Consciousness is more familiar to us than any other feature of our world, since it is the route by which anything at all becomes familiar. But this is what makes consciousness so hard to pinpoint. Look for it wherever you like, you encounter only its objects -- a face, a dream, a memory, a color, a pain, a melody, a problem, but nowhere the consciousness that shines on them. Trying to grasp it is like trying to observe your own observing, as though you were to look with your own eyes at your own eyes without using a mirror. Not surprisingly, therefore, the thought of consciousness gives rise to peculiar metaphysical anxieties, which we try to allay with images of the soul, the mind, the self, the "subject of consciousness," the inner entity that thinks and sees and feels and that is the real me inside. But these traditional "solutions" merely duplicate the problem. We cast no light on the consciousness of a human being simply by redescribing it as the consciousness of some inner homunculus -- be it a soul, a mind, or a self. On the contrary, by placing that homunculus in some private, inaccessible, and possibly immaterial realm, we merely compound the mystery.
Putting the point in that way makes it clear that, in the first instance at least, the problem of consciousness is a philosophical, not a scientific, problem. It cannot be solved by studying the empirical data, since consciousness (as normally understood) isn't one of them. We can observe brain processes, neurons, ganglions, synapses, and all the other intricate matter of the brain, but we cannot observe consciousness. I can observe you observing, but what I observe is not that peculiar thing that you know from within and that is present, in some sense, only to you. At least, so it would seem; if this is some kind of mistake, it is a philosophical and not a scientific argument that will tell us so.
This appropriation of the question by philosophy is apt to make scientists impatient. Surely, they will argue, if consciousness is real it must be part of the real world -- the world of space and time, which we observe with our senses and explain by science. But what part? First-person reports of conscious states are radically affected by brain damage, and the behavior that leads us to describe others as conscious originates in the nervous system, whose functions seem to be largely controlled by the brain. Common sense and scientific inference therefore both point to the brain as the seat of consciousness. So, scientists argue, let's study the brain and find out exactly which of its processes correspond to our conscious mental states. That way, they suggest, we will find out what consciousness is.
But will we? Unfortunately, the philosophical problem comes back at us in another form. How exactly do we discover a correspondence between consciousness and a brain process, given that consciousness is not something that we observe? And suppose we overcome that difficulty and produce a theory correlating conscious mental states with specific neurological events. This means that we have discovered what consciousness is only if we can advance from correspondence to identity. And that is precisely what so many philosophers doubt we can do. True, there are some who defend the view that conscious states are identical with brain processes, but they defend it on philosophical, not scientific, grounds. And their view is open to radical objections: for example, how can a state of one thing (a person) be identical with a process in another (a brain)?
If the neurobiologist Christof Koch, professor of cognitive and behavioral biology at Caltech, enters this territory with some trepidation, he nevertheless hopes to take possession of it in the name of science. The task, he believes, is to avoid getting lost in definitions and conceptual puzzles and instead to discover the "neuronal correlates of consciousness." He at once narrows that target, however, to "the minimal set of neuronal events and mechanisms jointly sufficient for a specific conscious percept." In other words, the object of study is not consciousness as such but "specific conscious percepts," in particular those involved in visual perception. Koch's ambition, nevertheless, is to integrate the analysis of vision into the more general program that he developed with the late Francis Crick, one of the discoverers of the structure of DNA, who contributes the foreword to the book. That program is to explain how consciousness evolved and identify the processes in the brain that carry it. The book gives a fairly comprehensive account of what neurobiology has to say about the higher functions of the brain. It is not surprising, therefore, that the writing is densely scientific and heavily referenced, with many digressions. But proceeding on the supposition that the science is correct, what do we make of the title? Does neurobiology in the style of Crick and Koch really take us further in the "quest for consciousness"? Or is it simply amassing more and more information about the brain, without telling us how brain and mind are connected?
First Person Singular
One of the problems, which constantly intrudes on Koch's argument but is never resolved, is that conscious mental states do not belong to a single category. We assume that all sensations are conscious (there is no such thing, for example, as unconscious toothache), that there is both conscious and unconscious thought, and that while desire may be unconscious, intention never is. But what do conscious mental states have in common? At times Koch seems to suggest that they are all "felt" by the subject, or that they each possess a particular subjective quality or "quale" that is observable only to the subject. But we don't feel our thoughts, and there is no subjective quale that distinguishes the belief that two plus two is four from the belief that three plus three is six, or the intention to sit down to supper from the intention to eat a steak. In the case of language-using creatures, we distinguish conscious from unconscious mental states through the "first-person" perspective. A state is conscious if the subject can truly confess to it, without having to carry out an investigation and on no basis other than understanding the words that he uses. Hence in other places Koch seems to take the first-person case as characteristic of consciousness, a procedure that deprives him of a clear basis for attributing consciousness to animals, who never confess to their mental states because they never confess to anything. This is serious, since the science on which Koch draws derives from examining the brains of mice and monkeys.
Crucial to the Koch-Crick approach is a thought experiment involving the idea of the unconscious zombie. This is a creature all of whose behavior issues by reflex action, mediated by the cortex, but who is not conscious of what he is doing. This creature feels nothing, has no inner "qualia" and -- presumably -- no first-person awareness of his own mental states. So what else does he lack? Or can he be exactly like us and lack only those things? Koch is of the view that a zombie would lack the capacity to plan for the future or to deal with multicontingency situations where complex choices must be made. Plotting, planning, and deciding, he says, are among the important functions of consciousness and point to a Darwinian explanation of why consciousness exists.
Such an argument will help in the "quest for consciousness" only if we can show how "feeling," "qualia," and the "first-person case" are connected to plotting and planning. If the connection is only contingent, then a zombie could possess all the functions of consciousness without the feelings. If the connection is necessary, then it must be established in some way other than by scientific inference. As it is, the reader is left at the end of Koch's book with the puzzle with which it began: granted that there are neuronal correlates of consciousness, what exactly are they correlated with? And what exactly do we mean by "correlation"?
To answer that question, I would suggest first that we dismiss the idea of purely subjective "qualia." The belief that these essentially private features of mental states exist, and that they form the introspectible essence of whatever possesses them, is grounded in a confusion, one that Wittgenstein tried to sweep away in his arguments against the possibility of a private language. When you judge that I am in pain, it is on the basis of my circumstances and behavior, and you could be wrong. When I ascribe a pain to myself, I don't use any such evidence. I don't find out that I am in pain by observation, nor can I be wrong. But that is not because there is some other fact about my pain, accessible only to me, which I consult in order to establish what I am feeling. For if there were this inner private quality, I could misperceive it; I could get it wrong, and I would have to find out whether I am in pain. To describe my inner state, I would also have to invent a language, intelligible only to me -- and that, Wittgenstein plausibly argues, is impossible. The conclusion to draw is that I ascribe pain to myself not on the basis of some inner quale but on no basis at all.
Of course, there is a difference between knowing what pain is and knowing what pain is like. But to know what it is like is not to know some additional inner fact about it, but simply to have felt it. We are dealing with familiarity rather than information. While one philosopher -- Thomas Nagel, a professor at New York University and author of The View from Nowhere, a fascinating study of subjectivity -- has placed great emphasis on the "what it's like" idea, suggesting that it describes a distinctive mark of conscious experience, the idea remains opaque to further analysis. "What it's like" is not a proxy for a description but a refusal to describe. We can spell it out, if at all, only in metaphors. Q: "What's it like, darling, when I touch you there?" A: "Like the taste of marmalade, harmonized by late Stravinsky."
Similarly, we are not going to get very far in understanding consciousness if we concentrate on the idea of "feeling" things. For there are conscious mental states that have nothing to do with feeling. We feel our sensations and emotions, certainly, just as we feel our desires. All of those mental states would once have been classified as passions, as opposed to mental actions -- thought, judgement, intention, deduction -- which are not felt but done. I can deliberately think of Mary, judge a picture, make a decision or a calculation, even imagine a centaur, but not deliberately have a pain in the finger, a fear of spiders, or a desire for more cake. Even if I could have a pain by willing it, or if I manage to suppress my desires, this does not mean that pains and desires are actions, but only that they are passions that I can affect through mental discipline, as a yogi might reduce his heart rate. Moreover, there are psychologists and philosophers who seem quite happy with the idea of "unconscious feelings." We may balk at the expression, but we know what they mean. It is possible to feel something without being conscious of the feeling. Feeling is a mark of consciousness only if we interpret "feeling" as "awareness." But what is it to be aware of something? Well, to be conscious of it.
How do we fight ourselves free from this tangle of circular definitions and misleading pictures? Two ideas seem to me especially helpful in explaining our sense of consciousness as a realm apart. The first is that of an emergent property. Mental states generally, and conscious states in particular, can be seen as emergent states of organisms. A useful analogy is the face in a picture. When a painter applies paint to a canvas, she creates a physical object by purely physical means. This object is composed of areas and lines of paint, arranged on a surface that we can regard, for the sake of argument, as two dimensional. When we look at the painting, we see a flat surface, and we see those areas and lines of paint, and also the surface that contains them. But that is not all we see. We also see a face that looks out at us with smiling eyes. In one sense, the face is a property of the canvas, over and above the blobs of paint; you can observe the blobs and not see the face, and vice versa. And the face is really there: someone who does not see it is not seeing correctly. On the other hand, there is a sense in which the face is not an additional property of the canvas, for as soon as the lines and blobs are there, so is the face. Nothing more needs to be added in order to generate the face -- and if nothing more needs to be added, the face is surely nothing more. Moreover, every process that produces just these blobs of paint, arranged in just this way, will produce just this face -- even if the artist herself is unaware of the face. (Imagine how you would design a machine for producing Mona Lisas.)
Maybe consciousness is an emergent property in that sense: not something over and above the life and behavior in which we observe it, but not reducible to them either.
The second helpful thought is one first given prominence by Kant and thereafter emphasized by Fichte, Hegel, Schopenhauer, and a whole stream of thinkers down to Heidegger, Sartre, and Thomas Nagel. The idea is to draw a distinction between the subject and the object of consciousness, and to recognize the peculiar metaphysical (Wittgenstein would say grammatical) status of the subject. As a conscious subject, I have a point of view on the world. The world seems a certain way to me, and this seeming defines my unique perspective. Every conscious being has such a perspective, since that is what it means to be a subject rather than a mere object. When I give a scientific account of the world, however, I am describing objects only. I am describing the way things are, and the causal laws that explain them. This description is given from no particular perspective. It does not contain words like "here," "now," and "I"; and while it is meant to explain the way things seem, it does so by giving a theory of how they are. In short, the subject is in principle unobservable to science, not because it exists in another realm but because it is not part of the empirical world. It lies on the edge of things, like a horizon, and could never be grasped "from the other side," the side of subjectivity itself. Is it a real part of the real world? The question begins to look as though it has been wrongly phrased. I refer to myself, but this does not mean that there is a self that I refer to. I act for the sake of my friend, but there is no such thing as a sake for which I am acting. (The parallel illustrates Wittgenstein's view of these puzzles as essentially grammatical.)
We can relate to conscious creatures in ways that we cannot relate to objects. Their behavior is the outcome of the way things seem to them and can therefore be altered by altering the way things seem. Giving them "food for thought" or -- in the case of more primitive animals -- "food for perception" and "food for belief," we also bend them to our purposes. Because they feel pleasure and pain, they can be rewarded and punished and so taught to behave in new ways. Everybody who has trained a dog or a horse in even the simplest task knows that consciousness is an essential intermediary in achieving the final result, and that there is nothing puzzling about this at all: consciousness is as much a part of the behavioral repertoire of the animal as eating and excreting. It consists in a set of functional connections between world and behavior, of a kind that leads us to identify a "point of view," a "way things seem" that distinguishes the creature with which we are dealing. This point of view is also the quickest and easiest channel to the springs of its behavior.
In referring to behavior, we don't have to accept the old behaviorist theory that mental predicates can simply be reduced to behavioral syndromes. When we interpret behavior as the expression of a conscious state, we are expressly situating it in an intuitively understood nexus of causal relations. The behavior of a man in pain is only superficially like the behavior of an actor who is pretending to be in pain. The sufferer really cannot stand on his injured leg, and the leg really is injured; the actor's behavior is voluntary, the sufferer's involuntary. And so on. All those judgments are hypotheses concerning the functional connections between world and behavior, and they form parts of a spontaneous theory that some philosophers have called "folk psychology."
Now, there are certainly "neuronal correlates" of consciousness, so understood: namely, all the electrical processes that are necessary to generate conscious behavior (among which, according to Koch, gamma waves -- oscillations recorded by an electroencephalogram in the 30- to 70-hertz domain -- are particularly important). Some animals exhibit these processes; some (insects, for instance) don't. To discover the source of these processes is, in a sense, to discover the seat of consciousness in the brain. But does this bring us any nearer to knowing what consciousness is? Suppose you came across a person who behaved and talked as you did, who related to you in all the ways that people relate to each other, and who one day -- to your astonishment -- unzipped the top of his head to reveal nothing save a dead kitten and a ball of string. Scientifically impossible, perhaps. But logically possible, and giving no grounds at all to deny that this person was conscious.
The Unselfconscious Dog
To put the point another way, consciousness is an emergent property of organisms. But it emerges from the total behavioral and neurological repertoire, not from brain processes considered in themselves -- just as the face in the painting emerges from the whole array of colored patches, not from the canvas that supports them, considered in itself. Of course, you cannot have the behavior without the brain, just as you can't have the painting without the canvas. In that sense there will be neuronal correlates of consciousness. But the discovery of these correlates does not tell us what consciousness is, nor does it solve the mystery of the subject, nor the equally perplexing mystery of the first-person case.
There is a difficulty that I have avoided, and which Koch too avoids, though incidental remarks show that he is aware of it. This difficulty arises from two radical ontological divisions in the realm of the mental. First, there is the division that separates conscious from unconscious creatures. We attribute perception of a kind to mussels and oysters -- but are they conscious? Should we feel remorse when we pry open the oyster and sting its wounds with lemon juice? We are inclined to say that such organisms are too primitive to admit the application of concepts like those of feeling, belief, and desire. Maybe that goes for insects, too, however much we may admire their amazing social organization and perceptual powers.
Secondly, there is the division that separates merely conscious creatures from self-conscious creatures like us. Only the second have a genuine "first-person" perspective, from which to distinguish how things seem to me from how they seem to you. The creature with "I" thoughts has an ability to relate to its kind that sets it apart from the rest of nature, and many thinkers (Kant and Hegel among them) believe that it is this fact, not the fact of consciousness per se, that creates all the mysteries of the human condition. Although dogs are conscious, they do not reflect on their own consciousness as we do: they live, as Schopenhauer put it, in "a world of perception," their thoughts and desires turned outwards to the perceivable world.
The difficulty is this: we want to say of human beings that their self-consciousness is a systematic attribute of their mental life, which affects everything that they think and feel. We want to say of dogs that their consciousness is a systematic attribute of their mental life too, since it distinguishes them categorically from mollusks and beetles. Yet similar mental states seem to exist at all three levels. The beetle sees things; so does the dog; so does the person. How is it that one and the same mental process -- visual perception -- can exist in three different ontological predicaments, so to speak: as a reflex link between visual input and behavioral output, as a conscious perception, and as part of the continuous and distinguishing sense of self?
That question has led some writers (the neuroscientist Antonio Damasio in his book Looking for Spinoza, for instance) to think of consciousness and self-consciousness as monitoring processes -- a move that comes dangerously close to the old homunculus fallacy. It is not as though my mind were just like a dog's, only with a self observing it, or a dog's just like an insect's, only with an internal monitor. Consciousness and self-consciousness are holistic properties, which emerge from the totality of a creature's physiognomy and behavior. We may discover organizations in the brain and nervous system that are biologically necessary for these features. But those "neuronal correlates" are no more likely to cast light on the mysteries of consciousness than the back of Leonardo's Mona Lisa can explain the mystery of her smile.
The conclusion to which I am tempted is not that there is no such thing as consciousness, but that there is nothing that consciousness is, just as there is no physical object that actually is Mona Lisa's smile.
Roger Scruton is visiting professor in the Department of Philosophy, Birkbeck College, London, and the author of more than 20 books, including Modern Philosophy and England: an Elegy. He farms in Wiltshire, England.
Copyright Technology Review 2005.