A.J. Ayer, answered from the texts and cited to the page.
Statements about physical objects are reducible — with substantial complication — to statements about actual and counterfactual sense-data.1 The empirical foundation is sense-experience: what is immediately given to consciousness in perception. Physical objects are not directly perceived; they are constructed from the patterns of actual and possible sense-data, and statements about them are translatable, in principle, into statements about what is and would be experienced under specified circumstances.
The strict 1940 phenomenalism in The Foundations of Empirical Knowledge faced genuine difficulties — the impossibility of completing the translation, the need for indefinite hypotheticals, the problem of the unobserved object — that were addressed across The Problem of Knowledge (1956) and resolved (so far as they can be resolved) in The Central Questions of Philosophy (1973).
The mature position preserves the empiricist core (no knowledge of physical objects independent of the sense- experience that grounds it) while admitting that the reductive translation programme cannot be carried through in strict form; the relation between sense-experience and physical-object knowledge is constitutive rather than translational.23
Naive realism (the view that we directly perceive physical objects) and the causal theory of perception (which posits unperceived material substrates causing our sense-data) both face difficulties the phenomenalist account avoids; Austin's Sense and Sensibilia (1962) attacked the sense-data framework but the attack succeeds against caricatures rather than against the considered position.45
It does not mean that physical objects are literally composed of sense-data, or that physical objects are fictions and only sense-data real. It means simply that statements about physical objects are somehow reducible to statements about sense-data, or, as it is sometimes put, that to say anything about a physical object is to say something, though not necessarily the same thing, about sense-data.Philosophical Essays, pp. 148–149
Whether the translation can even theoretically be carried out in any of these instances is another question. As we shall see, there are strong reasons for concluding that the phenomenalist's 'reduction' is not feasible; but its possibility cannot be excluded merely on the ground that it substitutes hypothetical statements at one level for categorical statements at another. A puzzling feature of Mr.The Problem of Knowledge, pp. 143–144
It is an empirical question whether the contents, say, of a visual-field are more accurately to be described as patches of colour or coloured 'objects'. But even if it is decided, on empirical grounds, that what is seen is, in some sense, a coloured 'object', it will still be a sense-datum, according to our usage.Philosophical Essays, pp. 147–148
John Stuart Mill, who held a phenomenalist position, summarized it by describing physical objects as 'permanent possibilities of sensation'.!' But a permanent possibility of sensation is not something that can very well be pictured.The Problem of Knowledge, p. 144
On any view, there is a problem about one's knowledge of other people's experiences, but for a theory of this kind it is especially acute. An alternative theory, which tends to find favour with amateurs of science, is that we do not perceive physical objects as they really are, but only the effects of their action upon us.Metaphysics and Common Sense, pp. 9–10
The phenomenalist account in The Foundations of Empirical Knowledge (1940) was a major contribution to the British sense-data tradition (Russell, Moore, Broad, Price, Lewis). Austin's posthumous Sense and Sensibilia (1962) was the major hostile attack and is widely held to have refuted the sense-data framework; Ayer's responses (in The Concept of a Person 1963 and across later writings) preserve the empiricist programme while modifying the strict reductive claim. The Central Questions of Philosophy (1973 Gifford Lectures) develops the modified position. Foster, Sosa, and Park engage the construction project carefully in the Hahn volume. The phenomenalist programme is widely held to have been superseded; Ayer himself acknowledged the difficulties while defending the empiricist core.