Colin McGinn
Colin McGinn; drawing by David Levine


It is autumn. All around, the leaves are blown. Yellow, and black, and pale, and hectic red. We may disagree a bit, perhaps, about the hectic red. For you, that is a bit over the top. Why invoke a fevered hue when this splendid maple leaf is just plain bright red? Pick up a leaf of another color. You think that is from a poplar? It is the leaf of an aspen. And so on: we know how to talk about these things, if they interest us.

How different it is with mental images (if that’s the right term). From now on I shall, in the spirit of Colin McGinn’s subtitle, just say “images” meaning “mental images,” “image” as sense 6 in the American Heritage Dictionary, “a mental picture of something not real or present.” That is sense 5 in the OED—in most of its meanings, there is nothing specially mental about “image.” Some people say they can visualize the leaf we held in our hands this afternoon, right down to the small black spots, and the tiny hole eaten by an insect. Others say they cannot think of doing that. I visualize the face of an elderly friend, crippled with arthritis, but grimly hanging on. Or do I? With what detail? Some people say they use images to help them remember intricacies. Others say they just remember. If they are able to form an image of the face, it is because they remember how it was: it is not that an image guides memory, but that memory produces an image, or the sense of imaging. We have no agreed way to talk clearly about such things.

And why should we? We need to be able to answer questions about the things around us, even leaves, on occasion. All peoples have evolved extraordinarily precise ways of settling issues about the things that matter to them. My images matter to me, but they do not matter much to us as a group. Many things about the mental life of others do matter. I want to know if I have hurt my neighbor’s feelings, or if a colleague on a committee has it in for me, or if the reason I got such-and-such an honor was that it was finagled by an old schoolmate whom I have not seen for decades. But it does not matter so much, or at all, whether the neighbor, hating me, has an image or not, or if the person of influence still has an image of me as the youth that he has not met since. The powers of visualization of my colleague are irrelevant to whether he will be able to vote against my proposals. We do not need to have a way to talk clearly about other people’s images.

Of course some people are more imaginative than others, and that matters—but not because of images. The other day I came across a few pages, previously unknown to me, by a writer whom I know very well. Probably dashed off in a moment, when asked to say some things to a group of architects. I was bowled over by the playfulness, by the depths, by the startling comparisons, by the shining ironies, by the sheer wit of the piece. When I say that, I am talking about a production. You may detest my writer, and say it is all flash and no substance. I laughed from time to time; you were offended and can tell me why. There we go again, talking about something we know how to talk about. Lots of imagination, I say.

Does that have anything to do with images? Maybe, but not obviously. I might have asked the writer, were he alive, “Did you visualize the cemetery, or the deflowering of the bride on the train, or the Persian carpet, etc., when you put them into your essay?” Maybe he replies: “Never, I just write the thoughts as they come to me.” Or: “Yes indeed, everything is present to my mind’s eye as my pencil skims the paper.” Both answers are interesting, but they do not affect the fact that my author is a man of great imaginative power, as well as one who has the skill to write and argue, to move in a handful of words from a joke to an abyss.

In short, we know how to recognize, admire, or fear the power of imagination. We know how to talk about it. But not about images. That must be why analytic philosophers have shunned them. Analytic philosophy depends on a steady, reliable way of talking. And now Colin McGinn, the very type of an analytic philosopher of mind, proposes to use exactly his sharpened skills in a reflection on images. Actually this is not quite such a surprise, for ever since “consciousness” became a hot subject in the Nineties, analytic philosophers have begun to say that they are going back to “phenomenology.” McGinn himself says he is doing the phenomenology of imagery. This does not refer to Husserl, Merleau-Ponty, and others who are identified with phenomenology. It is rather a case of what William James called “introspection,” the process of looking intensely into one’s own mind, long discredited among psychologists. So McGinn is both analyzing and introspecting in his innovative little book.1


Did not philosophers formerly talk much of images, as well as of the imagination? Yes, but remember we live in a world in which imaging as a mental act is not prized. Imagination, yes, but imaging, I suggest, no. Shamans and other priests once earned their living seeing what no one else could see. Did they see in the world of their mind, or out there? For us that is one big difference, but for many another civilization it did not matter one whit. Closer to home, take the arts of memory. Before the invention of printing, scholars had prodigious memories, and there was a whole raft of techniques for remembering. As a primer, you learned how to visualize a very complicated room, with lots of tables and niches. To remember a complex store of information, you mentally placed objects, people, even ideas, on standard places in this template room. Those skills are no longer needed.

McGinn is not one to lure us into thinking about the past. One of the delightful things about his new book is that none of his fellow philosophers of mind has had much to say about images, so he has no need to get into the dry business of referring and refuting. There are only two twentieth-century masters to whom he resorts, namely Wittgenstein (in scattered but powerful remarks) and Sartre (in a little book of 1940 called L’Imaginaire: Psychologie phénoménologique de l’imagination2). A class act, I’d say.

Nevertheless, loyal to his roots, he starts with Hume. Hume began by distinguishing what he called “impressions” and “ideas.” Impressions include, he said, “all our sensations, passions and emotions.” Ideas are “faint images of these in thinking and reasoning.” And what distinguishes the two? Hume was the model for every future positivist. There is, he said, nothing to distinguish them, in our consciousness, except the fact that impressions have a greater degree of force or vivacity. What on earth could he have meant by vivacity? Usually he can be read like a twenty-first-century thinker, but here he seems like a visitor from an older world, now lost to us, accessible only to scholars and then but dimly. McGinn will have none of that. Hume should be read without further interpretation. Hume was just wrong. McGinn is not quite as blunt about Hume as William James was—“The slightest introspective glance will show to anyone the falsity of this opinion”3—but he is much of the same mind.

McGinn gets to Hume on the first two pages of Chapter 1. We are reminded of the distinction between actually seeing something—having what McGinn calls a “percept” of it—and merely visualizing it within the mind. What, asks McGinn, is the difference between these two experiences? Rephrasing, in a not very obvious way, he asks, “In other words, how do percepts and images differ?” The classic answer, he replies, was given by David Hume—and then comes the quotation about impressions and ideas. Hume’s answer was wrong because the difference between percepts and images is not a matter of degree, of force or vivacity. It is a difference in kind.

And here is a curious fact about McGinn’s analysis. It is strictly within Hume’s framing of the subject. There is an initial dichotomy out there. Hume wrote of a distinction between impressions and ideas. McGinn does not use these words. He speaks of a distinction between percepts and images. This is a truncated version of Hume’s dichotomy. Hume’s impressions were very rich, spanning all the items that “make their first appearance in the soul,” starting with (to use Hume’s words, which may not have meant the same to him as to us) sensations, and proceeding through the passions to the emotions. McGinn’s percepts do not stray far from the sensation end of this range of experiences. McGinn writes as if the word “percept” picks up Hume’s impressions, and “image” picks up his ideas—without remainder. This is a little misleading, for passions and emotions have been dropped from percepts, and Hume’s “ideas” cover a lot more than images. Perhaps we should read Hume in a way that makes better sense of his words—when one of the most brilliant and careful thinkers in history says something obviously false on his second page, it may be our reading, rather than his words, that is at fault. But Hume will not concern us further. What does concern us is the persistence in McGinn of a dichotomy between “percept” and “image.”


As is so often the case in philosophy, it is the first step that counts. Here the first step is the dichotomy. In analytic philosophy, it is also the first word that counts. Here we have the word “percept.” It will not be familiar to most readers. I never spoke or wrote it in my whole life until now. McGinn does not define it. The OED teaches that it was invented by nineteenth-century philosophers, and that it had two senses. In the first sense, a percept is an object of perception, as one of them put it, “outside the organism of the percipient,” so not in the mind. Perhaps even the hectic leaf itself could be a percept, that very leaf we looked at this afternoon. But in the second sense the percept is the “mental product of perceiving,” and so definitely not the leaf in the hand, but a mental product. “Percept” could be a very handy word, even if you scrupulously avoided saying in which of these two senses you were using it. How disagreeable it is that the old bully appearance-and-reality is hiding just around the semantic corner waiting to bop you on the nose.

McGinn’s usage is on the mental product side. At any rate percepts are not objects seen, but what we see when we see objects. The leaf is different from an image of the leaf, but McGinn is not concerned with that contrast. It is rather the contrast between images and that visual sensation which we experience when we see the leaf. Philosophers have been trying to express some such notion forever. Hume was not given to footnotes, but we get one on his second page, in which he nervously wrote that by impressions he meant “merely the perceptions themselves; for which there is no particular name in English or any other language, that I know of.” There is still no agreed name in English or any other language for “merely the perceptions themselves.”

Or is there? Perhaps Neuroenglish, the speech of neuroscientists, is a usable dialect, and there the name for merely the perceptions themselves is “images.” How confusing. Philosophers (including me) fulminate at this usage but there it is, one might say, a linguistic fact. Obviously McGinn does not speak that dialect and wants to avoid it.

I apologize for all this, but it is the first step and first word that counts. I began by saying that we know how to talk about leaves but not about images of leaves. Here is something we really don’t know how to talk about: we have never, in any language, managed even to agree on a word for it! Do I hear some whining skeptic saying that there is a good reason for that, namely that there is nothing to talk about? Of course in any particular case, when there may be a mismatch between my experience and how things are, we have a wealth of words to use, starting with “it does not look like a lamppost,” “it appears to be a bit tilted,” etc. It is the overarching idea of percepts that the pathetic skeptic doubts. Skeptics are not too keen on an all-purpose bag of mental “images” either.

Let us move on. Once the distinction between percepts and images is stated, all rolls smoothly along. In fact, on first reading McGinn’s book, I found the opening part just great. It was only when I began to ask, what are we talking about, that the moments of gloom set in. For in the early chapters we have the best kind of analysis, brief, clear, usually modest, occasionally wry, always serious.

Here is how it goes. Hume was wrong. Percepts are not merely more lively than images. They are different in kind, and that for nine reasons, of which seven seem worth particular mention. First, we cannot will percepts into being: I cannot in general decide to see a leaf when looking at what I am holding in my hand, but I can decide to visualize a leaf in my mind’s eye. That is a metaphor that McGinn takes very seriously, by the way, and that gives him his title, Mindsight. Second, percepts are “informative,” a word that McGinn emphasizes, while images are not. Images contain only what we put into them. Wittgenstein and Sartre agree on this point, which McGinn explains in such a way as to avoid troubling counterexamples.

Perhaps more work is needed. Carpenters, architects, chemists, and math- ematicians all use diagrams. High-energy physicists live by Feynman diagrams. You can learn a lot by thinking about a diagram—drawn or imagined. I suppose that information is contained in the image of a diagram in a different way from the information McGinn, Sartre, and Wittgenstein were thinking of. And maybe scientists are not so good at imaging as you might expect. The indefatigably curious Francis Galton asked a great many more or less eminent Victorians to form an image of their breakfast table, and describe what happened as they did so. Being Victorian, they replied at studious length. Men of science, he found to his astonishment, describe themselves as less good at visualizing than anyone else does.4

Third, percepts lie in a visual field grounded in our anatomy—the field, what is included in what we see, is determined by where our eyes are. But images can be anywhere and nowhere. Fourth, percepts are filled in completely, saturated. Every point in a visual field, McGinn assures us, has some quality or other. Images may be vastly less determinate. This contrast seems clearer with the solid objects that McGinn uses as examples, chairs and the Eiffel Tower. When I see a chair in the room, the percept, McGinn asserts, is determinate. There are no gaps in the perceived chair; the perception of the leg is not smudged even if the yellow paint on the leg is peeling. One wonders about messier stuff. When I look at the multitude of leaves on my city street, is everything filled in in the same way, and with what qualities? The jumble of leaves on the street—all that messy blowing stuff out there—is filled in with qualities at every point. But is the (to use Hume’s version) “perception itself” filled in? Not for me. There is just too much there for my visual system to take it in. Speak for yourself.

Fifth, I need not and alas often do not attend to what I am looking at, but the percept is still there. But if I stop attending to an image, then aside from truly annoying cases of locked-in imagery (think of tunes that keep recurring in the mind), it is gone. McGinn’s seventh difference is new: I may need to recognize my friend in the crowd or that this leaf is from a poplar, not an aspen. In both cases I may be wrong. But an image of my friend just is one of my friend. It makes no sense to recognize her as such, for the image is brought into being as of my friend.

Query: Judith is told that X, whom she has not seen for some years, was present at a conference. She is astonished, for she had not noticed X in the room. She is good at visualizing. In her mind’s eye she runs across images of the people in that room, and cannot recognize X anywhere. Her husband understands, and says that X has changed beyond all recognition. He recognized X by X’s voice and intensity of expression, so no amount of mental scanning of faces would help. But might not Judith have said, suddenly, Aha, so that’s who it was, recognizing the face in her mind’s eye as X? (And she could have been wrong in the identification, just as in real seeing.)

McGinn’s ninth difference is tidier. I can be seeing a leaf and having an image of a friend. The image does not block the sensory system. Afterimages, in contrast, do interfere; if I have a vivid afterimage of bright green, I have trouble seeing the red maple. Hence afterimages get filed with percepts, not images. True images do not occlude percepts.


Having established beyond doubt that images are different kinds of items from percepts, McGinn addresses with care the resultant puzzles—why say that both percepts and images are both visual, when they are so different? McGinn’s book is indeed visual although there are nods to the auditory, for we extend the notion of mental pictures, by analogy, to other senses. It is a sorry consequence of our lack of agreed ways to talk about images that many observations degenerate into personal anecdote. I find that my images are not so clearly solely visual or solely tactile. There is a curious tactile quality to much that I myself visualize, as if in imagery I can hardly separate the sensory modalities. As I visualize a poplar leaf, I have a sense of what it would feel like to hold it in my hand. So what? I suspect if I convinced someone that this is true of their images, I would not be telling them how it is, but persuading them how to talk. My auditory experience and my visual experience seem pretty distinct: Will a younger person, whose imagery has been formed by the experience of rock video, feel the same way? Moreover there is what could mockingly be called a Heisenberg principle at work here. The more you introspect about your images, the more their character may seem to be modified by your reflection.

Almost everything in McGinn’s book depends on the first step, the basic need to file items as either percept or image. So when we get to the fascinating discussion of dreams, dreams get classified along with images. Are dreams so akin to what happens when we visualize a leaf or a lover? Well, we can learn to describe them that way, and that will be the last time that I ask you to reflect on how much this conception of images leads to finding the truth, and how much it is learning one way to talk. McGinn has some striking proposals. He starts with the sense that many have, that dreams involve an agent and a patient. There is the dreamer (who has the dreams) and the still-mysterious dream master, who creates the dreams (but who in another sense could also be called the dreamer). In ancient Greece the dream master stood above your head while you slept on the ground, and dreamed the dreams that were conveyed to you, dreaming. Many civilizations tell some such story, and so do I.

For example, I once ran a seminar for entering undergraduates, just twenty of them. We read Freud. I told them they should each keep a dream diary, though not to tell me about what they actually dreamed. You all dream almost every night, don’t you?—Oh yes. Next week, each and every one came in astonished. Not a single dream. Meanwhile I—I had so many fully remembered dreams the first night that I was Tristram Shandied out; it would have taken me two full days to write them down. Twenty-one dream masters were playing their familiar tricks on us. The master of dream masters was of course Freud, for whom McGinn does not much care, but he surely changed, perhaps forever, the way that people of our civilization dream; certainly he did for those of us who introspectively consider our dreams.

The notion of the dream master is familiar while the dream master himself remains elusive; but McGinn gives the notion a new twist. He compares dreaming to hypnotism. He suggests that it is as if our dream master hypnotizes us into a dream. Or conversely that hypnotism and lesser degrees of suggestion are to be understood as tapping into those aspects of the mind that make dreaming possible. The hypnotist somehow accesses the part of the brain that is used by the dream master, i.e., the part that allows someone to impose his or her wishes on another person. McGinn proceeds somewhat unnervingly into a flow of armchair psychology, but perhaps that is what is needed.

McGinn once refers with self-aware irony to the enthusiasm of philosophers twenty years ago (and still going strong) to describe the life of the mind as something like a calculus of beliefs and desires. But just as he is caught up in Hume’s dichotomous frame, so too he has not been fully emancipated from this curious vision of humanity. Thus he has a problem with dream beliefs. He thinks that the hypnotist first of all induces beliefs in the subjects, and that the dream master induces beliefs in the dreamer. We must believe what we dream when we dream it, for how otherwise could we sometimes be frightened by what we dream? If we did not believe it, we would not be afraid (although we might be upset by our capacity to produce bad dreams)!

Surely we should use the current folklore of neuroscience at this juncture: fear is deepest in the limbic system, the primitive part of the brain near the brain stem, and precedes all other emotions from an evolutionary point of view. In nightmares we experience terror, our heart races, we sweat. This is not because we believe something, but because the dream master is drawing on the limbic system, on our first being as animals, long before it made sense to speak of beliefs. More generally, many of us will feel less belief-ridden even in active life than McGinn would have it, and also suspect that belief just is not the right way to talk about the state of mind in dreams. Philosophers have attributed a vast structure of beliefs to people as part of a theory for analyzing action. They do not discover those beliefs by introspection that preceded the constitution of a theory.

After dreams, McGinn’s armchair reflections move to madness. His thoughts here appear strangely uncaring, even arrogant. Arrogant not because they ignore so much clinical psychiatric experience, but because he makes no attempt to listen to the mad, and to notice the immense variety of ways in which different people suffer. It is uncaring, for it is as if the mad are curiosities in Bedlam, all pretty much the same, rather than particular fellow human beings. McGinn observes, for example, that the auditory hallucinations of schizophrenia and other illnesses should not be called hallucinations: hallucinations are percepts that do not correspond to reality, while the hallucinations in the ward should be classified with images. (See, in effect, reasons 1–9 above.) Karl Jaspers is praised for having noticed this. The McGinn dichotomy relentlessly files the unfileable.

It is good to escape back to the analytical philosophy of mind. We return not via images but by way of the imagination. If you suspect that imagination has not all that much to do with images, but can often work without images, you may detect a radical split in the book here. No matter: what is said is very interesting. McGinn beautifully puts together and discusses the philosophies that Russell and Wittgenstein advanced early in the twentieth century. Russell thought that sentences got meanings by their parts being matched with items in the world that we are acquainted with—almost as if you need only know what is the case, e.g., that this is a red leaf, in order to grasp meanings. Wittgenstein taught that you have to know what might be the case—you might be seeing a leaf or not, and the choice is crucial. Meaning is not a matter of actuality but of possibility. Possibilities—that means imagining what is other than what is the case. So a proper understanding of meaning demands reflection on the imagination. And for McGinn this includes the acts of imagining and of imaging. McGinn does not resist the sensible notion of mental acts as so many of his colleagues do.

That is just the beginning of a very quick speculation that he puts forward, which goes like this: the evolutionary discovery that made language possible was the imagination. And contrary to my skepticism, imagination starts with images. That which is in the mind differs from what is before the eyes. The primary power of the imagination is to envisage as possible what is not the case. What better mechanism to start with than visualizing the face of the lover while looking at the leaf? Thus McGinn leads us to speculation far more stimulating, far more imaginative, than most of what passes for evolutionary psychology.

Is the imagination something new with human beings? Hume, too, thought that “animals are but little susceptible either of the pleasures or pains of the imagination.” But one must be wary. Rousseau had an idea parallel to McGinn’s, that desire requires imagination. Desire requires the sense of possibility beyond the present, and Rousseau is wonderfully astute about how desire differs from need. Less astutely he said of the savage that “his imagination paints no pictures.”5 Are we sure that the imagination of a dog paints no pictures? Olfactory images, anyone? We have seen Tray asleep, making noises and jerking as if dreaming—dreaming sights, dreaming smells.

I happen not to follow McGinn. If there is such a thing as an objective question whether animals form images, I have no confidence in its answer. I also doubt the giving priority, first to imagination, then to meaning. In my opinion, the aspect of imagination and talk on which McGinn fixes is so closely tied to his theory of the importance of grasping possibilities in order to grasp meanings that it makes little sense to think of an imaginative faculty evolving first. At best it would be language and imagination evolving together. And that may be too close to tautology: we cannot confidently say of a being that it has imagination until we can talk with it. But it is rare and wonderful to end a book with a new idea. Let us hope it is quickly developed and refined with vigor before it lapses into scholasticism.

This Issue

April 7, 2005