Where Does Certainty Get Us? Moore's Hands and Our Dreams

Part of the Philosophy series.

Every night, most of us live for a while in a world that is entirely convincing and entirely untrue. We see our hands, walk through rooms we know, and doubt none of it until we wake. This fact came back to me in the summer of 2024, in a Metaphysics course at NYU, when G. E. Moore’s famous answer to the skeptic, who doubts that we can know anything beyond our own experience, came up in discussion. Moore held up one hand, then the other: here is one hand, and here is another, and so at least two external things exist. He could not prove the premise, he admitted, but he knew it.

Moore never said that a feeling proves anything, but his gesture has outgrown him and become, in other hands, the appeal to common sense: if something is obvious to the senses, no clever argument can touch it. Samuel Johnson gave it its folk form when he kicked a stone to refute Berkeley. It is this broader appeal, and not Moore himself, that I want to question. My position, then and now, rests on a single thought: what seems real to the senses can also be illusory, and our dreams show this every night. If that is so, the felt certainty of the senses cannot be the measure of what is true. The argument has three steps.

The argument

First, in my dreams it seems to me, exactly as it does now, that here is a hand, and nothing in that seeming marks it as a dream. It is immediate, and it comes from the senses, not from reasoning.

Second, in the dream that seeming is false, and not as a rare accident. Every dream I remember that I failed to recognize as one was a stretch of being wrong without suspecting it.

Third, a sign that is present just as fully when it is wrong as when it is right adds no evidence of its own about which case one is in. Whatever weight “here is a hand” carries against an argument must come from elsewhere, from how it fits with everything else I know. The same holds for the other side, since dreams make bad reasoning feel compelling too. The lesson is not that argument defeats the senses, but that neither side’s feeling decides. Fit does.

Yet how do I know that the dream was wrong? What, if not another certainty, does the correcting? I know it only because I woke and the dream did not fit with my bed, my memories, or the morning. Waking corrects the dream, but not by feeling more certain, for the dream felt just as certain. It wins because the dream fails to cohere with what follows, and waking life does. This fit, however, is available only afterwards, and in the moment, as false awakenings show, a dream can counterfeit it. So what makes my present belief that I am awake knowledge is not any mark I can inspect but the plain fact that I am awake.

When I made this argument in class, Will, our instructor, agreed, and said that there are many arguments against the validity of Moore’s move. One does not need a course in philosophy to arrive at the doubt; one needs only to have slept.

Why this is not Descartes again

Descartes had the dreams as well. In the First Meditation he is sure that he is sitting by the fire, stretching out his hands, and then remembers having been convinced of just such things while asleep. From this he finds no reliable way of telling waking from sleeping, and from that he supposes that he may be dreaming now. My argument shares his record and even his finding, and declines only the last step. I am not doubting that I am awake. I am asking whether the feeling of being certain is what tells me so. That is also why Wittgenstein’s objection to Descartes misses me. In On Certainty he called “I may be dreaming” senseless, since if I am dreaming the remark is being dreamed as well. I never make that remark. I argue from the waking side, about dreams that are already over.

The replies

Moore: waking certainty is of a higher grade. In “Certainty,” a lecture of 1941, Moore granted the skeptic’s conditional, that unless he knew he was not dreaming he could not know he was standing up, and ran it the other way: since he knew he was standing up, he knew he was not dreaming. His present experience and his memory of the last few minutes might, he suggested, be enough to show this, though he granted that he could not prove it. At its strongest, the reply holds that waking certainty is reflective, whereas a dreamer is carried along.

Dreaming usually dulls the faculty that checks, but not always. In a false awakening one “wakes up,” remembers going to bed, reflects on the dream, and feels sure of being awake, all while still asleep. They are rare, but they need not be common. Their significance is that when the reflective check fails, it feels exactly as it does when it succeeds. Hobbes pressed this against Descartes’s own test for waking: one can dream that the dream fits with everything that came before. That is why the fit I rely on is worked out afterwards.

Sosa: dreams are not beliefs. Ernest Sosa argues that when we dream we do not believe but imagine. If he is right, I have no record of false beliefs, only of vivid imagining, and my waking belief in my hands is safe, since I could not easily have held it while it was false. Sosa would reject my second step if it were a claim about belief. It is a claim about seeming, however, and his view grants what I need. He concedes that a vivid dream is “subjectively very much like its corresponding reality,” so the two feel alike from the inside. What makes the waking belief knowledge, on his account, is its safety, the fact of being awake, and not how sure it feels.

Pryor: seeming is enough. James Pryor’s dogmatism is the best modern defense of Moore. When it looks to one as if something is so, one thereby has some justification for believing it, without needing any further reason. I think this is right, as far as it goes. The dreamer, however, has the same seeming, and so the same justification, and is wrong. Dogmatism protects the right to believe what the senses present. It does not make the seeming a mark of what is true, and that is all I deny.

Williamson: the waking have better evidence. Timothy Williamson holds that one’s evidence is what one knows. Awake, I know that here is a hand, and the dreamer does not, so our evidence differs however alike we seem. I can accept this, since Williamson also holds that one is not always in a position to know what one knows. The difference in evidence is real, but nothing in the feeling announces it.

Carnap: the outside question is not a real one. Carnap, one of our course readings, held that questions of existence make sense only inside a framework, a shared way of speaking and checking. Inside the everyday framework of things, “here is a hand” is settled by looking and touching. The question whether the world of things as a whole is real, asked from outside every framework, is not a theoretical question but a choice of language. Discovering that I dreamed is a correction inside the framework, so the dream never reaches the outside question, and there I think he is right. But the inside question was settled not by looking, since the dream had looking too, but afterwards, by setting the dream against everything else. Even within the framework, the feeling does not decide.

What this does not show

None of this shows that the world is unreal or that I am dreaming now. My hands are here, and I will go on using them. Nor does it show that I fail to know I have hands, for if knowing is a matter of safety or reliability, I very likely do. Nor does it touch Moore, who was right that knowing does not require proof, and usually right that an argument with an absurd conclusion hides a bad premise. What fails is the slogan: that what seems obviously real to the senses is thereby settled. The hand is real. The slogan was wrong only about what vouches for it.

What certainty does get us

If the felt certainty of the senses cannot tell us what is true, is there anything it can tell us? What did the dream leave standing?

Moore himself showed where to look. In “The Refutation of Idealism” he argued that every sensation has two elements: its object, which varies, and consciousness, “in respect of which all sensations are alike.” When we try to look at consciousness itself, “it seems to vanish,” as if it were “diaphanous.” Yet it can be distinguished, he added, “if we look enough, and if we know that there is something to look for,” and from that distinction he drew a realist conclusion, not mine. Descartes comes near this too. In the Second Meditation, because he may be dreaming, he cannot be sure that he sees light or feels heat, yet he certainly seems to see and to be warmed, and this cannot be false. What he finds there is a thinking thing busy with doubting and willing, which is more than I will claim, and he holds that it exists only as long as it thinks, a caution I will come back to.

The central thought, then, is this: everything on the side of what is seen can change between dreaming and waking, the hand, the room, even the person I take myself to be, for in some dreams I am someone else. What does not change is that whatever appears, appears to awareness. The dream can replace the hand and the self that owns it, but not the fact that it was seen. This is not a stronger feeling of certainty, for the dreamer felt just as sure of everything it saw. It is a claim of a different kind, one that no waking can correct: to doubt it is already to have an experience, and what waking corrects was itself experienced.

Hume looked inward for a self and found only perceptions. Moore thought attention could catch awareness, but I would put it differently. Awareness is not known as one more perception but in their being perceived at all. That is also the sense in which I call it unchanging. It has no features of its own that could vary, and so nothing by which last night’s awareness could be counted as different from this morning’s. Everything that varies belongs to what appears to it. Dreamless sleep remains the hard case, the one where Descartes’s caution bites, and I cannot report on it. I can only note that I wake able to say that I slept well and knew nothing, and it seems to me that something must have been there for there to have been nothing to know. That is my view, and the dream argument does not prove it.

Why, then, call this real, rather than merely certain? Not because I am surer of it, for being sure settles nothing. I call it real by the same test that let waking correct the dream: what survives every change of state has the better claim to the word. The experiencer, the psyche, what I have been calling awareness, survives every state I have known. The hand comes and goes with the state I am in, and the one to whom it appears does not. The hand is real, and not irrelevant, but it is not what endures.

Perhaps this is what our dreams have to teach. The waking world is neither to be dismissed as one more dream nor to be mistaken for the ground of our certainty. It is to be lived in fully, while the one who receives it is recognized as the one thing that neither state alters. Each dream takes away a world we were sure of and leaves behind the one to whom it appeared, and to notice this once a night is to begin noticing it by day.