In this paper I wish to set forth as plainly and simply as possible a theory of the good which finds little modern vogue, but which nevertheless seems to me more plausible than any of those now current. The perfection theory of the good can claim weighty historical roots, especially in Plato and Aristotle; but I shall proceed systematically rather than historically. The exposition of the theory will be presented in three parts: first, a statement of a generic definition of perfection; second, an attempt to give more specific content to the definition; and third, an indication of my reasons for upholding the truth of the theory.

I would like to make some further clarifying remarks about the nature of learning machines, or finite automata as they are more generally known these days. It is clear from much that has recently been written on this subject that there are still many misunderstandings about their capacity and significance.

There is probably no student of modern philosophy, and certainly no listener to the Third Programme, who has never received the warning that he must on no account deduce an “ought” from an “is.” This prohibition, it is claimed, is securely based in established and unchallengeable principles of logic. Professor Flew was speaking for many others when he said, in the course of a broadcast entitled “Problems of Perspectives”, “I think it is very important indeed to make as clear as we can and to underline with all possible emphasis that this is a point of inexorable logic”. And Professor Popper, to take but one other example, has expressed himself no less trenchantly: “Perhaps the simplest and most important point about ethics is purely logical, I mean the impossibility to derive nontautological ethical rules—imperatives, principles of policy, aims or however we may describe them—from statements of fact”—a view that is fully endorsed by Mr. Hare in his Language of Morals.

The word “objective” is of course the trouble–maker here, Miss Smith assumes that if an aesthetic statement is held to be objective (or to have an objective reference) then it is the physical existence of the work of art (the picture or the sound of the music) that constitutes the objectivity: i.e. if a work of art is exteroceptively perceivable, then an aesthetic statement involving it is objective. Some writers, however (usually philosophical idealists) have held that in genuine works of art there is manifested an ultimate spiritual Reality (it might be called God) which we apprehend when we appreciate such works. On this theory, an aesthetic statement (“This picture is beautiful”) has an objective reference if the subject of it (the work of art) succeeds in expressing or communicating such a supersensible Reality: if it fails to do so, then the statement is subjective, i.e. it can be analysed completely into a statement about our feelings or emotions (e.g. “I like this picture”).

The central task to which contemporary moral philosophers have addressed themselves is that of listing the distinctive characteristics of moral utterances. In this paper I am concerned to propound an entirely negative thesis about these characteristics. It is widely held that it is of the essence of moral valuations that they are universalisable and prescriptive. This is the contention which I wish to deny. I shall proceed by first examining the thesis that moral judgments are necessarily and essentially universalisable and then the thesis that their distinctive function is a prescriptive one. But as the argument proceeds I shall be unable to separate the discussion of the latter thesis from that of the former.

IN delivering this lecture I am to speak on some aspect of truth. The practice of examining the various contexts in which a word may be used, in order to disclose what its usages have in common as a clue to its meaning, is of respectable antiquity. If I indulge this practice I find an “embarras de richesse” in modern philosophical literature under two headings, the truth of analytic propositions and the truth of synthetic propositions: the first deals with criteria of consistency and the second with criteria of verifiability.