Ayer’s Language, Truth and Logic

A.J. Ayer Language Truth and Logic Victor Gallancz 1946

The First Edition was 1936, but the 1946 addenda, treated here as an appendix, is significant.

I have a 2001 Penguin Books edition, which includes a 1946 appendix. According to the back cover:

Ayer argued that factual propositions were only meaningful if they could be verified by direct experience. This enabled him to reject metaphysical claims … as, literally, nonsense. [This book] became a classic text, which not only revitalized British Philosophy but which set a new standard for philosophical debate.

Preface to the First Edition

The views which are put forward in this treatise derive from the doctrines of Bertrand Russell and Wittgenstein, which are themselves the logical outcome of the empiricism of Berkeley and David Hume. Like Hume, I divide all genuine propositions into two classes: those which, in his terminology, concern ‘relations of ideas”, and those which concern ‘matters of fact’. The former class comprises the a priori propositions of logic and pure mathematics, and these I allow to be necessary and certain only “because they are analytic. That is, I maintain that the reason why these propositions cannot be confuted in experience is that they do not make any assertion about the empirical world, but simply record our determination to use symbols in a certain fashion. Propositions concerning empirical matters of fact, on the other hand, I hold to be hypotheses. which can be probable but never certain. And in giving an account of the method of their validation I claim also to have explained the nature of truth.

To test whether a sentence expresses a genuine empirical hypothesis, I adopt what may be called a modified verification principle. For I require of an empirical hypothesis, not indeed that it should be conclusively verifiable, but that some possible sense-experience should be relevant to the determination of its truth or falsehood. …

As for the propositions of philosophy themselves, they are held to be linguistically necessary, and so analytic. And with regard to the relationship of philosophy and empirical science, it is, shown that the philosopher is not in a position to furnish speculative truths, which would, as it were, compete with the hypotheses of science, nor yet to pass a priori judgements upon the validity of scientific theories, but that his function is to clarify the propositions of science, by exhibiting their logical relationships, and by defining the symbols which occur in them. Consequently I maintain that there is nothing in the nature of philosophy to warrant the existence of conflicting philosophical ‘schools’. And I attempt to substantiate this by providing a definitive solution of the problems which have been the chief sources of controversy between philosophers in the past.

Truth and Probability

 [We] ought, perhaps, to justify our assumption that the object of a ‘theory of truth” can only be to show how propositions are validated. For it is commonly supposed that the business of the philosopher who concerns himself with ‘truth’ is to answer the question ‘What is truth r” and that it is only an answer to this question that can fairly be said to constitute a “theory of truth”. But when we come to consider what this famous question actually entails, we find that it is not a question which gives rise to any genuine problem; and consequently that no theory can be required to deal with it.


In practice we assume that certain types of observation are trustworthy, and admit the hypothesis that they have occurred without bothering to embark on a process of verification. But we do this, not from obedience to any logical necessity, but from a purely pragmatic motive, the nature of which will shortly be explained.

When one speaks of hypotheses being verified in experience, it is important to bear in mind that it is never just a single hypothesis which an observation confirms or discredits, but always a system of hypotheses. Suppose that we have devised an experiment to test the validity of a scientific ‘law’. The law states that in certain conditions a certain type of observation will always be forthcoming- It may happen in this particular instance that we make the observation as our law predicts. Then it is not only the law itself that is substantiated, but also the hypotheses which assert the existence of the requisite conditions. For it is only by assuming the existence of these conditions that we can hold that our observation is relevant to the law. Alternatively, we may fait to make the expected observation. And in that case we may conclude that the law is invalidated by our experiment. But we are not obliged to adopt this conclusion. If we wish to preserve our law, we may do so by abandoning one or more of the other relevant hypotheses. We may say that the conditions were really not what they seemed to be, and construct a theory to explain how we came to be mistaken about them; or we may say that some factor which we had dismissed as irrelevant was really relevant, and support this view with supplementary hypotheses. We may even, assume that the experiment was really not unfavourable, and that our negative observation was hallucinatory. And in that case we must bring the hypotheses which record the conditions that are deemed necessary for the occurrence of a hallucination into line with the hypotheses which describe the conditions in which this observation is supposed to have taken place. Otherwise we shall be maintaining incompatible hypotheses. And this is the one thing that we may not do. But, so long as we take suitable steps to keep our system of hypotheses free from self-contradiction, we may adopt any explanation of our observations that we choose. In practice our choice of an explanation is guided by certain considerations, which we shall presently describe. And these considerations have the effect of limiting our freedom in the matter of preserving and rejecting hypotheses. But logically our freedom is unlimited. Any procedure which is self-consistent will satisfy the requirements of logic

It is only tautologies that are certain.

[We] test the validity of an empirical hypothesis by seeing whether it actually fulfils the function which it is designed to fulfil. [The] function of an empirical hypothesis is to enable us to anticipate the future.

[To] be rational is simply to employ a self-consistent accredited procedure in the formation of all one’s beliefs. The fact that the procedure, by reference to which we now determine whether a belief is rational, may subsequently forfeit our confidence, does not in any way detract from the rationality of adopting it now. For we define a rational belief as one which is arrived at by the methods which we now consider reliable.

To say that an observation increases the probability of a hypothesis … is equivalent to saying that the observation increases the degree of confidence with which it is rational to entertain the hypothesis [where] the rationality of a belief is defined, not by reference to any absolute standard, but by reference to part of our own actual practice.

I suspect … that those who object to our pragmatic treatment of history are really basing their objections on a tacit, or explicit assumption that the past is somehow ‘objectively there’ to be corresponded to – that it is ‘real’ in the metaphysical sense of the term.


This was the Introduction to the 1946 edition.

In the ten years that have passed since language, Truth and Logic was first published, I have come to see that the questions with which it deals are not in all respects so simple as it makes them appear; but I still believe that the point of view which it expresses is substantially correct. [There] are a number of points that seem to me to call for some further explanation, and I shall accordingly devote the remainder of this new introduction to commenting briefly upon them.


[An] ‘observation statement’ [designates] a statement ‘which records an actual or possible observation’.

[A] statement is directly verifiable if it is either itself an observation-statement, or is such that in conjunction with one or more observation-statements it entails at least one observation-statement which is not deducible from these other premises alone [and it] is indirectly verifiable if … first …in conjunction with certain other premises it entails one or more directly verifiable statements which is not deducible from these other premises alone; and secondly, that these other premises do not include any statement that is not either analytic, directly verifiable, or capable of being independently verifiable as indirectly verifiable.

[The] principles of verification [requires] that of a literally meaningful statement, that is not analytic, that it should be directly or indirectly verifiable … .

For the rest I can find no better way of explaining my conception of philosophy than by referring to examples: and one such example is the argument of this book.

My Comments

Ayer’s book is written for those familiar with the mainstream British philosophy of the 1930s, not for us. It came to be regarded as popularising ‘logical positivism’ in Britain. But its seems to me that it starts from position of respectful pragmatism that we are free to discount. But still, it suggests some interesting insights.

Ayer’s views

Ayer, rightly, notes that any non-tautological claims to ‘truth’, ‘probability’, ‘rationality’ and ‘logic’ can only be relative to our best understanding, and not absolute. Thus we may read other’s empirical claims (even those of our past selves) as relative to that understanding, and so should regard them as implicitly ‘uncertain’.

Ayer’s pragmatism leads him to suppose that it is rational to act as if one believed claims ‘arrived at by the methods which we now consider reliable‘, without regard to any implicit uncertainty. One should stop believing them when one notices an anomaly which cannot be explained away. As he notes, if we are strongly committed to our beliefs we may  go to great lengths to explain away or discount any anomalies. This seems unfortunate.

Ayer doesn’t discuss an alternative approach in which:

  • One acknowledges anomalies and regards such claims as unreliable, seeking to develop alternative hypotheses and to gather diagnostic observations, to resolve the uncertainty as best one can.
  • Even when there are no apparent anomalies, one actively:
    • Considers the broader implications of the claims (whether regarded as ‘true’ or merely ‘probable’) and seeks out potentially contrary evidence.
    • Seeks out and considers  alternative claims that are consistent with the observations according to one’s methods.
    • Seeks out and considers alternative potentially credible methods that may admit alternative evidence or hypotheses.
  • One seeks to develop strategies that don’t just take account of the predicted or ‘expected’ outcomes, but which also take account of the possibilities above and seek to avoid avoidably bad outcomes, seeking more ‘evidence’ where appropriate.

That is, Ayer does not look out for problems that may arise from his beliefs, and so may continue to act ‘rationally’ on them even when there are (unnoticed) anomalies. This ‘closed eyes closed mind’ approach may often be considered ‘pragmatic’, but it seems to me that a more open-minded approach would comply with all of Ayer’s requirements (e.g. ‘verifiability’) apart from discounting uncertainty, and as such might reasonably be regarded as a logical next step beyond Ayer’s narrow pragmatism.

A possible disadvantage of this broadened view of pragmatism is that one can never absolutely justify one’s actions. But then is justifying them on the basis that ‘it seemed like a good idea at the time’ really much better, except possibly psychologically?

In more detail ..


Ayer writes:

[The] function of an empirical hypothesis is to enable us to anticipate the future. … [To] be rational is simply to employ a self-consistent accredited procedure in the formation of all one’s beliefs. [We] define a rational belief as one which is arrived at by the methods which we now consider reliable.

Such beliefs are regarded as having been verified and true, but I only follow Ayer in so far as we can regard them as having been verified according to accredited method, by which he presumably means ‘generally accepted, orthodox’ [C.O.D.] But accepted by who? Our selves? Our peer group? Our ‘betters’? Our countryfolk? The ‘West’? The world as a whole? If ourselves, then to  be rational it seems enough that we believe in our rationality. If others, then it seems enough that we believe that they believe in our procedures. In any case, as Ayer acknowledges, logically, rationality does not eliminate the grounds for uncertainty.

Suppose that our orthodox ‘system of hypotheses’ or scientific law has not always been reliable. It will often be pragmatic to act ‘as if’ it were reliable. But why should we consider it to be as reliable as we would if we had not experienced any problems? We may consider its predictions ‘probable’ in Ayer’s sense, but why should we consider them [absolutely] reliable?

For example, I may suppose that there is some reality ‘out there’ that we intend to make statements about, and that in at least some cases, the more a question is tested (‘verified’ for Ayer) the more probable (in Ayer’s sense) it becomes and the more it is ‘pragmatic’ to rely on it, at least in the short-term.

Thus Ayer’s view seem to be that one ‘should’ form beliefs and then act upon them, even though we may not be entirely confident in either the ‘truthfulness’ of the observations or of  ‘the procedure,  by reference to which we now determine whether a belief is rational’.  Alternatively, if we recognize that such observations, procedures or accreditations ‘may subsequently forfeit our confidence’, how is it that we are entitled to consider them reliable?

An alternative application of Ayer’s presupposition is that we should learn and orthodox procedures and views, and at least accept what is generally accepted unless and until it is contradicted by our own experience. Hence, it seems, the only test that we really need to apply is whether or not something is ‘generally accepted’.

Thus in the initial book Ayer presents us with a conundrum. The solution, in the appendix, is that we regard a statement as verified if it is useful in the sense of being accredited as correctly predicting something that we couldn’t otherwise predict. I think this does capture something of how many people do think. But is it wise?

I tend to agree with Keynes’ Treatise on Probability, that some caveat is needed. For example, if the procedure for establishing rationality has a good track record. It may be significant if the procedure is ‘generally accepted, orthodox’, but this hardly seems decisive. In economics, for example, it seems to me that there is a poor track record, and no grounds for supposing that what is ‘generally accepted, orthodox’ is going to be reliable. But I do think the notion of probability is a useful one. We do not suppose that a probability estimate is necessarily reliable, only that it reflects what we currently consider to be best practice. It is a separate question as to whether the best current practice is appropriate to the case at hand. Thus it is quite possible that those who were predicting growth of about 2% through 2008 were fully justified in Ayers sense, and knowing now that there was a financial crash is no grounds for criticism. What – with hindsight – ought to have been questioned was the basis for those expectations. Were there empirical grounds for believing them to be adequate? In Ayer’s terms this is a question about how the ‘pseudo-mathematical’ procedures came to be accredited. My view is that it is not enough to use a procedure unless and until it ‘forfeits our confidence’: we should be actively critiquing the procedure, so that we reject or amend it as early as is practicable.


Ayer regards ‘evidence’ as justifying an increase in the probability accorded a claim, and ultimately the status of being ‘true’. He does not explain what how evidence is so used. It seems ‘rational’ (in his sense) to judge the impact of the evidence according to some accredited procedure, unless and until one is forced to face some contradiction. This may be a good description of common practice, but what if the current orthodoxy is wrong? Science seems better than this, in not dismissing as undiscussable alternative theories, even if they rely on alternative (but still credible) procedures.

Ayer seems to suppose that it is only worth considering evidence that one currently believes to be relevant. But what if one is wrong?


What I take from Ayer’s argumentation is that ordinary logic, expressed in ordinary language, should only ever be taken as expressing a belief, albeit one that is ‘generally accepted’ or orthodox. This seems right. Whatever I write will initially be read in an orthodox fashion which mere words cannot escape from. If others have had experiences which make them doubt some orthodoxy, then maybe I can communicate with them. But can mere words communicate beyond experience?

It seems to me that ordinary language and logic cannot. But maybe specialised languages and logics, such as those developed by Russell and Whitehead,  can? But to whom? Perhaps a dissatisfaction with what is generally accepted in one’s particular culture is needed, but then it seems to me that one does need, and maybe can develop, a language which transcends the current orthodoxy. Isn’t this what happens in science?


On the other hand, I think that in the appendix Ayer does give a good ‘standard’ against which to judge methods. They should have:

  1. Criteria as to what counts as a legitimate ‘observation’.
  2. Criteria as to what one can talk about.
  3. Criteria as to what counts as legitimate reasoning.
  4. Criteria as to how such a system of hypotheses can be accredited in practice.

It seems to me that best practice in theology, astrology and astronomy all fit this mould. The key to Ayer’s view is (1) and (2), but (4) seems to me of broader significance. For example, much ‘science’ and ‘mathematical modelling’ is part of (3), yet Keynes, for example, is clearly concerned with the logic of (4). So Ayer would seem to raise the question as to what is the proper role of logic and mathematics in empirical activity (including sciences)? (A good question.)

Literal verification

Ayer’s use of the term ‘literal’ also seems significant. For example, is science literally verifiable? If so, then we have to bound the application of such science with (1)-(4) in mind. Thus there will be many issues outside of its proper scope where it should not be taken literally. In practice, though, science is often applied literally outside of its scope, and hence in Ayer’s terms, cannot be relied upon. Yet we do.

An improvement would be to try to understand what the science ‘is really saying’, as genuine empirical propositions, and to acknowledge the ‘speculative’ gap requiring our judgement. In physics, for example, theories are ‘accredited’ by reference to characteristic experiments, so to understand physics would mean not just knowing what the current ‘laws’ are and how to apply them, but in understanding the critical experiments and how they seem to justify the laws, so that one would recognize when one had a novel situation, not covered by the ‘laws’. Thus, particle physicists don’t simply assume that their theories will be verified by experiments at ever higher energies. Thus, perhaps an addendum to the above criteria is required:

  • A characterisation of what has actually been taken account of in accrediting the theory.


Much of Ayer is most readily understood from the perspective of a single theoriser. Even if we have a theory, such as astrology, that meets the above criteria (perhaps generously interpreted) their remains an important question: to what extent is any success due to theory, and how much to its application? What we seem to need is something like:

  • A characterisation of the sort of person and the sort of circumstances for which the ‘accreditation’ applies.

Strategic Thinking

Ayer acknowledges uncertainty, and the likelihood of unintended consequences, and can easily read as consistent with the view that one should always take things ‘one step at a time’. For those for whom such a view is orthodox, he doesn’t seem to caution against it. Yet sometimes (as in game theory) it makes no sense to try to justify individual acts, only long-term strategies.

Ayer’s approach seems reasonable when what is ‘accredited’ and orthodox is reliable, but unfortunately there seem to be many dangerous beliefs that are accredited and orthodox within particular cultures. (Intellectually, I know that I should add ‘including my own’, but of course we all know that we have now reached high point in our culture, and the end of its need to develop further.)


Ayer does urge a ‘pragmatic treatment of history’. It seems to me that Ayer’s pragmatism was consistent with the attitudes that led to World War only 3 years later. But maybe it could work with a reformed orthodoxy?


Taken too literally, Ayer’s conclusion seems to be that whatever is literally verifiable and considered to be sufficiently probable is ‘true’. But this seems to depend on his pragmatic assumption, that one should act – ‘rationally’ – on one’s beliefs, ignoring any uncertainty. This seems dangerous. On the other hand, Ayer’s work continues to provide some useful ‘food for thought’.

I would rather say that any written claim can only be ‘true’ relative to some assumptions, and it seems a good idea to make those as explicit as one can. One implicit assumption seems to be that the experience on which one’s methods were accredited remain adequate. A way to challenge this would be to make the argument for accreditation, including the logic, more explicit, and to consider ways in which the current circumstances may be novel. Such inventions may be not rational, in the sense that one has to invent possible explanations for the current situation for which there is currently no clear evidence, but they may still be reasonable and useful.

One way to ‘square’ this with the broader view of pragmatism, that one’s actions reveal one’s effective beliefs is that acting with ‘full confidence’ reveals an unconditional belief, whereas acting, as suggested above, in full recognition of the inevitable uncertainties and with at least a ‘promise to a future self’ to attempt to watch out for and deal with the ‘unintended consequences’ reveals a ‘belief that it is appropriate to act in this way’ without entailing a commitment to the ‘truth’ of whatever one’s immediate actions suggest as being one’s ‘absolute beliefs’.



Dave Marsay

%d bloggers like this: