It is certain, therefore, that we do commonly use the word "existence" in a sense, in which to say "A exists" is not to say "A is perceived," or "A is systematically connected with other real things," or "A is purposive." There is a simpler sense than any of these—the sense in which we say that our own perceptions do exist, and that Sindbad's perceptions did not exist. But when I say this, I am by no means denying that what exists, in this simple sense, may not always also exist in all the others; and that what exists in any of them may not also always exist in this. It is quite possible that what exists is always also perceived, and that what is perceived always also exists. All that I am saying is that, even if this is so, this proposition is significant—is not merely a proposition about the meaning of a word. It is not self-contradictory to suppose that some things which exist are not perceived, and that some things which are perceived do not exist.
But, it may be asked: What is this common simple sense of the word "exists"? For my own part, it seems to me to be so simple that it cannot be expressed in other words, except those which are recognised as its synonyms. I think we are all perfectly familiar with its meaning: it is the meaning which you understood me to have throughout this paper, until I began this discussion. I think we can perceive at once what is meant by asserting that my perception of black marks on a white ground is "real," and that no such perception as Sindbad's of a Roc was ever "real": we are perfectly familiar with the property which the one perception is affirmed to possess, and the other to be without. And I think, as I have said, that this property is a simple one. But, whatever it is, this, which we ordinarily mean, is what I mean by "existence" or "reality." And this property, we have seen, is certainly neither identical with nor inclusive of that complex one which we mean by the words "is perceived."
I may now, then, at last approach the main question of my paper. Which among the "contents" which I observe will give me reason to suppose that my observation of some of them is generally preceded or accompanied or followed by the existence of certain particular perceptions, thoughts or feelings in another person? I have explained that the "contents" which I actually observe may be divided into two classes: on the one hand, those which, as we commonly say, we perceive "through the senses"; and, on the other hand, my perceptions of these last, my thoughts, and my feelings. I have explained that if any of these observed contents are to give reason for a generalisation about what exists, they must exist. And I have explained that with regard to both classes of "contents" I am using the word "exist" in precisely the same sense—a sense, in which it is certainly not self-contradictory to suppose that what is perceived, does not exist, and that what is not perceived, does exist; and, in which, therefore, the assumption that a red patch with gold letters on it exists, is a different assumption from the assumption that my perception of a red patch with gold letters on it exists; and the assumption that my perception of a red patch with gold letters on it exists, is a different assumption from the assumption that a red patch with gold letters on it exists.
What, then, that we observe, can give us any reason for believing that anyone else has certain particular perceptions, thoughts or feelings? It has, I think, been very commonly assumed that the observation of my own perceptions, thoughts, and feelings, can, by itself, give me such a reason. And I propose, therefore, to examine this assumption. If, as I hope to show, it is false; it will then follow, that if our own observation gives us any reason whatever, for believing in the existence of other persons, we must assume the existence, not only of our own perceptions, thoughts and feelings, but also of some, at least, among that other class of data, which I may now, for the sake of brevity, call "sense-contents"; we must assume that some of them exist, in precisely the same sense in which we assume that our perceptions, thoughts, and feelings exist.
The theory which I propose to examine is, then, the following. My observation of my own thoughts, feelings, and perceptions may, it asserts, give me some reason to suppose that another person has thoughts, feelings, and perceptions similar to some of mine. Let us assume, accordingly, that my own thoughts, feelings, and perceptions do exist; but that none of the "sense-contents," which I also observe, do so. Where among my perceptions am I to look for any which might conceivably give me a reason for supposing the existence of other perceptions similar to my own? It is obvious where I must look. I have perceptions which I call perceptions of other people's bodies; and these are certainly similar in many respects to other perceptions of my own body. But I also observe that certain kinds of perceptions of my own body are preceded by certain other perceptions, thoughts, or feelings of mine. I may, for instance, observe that when I perceive my hand suddenly catch hold of my foot in a particular way, this perception was preceded by a particular kind of feeling of pain. I may, perhaps, observe this often enough to justify the generalisation that the perception of that particular motion of my body is generally preceded by that particular feeling of pain. And in this way I may perhaps have reason for quite a number of generalisations which assert that particular kinds of perceptions of my own body are generally preceded by other particular kinds of perceptions, thoughts, or feelings of my own.
But I may also, no doubt, have the perception, which I call the perception of another person's hand catching hold of his foot, in a manner similar to that in which I have perceived my own hand catch hold of my own foot. And my perception of another person's hand catching hold of his foot may undoubtedly be similar in many respects to my perception of my own hand catching hold of my own foot. But I shall not observe the same kind of feeling of pain preceding my perception of his hand catching hold of his foot, which I have observed preceding my perception of my hand catching hold of my foot. Will my generalisation, then, give me any reason to suppose that nevertheless my perception of his hand catching hold of his foot is preceded by a similar feeling of pain, not in me but in him? We undoubtedly do assume that when I perceive another person's body making movements similar to those which I have observed my own body making, this perception has generally been preceded by some feeling or perception of his similar to that which I have observed to precede my perception of similar movements in my own body. We do assume this; and it is precisely the kind of generalisation, which, I have insisted, must be admitted to be true. But my present question is: Will such observations as I have described give any reason for thinking any such generalisation true? I think it is plain that they will not give the slightest reason for thinking so. In the first place, all the perceptions which I call perceptions of another person's body differ very considerably from any of those which I call perceptions of my own. But I am willing to waive this objection. I am not offering any theory as to what degree of likeness is sufficient to justify a generalisation: and therefore I will allow that the degree of likeness may be sufficient. But there remains an objection which is, I think, quite fatal to the proposed inference. This objection is that the inference in question plainly does not satisfy the third condition which I suggested above as necessary, wherever any generalisation is to be justified by observation. I am willing to allow that my observations of the fact that my perception of a certain movement in my own body is preceded by a certain feeling of pain, will justify the generalisation that my perception of any such movement, whether in my own body or in that of another person, is generally preceded by a similar feeling of pain. And I allow, therefore, that when I perceive a certain movement in another's body, it is probable that the feeling of pain exists, though I do not perceive it. But, if it is probable that such a feeling of pain exists, such a feeling must stand in the same relation to my perception of the movement in another person's body, in which a similar feeling of pain has been observed by me to stand to my perception of such a movement in my own body. That is to say the only kind of feeling of pain, which my observations do justify me in inferring, if (as I admit they may) they justify me in inferring any at all, is a feeling of pain of my own. They cannot possibly justify the belief in the existence of any such feeling except one which stands to my perception in the same relation in which my feelings do stand to my perceptions—one, that is to say, which is my own. I have no more reason to believe that the feeling of pain which probably precedes my perception of a movement in another person's body can be the feeling of another person, than, in my former example, I had reason to suppose that the hen, whose existence probably preceded that of a given egg, could be a hen, which had never been near the egg in question. The two cases are exactly analogous. I observe a feeling of pain of my own preceding a perception of my own. I observe the two, that is to say, as standing to one another, in those relations (whatever they may be) in which any perception of mine stands to any other thought, perception or feeling of mine, and which are, at all events, different from any relation in which a perception or feeling of another person can stand to one of mine. I never perceive the feeling and the perception as standing in any other relation. In any case, therefore, where I do observe something like the perception, but do not observe the feeling, I can only be justified (if justified in inferring any feeling at all), in inferring an unperceived feeling of my own.
For this reason I think that no observations of my own perceptions, feelings or thoughts can give me the slightest reason for supposing a connection between any of them and any feeling, perception, or thought in another person. The argument is perfectly general, since all my perceptions, feelings and thoughts do have to one another those relations in virtue of which I call them mine; and which, when I talk of a perception, feeling or thought as being another persons, I mean to say that it has not got to any of mine. I can, therefore, merely from observation of this class of data never obtain the slightest reason for belief in the existence of a feeling, perception, or thought which does not stand in these relations to one of mine—which is, that is to say, the feeling, perception or thought, of another person. But how different is the case, if we adopt the hypothesis, which I wish to recommend—if we assume the existence of that other class of data which I have called "sense-contents!" On this hypothesis, that which I perceive, when I perceive a movement of my own body, is real; that which I perceive when I perceive a movement of another's body is real also. I can now observe not merely the relation between my perception of a movement of my body and my own feelings, but also a relation between a real movement of my body and my own feelings. And there is no reason why I should not be justified in inferring that another person's feelings stand in the same relation to the real movements of his body, in which I observe my own feelings to stand to similar real movements of mine.
But there is another argument which may still be urged by those who hold that my own perceptions, thoughts, and feelings, by themselves, may be sufficient to justify a belief in the existence of other persons. It may be said: "Our observation of our own perceptions may be sufficient to verify or confirm the hypothesis that other persons exist. This hypothesis is one which "works." The assumption that other persons have particular thoughts, feelings and perceptions enables us to predict that they will have others and that our own perceptions will be modified accordingly: it enables us to predict future perceptions of our own; and we find that these predictions are constantly verified. We observe that we do have the perceptions, which the hypothesis leads us to expect we should have. In short, our perceptions occur just as they would do, if the hypothesis were true; our perceptions behave as if other persons had the perceptions, thoughts and feelings which we suppose them to have. Surely, then, they confirm the truth of the hypothesis—they give some reason to think it probably true?"
All this, which I have supposed an opponent to urge, I admit to be true. I admit that the fact that an hypothesis works may give some reason to suppose it true. I admit that my perceptions occur just as they would do, if other people had the perceptions which I suppose them to have. I admit that that assumption enables me to make predictions as to future perceptions of my own, and that I observe these predictions to come true. I admit all this. But I admit it only in a sense in which it in no way conflicts with the position which I am maintaining. The words, which I have put into the mouth of a supposed opponent, may, in fact, mean three different things, which it is worth while to distinguish. In two of those meanings, which I shall admit to be true and which are what make them seem plausible, they do not deny what I assert. Only in the third sense are they an objection to my position: and in that sense they are false.