Philosophical Connections: Ayer

The son of a French-Swiss timber merchant and a Dutch Jewish mother, Sir Alfred Ayer was educated at Eton (where he was a militant atheist), Christ Church, Oxford, and at the University of Vienna. In Vienna he attended meetings run by Carnap and other logical positivists. His espousal of this philosophy and his publication of Language, Truth and Logic in 1936 shook the conservative establishment at Oxford, to which he had returned as lecturer at Christ Church three years earlier. After war service he became Dean of Wadham College, Oxford and in 1946 was appointed Grote Professor of the Philosophy of Mind and Logic at London University. In 1949 he was back again at Oxford as Wykeham Professorship of Logic and a Fellow of New College. He was elected Fellow of the British Academy in 1952 and was knighted in 1970. He was President of the Humanist Association for six years, and was also an enthusiastic football fan (Tottenham Hotspur).

PHILOSOPHY OF LANGUAGE AND LOGIC

[1] Ayer [Language, Truth and Logic, ch. IV] distinguished two classes of significant propositions. (1) Synthetic propositions, which concern matters of fact, are testable by observation, and thus are known a posteriori. (2) The formal propositions of logic and mathematics, on the other hand, which have nothing to say about the world, are said to be analytic and therefore tautological, in that they are true 'conventionally' by virtue of the meanings of their constituent symbols, and are thus known a priori. Any propositions which do not fall into either of these two classes Ayer held to be 'metaphysical' and non-sensical [a]. Underlying this position was his commitment to the verification principle: a proposition is significant if and only if it is either a tautology or empirically verifiable. Ayer in fact distinguished between a strong and weak form of the principle. As the strong form — that a proposition is meaningless unless its truth can be conclusively established by observation — seems to rule out universal laws or propositions about the past, he confined himself to the weak form, which requires only that some observation should be 'relevant' to the determination of a proposition's truth or falsity [b]. In his introduction to the second edition of Language, Truth and Logic, in response to criticism, he modified his statement of the verification principle still further. He first of all distinguished between sentences, statements, and propositions. Sentences are grammatically significant sets of words or symbols; and all indicative sentences, whether or not they are literally meaningful, express statements. Statements which are literally meaningful he termed 'propositions'. (For the sake of brevity he applied the principle of verification to statements rather to the sentences which express them.) He defined a proposition as "a class of sentences which have the same intensional [N.B. 'intenSional'] significance for anyone who understands them" [ch. V]. He rejected the view that propositions were real entities or the objects of 'intentional attitudes' [N.B. 'intenTional'] [c] [See also Central Questions of Philosophy, ch. IX]. Recognising the vagueness of his usage of the term 'relevant' he proposed a version of the principle expressed in terms of deduction of an observation-statement in conjunction with certain other premisses, without its being deducible from them alone. But as this allowed meaning to any indicative statement whatsoever he reformulated the principle once more, now distinguishing between direct and indirect verifiability. As to the status of the principle itself, it was pointed out by critics that it was neither a tautology nor a deduction from observation-statements. Ayer therefore said that he regarded it (in the weak form) not as an empirical hypothesis but as a stipulative definition [LTL, 2^nd edn, Introduction]; and that it perhaps should have been interpreted as providing a means to demarcate literal sense from nonsense [CQP, II A] [d].

Having, as he thought, eliminated metaphysics Ayer supposed he could pursue philosophy as an exercise in analysis, in the sense that its job is to investigate the logical relations and intertranslatability between various classes and levels of statements (object language, sense-experience language, scientific statements, 'common sense' ones). Philosophy is therefore now to be understood properly as a "department of logic" [LTL, ch. II]. Underlying Ayer's general approach, particularly in his later writings, is the view that philosophy is a quest for justification [e]. He was therefore critical of the methods and assumptions of 'ordinary language' philosophy, which he felt is not particularly illuminating or helpful in dealing with philosophical problems. And while he allowed [The Problem of Knowledge, ch. II, v] that words must be used in accordance with a set of rules if they are to be given descriptive meaning, he said it is open to individuals to use rules of their own without it being essential that they should be capable of being publicly checked. In principle Ayer therefore allows the possibility of a 'private language' [f].

KNOWLEDGE

[2] In his account of perception Ayer initially [Foundations of Empirical Knowledge and 'Phenomenalism'] promoted a theory of linguistic phenomenalism. He argued that observation cannot help us to decide between 'realist' and 'sense-datum' theories. He saw it as a question of 'convenience' whether we adopt the realist view that material objects can possess different colours at the same time or the sense-datum view that they cannot. It comes down, in the last analysis, to a decision to use a technical 'sense-datum' language, into which sentences about material objects can be translated, at least in principle [FEK, I] [a]. The sense-datum language, Ayer argues, is best able to deal with the problem of illusion. He accepted, however, that the latter can never formally specify material objects precisely, so that we cannot analyse statements about, say, a table into a set of statements about sense-data. [See Problem of Knowledge, ch. III, vi.] Nevertheless the attempt to construct assertions about material objects out of empirical data is still legitimate.

In the Problem of Knowledge [ch. III, vii] Ayer also considered the issue of perceptual knowledge in the context of his examination of justification in the face of sustained assaults by sceptics [ch. II]. In earlier writings [for example, Language, Truth and Logic, Introd; and 'Basic Propositions'] he had supposed there to be 'basic propositions' (such as 'This looks to me to be red') which are 'incorrigible'. He had already criticized the claims of the 'cogito' [LTL, II; see also PK, II, iii], but he also later came to reject the empiricists' claim to incorrigibility [PK, II, vi; see also FEK, II, 8] and agreed that the sceptic is correct in denying that sense-datum statements are either equivalent to or are a logically conclusive proof for material object statements. However, he says the former can be used for judging the latter — this constitutes 'justification'. Indeed, having knowledge as such was regarded by Ayer as having the right to be sure about one's belief in the truth of a statement; and it is a matter of how strict we wish to make the criteria. But the logical possibility of error cannot be ruled out. And this could occur even in one's own 'private language' [see 1g], for making mistakes in assessing one's experience need not be the same as making mistakes in one's choice of the correct words to describe it. [See PK, chs 2, v; on sense-data see also Metaphysics and Commonsense, 'A Reply to Austin', and CQP, ch. IV A.]

The notion of justification is discussed further in The Central Questions of Philosophy [ch. VIII C]. Here Ayer says that a belief is justified if it accords with the available evidence, or is derivable from some wider generalization [b]. However, this raises the problem whether it is sufficient that the propositions used to justify a belief are true. This would give us a criterion for justification, but it does not require us to know that the criterion has been complied with. Is it therefore necessary that we should have a good reason to believe that they are true? If so, we run the risk of an infinite regress. We can stop this only by making special rulings on the basis that there is evidence of a given strength in its favour.

In Problem of Knowledge [ch. II, ix] and Central Questions [III E] Ayer identified three standpoints characterizable by their different responses to the sceptic. (1) The naive realist denies there is a gap to be bridged between sense-impressions and physical objects: we perceive such objects directly. (2) The reductionist (including early Ayer) accepts the sceptic's objections on this point but argues that statements about physical objects are translatable into statements about sense-impressions. If the sceptic opposes this claim, it is open to those who adopt a 'scientific' approach to argue (3) that the existence of physical objects is a probable hypothesis which one is justified in accepting because of the way it accounts for our experience. Even if this claim is rejected, it can be argued that justification need not be confined to either deductive or inductive procedures as the sceptic seems to suppose.

Ayer's later theory of perception [The Origins of Pragmatism] is based on a distinction between a realist theory of being, which gives primacy to physical objects, and an 'empiricist-sensory' theory of knowledge — he now refers to our experiences as 'qualia' [c]. Priority in knowledge is thus no longer coincident with what is supposed to be prior in being. He argues further that out of the spatio-temporal relations which are given to us directly in our sense-experience we can construct a system of physical space and time. The ontological issue now is whether the realm of physical objects is best understood in terms of naïve, commonsense 'realism' (tables and chairs, and people) or of science (atoms and molecules, etc) [see 3c].

Ayer's general methodological approach is also apparent in his treatment of problems of the self, other minds, and memory. Earlier [LTL, Introd. and ch. VII] he had argued for a theory that statements about other people's minds are translatable or 'constructible' into statements about their bodies. However this leads to the problem that we must consistently regard statements about our own minds as being equivalent to statement about our own bodies. Rejecting this behaviourist analysis Ayer therefore argues [CQP, ch. VI E; see also 'One's Knowledge of Other Minds'] that the accounting for the behaviour of others by analogously attributing to them conscious thoughts, sensations, emotions, purposes, like those I directly experience in myself, is a consequence of accepting a "whole body of theory". One does not have to rely on inductivism [as he does in The Problem of Knowledge]. Similarly it is possible that my memories and therefore statements about the past are mistaken. But notwithstanding the logical possibility that an occurrence of a memory experience may be logically consistent with the non-existence of the previous event of which it purports to be a memory, the realistic view that memory is trustworthy is simpler [d], "besides supplying the goods in which the other theory [namely, memories of an unreal past] trades".

METAPHYSICS

[3] Ayer's early logical positivist philosophy was iconoclastic and explicitly anti-metaphysical — in virtue of the verification theory of meaning. However, his later writings [see especially Metaphysics and Common Sense] suggest he had acquired a more sympathetic attitude. He now regarded metaphysics as an attempt to build a system which would accommodate contemporary scientific concepts and principles. Metaphysics was thus seen as a kind of revisionary and constructive analysis [a], and as having some explanatory value as a 'secondary system' [CQP, III C] [see below 3c]. In Language, Truth and Logic he seemed to be committed to a 'neutral monist' ontology. Central to this was the claim that 'entities' such as physical objects and minds could be turned into logical constructions (namely, 'sets of experiences) by a process of reductive analysis and thereby be shown to be 'unreal' [b]. However, he came to adopt the view that one must choose between one of two supposedly conflicting realist 'systems' — the scientific account of physical objects (in terms of particles, and so on) and the common sense or realist view that the world consists of tables, trees, and the like. In Central Questions [ch. VE] he discusses the scientific theory in terms of primary and a secondary system. The primary system includes the data (of immediate perception) and the propositions that support the theory, while the theory as such belongs to the secondary system (which also includes 'entities' which cannot be identified with objects of the primary system but which are conceptual tools for arranging the primary facts). In the case of conflict the one system is treated as concerned with fact, the other as explanatory. Because the distinction between fact and theory is only relative, Ayer thinks we have some choice as to where to draw the line, but says a "reasonable decision" can be made. The question of what actually exists is an empirical one to be considered within the framework of the theory which supplies criteria for answering it [c]. [See also CQP, VII B. and The Origins of Pragmatism.]

On the question of minds, Ayer [CQP, VI B] rejects claims that there are mental substances or transcendental egos. We have no empirical or other grounds for asserting the existence of such entities. There is no self other than inter-relations of experiential data. The claim to self-consciousness, he says, is just the claim that one's present or past experiences are one's own. Moreover, 'I' and 'my body' are not simply substitutable. Rather, "my body is grammatically represented as one of my possessions". However, he allows that grammar is not always a safe guide to the facts [d]. The use of the personal pronoun commits us to no more than is strictly necessary for the establishment of one's self-identity; and experiences suitably related to one's body and to each other are sufficient or this purpose. Our sensations, perceptions, and thoughts are indeed mental acts: but for Ayer they do not have 'intentional' objects; rather they are to be understood as predicable — to 'persons' regarded as physical objects (which, like other physical objects are the "product of theory"). Thus Ayer does not deny we have an inner private life; and he rejects 'physicalist', that is, behavioural or dispositional theories as well as functionalist and identity theories of mind [CQP, VI D; see also Philosophy in the Twentieth Century, passim] [e].

METHODOLOGY

[4] The distinction between primary and secondary systems is employed also in Ayer's analysis of explanation [CQP, VII] and causation [VII C, VIII D]. He says that the ascription of causality involves nothing more than a de facto correlation or constant conjunction of facts, and that this implies reference to a generalization of a law-like character in that it can be 'projected' over undetermined or imaginary instances. (He argues for causal terms to be considered as facts rather than as events, because in this way they can accommodate 'negative' causes, and they fit in better with the complexity of usage to cover states of affairs at different observational or theoretical levels.) Correlations occur at the primary level while projections take place at the explanatory secondary level [a]. However, he argues that not all constant conjunctions are causal: some correlations are accidental, and many may relate to human decisions and actions, and the generalizations we are 'disposed' to project need be no more than "generalizations of tendency". The difference between accidental generalizations and generalizations of law consists in a difference in our attitude towards them. Generalizations of fact are more 'vulnerable' to new information in a way that generalizations of law are not. But the latter entail 'unfulfilled conditionals'. A particular event upon which we base a projection is said to be the cause and a 'necessary' condition only in the sense that the behaviour would not have occurred without the event (that is, the corresponding statement is a counterfactual conditional). In general, rejecting the distinction between motives and causes, Ayer considers the explanatory model of the natural sciences can be extended to the human sciences [see 'Man as a Subject for Science'] [b]. In universal generalizations, he suggests, there are various features which lead us to single out one factor rather than another as the cause (for example, the results of our own actions, the role played by events in a wider explanatory system, and so on). Otherwise identification of such factors is arbitrary. In the last analysis, cause and effect have their place only in our imaginative arrangements and extensions of the facts at the primary level.

Ayer's account of explanation requires theories to fit the facts, that is, if they are to have any explanatory value they must be empirically testable. This leads to the problem of induction [CQP, VII A, VIII A] (though he says that simple inductive procedures — achieving universal hypotheses by generalizing from observed instances — have a part to play in comparison with the advancing of theories by connecting events in novel ways). The assumption that event A will be followed by event B (it having done so on a previous occasion) presupposes that nature is uniform. However, he argues [Probability and Evidence and CQP, VIII A] that in formulating the principle we must avoid making it either too strong or too weak. If it is made too strong, we force it into a deductive form (premisses, conclusion) which will be invalidated by any exception. If it is too weak, it will be consistent with any sequence of events. It is legitimate, indeed necessary, to seek for 'backing', but Ayer accepts that the possibility of error cannot be eliminated [c]. As noted above [see 2b], there is a problem with attempts to justify our beliefs. There is also a problem with evidence. Confirmation procedures are beset with paradoxes (such as Hempel's). Even the falsificationist approach is inadequate [CQP, VIII C] [d]. It is an inductive step, Ayer says, to assume that a theory which has passed a variety of testing procedures leading to increasing degrees of corroboration is a better guide to the future than one which as not been tested or which has but has been found wanting. Morever, we do require hypotheses to be confirmed. Consider the hypothesis that malaria is contracted as a result of a mosquito bite. According to the falsificationist we should not be interested in a person bitten by the insect if we did not know whether he had malaria, because his case could not falsify the hypothesis. But, according to Ayer, if the experimenter discovered that the subject had contracted the disease this would be taken as confirmation.

The concepts of backing and testing are connected with that of probability; and this is also relevant to Ayer's concern with the assessment of evidence in the context of his critique of scepticism and his analysis of justification. [See CQP, VIII, B; also Probability and Evidence.] To explain probability Ayer discusses three senses of the term used in different kinds of statements, and which he says must be kept distinct. (1) A priori statements such as those about, say, the throwing of dice. These relate to the mathematical calculus of chance and has nothing to say about the likelihood of actual events. (2) Statistical statements, such as that about the probability that an unborn child will be a boy. These relate to a frequency theory referring to classes of events, and therefore are not helpful in individual cases, as an individual may belong simultaneously to different classes and this gives rise to problems of choice. (3) Statements of 'credibility'. These may be based on statistics. But Ayer rejects the view that they can be understood in terms of a logical relation; for (i) they then become analytic; (ii) the probability may vary according to the evidence. There can be no grounds for deciding between different statements all of which are logically true. How then is evidence assessed? Can we be sure it has all been gathered in a given explanatory situation? Is there not an arbitrary element involved? Ayer argues that "'p' is probable" is to be understood as meaning that it is reasonable to believe 'p' rather than as a qualified assertion of 'p' [e]. We must therefore appeal to all relevant factors, negative or positive. But this gives rise to the problem of the reliability of sample instances. Are they fair or deviant? Ayer says they must be supposed to be fair; we can go no further.

Ayer's conclusion from his discussion of paradoxes and probability that, whatever the evidence, we still have some latitude in our choice of the hypotheses we are going to project. We can offer different criteria or standards of rationality, but in the last analysis the only test is whether the method of choosing hypotheses 'works' — whether the past proves to be a successful guide to the future [f]. And by 'success' he means the likelihood of our being correct, as measured in terms of the theories we accept.

ETHICS/ AESTHETICS

[5] Ayer's views on ethics were first set out in Language, Truth and Logic [ch. VI] and generally remained unchanged in later essays [see, for example, 'On the Analysis of Moral Judgements']. (He notes that what he will say about statements of ethics applies to statements of aesthetics, and indeed to all 'judgements of value'.) He starts by dividing ethical statements into four classes: (1) propositions which express definitions of ethical terms or judgements about such definitions; (2) propositions which describe phenomena of ordinary experience; (3) "exhortations to moral virtue"; and (4) actual ethical judgements. He then confines his discussion to the first class, which he says are the only propositions properly belonging to ethical philosophy. (The others are either propositions of psychology or sociology or are not really propositions at all.) Ayer denies that 'normative' sentences (for example, 'x is wrong') are equivalent to sentences expressing any kind of empirical propositions and so he rejects two kinds of naturalistic moral philosophy — subjectivism and utilitarianism [a]. According to one version of the former, to call an action right, or a thing good, is to say that it is generally approved. But Ayer argues that no self-contradiction is involved in saying that some generally approved of action is not right, or a thing not approved of is good. Similarly, in the case of utilitarianism, there is no self-contradiction involved in asserting that it is sometimes wrong to perform the action which would actually or probably cause the greatest happiness. He also rejects 'intuitionist' or absolutist' theories, as they provide no criterion for determining the validity of moral judgements. Although according to such theories moral judgements are held to be synthetic, they are also regarded as not empirically verifiable. However, Ayer agrees with the absolutists that fundamental ethical concepts (good, duty, obligation) are unanalysable. However, he regards them as 'pseudo' concepts [b], in that they add nothing to the factual content of statements such as 'You stole the money' in 'You were wrong to steal the money'. His own position is that ethical and aesthetic statements have no objective validity. Ethical and aesthetic concepts not only serve to express feeling; they also an emotive function, namely, to arouse feelings, for example, of moral disapproval, so as to stimulate action. And some are used in such a way as to give the sentences in which they are used the effect of commands [c]. In cases of dispute, argument is not about values but about questions of fact — whether, for example, a person has misjudged the consequences of an action, or the motivation of the agent, and so on.

CRITICAL SUMMARY

Although in the course of his long career Ayer moved some way from the youthful radicalism of Language, Truth and Logic, he remained firmly committed to the basic assumptions of empiricism. Philosophy is regarded as essentially the "logic of science" — the analysis of interrelations and translatability between different classes and levels of statements. However, the modifications he introduced are significant. He came to be less concerned to eliminate metaphysics than to investigate how knowledge claims might be justified in the face of scepticism; and this led him to examine the concepts of evidence and probability. His broadly pragmatic approach and appeal to 'reasonableness' allowed it to be a matter of choice how strong one's criteria for knowledge should be. His linguistic phenomenalism gave way to a more physicalist approach — later grounded in his distinction between primary (factual) and secondary (explanatory) systems. Persons too are physical objects, the 'self' being understood in terms of interrelated experiential data. He preserved the privacy of 'inner' life against behaviourism and dispositional theories but also rejected Cartesian or transcendental accounts. Generally the question of what there is is a matter to be determined by criteria appropriate to the theoretical framework within which one is by choice operating.

Perhaps the main objection to Ayer's general approach comes from linguistic philosophers. They criticize his ready acceptance that our conceptual scheme may be modified as utility requires without any recognition of the alleged primacy of our everyday publicly validated discourse, and also his conviction that any change has to be effected in conjunction with our scientific theories. The distinction he made between primary and secondary systems, and within which he accommodated his account of physical objects and minds, is thus rejected. It has also been claimed that Ayer is too firmly wedded to the distinction between logical and factual statements. Recent work has raised questions about the supposed a priori-analytic-necessary and a posteriori-synthetic-contingent parallelism. And although he had been receptive to Quine's ontology, Ayer never responded to his critique of the analytic-synthetic distinction itself. His ethical theory has also elicited considerable critical response. It has been said that while ethical statements do have an evaluative function they also have both prescriptive and descriptive functions. Even if this alternative account is equally contentious, most philosophers would accept, as against Ayer, that ethical statements are more than just expressions of feelings. Moral terms have 'meaning' (however this may be construed). It is also questionable whether the notion of a moral dispute can have purchase given Ayer's premisses. Is the dispute just about, say, misunderstanding or motives, or about consequences?

READING

Ayer: [of many writings] Language, Truth and Knowledge (1936; 2^nd revised edn 1946); The Foundations of Empirical Knowledge (1940); Philosophical Essays (1954) — includes 'Phenomenalism' (1947-8), 'On the Analysis of Moral Judgements' (1949) 'Basic Propositions' (1950), and 'One's Knowledge of Other Minds' (1953); The Problem of Knowledge (1956); The Concept of a Person and Other Essays (1963); Metaphysics and Common Sense (1969) — includes 'Man as a Subject for Science' (1964) and 'Has Austin Refuted the Sense-data Theory?' (1967); Probability and Evidence (1972), and The Central Questions of Philosophy (1973).

Studies

J. Foster, A.J. Ayer.

O. Hanfling, Ayer.

Collections of essays

T. Honderich, T., Essays on A.J. Ayer.

L. E. Hahn (ed.), The Philosophy of A. J Ayer.

Philosophical Connections

Compiled by Anthony Harrison-Barbet

AYER

PHILOSOPHY OF LANGUAGE AND LOGIC

KNOWLEDGE

METAPHYSICS

METHODOLOGY

ETHICS/ AESTHETICS

CRITICAL SUMMARY

READING

CONNECTIONS

Ayer