical positivism and falsificationism: Ayer, Popper
It is perhaps unnecessary to make any connection between A.J.Ayer
and Karl R.Popper other than to point out that they both had great
influence on Western philosophy during the middle part of this
century, an influence that has continued to this day. However, a
common historical and intellectual connection is the Vienna Circle;
this was a group that met in Vienna during the 1920s and 1930s and
developed the philosophy of logical positivism, which was intent on
setting philosophy on a sure footing so that the scope of its tasks was
clear. Logical positivism, by way of a theory of meaning, involves the
elimination of much of traditional philosophy, in particular
metaphysics and also theology, as literally meaningless. What this
amounted to was the view that the investigation of any substantial
facts about the world was the province of science alone, not
philosophy, which could be concerned only with conceptual
elucidation and the linguistic task of precise definition. Both Ayer
and Popper attended the meetings of the Vienna Circle, but whereas
Ayer initially became a powerful advocate of its views, Popper,
although deeply interested, like the Vienna Circle, in the philosophy
and methodology of science, was critical of logical positivism.
Popper aims to demarcate science from non-science so as to
understand better the nature of scientific knowledge. Non-science
includes pseudo-science: areas which are not scientific but claim to
be so. It does not follow from this that what is non-science, including
pseudo-science, is thereby literally meaningless, as logical positivism
supposed, or even that it is untrue. Ayer has always had a great
interest in the problem of meaning, which Popper regards as a
largely fruitless field of philosophical investigation if regarded as an
end in itself. What perhaps unites Ayer and Popper, although they
259
are by no means alone in this, is their view that the heart of
philosophy is epistemology, and in particular the nature of empirical
knowledge.
Ayer
Alfred Jules Ayer (1910–89) was educated at Eton and Christ
Church, Oxford; his tutor in philosophy at Oxford was Gilbert Ryle
(1900–76). After graduating, he thought of going to Cambridge to
study with Wittgenstein; instead he went to study in Vienna in 1932
in order to find out more about the logical positivist philosophy of
the Vienna Circle. After a short period in Vienna he returned to
Oxford and became a lecturer in philosophy at Christ Church. In
1936 he published Language, truth and logic. While we must allow for
differences within the logical positivist movement, Language, truth
and logic states clearly what is essential to the doctrine of logical
positivism. In 1940 he joined the Welsh Guards and worked for
most of the war in military intelligence. He returned to Oxford in
1945 to become Dean of Wadham College. From 1946 to 1959 he was
Grote Professor of the Philosophy of Mind and Logic at University
College London. From 1959 until his retirement in 1978 he was
Wykeham Professor of Logic in the University of Oxford. In 1970 he
was knighted. Although he came to reject the most radical
proposals of logical positivism, Ayer remained a close follower of
the British tradition of empiricism and logical analysis. It was
Wittgenstein’s Tractatus that set Ayer on the course which led to
Language, truth and logic. However, the greatest influences on Ayer
were Russell and Hume. He continued to admire Bertrand Russell,
regarding him as probably the greatest philosopher of the twentieth
century; and, like Russell, he was an enthusiastic atheist. Ayer also
became interested in the American pragmatists, such as William
James (1842–1910). Again, like Russell, Ayer was a passionate
advocate of reason, and thought that intellectual honesty demanded
that we seek sufficient evidence for any beliefs that might be
proposed for acceptance.
The motivation for logical positivism stems from two connected
lines of thought: (I) the unity of science, and (II) the elimination of
metaphysics. In short, this amounts to the view that really all science
forms a single system; it alone is able to give true characterizations of
the nature of the world which can in the end be exhaustive. The unity
of science means that all branches of scientific inquiry have a
common epistemological basis: it is that determining the truth or
falsity of scientific theories about the nature of the world depends
entirely on an appeal to the evidence of experience and observation.
The elimination of metaphysics complements this, because
260 Logical positivism and falsificationism
metaphysics commonly supposes there is some way of determining
the nature of the world—perhaps its real or essential nature beyond
appearances—other than by an appeal to experience and observation.
The apparent assertions by metaphysics about the nature of the
world are, according to logical positivism, not true or false, but
nonsense—literally meaningless. With the elimination of metaphysics
as a source of knowledge about the world, science is unified as a
system of factual propositions, that is, statements whose truth or
falsity and, indeed, meaning depend on their being open to the test
of the facts of experience.
Propositions are what is determinately true or false: that is, they are
literally meaningful. Propositions are what literally meaningful
indicative sentences (sentences which grammatically appear to state
things) of any particular language express; this is important because
sentences of different languages can express the same proposition, as
in it is raining and il pleut. The criterion for a sentence is that it is
grammatically well formed, that is a necessary condition for it to be
meaningful, otherwise it is mere gibberish, such as “foot a fight will”.
The logical positivists argue that many grammatically well formed
sentences do not express genuine propositions, although being
grammatically well formed sentences they may appear to do so.
Sentences that appear to express a proposition, whether they do so or
not, Ayer calls putative propositions or statements. The logical
positivists argue that all genuine propositions are either analytic/
tautologies or verifiable by experience; statements—that is, indicative
sentences which appear to express propositions—which are neither
analytic nor verifiable by experience are literally meaningless or
nonsense. Sentences and statements that do not express genuine
propositions may be meaningful in some other way—they may have
poetic or emotive significance—but they are not literally meaningful. If
a statement is literally meaningless, then the question of its truth or
falsity cannot arise.
It has to be the case that a distinction is made between sentences
being meaningful in some broader sense than literally meaningful
because otherwise the criterion of literal meaningfulness would have
no possible application; in order to discover if a statement is analytic
or empirically verifiable, we already have to understand what it
means.
A sentence expresses an analytic proposition if, and only if, its truth or
falsity follows solely from the definition of the terms it contains. Thus
“All bachelors are unmarried” is analytic, since the predicate
“unmarried” is part of the definition of “bachelor”; establishing the
truth or falsity of the proposition consists in merely unpacking the
definition of its terms. The truth or falsity of analytic propositions
depends entirely on the meaning of the symbols in the sentence the
proposition expresses. Analytic propositions are true or false, and can
Ayer 261
be known to be so, a priori, that is, independently of the evidence of
experience; they are also devoid of factual content as they make no
claim about the world; their truth or falsity is compatible with any
evidence of experience whatsoever. That which is necessary is that
which must be and cannot be otherwise. If an analytic proposition is
true, it is necessarily true—it must be true and cannot be false. If an
analytic proposition is false, it is necessarily false—it must be false and
cannot be true. The denial of a true analytic proposition implies a
logical contradiction.
A sentence expresses an empirically verifiable proposition if, and only
if, some possible experience is relevant to determining its truth or
falsity. The truth or falsity of such empirically verifiable or factually
significant propositions cannot be determined merely by examination
of the definition or meaning of the symbols in the sentence the
proposition expresses. Thus “The cat is on the mat” is a factually
significant proposition; its truth or falsity does not follow from the
meaning of the terms it contains—it is not an analytic but a synthetic
proposition; its truth or falsity can only be determined a posteriori by
consulting experience. That which is contingent is that which may or
may not be: that which could be otherwise. If an empirically verifiable
proposition is true, then it is contingently true—it is true, but could
have been false. If an empirically verifiable proposition is contingently
false, it is false, but could have been true. The denial of an empirically
verifiable proposition never implies a logical contradiction.
The two classes of analytic and empirically verifiable statements are
mutually exclusive and collectively exhaustive of all literally
meaningful statements: they are the totality of genuine propositions.
That is, it is a necessary and sufficient condition for a statement to be
literally meaningful, and so capable of being true or false—a
proposition—that it be either analytic or empirically verifiable. Put
another way, a statement is a genuine proposition if and only if it is
either analytic or empirically verifiable, otherwise it is nonsense.
Metaphysics generally attempts to describe the essential structure of
reality: what the real world must ultimately be like according to
intellectual argument, although it may appear otherwise. Plato speaks
of fixed “Forms” beyond the flux of experience and space and time,
but accessible to the intellect, defining the “whatness” of things;
Leibniz speaks of non-spatial “monads” as the indivisible,
indestructible substance of the world which remain the same through
all natural change; Hegel speaks of the fully real as “The Absolute”,
the universe as ultimately a self-thinking totality. There are also
theological statements asserting the existence and nature of an eternal
transcendent God outside space and time.
Metaphysics, with theology, is eliminated as literally meaningless
because what it characteristically proffers as propositions are not
genuine propositions at all. The need to be clear about what are
262 Logical positivism and falsificationism
genuine propositions arises from the fact that we are misled by the
surface appearance of statements in metaphysics into thinking they
express propositions; but we know they do not express propositions
because they do not say anything whose truth or falsity can be
determined in the only two ways possible: by their being analytic or by
their being empirically verifiable. Metaphysics is disposed of not
because it is false, but because it is composed of statements which are
largely nonsense; it may appear to be composed of propositions—
statements that can be true or false—but really it is composed of
statements incapable of being either true or false because their truth or
falsity cannot be established even in principle by the only two ways
possible. If we are to say that any of the statements of metaphysics is
literally meaningful, then it must be translatable into statements that
are analytic or empirically verifiable. However, if a statement is
analytic, it tells us nothing about the world, and if it is empirically
verifiable, then it ceases to be a metaphysical statement at all, but
merely becomes part of the body of scientific theory testable by
observation. Neither translation is congenial to the metaphysician who
wishes to contend that his statements both say something about the
world—are factually significant—and cannot be settled by empirical
verification; but it is impossible, Ayer argues, that both these
conditions can be simultaneously satisfied. Indeed, metaphysics often
claims to speak of the world behind or beyond the world as it appears.
Either a statement says something about the world, in which case it is
empirically verifiable, or a statement says nothing about the world; no
statement can be about the world and not be empirically verifiable.
Therefore metaphysics, which purports to produce truths and
refutations of falsehoods about the nature of the world or reality in
statements which are empirically unverifiable, is impossible; it
produces only literal nonsense. Metaphysics makes only literally
meaningless assertions and raises spurious questions; it is, in short,
composed of meaningless pseudo-propositions which have the
appearances of genuine propositions. It follows that there can in reality
be no genuine disputes between metaphysicians: if “p” is a
metaphysical statement, it is literally meaningless, but then “not-p” is
also meaningless.
Logical positivism holds that all a priori propositions are analytic
and, although necessary, are necessary only because they are factually
empty: they say nothing about the world, but reveal only the
conventional meanings of words. All a posteriori propositions are
synthetic and contingent, but they are, whether true or false, factually
informative: they say something about the world. Contrary to the view
of a philosopher such as Kant, there can be no a priori, necessary
propositions that are synthetic. These considerations can be
summarized in the following diagram:
Ayer 263
All genuine propositions—that is to say, all propositions—fall into
either, but not both, of the shaded areas: A and D. No propositions fall
into the unshaded areas: B and C. There are therefore only two classes
of genuine propositions:
A: a priori/analytic/necessary.
D: a posteriori/synthetic/contingent.
All statements that fail to fall into the classes A or D are not propositions
at all; they are incapable of being true or false—they are not
literally meaningful—although they may be meaningful in some
other way.
Thus, in so far as metaphysics does contain literally meaningful
propositions, it consists either of analytic propositions, which tell us
nothing about the world, whose truth or falsity can be determined a
priori, or synthetic propositions, which do purport to tell us something
about the world, whose truth or falsity can be determined only a
posteriori. There is no special class of metaphysical propositions which
are at once a priori and tell us something about the world: no facts can
be known a priori.
All the statements of logic, mathematics and geometry express nonempirical,
non-factual, propositions, that are a priori valid and
necessary in virtue of their being analytic or tautologies: their truth
depends solely on the meaning of the symbols of which their
statements are composed. They are also devoid of factual content; the
reason such truths are necessary is just that they do not make any
assertions about the world that could be confuted or confirmed by the
evidence of experience. We do not have to suppose, in order to explain
our a priori knowledge of necessary truths, that the truths refer to some
realm of entities transcending experience. All a priori analytic truths—
including those of logic, mathematics and geometry—are not about
anything at all, but simply reflect the meaning we have chosen to give
to linguistic signs.
Philosophers such as Kant have argued that there is a special class
of propositions which are a priori synthetic and necessary. Kant
accepted that propositions such as “All bachelors are unmarried” are
analytic, necessary, their denial implying a contradiction; the concept
of the predicate is implicity contained in the concept or definition of
the subject, so to assert that someone is a bachelor, but not unmarried,
is a logical contradiction. Such propositions, Kant agreed, tell us
nothing about the world. However, Kant thought that the propositions
264 Logical positivism and falsificationism
of arithmetic and geometry were at once a priori and synthetic. He then
felt obliged to construct an elaborate philosophical system in order to
explain how this was possible. How could a proposition which is
synthetic, so that its denial does not entail a logical contradiction, be
true, and be known to be true, a priori! It appeared to Kant that
arithmetical propositions such as 7+5=12 were known a priori, and
were necessary truths, and yet were synthetic because it was possible to
think of 7+5 without thinking of 12. Ayer argues that this is a purely
psychological point. Kant’s explanation for our knowledge of synthetic
a priori truths is that they characterize the form we impose on the
matter of sensation and so are valid for the world only as it appears.
Ayer thinks such an explanation quite unnecessary: the truth of 7+5=12
and the a priori knowledge of that truth depend entirely on the
conventional definition of the terms in it, and it is thus quite
independent of empirical evidence or, a priori. The same argument
applies to geometrical truths; such truths are not a description of
physical space, they merely unravel whatever definition of the terms
we started off with. Logical propositions such as “Either p or not-p” are
true regardless of any facts of experience and depend for their truth
entirely on the meaning of the signs composing them; they are
tautologies because they always come out true regardless of what
propositions are substituted in them provided the substitution is done
uniformly. It follows that such analytic propositions, although
necessary, are trivially true or devoid of factual content. The
proposition “either it is raining or it is not raining” tells us nothing
whatsoever about the weather, and is true independently of whatever
the facts about the weather are; its truth excludes nothing at all.
If it is the case that all a priori propositions are analytic, how do we
explain the usefulness of logic, mathematics and geometry, and their
ability to surprise us? The explanation lies entirely in the limitations of
our intellect. In the case of complex analytic propositions we are, as a
matter of fact, intellectually incapable of seeing at once all the
consequences of the definitions we adopt. To an intellect of sufficient
power, the complex prepositional theorems of logic, mathematics and
geometry would be of no more interest than “A=A” is to us. The
interest for us of analytic propositions is that we cannot always see
immediately everything that our definitions imply.
This brings us to what for Ayer is the function of philosophy.
Philosophy cannot determine the nature of reality, as metaphysics
would suggest. Any proposition concerning the nature of reality would
be a factual scientific or common-sense proposition whose truth or
falsity could be established only by the test of experience and not by
philosophy as such. The function of philosophy, once it is demonstrated
that metaphysics is literally meaningless, is analysis and clarification.
Analysis is a branch of logic and consists of giving precise definitions of
concepts, or presenting the logical consequences of definitions, of terms
Ayer 265
used in science and common sense; thus all the propositions of
philosophy are analytic. The function of philosophy is to translate talk of
one sort into logically equivalent talk of another sort, an activity which
has purely linguistic significance. Philosophy itself can produce no new
factual knowledge about the world but can only deduce the logical
consequences of propositions whose truth or falsity, if they are not
analytic—and so devoid of factual content—is determined by the facts.
It is important to establish more exactly what is meant by empirical
verifiability in order to determine which non-analytic statements are
propositions. Such propositions must in all cases be capable of being
verified or falsified by experience. It is necessary, however, to make
two sets of distinctions here:
(a) verification in practice
(a’) verification in principle
(b) “strong” or conclusive verification
(b’) “weak” or probabilistic verification.
In both cases Ayer says he adopts the more liberal of the two
alternatives, (a’) and (b’). The reason for this is that (a) would entail
denying as literally meaningful all sorts of empirical propositions
because we could not in fact verify them. Thus the proposition “There
are mountains on a particular planet on the other side of the galaxy” is
not a proposition which I could in fact verify; perhaps it never will be
verified; nevertheless we know what would verify the proposition; we
can conceive of certain logically possible observations which could in
principle be made which would verify or falsify the proposition. There
would be an inevitable tendency for (a) to lead to solipsism whereby
my possible knowledge extended only as far as propositions
describing my actual private experiences. Adopting (b) would also
prove or exclude too much, for no empirical proposition can be
conclusively verified or falsified; empirical observations can only render
the truth or falsity of a proposition more or less probable. One reason
for this is that, whatever empirical proposition we take, the conclusion
or import we draw from observations relevant to determining the truth
or falsity of the proposition will always depend on assuming the truth
of certain other propositions describing the circumstances of the
observation. But then the truth or falsity of these other propositions
describing the initial conditions of the empirical test could themselves,
if they are factually significant, be tested by experience, and so on.
Also most of the propositions of natural science of the form “All A is
B” would be rendered literally meaningless if we adopted (b) because
we could not even in principle examine what is an open infinite class
of cases; there may always be cases we have not examined, and there is
no way of demonstrating that there are not such cases. In short, Ayer
thinks all empirical propositions are hypotheses because there is no way
of absolutely confirming or refuting such propositions.
266 Logical positivism and falsificationism
Ayer admits that empirical hypotheses do not confront experience
singly, but only as part of a system of propositions. Thus if an
observation appears to verify or refute a given hypothesis, it is always
logically possible for us to refuse to admit to the significance of the
observation by modifying the other hypotheses that gave the
observation its significance as evidence of a particular sort. Take the
proposition “All trees have leaves”; suppose we test the truth or falsity
of this proposition by making observations; whatever observations we
make, they always depend on certain other empirical hypotheses
connecting the observation and the proposition under test; for
example, that we are not suffering from an illusion, or we have
correctly identified something as a leaf. Some of the logical positivists
argued that there is a class of isolated “basic propositions” about
which it is impossible for us to be mistaken, and which can be
conclusively confirmed or refuted by experience because they refer
only to immediate experience. Ayer initially thought that any factually
significant proposition involves using general classificatory terms
(such as “red”) which it is always possible to misapply, and so no
factual proposition can be conclusively verified or refuted, since we
can always find out we have made a mistake in the light of further
evidence.
Thus, according to “weak” verifiability (b’), a genuine proposition—
a statement capable of being true or false—if it is not analytic, is an
empirical hypothesis the truth or falsity of which experiences could, in
principle, render more or less probable. The purpose of formulating
scientific theories is essentially predictive and pragmatic: it is therefore
the very meaning of rational behaviour that we adopt those theories
and methods which function to enable us to anticipate and control the
course of our experiences. The function of theories, and the purpose of
testing them, is to produce theories which are more efficient
instruments for describing and anticipating experiences. Whether a
theory will be successful in this way can be revealed not by a priori
argument but only by its success in practice, but it is always logically
possible that it may fail in cases we have not observed.
The “weak” verification principle thus states that all literally
meaningful non-analytic statements are in principle verifiable by being
rendered more or less probable by propositions which describe specific
experiences; all other statements, apart from analytic ones, are literally
meaningless. So all statements which are not analytic propositions and
cannot be verified by experience are literally meaningless: they do not
express a proposition at all. The verification principle gives a criterion
for distinguishing the literally meaningless from the literally
meaningful.
The attempt to give a precise formulation of empirical verifiability
leads Ayer into difficulty. Ayer’s initial version of the “weak”
verifiability principle is: a non-analytic statement is a genuine factual
Ayer 267
proposition—and thus not literally meaningless—if we can deduce
from it, along with certain other statements describing the conditions
under which relevant observation could take place, some experiential
proposition which refers to actual or possible experience (sensecontents),
which cannot be deduced from those other statements alone.
This formulation is, however, faulty as it excludes nothing as a literally
meaningful proposition. If N is any statement you like, even one that is
meaningless or metaphysical, and O is an experiential proposition,
then O is deducible from [(if N then O) and N], without being
deducible from O alone. This means that N would, by the criterion, be
verifiable and hence a literally meaningful proposition even though it
can be any statement at all. If we say that the “other statements” must
be themselves factually significant, then we have got no further, since
distinguishing factually significant statements was the point of the
criterion, and we cannot assume we can distinguish which statements
are factual. Ayer tries to rectify this fault, but he does not succeed in
discovering a precise formulation that includes and excludes just what
he wants.
One way of avoiding such problems would be to adopt the “strong”
verification principle (b). In this case it is not just a matter of some
empirical evidence being deducible which would be favourable or
unfavourable to the truth of a proposition. “Strong” verification
demands that the whole content of empirical or factual propositions,
when fully analyzed, be expressible in wholly experiential
propositions or observation-statements. Indeed, sometimes Ayer does
seem to be working with the “strong” verifiability principle, whereby
any genuine non-analytic proposition must, if we are to understand it,
be translatable into propositions which describe only actual or possible
experiences: sense-contents. A statement is then a factually significant
proposition if and only if it can be completely defined as a logically
equivalent set of purely experiential propositions which entails the
original proposition and is entailed by it; the two statements are thus
identical. The literal meaning of any factual proposition is then no
more or less than a set of propositions describing some actual
(categorical) or possible (hypothetical) experiences. The thinking
behind this is that understanding the meaning of factually significant
statements involves having, at least in principle, access to experiencing
the factual conditions under which the proposition which expresses
the statement would be true; that is, experience in principle of the
truth-conditions of a proposition is required to understand the literal
meaning of the statement it expresses. All factually significant
propositions, such as “I am now sitting in front of a table”, are
abbreviations for a complex of propositions describing sense-contents
alone. If any part of a statement appears to refer to something that is
not even in principle a feature of actual or possible experience, then we
can be sure that that part of the statement is without factual
268 Logical positivism and falsificationism
significance, and is meaningless unless it is analytic: that part is literal
nonsense, what we say is literally “sense-less”. Only by expressing a
non-analytic statement using symbols which wholly stand for sensecontents
are we able to make literally intelligible what it is we are
talking about.
It is surely this “strong” notion of verifiability that leads Ayer to
various forms of philosophical analysis and reductionism. Such
analyses are epistemological and are ontologically neutral. We find this
reduction at work, for example, in his analysis of the concepts of a
material object and of causation. In the case of material objects Ayer is
led to phenomenalism: statements about material objects, if they are
meaningful at all, must be wholly translatable into experiential
propositions which do not mention material objects; what we mean
when we talk about “material objects” is nothing more than some set
of actual or possible sense-experiences. Such a translation defines
“material object”. This disposes of the problem of the existence of the
external world arising from our making inferences from propositions
concerning our experiences to propositions referring to material
objects, because there is no gap in the end between experiences and
material objects: to talk of material objects is just to talk of certain
ordered collections of actual or possible experiences, and the set of
propositions describing particular sense-contents is identical to a
proposition describing a material object. The same analysis applies to
causation. Ayer agrees with Hume that “C causes E” is not a logical
relation: if “C causes E” is a non-analytic, factual, proposition then to
assert C occurs but deny E occurs is never a logical contradiction. To
say that “C causes E” is to say no more than that “whenever C, then,
under certain circumstances, E”; there is nothing further in our
experience, and indeed nothing further at all, to which the concept of
the “necessary connection” of C and E could correspond. Causality
amounts to no more than the definition “invariable association in a
potentially infinite number of possible cases”. Generally, to avoid
talking literal nonsense one must specify what feature of actual or
possible experience the talk describes.
The “self” is also not meaningfully identifiable with any nonexperiential
soul or mental substance, but is, like a material object
logically constructed out of sense-contents. The way in which we think
of the minds of others presents problems, however, because we have in
principle no access to their sense-contents, but only to their behaviour.
This produces an incoherent asymmetry whereby the ascription of
mental states to myself is phrased in “mental” sense-contents, whereas
its ascription to others is phrased in “physical” or “behavioural” sensecontents.
Logical positivism has a dilemma. The problem with adopting
“strong” verifiability is that although it excludes statements that Ayer
wishes to regard as literally meaningless, it also excludes statements he
Ayer 269
would wish to regard as meaningful. Ayer came to think later that the
complete reduction of propositions about material objects to sensecontents
was not possible, because no finite set of propositions
referring to sensory experience was ever logically equivalent to a
statement referring to a physical object. No finite set of observationstatements
can give the necessary and sufficient conditions which
would constitute the truth that X is a physical object, since further,
logically possible evidence—further experiences—may show we must
have been mistaken. So no finite set of propositions referring to senseexperiences
can conclusively verify the proposition that X is a material
object. Hence the problem with “strong” verifiability is that it implies
that most, perhaps all, of the statements of natural science are
meaningless. The problem with “weak” verifiability is that although
plausibly it permits the statements of science and common sense as
literally meaningful, factual, propositions, it fails to exclude those
statements which Ayer wishes to regard as metaphysical and
meaningless.
Take, for example, the statement “God exists”: the same
considerations apply to “God does not exist”. Ayer wants to say that
such an assertion is literally meaningless rather than false. But it is not
excluded by the “weak” verification principle, for someone might
admit that a particular experience was evidence for or against the
existence of God—thereby qualifying “God exists” as a literally
meaningful proposition—without thereby having to admit that what is
meant by “God” and “His existing” is wholly exhausted by those
evidential experiential propositions. Only by adopting the “strong”
verification principle is there hope of identifying “God exists” as
literally meaningless and so eliminating it. However, no sophisticated
religious believer is likely to admit that what he means by God existing
is nothing more than some actual or possible sense-experiences—for
example, the observed intricateness and orderliness of nature—even if
he might admit it as evidence of God’s existence.
Ayer’s analysis of apparent ethical and aesthetic statements—
“statements of value”—concludes they are not genuine propositions at
all; they are without literal meaning. They are not factual synthetic
statements, but rather expressions of feelings of approval or
disapproval, which may affect others so they feel the same way. Value
statements are not about anything—they do not even describe the fact
that there is a subjective psychological state which constitutes a
feeling—rather, they are an expression of feeling, akin to a cry of pain
or grunt of satisfaction. Expressions of value are therefore neither
rational nor irrational: they are just a piece of non-rational behaviour.
Since value statements are incapable of truth or falsity, then no two
value statements can conflict. If we argue with someone over value, it
must be over what are the facts concerning the situation which
prompted our feeling.
270 Logical positivism and falsificationism
A further problem that arises with the “verification principle” which
lies at the heart of logical positivism is the logical status of the
principle itself. For the statement “Every genuine proposition must be
either analytic or empirically verifiable” appears itself to be neither
analytic nor empirically verifiable, in which case it is self-defeating
and the “verification principle” is literally meaningless and incapable
of truth or falsity. Logical positivism is not the first or the last
philosophy to saw off the branch on which it is sitting. One response to
this is to say that the principle is not a statement, but a prescriptive
rule which we ought to adopt. But the problem with that is there is no
way of showing why the rule should be adopted.
Popper
Karl Raimund Popper was born in Vienna in 1902. Although his
parents were Jewish, they were baptized into the Protestant Lutheran
Church before their children were born. The circumstances in which
he was brought up were bookish and intellectual. His father was
doctor of law of the University of Vienna and, as well as practising as
a lawyer, he was also a scholar. With this background Popper began
reading early about philosophical, scientific and political matters. In
1918 he enrolled at the University of Vienna and sampled a wide range
of lecture courses, but concentrated his attention on mathematics and
physics. After university he taught mathematics and physics in
secondary schools. During this time he took a keen interest in leftwing
politics, although his later work was greatly concerned with the
totalitarian dangers of socialist and Marxist mass collectivization and
of the belief in inevitable laws of historical development. His
resistance to doctrines claiming access to final truths and dogmatism
led him to favour individualism and piecemeal evolutionary social
change rather than grand revolutionary change, also tentative
solutions to social problems against a background of the greatest
possible freedom for the expression of opinion and criticism which is
characteristic of an open society. The chief culprits attacked by Popper
are Plato, Hegel and Marx.
Popper had contacts with the logical positivism of the Vienna Circle,
but he was never a logical positivist, and instead became one of its
critics, despite a common interest in the methods of science. The root
of Popper’s criticism was that questions of meaning were of relatively
little importance; what concerned him was the status of theories and
their testing. The logical positivists held that, apart from the
propositions of logic and mathematics, all literally meaningful
statements were empirical and scientific. Popper never held that all
non-logical statements that were not scientific were meaningless.
Popper’s “criterion of demarcation” was, unlike the logical positivists’
Popper 271
criterion, concerned with the distinction not between the meaningful
and the meaningless but between science and non-science. Nonscience
includes pseudo-science, which consists in intellectual
activities that claim to be scientific, but are not.
Before the Second World War, Popper left Austria, and from 1937 to
1945 he taught philosophy at the University of New Zealand. He came
to England in 1946. He remained on the outside of philosophical
activities as practised in both Oxford and Cambridge, and received
greatest intellectual sustenance from those who were not primarily
philosophers such as the art historian E.H.Gombrich and the
economist and political theorist F.A.Hayek. In 1949 Popper was made
Professor of Logic and Scientific Method at the London School of
Economics; and this position he held for the rest of his university
career. He was knighted in 1965. Popper’s work has been enormously
influential in the philosophy of science, and on the methodology of the
social sciences.
It is possible to identify three important connected strands of
thought in Popper’s philosophy: (a) the solution of the problem of
induction, (b) the problem of demarcating science from non-science, (c)
the importance of maximizing criticism and maintaining a “critical
attitude” as essential for rationality and vital for the growth of
knowledge.
The essential nature of philosophy involves the critical questioning
of fundamental assumptions that we might otherwise take for granted;
this is obviously connected with point (c). Points (a) and (b) are also
connected with this because it has been thought that what
distinguishes science from non-science is the inductive method: the
extent to which the truth of its propositions is derived from and
justified by their origin in the facts of experience. The ideal picture that
this inductive model of science evokes is its beginning by collecting
pure or presuppositionless observations which give the facts, in a
passive, unprejudiced, neutral manner; then from the repetition of
these observations certain patterns begin to emerge which lead to the
framing of universal hypotheses connecting particular observed
phenomena; these hypotheses are then, by further experimental tests,
proved true, or at least confirmed as highly probable. The aim is to
pick out, from the many features repeatedly observed, the necessary
and sufficient conditions for the event to be explained; that is, the aim
is to identify the cause of the event by identifying that feature of the
situation that is always present when the event to be explained occurs
and is never present when the event to be explained does not occur.
Popper argues, with others, that there are at least two major
problems that such a view of science encounters,
(i) The first problem is that there are no presuppositionless, neutral,
raw observations free of theoretical content. All observation
272 Logical positivism and falsificationism
involves some identifying, and therefore theory-loaded, idea of
the nature of the thing observed that already determines and
presupposes the kind of thing observed, which therefore
necessarily pre-empts any conclusion derived from observation.
To observe at all necessarily involves theoretical presuppositions
about what we are observing. We always when observing observe
something as a so-and-so which carries with it theoretical
implications which often take us beyond the bare content of the
observation. For example, the assertion “Here is a glass of water”
carries with it theoretical assumptions about the behaviour of
entities denoted by “glass” and “water”, assumptions with
implications beyond the evidence of present observations;
indeed, Popper says that such a statement is unverifiable, because
the universal law-like behaviour implicit in denoting terms such
as “glass” and “water” is not reducible to any finite class of
experiences. Another point is that when we identify two events as
a repetition of the same event, we are necessarily picking out
some respect in which they are similar, and ignoring other
respects in which they differ; they must differ or they would not
be two distinct events. Observations, to be possible at all, always
involve the selection, implicitly or explicitly, of certain of the
features of our environment and the rejection of others; the
possible range of things we could make note of is infinite, so we
are forced to be selective. What we choose to observe is guided by
theoretical interests.
(ii) The second problem is that of inductive inference; Popper
characterizes this as “Hume’s problem”. In valid deductive
reasoning it is not possible for the premises of the argument to be
true and the conclusion false; necessarily if the premises of a
valid deductive argument are true, then the conclusion is true. To
assert the premises and deny the conclusion of a valid deduction
is to contradict oneself. A deductive argument involves the claim
that the premises present conclusive grounds for its conclusion.
Thus if it is the case that “All men are mortal” and “Socrates is a
man”, then “Socrates is mortal”. Inductive arguments are not
conclusive in this way: the premises can be true, yet the
conclusion false.
The theories of science are characteristically universal
propositions of the form “All As are Bs” which go beyond the
evidence of experience; the proposition does not follow from any
finite number of observations of As and Bs—which give
propositions of the form “Some As are Bs”—for there is no logical
contradiction involved in the assertion that the next observed A
will not be a B. From this it follows that no universal scientific
proposition can be proved to be true. Scientific laws always
transcend experience. The inference from experience to universal
Popper 273
laws, or more generally to unobserved instances, is neither a
logically valid deductive argument nor an inference that could be
justified by experience, for the argument from “inductive
inferences have worked in the past” to therefore “inductive
inferences will work in the future” is itself an inductive inference,
so any such attempted justification would be circular. An
inductive inference could be made valid on the assumption that
regularities or uniformities observed in the cases we have
observed hold in cases we have not observed. But this assumption
is not a logical a priori truth such that its denial implies a
contradiction or such that it can be justified by experience. We
might say that uniformities have been found to hold in all cases
we have observed, therefore uniformities will hold in cases we
have not observed; but that evidence from cases we have
observed can be evidence for cases we have not observed is
exactly what the uniformity principle justifies, so such evidence
cannot be used to justify the uniformity principle itself.
It will not help to fall back on probability, for we can still ask
why we think the observation of certain cases should even make
more probable events we have not observed. We can say further
that no finite number of observations can make a universal
statement of the form “All As are Bs” more probable by the
frequency theory of probability; the class of examined cases is
always finite, and the class of unexamined cases is potentially
infinite, so that the probability of the universal statement “All As
are Bs” will always approach zero. Even if we restricted the range
of our general statement, we could still not be sure that the next,
ninety-ninth out of a hundred, A will be a B, on the basis of
observing past As and Bs, since “A and not-B” is never a logical
contradiction.
Popper rejects induction both as a fruitful method of formulating
scientific theories, and as a logic for justifying theories. He claims to
have solved the problem of induction, but he does not so much solve it
as sidestep the problem; he does not give or seek a justification for
induction, rather he substitutes a different scientific methodology that
is independent of induction, but does the same job as induction in
allowing us rationally to prefer one theory to another on empirical
grounds. Popper maintains the empiricist principle that it is only by
observation and experiment that we may rationally decide to accept or
reject scientific theories. Such decisions cannot be justified a priori. This
leads on to the heart of Popper’s philosophy, and the idea that what
distinguishes science from non-science is not induction as a method or
a justificatory logic, but that science consists of theories which are both
logically self-consistent and such that they can in principle be falsified or
refuted. Popper uses the terms “hypotheses”, “conjecture”, “theory”
274 Logical positivism and falsificationism
and “scientific law” interchangeably. The logical basis for this is quite
simple, and derives from the deductive principle of modus tollens:
Roughly this says that if asserting p entails asserting q, and q is false,
then p is also false. We can substitute in this formula, H, standing for
some universal scientific hypothesis, for p, and e, standing for an
observation-statement, for q. The observation-statement e is deduced
from H. We then have the following.
The essential point to notice is that this indicates a logical asymmetry
between verification and falsification: while it is the case that no finite
number of observations can ever prove the truth of a universal
scientific theory, logically only one case is required to contradict a
theory’s universal assertion in order for it to be falsified or refuted.
What is distinctive about scientific theories is not that they can be
proved true, or even made more probable, but that they are testable,
that is, they can be falsified. So from the universal proposition “All As
are Bs” (H), we can deduce the proposition that “It is not the case that
some (even one) A is not a B” (e); if we observe “Some (at least one) A is
not a B” (not-e), then it follows purely as a matter of deductive logic
that “All As are Bs” is false (not-H). The assertion “All swans are white”
is falsified by the observation of a single non-white swan which entails
that “Not all swans are white”. Thus a theory is falsifiable if and only if
there is some observation-statement deducible from it, which, if false,
would falsify the theory. A genuine scientific theory must exclude some
logically possible state of affairs by specifying more or less exactly what
the state of affairs will be: it must not be compatible with all logically
possible evidence. More exactly what is deducible from a scientific
theory is at least one “basic statement” which is a potential falsifier;
such a statement will be a singular observation-statement that refers to
some publicly observable event. This excludes pure existential
statements of the form “Some A is a B” from being scientific because
they are untestable; no possible evidence can ever refute them as there
is, so to speak, always somewhere we have not looked.
Popper was impressed by the contrast between the theories of
Marxism and Freudian psychoanalysis on the one hand, and Einstein’s
theories on the other. According to Popper, Marxists and Freudians saw
everywhere confirmation for their theories, whereas Einstein made an
effort to formulate a very specific observable prediction which followed
Popper 275
from his theory concerning the bending of light, which, if it failed to be
upheld by observation, would have refuted the theory. What is at issue
here is not the psychological fact, if it is one, of the reluctance of Marxist
and Freudian defenders to admit evidence refuting their theories, but
rather the nature or logical structure of Marxist and Freudian theories
themselves which rendered them immune from falsification. Popper’s
suspicion was that Marxist and psychoanalytic theories were only
“confirmed”, and seemed to explain everything, because they were,
through reasons of vagueness or devices designed to explain away
counter-evidence, irrefutable. Such theories are anathema to the proper
critical scientific attitude. That is not to say that Marxist and Freudian
theories were meaningless, or even that what they said was untrue,
rather the theories were not scientific in that they were highly untestable,
that is, difficult, if not impossible, to falsify. The theories were constantly
hedged around with caveats or qualifications, so that apparent counterevidence
was no longer a deducible consequence of the theories. For
Popper this indicates that the holders of these theories were not adopting
the proper critical scientific attitude. But far from pre-scientific myths
being meaningless, Popper says they can often be modified to form the
basis of later scientific theories and so become testable by experience.
A further point concerns a comparison of Newton’s and Einstein’s
theories. Popper argues that despite the fact that Newton’s theory can
be massively confirmed by observation, this is not enough to establish
its truth. He holds the view that discrepancies emerged in Newton’s
theory, between its predictions and observations, which led to the
development of Einstein’s competing theory despite the enormous
confirmation of Newton’s theory.
Having explained the logic of Popper’s philosophy of science, it is
necessary to distinguish this from the methodology or practice of
falsificationism. While the logic of falsification is quite simple, the
methodology is a good deal more complex. This arises because,
although it is clear what would, logically speaking, constitute the
refutation of a scientific theory, determining whether a theory is in fact
refuted is quite a different matter. Not only is it the case that there are
various reasons why it is difficult to determine if a refutation has taken
place, but Popper also acknowledges that there are various ways in
which an apparent refutation can always be avoided. These
considerations require that we adopt certain methodological rules so as
to maximize the possibility of scientific progress, although there is no
method that can guarantee it.
There are various problems that arise in attempting the actual
falsification of a theory by critical discussion, observation and experiment,
(i) It is always possible to doubt that the observation we have
made is correct—we may have made an observational error.
This introduces the problem of the empirical base: if we cannot
276 Logical positivism and falsificationism
be certain of the truth of the observation-statements we use to
test our theories, we cannot be certain our theories are refuted
by them. Popper admits that there are no indubitable
observation-statements; all observation-statements themselves
have some theoretical content and are open to further testing.
But this does not lead to a vicious infinite regress, because
although all empirical statements are potentially testable, they
can be provisionally or conventionally held as true, and so
used to test or falsify theories for which they are potential
falsifiers. If they are doubted, further tests can always be
carried out. There are no ultimate empirical foundations,
(ii) This problem concerns the fact that scientific theories are
always tested in groups. In testing any theory it is necessary
that we describe the initial conditions by a set of auxiliary
hypotheses; that is, certain other theories are involved which
act as assumptions concerning relevant circumstances of the
test; these also give the falsifying significance to the
observation deduced from a theory. For example, in making an
observation we might assume that light travels in a straight
line. Thus the falsifying modus tollens formula becomes more
complex.
Strictly speaking, all we can say in this complex situation is
that some element in the totality (H+h) is refuted—is shown to
be false—and that need not be the theory H under test, but
could instead be one or more of the auxiliary hypotheses h.
What can be said here is that the auxiliary hypotheses are
themselves open to testing.
(iii) Closely connected with point (ii), it is always possible to adopt
ad hoc hypotheses so as to evade refutation. By ad hoc
hypotheses is meant hypotheses adopted for no other purpose
than to avoid refutation. For example the theory “All bread
nourishes” can be immunized against refutation by the
example of some poisonous bread in France by tacking on to
the proposition “All bread nourishes” the expression “except
in France”. Another ad hoc method of evading refutation is
simply to define away apparent counter-evidence; thus if “All
As are Bs” is presented with the evidence of an A that is not a
B, it can be said that if we seemed to observe an A that was not
a B, then it could not have been an A that we observed at all;
this makes being a B part of the identifying definition of an A.
So we might say a non-white swan is not a swan at all.
Popper 277
The adoption of ad hoc hypotheses and definitional
manoeuvres Popper regards as intellectually dishonest. We
must therefore adopt some methodological rules so as to avoid
adopting ad hoc hypotheses. Partly this is achieved by the
methodological principle that if we modify a theory with the
addition of some new hypotheses so as to avoid refutation,
there must be some consequences that can be deduced from
the original theory and the new additional hypotheses that
were not deducible from the original unmodified theory. In
other words, the additional or modified hypotheses must form
a new hypothesis which is testable in some way the original
hypothesis was not: they must be independently testable. Thus
we reject as ad hoc “All bread nourishes, except in France”, since
it has no new testable consequences which are not also a test of
“All bread nourishes”; the reverse is not the case, since there
are testable consequences of “All bread nourishes” which are
not also testable consequences of “All bread nourishes, except
in France”.
It is clear that some hypotheses are more testable or falsifiable than
others. The theory that “All planets move in loops” (H1
) is less
falsifiable than “All planets move in ellipses” (H2
), because H1
is less
specific about what evidence would refute it. To put it another way, H1
excludes less than H2
: its truth is compatible with a far greater range of
possible observations. H2
not only says that the planets move in closed
loops, but also specifies the exact kind of loop that is involved. Thus
we can say that all the observations that would falsify H1
would falsify
H2
but some observations that would falsify H2
would not falsify H1
; if
the planets moved in anything but ellipses, H2
would be false, while as
long as they still moved in some kind of loop H1
would be true. Popper
expresses this point by saying the greater the information content of a
theory, the more falsifiable it is: it tells us more about the way the
world is by excluding as being the case more logically possible states of
affairs. The information content increases with the set of statements
which are incompatible with the theory.
Popper also notes that the falsifiability and the information content
of a theory are in inverse proportion to its probability. The information
content of a tautology—for example, “Either it is raining or it is not
raining”—is zero, and its probability is at the maximum of 1. The
probability of H2
(“All planets move in ellipses”) being true is far less
than the probability of H1
(“All planets move in loops”) being true
because the class of potential falsifiers of H1
is a proper subclass of the
potential falsifiers of H2
. For example, “The planets move in a straight
line” would falsify both H1
and H2
, but “The planets move in a circle”
would falsify H2
but not H1
because a circle is a kind of loop.
Popper’s overall position is then that we make progress in our
278 Logical positivism and falsificationism
knowledge, and approach the truth, by a process of trial and error.
Popper gives the following evolutionary view of the growth of
scientific knowledge:
P
1
® TT1
® EE1
® P
2
Here P1
designates a problem, for which we propose the tentative
theory TT1
; we then try to eliminate false theories by testing them
severely and subjecting our theory to critical discussion, EE1
; then P2
is
the problem-situation as we emerge from our attempted solution to
our problem, and so on. Science makes progress by conjecture and
refutation; we learn from our mistakes. We start with problems, not
with neutral observation: that is, we start with the failure to explain
some phenomenon. No mere observation constitutes a problem; we
have a problem only in the light of some existing theory which fails to
explain an observation. We try to solve the problem not by proposing
the most probable theory—for more probable theories have less
information content—but by proposing bold conjectures or guesses
which, because they are highly specific and precise in what they say
about the world, are highly falsifiable; we can then test these theories
in severe and crucial tests. The tests are severe because what the theory
entails is incompatible with a very wide range of possible
observations. Intuitively we can see that the severity of a test will
increase with its improbability. A new theory will be bold and
improbable (unlikely) and its tests severe because it involves rejecting
part of the background knowledge of scientific theories of its historical
time. For example, Einstein’s theory was bold relative to the theoretical
background assumptions of its time because it contradicted the
background assumption of its time that light travels in straight lines.
It is significant that in Popper’s falsificationism the source of a
scientific theory is totally irrelevant to whether it is scientific or not. A
theory is scientific if and only if it is falsifiable; it is quite unimportant
whether the theory arises from laboratory observation or an
inspirational blow on the head. One method might as a matter of fact
be more fruitful as a means of producing good theories than another;
but that is irrelevant to the question of whether a statement is scientific
or not, and, if it is scientific, how good a scientific theory it is. Science
has no mechanical method by which it can make progress; Popper’s
philosophy gives free rein to imaginative bold speculation. Good
science requires just as much imagination as any of the arts. Popper
says we do not in fact come to the world as passive or neutral
observers, but are born with certain natural expectations or
dispositions that operate in the same way as consciously constructed
theories. Indeed all animals are in their behaviour acting out innate
solutions to problems. But while these innate “theories” might be
psychologically a priori, that does not mean they are a priori valid. The
main difference between man and other animals is the extent to which
Popper 279
man can allow his theories to die rather than dying himself; man can
adopt new theories rather than hanging on to his theories and dying
with them. One sees the point of this in considering the way a wasp
unremittingly batters at a glass window and so “fails to solve” the
“problem” of escaping.
Normally we will not be in the situation of testing one theory in
isolation, but will have to choose between a number of competing
theories. Even if we find an observation that falsifies a theory we will
not reject it unless we have some better theory with which to replace it.
Indeed, Popper’s methodological rules demand that we do not hastily
reject a theory after a single falsifying instance, but only after frequent
and rigorous falsification has taken place, and we have a better theory
with which to replace it. The choice between competing theories
should be made in the following way: theory T
2
should be preferred to
T
1
if T
2
solves all the problems that T
1
solves and it solves the problems
T
1
failed to solve (that is, where T
1
was refuted), and it offers solutions
to some additional problems about which T1
says nothing, thus
allowing the further possibility for refutation. To put it another way, we
should choose the theory that explains all the previous theory explains,
explains what the previous theory failed to explain, and offers an
explanation for further phenomena not explained by the previous
theory. The satisfaction of these conditions effectively rules out our
new theory being merely the old theory plus some ad hoc hypotheses
which serve only to avoid the apparent refutations or failures.
Popper’s philosophy of science can be summarized in the following
way. Knowledge progresses by proposing bold explanatory theories,
that is, explanations with a high information content that are highly
falsifiable, by subjecting those theories to severe and crucial tests and
by the replacement of falsified theories by better theories. We can be
said to replace a theory by a rationally preferable, better theory, even if
the old theory has not been conclusively falsified, when the new theory,
provided it has not been falsified, is able to explain all that the old
theory explained, and things the old theory failed to explain, and
offers as well explanations for things for which the old theory offered
no explanation. That is, the better theory T
2
will contain T
1
as an
approximation. If any falsification of T
1
would be a falsification of T
2
,
but not vice versa, then T
2
is rationally preferable to T
1
provided T
2
has
not been falsified. This means we choose the theory which is more
falsifiable—has more information content—provided that theory has
not been falsified. We can make our assessment of theories only from
the position of the current historical state of critical discussion.
If a theory survives continuous attempts to falsify it by severe tests,
it can be said to be highly corroborated. That is not to say its truth has
been conclusively established, or even made more probable. The
corroboration of a theory at a certain time is essentially a report on its
degree of testability, the severity of the tests to which it has been
280 Logical positivism and falsificationism
subjected, and the way it has stood up to those tests. The corroboration
of a theory will increase with its falsifiability, provided it is not
falsified, because the more falsifiable it is the more severe the tests it
can potentially survive. It can then be subjected to further severe tests.
Popper is quick to deny that corroboration reintroduces the notion of
induction, for he says that the corroboration accorded to a theory does
not say anything about its reliability in the future or anything about its
future performance. The less the probability of a theory, the higher its
degree of potential corroboration can be. A less probable theory can
pass more severe tests and so can be more highly corroborated. A
theory that has been well corroborated can be provisionally accepted. If
there is more than one theory covering the same ground, it is rational
to choose the best corroborated theory because that has been most
severely tested. This again gives an account of rational preference
between theories: T
2
is preferable to T
1
if T
2
survives all the tests T
1 survives, survives the tests T
1
fails, and goes on to explain further facts
which are testable consequences of T
2
, and if T
2
has not yet been
refuted.
It is clear from Popper’s position that we can never establish that a
theory is true. He says that we can never “know” in the sense of
conclusively establishing a theory to be true so that there is no
possibility of our being mistaken. In this sense Popper is a fallibilist:
we can never be certain that we have found the truth. All our theories
are conjectures or guesses which are open to testing; we can then
perhaps say that some conjectures are better than others because they
have stood up to tests better.
Since we are interested in the truth we shall be interested in
eliminating a theory which we discover to be false, for that way we
might hit upon a theory that is true. Popper is absolutely clear in
distinguishing whether a theory is objectively true or false as a matter
of correspondence, or failure of correspondence, with facts (p is true if
and only if it corresponds to the facts), from our knowing if p
corresponds to the facts. Popper takes from the logician Alfred Tarski
(1902–83) the definition of truth: “‘p’ is true if and only if p”. Every
unambiguous statement is either true or false, and there is no third
possibility; but determining when a proposition corresponds to the
facts is quite a different matter, and Popper thinks we are never in the
position to say that we have established or justified the truth of a
theory. However, the correspondence definition of truth can act as a
regulative principle: it is something we can aim at and get nearer to.
Indeed, as the corroboration of a theory increases, it is reasonable to
conjecture that we are getting nearer the truth. The extent to which a
theory approaches the truth Popper refers to as its verisimilitude.
Popper derives the notion of verisimilitude from the information
content of a theory: the content of a theory T is all those propositions
entailed by it. The content of T can then be divided into its truthPopper
281
content (the class of all true statements entailed by T) and its falsitycontent
(the class of all false statements entailed by T). The
verisimilitude of T is its truth content—minus its falsity—content.
Assuming that theories T
1
and T
2
are comparable, then T
2
has greater
verisimilitude than T
1
if its truth-content is greater than T
1
, but its
falsity-content is less than T
1
, or the falsity-content of T
2
is less than T
1 but the truth-content of T
2
is greater than T
1
. If more true statements,
but not more false statements, follow from T
2
than T
1
, then T
2
is nearer
the truth.
If T
2
entails all the true statements entailed by T
1
, and T
2
entails
some true statements not entailed by T
1
, and T
2
does not entail more
false statements than T
1
, then it is reasonable to say that T
2
is nearer the
truth than T
1
:T
2
has greater verisimilitude even if it is false. Thus we
can rationally prefer T
2
to T
1
if we are in pursuit of the truth even if T
2 is false, provided that the falsity-content of T
2
is not too great.
The verisimilitude and the degree of corroboration of a theory are
connected. If we compare the corroboration of two theories and
determine that all the tests passed by T
1
are also passed by T
2
, and that
T
2
passes some tests that T
1
does not pass, and that T
2
does not fail
more tests than T
1
failed, then it is rational to prefer T
2
to T
1
because T
2 can be conjectured to have greater verisimilitude: it is nearer the truth.
T
2
will be more testable than T
1
: it will have a greater information
content; it will say more about the way the world is. Although we have
not established the truth of T
2
—indeed, as we are fallible, it is likely to
be false—we can express a rational preference for T
2
as being better
corroborated than T
1
and nearer to the truth than T
1
. T
2
is more testable
and survives more tests than T
1
.
From Popper’s acceptance of the correspondence theory of truth it
can be seen that he is a metaphysical realist. He thinks that our
theories, if true, refer to a reality which is independent of mind and
our theories. However, he agrees that such metaphysical realism does
not take us very far except as a regulative idea, for we still have to
determine when our theories correspond to things as they really are.
We cannot “look around” our theories to reality, but can only take to be
reality what our best theories in the light of current critical discussion
and testing say reality is. Popper thinks it unlikely that we will ever
discover “the truth” about the world. Popper is opposed in science to
instrumentalism, which asserts that scientific theories do not refer to
real entities which explain the course of our observations, but are
rather useful devices which posit whatever is required—without
maintaining its reality—for predicting accurately the course of our
experience. On this view scientific laws are rules rather than truths.
Popper also opposes essentialism, which maintains that we can
discover an ultimate reality in terms of which everything else is
explained. This attitude he sees as stultifying to the pursuit of ever
better explanations. Popper takes a middle course in which science is a
282 Logical positivism and falsificationism
genuine attempt to explain some real state of affairs which is known or
assumed to be true by some other real state of affairs that is unknown
and requiring discovery, the truth of which can be tested
independently of the phenomena to be explained; but there is no end
to the depth to which we can progress in pursuit of explanations.
When Popper talks about “knowledge” he is not referring to finally
established, or justified, truths. He also emphasizes that when he talks
about “knowledge” he is talking about knowledge in the objective
sense. He intends by this to make a distinction between any person’s
subjective knowledge and objective knowledge as it is formulated in
language and existent in books and journals in libraries and research
institutions open to public inspection and testing. Scientific
knowledge is objective in this sense. Objective logical relations exist
between statements which are formulated in language, regardless of
whether anyone is actually aware of them or not. What individual
scientists believe is relatively unimportant compared to the objective
growth of knowledge. The error of what Popper terms “belief
philosophy” is that it tries to see knowledge as an especially sure kind
of belief.
Popper, in fact, makes a distinction between three interdependent
worlds: World 1 is the physical world; World 2 is the subjective mental
world; World 3 is the objective world of theories, mathematics,
literature, art, and the like, within which there exist objective logical
relations—objective, that is, in being independent of the awareness of
individual minds. The objects of World 3 are developed by World 2
minds, often in response to problems perceived in World 1; but once
formulated, they have an objective status transcending the intentions
of the individual. Yet it is knowledge in the subjective sense—what the
individual person can know—on which traditionally philosophy has
concentrated, the notion being that it is only from what an individual
mind can really know that any further knowledge claims can be
justified. Yet most human knowledge in the objective sense is not
known by anyone in the subjective sense. Human knowledge,
especially scientific knowledge, almost entirely consists of knowledge
without a knowing subject. Popper’s World 3 has some similarities to
Plato’s realm of objective Forms; however, a vital difference is that
Popper’s World 3 is by no means fixed, but constantly changes and
develops as knowledge grows and progresses through the critical
examination of the knowledge we already have.
Popper 283
continued to ...