a) An Outline of the Argument

The argument from hallucination is supposed to show that we never see perceptual objects but see only sensations (sense-impressions). This conclusion is so obviously false that we know that the argument cannot possibly be sound. The value of the argument consists, in my opinion, in the fact that it highlights a very important ontological puzzle. As an argument for idealism or phenomenalism, it is doomed from the start.

The argument has two parts. The conclusion of the first part is that in a hallucination, a person sees sensations rather than perceptual objects. Let us follow Ayer’s formulation of the argument for the details (A. J. Ayer, The Problem of Knowledge, pp. 98-99). Macbeth had a hallucination in which he saw a dagger. But, of course, there was no dagger before him. Since there was no dagger in front of him, he could not really have seen one. So, what did he see? Ayer answers: “if we are to say that he saw anything, it must have been something that was accessible to him alone, something that existed only so long as this particular experience lasted; in short, a sense-datum.” What we see in hallucinations, therefore, are sense-impressions.

In the second part of the argument, we are supposed to be convinced that what is true for hallucinations holds for all perceptual situations: In all perceptual situations, we perceive sense-impressions. The crucial premise is that ordinary perceptual situations do not differ phenomenally from hallucinations and must, therefore, yield under analysis the same ingredients. It follows presumably from this assumption that even in normal situations, we perceive nothing but sense-impressions. For example, when you see a dagger before you on the table, you, just like Macbeth, are seeing not a perceptual object, but merely a visual sensation. What we consider to be a real dagger, in distinction to Macbeth’s imaginary one, is therefore really a sense-impression (or a bundle or family of sense-impressions).

It is clear that the phenomenalist who accepts this argument faces the same urgent problem as Berkeley: If the real dagger is just another sense-impression, how does veridical perception differ from hallucination? And the only half-way plausible answer is Berkeley’s answer: These two kinds of situation can only differ in the degrees of their coherence. But this answer is quite obviously false. We know that in the case of veridical perception there is a dagger in front of you on the table, while in the hallucination there is no dagger before you. And this proves again that the argument from hallucination cannot possibly be sound.

There can be no doubt that Macbeth sees a dagger. There can be no doubt that he experiences an act of seeing a dagger. And there can also be no doubt that there is no dagger before him to be seen. The first part of the argument from hallucination challenges us to resolve this apparent contradiction. The proponent of the argument resolves it by claiming that the act of seeing does indeed have an object, but that that object is not a dagger. It is true, so he reasons, that Macbeth sees something. But it is also true that what he sees is not a dagger. What is it? It is a sense-impression (or a family of sense-impressions). From our point of view, quite independently of the rest of the argument, this conclusion is an absurdity. According to our view, one can no more see a sense-impression than one can see a pain. And what holds for seeing, holds for perception in general: One can only perceive perceptual objects, never sensations. What Macbeth sees, therefore, cannot possibly be a sense-impression. Does he then see a perceptual object? In one sense he does, in another, he does not. But there is no mystery. Since we do not want to mislead someone who does not know the story, we cannot just say that he saw a dagger and let it go at that. There was no dagger; Macbeth had a hallucination. But there occurred a seeing of a dagger and not, for example, an experience of a shrill auditory sensation or the seeing of a pink rat. How, precisely, we make clear to someone that even though there occurred the seeing of a dagger, there was no dagger before Macbeth, depends on the situation. What is important is the fact that we all know that Macbeth could have seen a dagger, even when there was no dagger. We know that one can see a dagger when there is none to be seen. We know that this kind of experience can occur. But does not our insistence that Macbeth experienced an act of seeing imply that some object or other must have been seen, even though this object cannot be a dagger? No, it does not! A mental act of seeing can occur, even when its object does not exist. “It is only if we artificially combine the decision to say that the victim of a hallucination is seeing something,” Ayer correctly observes, “with the ruling that what is seen must exist, that we secure the introduction of sense-data” (Ayer, 1956, P. 99).

From the first part of the argument we learn the profound lesson that even though every mental act has an object, not all of these objects exist. Macbeth sees something, a dagger. His act of seeing, we say, has an object. But this object does not exist. There is no dagger in front of him. Could anything be plainer than that what he sees, the object of his seeing, does not exist? Is it not a mere truism that what we see on certain occasions is not there; that what we believe is sometimes not the case; that what we desire often does not come to pass? The proponent of the argument from hallucination denies these truisms.

According to our analysis, there occurs a mental act of seeing a dagger, but there is no dagger. What about a veridical situation? Well, in this case, too, there is the seeing of a dagger. The person in this situation experiences a mental act of seeing a dagger, just as Macbeth does. But now there also exists a dagger. The two situations, we may say, agree phenomenally but differ ontologically. If we describe the two minds, there is no difference. But the two perceptual situations consist not only of a mind: There is also extra-mental reality to be considered. And in regard to this reality, the two situations differ as day differs from night. In the one case, there is no dagger; in the other, there is one.

What gives strength to the argument from hallucination is the contention of its first part that Macbeth must be seeing sense-impressions, since there is no dagger to be seen. We have exposed the hidden premise of this line of reasoning, namely, the assumption that the objects of mental acts must exist. Put differently, the assumption is that acts cannot have nonexistent objects. But we cannot let matters rest by denying this assumption. There is more to the story of nonexistent objects. We have to face a genuine ontological dilemma.

b) The Nexus of Intentionality

There are at least two distinct reasons why it has been held that the objects of mental acts always exist; one profound, the other rather shallow. The shallow one is that we would not tell an unsuspecting person that Macbeth saw a dagger when he merely hallucinated. From the fact that we would mislead a person by telling him that Macbeth saw a dagger when he had a hallucination, one somehow concludes that Macbeth could not really have seen a dagger, but must have thought that he saw a dagger. It does not take a great amount of ingenuity to untangle this small problem of communication. We merely have to tell the third person that Macbeth had a hallucination, that what he saw did not exist. That he saw something is then quite obvious to the third person.

The profound reason goes back to the proper analysis of the nature of intentionality. This analysis presents us with two tough dilemmas. When you see a real dagger before you, then you are somehow related to the dagger; when you believe that the earth is round, you are related to this fact; and when you are afraid of spiders, you are related in some fashion to spiders. There is a very strong suggestion that the intentionality of mental acts is relational in character. All mental acts, one may conclude, are somehow related to their objects. Brentano’s thesis of intentionality, the thesis that every mental act has an object, is but the thesis that every mental act is related to an object. Let us call this peculiar relation “the intentional nexus.” However, this obvious analysis of the nature of intentionality runs immediately into the problem of nonexisting objects. How can there possibly obtain a relation between Macbeth’s act of seeing, on the one hand, and a dagger, on the other, if there is no dagger that can serve as the second term of the relation? How can there be a relation between a mental act and something that does not exist? Is not a relation between an existent, the act, and something that is not there at all, the object, an ontological absurdity? These rhetorical questions arise from the conviction that a relation cannot have nonexistent terms.

But if it is a law of ontology that all relations have existent terms, then we must search for a nonrelational account of intentionality. It will not do to say, as some philosophers have, that the intentional nexus holds in veridical perception, but not in Macbeth’s case. This would not be an account of the feature of intentionality with which we are concerned. Macbeth’s seeing has an object in the very same sense in which a veridical seeing has an object. In the relevant sense, every act has an object. Every seeing is a seeing of something; every belief is a belief in something; and every desire is a desire for something. Our task is to make ontological sense of this common feature of all acts. What other possibility is there? I do not believe that any account other than the relational one is plausible. But I cannot show this here (see my The Categorial Structure of the World, pp. 189-203; and also my “Nonexistent Objects Versus Definite Descriptions,” pp. 363-77). I shall mention, though, that I am convinced that a nonrelational analysis of intentionality must lead to solipsism. If the mind is not related in some fashion or other to the world, then it cannot know the world. Nor, it seems to me, can it even know itself unless it is so related to itself. Here then is the first of the two dilemmas I mentioned earlier, a dilemma as threatening as any I know of: A relational assay of intentionality confronts us with the problem of nonexistent objects, while a nonrelational one leads to solipsism.

I embrace the first horn of this dilemma. Thus I have to face the problem of nonexistent objects. What is at stake, we have already seen, is the truth of the ontological hypothesis that all relations, including the intentional nexus, have existing terms, that all relations are “normal.” Now, this hypothesis is either true or it is false. If we accept a relational analysis of intentionality, we are therefore faced with a second dilemma. If we assume it to be false, as I do, then we must admit that there are “abnormal” relations, relations which can connect existents with nonexistents. On the other hand, if we accept the hypothesis, then we have to say that Macbeth’s hallucinatory dagger, all appearance to the contrary, really exists. Since I cannot bring myself to believe that hallucinatory objects exist, I prefer to be impaled by the first horn of the dilemma: I must reject the alleged law about relations. I hold that the intentional nexus is different from ordinary, “normal,” relations. Spatial, temporal, and other relations are normal: They hold only between existents. If A is to the left of B, then both A and B exist. If A occurs later than B, then both A and B exist (but not, of course, at the same time). And if A is the father of B, then both A and B exist. But when Macbeth sees the dagger, then only his seeing exists, the dagger does not. More precisely, since I hold that all perception is propositional, when Macbeth sees that there is a bloody dagger dangling in front of him, then his act of seeing exists, the state of affairs seen, however, does not exist.

The great strength of the argument from hallucination derives from its hidden premise that there are no abnormal relations and, in particular, that the intentional nexus is not abnormal. Macbeth sees something. There can be no doubt about this point. What he apparently sees, a dagger, does not exist. There can be no doubt about this point either. Since his act of seeing is related to its object by a normal relation, it must have an existing object, and what could that possibly be under the circumstances except a bundle of sense-impressions? In this situation, what else exists and is “private”?

One may try to rob the argument of its force by inventing a diluted kind of being. Macbeth, one might argue, does not see a bundle of sense-impression, but sees a dagger. However, this dagger, though it does not exist, has being of a lesser sort. It is not just nothing. Since the dagger has some sort of being, the intentional nexus is not “abnormal.” It holds between two beings. We seem to have escaped from both horns of the dilemma: We have to hold neither that the intentional nexus is abnormal nor that what Macbeth sees is a bundle of sensations. But this impression is mistaken. It is created by a semantic trick. An abnormal relation, we must recall, is a relation which does not connect existents. The intentional nexus between Macbeth’s act of seeing and the dagger with diluted being is therefore not normal but abnormal. The view under discussion does not escape from the fate of having to assume the existence of an abnormal relation. It merely tries to hide the weirdness of the intentional nexus by the purely gratuitous introduction of a new kind of being.

We can see what is not there and believe what is not so. These are the facts. The argument from hallucination, we have seen, denies these facts. We accept them. But we have also seen that a difficult dilemma may force one to take the other side. Common sense triumphs in the end, but not without having paid some respect to philosophical sophistication.

c) Nonexistent Perceptual Objects

Perceptual acts are propositional. This is one of the most important theses of my view. Precisely speaking, when Macbeth sees the hallucinatory dagger, there occurs an act which intends some such state of affairs as: This is a bloody dagger. It is true, of course, that Macbeth sees a dagger. I am not trying to deny this truism. But I wish to insist that he sees a dagger because he sees that there is a dagger in front of him or that this is a dagger. It is true that Johnny sees an elephant when Johnny sees, for example, that an elephant is balancing himself on his hindlegs, or that an elephant is entering the ring, or that an elephant is splashing water over his back. What holds in this sense for seeing, holds in my opinion for all acts of perception.

I could have said that all perceptual acts are judgments. But this would have been misleading. A judgment, one usually explains, is something that can be true or false. A mere “presentation” (idea, notion, concept), on the other hand, cannot be true or false. And neither can a wish or a desire, even though these mental acts are propositional. But a perception is either true or it is false. When Macbeth sees that there is a dagger in front of him, he is mistaken. There is no dagger before him. And when Johnny sees that the elephant is balancing on his two hindlegs, his perception is correct. Yet, a perception is not a judgment, although they share this feature of being true or false. Perceptions and judgments form two distinct kinds of mental act. When you see that the elephant is eating peanuts, you experience a mental act which is totally different from the one you experience when you merely think (judge) that he is eating peanuts.

It is important to emphasize this difference because it has often been held that a perception is nothing but a judgment which is based on sensations (sense-impressions). The difference between seeing the elephant eat peanuts and merely thinking about it is not just that in the first case there occur certain visual sensations which are absent in the second situation. No, the crucial mental acts are different in the two situations: In one situation, there occurs an act of seeing, and in the other, an act of judging (asserting, assuming, etc.). We have already pointed out how the mistaken view that perception consists of sensations plus judgment invites skepticism. We can now understand why it has been so popular. Since perception is propositional and, moreover, true or false, it is easily confused with judgment. But it is also clear that perception somehow involves the senses. Perception, one erroneously concludes, consists both of judgment and the experience of sensations.

Macbeth sees a dagger when there is no dagger in front of him. What he sees is not so, it is not the case, he is mistaken. Put philosophically, the state of affairs which his act of seeing intends is not a fact. In general, a perceived state of affairs may or may not be a fact. Let us say, equivalently, that it may or may not exist. The intentional nexus is abnormal, as I put it earlier, in that the perceived state of affairs may not exist. The “object” intended, a certain state of affairs, may not exist. When we first talked about Macbeth’s hallucination, though, we called the dagger the nonexistent object of his seeing. We must keep these two notions of object apart. In one meaning, the object of a mental act is whatever is intended by the act; whatever the intentional nexus connects with the act. In perception, this object is always a state of affairs. But this object, the state of affairs, may involve a perceptual object. In the case of Macbeth’s hallucination, this second object is of course the dagger. In his case, the state of affairs seen does not exist, and neither does the object seen. The dagger does not exist, because the state of affairs seen, that this is a dagger or that there is a dagger, does not exist. The elephant, on the other hand, exists because what Johnny sees is a fact: There is an elephant balancing on two legs.

These considerations lead us to distinguish between hallucination and mere illusion. In Macbeth’s case, there was nothing in front of him at all at the place where he saw the dagger. But assume that he had mistaken a candle for a dagger. He thinks he sees a dagger in front of him when in reality there is a candle at that place. In this case, too, his perception is mistaken. But it is mistaken, not because there is nothing in front of him where he sees a dagger, but because he mistakes something that is in front of him for a dagger. A perception that this is a dagger can be false either because this is not a dagger but something else, or because there is nothing there at all. The state of affairs This is a dagger may not be a fact, either because this is not a dagger, or else because there is no this at all. In the former case, one usually speaks of an illusion; in the latter, of a hallucination.

And this brings us back to the argument from hallucination. It is claimed that there really is a this which Macbeth sees, and that this this is a sense-impression. But Macbeth is mistaken, for he believes that this sense-impression is a dagger. What he sees, therefore, is in a sense there, and, in another sense, not there. The sense-impression exists; the dagger does not. According to this analysis, Macbeth’s hallucination is more like an illusion, as we have explained it. But notice that this analysis does not fully escape from the problem of nonexistent objects. Macbeth comes to believe, mistakenly, that this is a dagger when in reality it is a sense-impression. (Have we here one of the sources of the popular philosophical view that perception is nothing but acquired belief?) Now, what does this act of believing intend? Obviously, the nonexistent state of affairs that this is a dagger. In this case, the intentional nexus connects a belief with a state of affairs which does not obtain, which is not the case, or, in short, which does not exist. Thus one cannot escape from the problem of nonexistent objects as long as one admits that there are false beliefs, false assertions, false judgments or, given our conception of perception, that there are false perceptions.

We insist, as I have said before, that sensations (sense-impressions) cannot be seen. Macbeth cannot possibly be seeing a sense-impression when he hallucinates the dagger. Of course, at that moment, he experiences a certain sensation of a certain shape, size, and color. And he probably experiences also a great number of other sensations. But what he experiences, in our sense of the term, is not what he sees; and what he sees is not what he experiences. Neither in veridical perception nor in hallucination do we ever see sensations.

What one perceives may not be so; what one perceives may not be there. Perception raises the same problem as false belief: How can the mind be related to what is not the case? But it also raises an additional problem. How can the mind be related to things that do not exist? In perception, something is directly given to us; something appears in person before us, so to speak. This makes perception quite different from mere belief or mere assumption. You cannot, precisely speaking, believe that this is a dagger. You can only see it. But you can believe, for example, that there is a dagger before you. Individual perceptual objects appear only in perception directly, in person, before us. What cases of hallucination teach us is that even though individual things may be directly presented to us in perception, these things may not exist.