336. A. J. Ayer to Harrod , 26 December 1933 [a]

[Replies to a letter not found; follows on from 332 ]

Trevin Towers, Eastbourne #

26 December 1933

Dear Roy,

I have been so slow in answering your letter that you have probably forgotten exactly what you said. Therefore I will deal with your objections generally rather than one by one and hope, if not to convince you, at least to make my own position clear.

You accuse me of not facing up to the difficulty about probability. If I maintained that prima facie propositions were senseless unless they were conclusively verifiable you would have an unobservable case here. But I do not maintain this, though I admit that my unfortunate use of the word verification is calculated to make you suppose that I do. [1] I do not think that a finite series of experiences [b] can even in principle conclusively establish the truth of an empirical proposition. There will always remain the bare possibility that it is false. But the truth of every significant proposition can be made more probable--can be confirmed [c] in experience--and nothing but experience is relevant to establishing its truth. You can not have a significant non-tautological proposition, I hold, which no conceivable experience would go any way to verify nor is there anything rather than experience (& ultimately "my" experience) by which the truth of a proposition is rendered more probable.

I lay stress on this phrase conceivable [d] experiences. [e] I do not deny that a great many--indeed the vast majority--of significant propositions will not actually be verified in my future experience. But these are all such as possible--i.e. imaginable--experience [f] could verify, in my loose sense of verify in which a proposition is verified in any experience which makes its truth more probable. To understand a proposition I must know what experiences would verify it. And then the proposition is equivalent to propositions describing those experiences.

I think my admission that empirical propositions can never have their truth made more than highly probable dispenses anyhow of one of your objections. You said that on my view the proposition "if I move away from the fire I shall feel colder than if I stand by it" would be meaningless. Now it is <certainly> [g] the case that I am not justified in being absolutely sure of the truth of this proposition, inasmuch as I can not simultaneously be near the fire and further away from it. But I can find empirical evidence to confirm its truth--e.g. the feeling of heat when I am sitting near it combined with the memory of having felt cold when I was standing away from it--evidence that I said I was cold or the evidence of people who touched me etc.--combined with the observation that circumstances have not altered so that when I move away again I am likely to feel as before. All this needs to be put more precisely but it is sufficient to indicate the sort of experience whose relevance in such a case saves the proposition from being meaningless.

But the main point of difference between us still remains untouched. I do not think I can add much to what I said in my previous letter and in the original article. [2] I do not deny that "other minds" exist or that they have emotions similar to mine--I think the truth of these propositions is as highly probable as that of almost any other--but I think they are translatable into propositions describing the experiences actual and possible on account of which they become so highly probable and these experiences are all of them observations of other people's bodies. I can not see that my recollection that when I felt angry my body had an appearance very similar to the appearance of your body now is itself one of the experiences that go to verify "you are angry". The most I concede is that it may cause the use of the same word angry to refer to my feeling and your bodily state. I tried to show that the theory of logical behaviourism was "psychologically unsatisfying to most people" in order to explain its almost universal rejection. [3] But of course those who disbelieve it can always answer that it is rejected not because it is psychologically unsatisfying but because it is seen to be false. But as you said the question whether the theory is psychologically satisfying or not is not relevant to the argument by which I seek to prove it true.

In conclusion you criticise me for attempting to re-introduce a spurious certitude into philosophy. But I am not sure that certitude in philosophy is necessarily spurious. We have given up, admittedly, the attempt to depict reality in metaphysical systems, and we recognise that the propositions of natural science which do depict reality in a sense can never become more than highly probable: but perhaps the propositions of the new philosophy may aspire to certainty first because they are not competing with the propositions of science, second [h] because they are not descriptions of reality in the same sense. The propositions of philosophy purport to give our meaning--to give an analysis of certain very general expressions that we use. It might be said that our meaning being something wholly within our control can be known by us with certainty 1 . On the opposite side we have Ramsey's view, which perhaps you share, that the analysis of meaning is a psychological as well as a logical problem. [4] And if that is so then the propositions of philosophy would be a sub-class of the propositions of analytical science, the science of psychology and their truth could never even in principle be established with certainty. I am inclined to think that Ramsey is right but I have not made up my mind about this. I should be glad to have your views on this question. Do you agree with Moore that philosophy is not concerned with justifying our beliefs but with analysing what is believed. [5] And what do you think logical analysis consists in? I think that for the growing number of philosophers including myself who say that philosophising consists of the analysis of expression <+> [i] not make this at all clear.



  1. 1. Letter 332 , [jump to page] .

    2. A. J. Ayer, "The Case for Behaviourism", The New Oxford Outlook 1:2, November 1933, pp. 229-42.

    3. On the criticisms to behaviorism see in particular "The Case for Behaviourism", pp. 229-30.

    4. F. P. Ramsey, "Facts and Propositions", Aristotelian Society Supplementary Volume VII, 1927, pp. 153-70.

    5. Moore's paper, later cited in this connection in Ayer's Language, Truth and Logic (1936, p. 52), is referred to in note 12 to letter 332 .

    1. a. ALS, ten pages on five leaves. The envelope is addressed to R. F. Harrod, Esq, 51 Campden Hill Square, London W8; it is marked "Probability from A. J. Ayer" in Harrod's post-war handwriting. In HP IV-22-65/3.

      b. Ms: «experience».

      c. Ms: dashed underlining.

      d. Ms: dashed underlining.

      e. Ms: there is no full stop.

      f. Ms: «imaginable experience».

      g. The only legible letters are "cert" and the final "tly", which suggest a misspelling for "certainly".

      h. Ms: «first».

      i. Two illegible words (perhaps: <or what>; but this solution does not seem to make grammatical sense).

1. on this view philosophical propositions would be not so much statements of fact as declarations (they can not be simply this) of our intention as to use words in a certain way. It would be implied of course that this was the way in which we had been using them hitherto but this the philosopher assumes & does not attempt to prove. The proof of it is left to the historian or the psychologists and their statement admittedly cannot rise above probability [Ayer's note on top of page].

Welcome page

top of page

Return to index of this section

Go to previous page

Go to next page