I think that one of the main objections to be made to Ayer's verifiability criterion is simply the mechanical way in which it is designed to work: supposedly, a philosopher need not study, for example, how religious assertions are used, nor what sorts of illumination their users take themselves to be shedding on the human condition; instead, Ayer imagines that we can test them in a simple way that requires us to do no exploration whatever. This, surely, is hubris; and that accounts for much of the very angry and passionate reaction which Ayer's criterion provoked.

The problem is that Ayer is taking as his inspiration or model a very simple notion of one's describing a familiar sort of object before him. But when one turns to other enterprises of much greater complexity and nuance, this simple model just doesn't help. If a music critic, e.g., says that a pianist missed a particular note, that will fit the simple model; but if he says that the last movement was played lifelessly or humorously or mechanically, although we don't know exactly how to verify it (specifically, what patterns must the tape recording reveal?), still, it surely is a claim about the performance—and it's an objective claim at that!

When one thinks about it in detail, the notion that one must be able to say how to verify a claim before he can tell whether it is meaningless is rather incredible. Take, for example, a perfectly meaningful claim: "Once upon a time, the ancestors of the whale came out of the water and lived on land; but eventually, their progeny returned to the sea." Now in fact, the claim is verifiable: but I suspect that we could see that the sentence is meaningful without having devised any way at all to test for it. Take the last part of the claim: one might look for the vestiges of rear legs in the whale—and indeed, bones of the right sort are found there. Take the first part: one might look at the development of the fertilized egg (going through stages including a creature with a tadpole fin and gills) as a sort of rehearsal of its ancestral development—and indeed, that seems to be true also. But to suppose that one must have fixed his attention on these quite ingenious means of verification as a condition of his recognizing the sentence to be meaningful, is monstrous.

For that matter, even Ayer has trouble sketching exactly how we might go about verifying sentences that even he concedes are meaningful; he benignly assumes that we would have some sort of idea about their verification if we were pressed. But now consider the sorts of thing which Elizabeth Anscombe has called "institutional facts": their descriptions seems to work like factual assertions, yet we really have no idea how to reduce them solely to strictly "empirical" terms. An agreement, for example, is not the making of a mark with one's hand; one must have signed voluntarily and intentionally, whatever those can mean in sensory terms. And although we must, perforce, be able to "tell" or determine whether someone has made a promise by using our senses, yet there is no obvious way of defining "the making of a promise" in sensory terms: the things our senses detect enable us to tell that one has made a promise in virtue of our antecedently having learned what a promise is, and how promises are customarily made.

There is at least an analogy with the case of making statements about institutional facts and aesthetic and ethical judgments—and perhaps there is something stronger than merely an analogy. Isn't it a fact that Shakespeare's plays are better than many others—(choose for comparison the hack you detest most)? And it certainly looks as though we can account for such judgments by showing all sorts of objective features of Shakespeare's plays—compellingly written verse and prose, a stunning use of imagery and analogy, characterizations which highlight aspects of genuine types of people, situations which are realistic in type. And whether or not one needs to bring in an additional cognitive faculty, nevertheless, by reading Shakespeare's plays together with many others', one simply learns of the objective superiority of his plays. When we turn to ethics, it will be a mistake to concentrate on assertions such as "murder is wrong." Instead, look at nuances such as this: people grade types of murder into most horrible to less serious—e.g., as reflected in first degree, second degree, and manslaughter; and even a person who is not schooled in legal or quasi-legal vocabulary nevertheless marks these same distinctions when he compares pairs of examples of murder. Moreover, by drawing his attention to certain facts which we recognize as morally relevant, you can very often get him to change his mind. Thus, there seems to be recognized a systematic but very complicated connection between an act's having certain non-moral characteristics and its being appraised in specific moral terms; but it's very difficult to see how Ayer's theory can plausibly account for this. If we restrict our attention for the moment to a trained and sensitive moralist, then it begins to look as though from the fact that he experiences a so-called moral emotion toward some act, that we might indeed predict that the act has one of a certain range of characteristics. And that begins to suggest that such a "moral emotion" works like some sort of complex barometer of complexes of non-moral characteristics. (Notice that throughout I speak of "non-moral" characteristics because many of the characteristics will be "institutional facts," and it is not at all clear, as I mentioned earlier, just how or how adequately Ayer's radical empiricism can account for literally meaningful sentences which describe such "facts.")

We might even be able to make something of the case for religious statements' being meaningful. I take it that, among saying other things, the Old Testament says something like this: "Through our forebears, God has made a promise to us—that if we keep his laws, he will look after us and ameliorate or prevent catastrophes from occurring to us." Well, that's a sort of hypothesis, isn't it? Even if we can't see God, nevertheless we have some idea what is claimed—a relationship between our conduct and what happens to us, established by some creature who watches. And although there is no easy way to verify this, one is not completely at a loss to understand what is being implied. (Of course, it doesn't say how great a failure in one's obedience will stop God's intercession, and it doesn't specify exactly the nature of the potential disasters which might be stayed. Still, when Tommy Smothers says to his brother, "Mom always loved you best," even though this is surely verifiable, no particular act of preferential treatment or behavior is implied.)

Of course, there are many other religious assertions for which we might not be so confident that we could have an idea what they implied. Well then, perhaps some of them are meaningless: but the point is that this would seem to require its being established through a thorough investigation, and not at the moment we first consider the sentence, bare and naked on the page. If, when we consider a sentence which is certainly meaningful—such as the one I mentioned earlier about the whale—we can't reasonably be expected to be able to specify how it could be verified, then we should not expect the fact of a given sentence's being meaningless to be evident upon mere examination of its sequence of words. To borrow an example from Wittgenstein, we would initially react to the sentence, "I feel a terrible pain in the table" as to an absurdity: yet suppose it transpired that whenever you poked the table with a knife, even though I was blindfolded, I screamed; and when you smeared salve on the table, quite without my knowledge of your having done so, I volunteered the comment that the pain was getting better just on its own; then would the assertion still look absurd?

The same point holds about intellectual intuition—Ayer's bête noire: maybe there isn't anything answering to a cognitive faculty which is independent of the established five senses; but how could we know this at the outset, and without the necessity of examining each claimed instance as it arose?

Finally, Ayer's guiding intuition seems to be correct—that a sentence's being meaningful does seem to be connected with whether anything could possibly bear on its being true. But the difficulty with Ayer's theory is that a sentence can be perfectly meaningful with out our knowing or guessing what things do bear on its being true. And one can recognize that a sentence is meaningful without knowing how to begin to verify it. Think for a moment about the marvelous poem about the Nature of Things by Lucretius: he spelled out, beautifully and plausibly, an atomic theory of the material universe—and yet it was wholly an exercise in metaphysical speculation. And while it would be silly to credit him with atomic physical theory, still, it does at least appear that we can credit his sentences as meaningful. For one is inclined to say that his sketch turned out to be verifiable even though Lucretius couldn't himself have foreseen any of the details of that verification. And, again, while ethical judgments are indeed not exactly like factual ones, nevertheless there is sufficient analogy to raise the question whether ethics is a cognitive enterprise. Although it may turn out not to be so, it's hard to see how someone could responsibly conclude this without a detailed investigation of its features.