The visual cortex is spread out over the back of the brain, at the back of the occipital cortex, and it's broken up into many regions: V1, V2, V3, V4, and so on. The fascinating feature of these regions is that they preserve the basic pattern or mosaic of cells in the eye's retina, so that an image on the retina is actually visible, you might say -- or is at least discernable as an image -- in these many regions. On the other hand, such an image is hugely distorted because of the vast discrepancy in grain levels between the retina's high-resolution fovea and its low-resolution parafovial regions. Neurons connected to the fovea take up much more space, so if you were to color in this distorted image, the resulting picture would look shockingly different from the way the world seems to us.

Acknowledging this incongruity is critical, for it jolts us out of a terrible conceptual habit: namely, our tendency to think of vision as if it were television, as if the task of vision were to create in our brains a picture of the world. This is the fallacy of the so-called Cartesian theater, a popular but erroneous conception of consciousness.

By studying the visual cortex, you learn that the job of perception is not one job but many. And these jobs get parceled out. Different regions -- different retinotopic maps -- specialize. Moreover, they don't send their results to one central processing area; they send their results all over the place. Everything is connected to everything else, and the connections run back and forth. The metaphor of "inbound" information -- sensory data coming from the outside world to the brain's central showroom -- has to stop as soon as you reach V1. There's no good way of defining "more central" in the brain than that. Neural fibers go every which way.

There's nothing problematic with this decentralization, by the way. It's just what we should expect. Indeed, the reason I particularly delight in the visual cortex is that it rubs your nose so firmly in the unacceptability of the traditional, centralized vision of consciousness.

The key insight about vision and the brain is not that there are, say, hand-eye coordination circuits -- though there are, of course. It's that there are lots of these systems which use the eyes in one way or another. The wrong approach is to try to simplify that truth into a conception of the eyes as simple receivers of some input which will subsequently get packaged and sent to some place to get appreciated. That's a fundamental error -- albeit an extremely attractive, seductive error. Simply confronting what we now know about the functional neuroanatomy of the visual cortex seems to me the first step in forcing us out of these bad conceptual habits.

It's no accident, I think, that a rough synonym for "I understand" is "I see." Vision seems to be the dominating modality for human thought. What's tantalizing and wonderful about vision is that it is both too familiar and too strange. We can't get away from it; it's too close to us. It's very hard to get the distance from it that you need to make sense of it. And as soon as you do get the distance, you're inclined to think you've left something crucial out. This inclination is at the heart of the so-called first-person (as opposed to the third-person) theory of consciousness. As an embattled defender of the third-person approach, I have to contend on an all-too-regular basis with thinkers who are just sure in their bones that as soon as you adopt a resolutely third-person approach to something like vision, you don't just leave out a tiny bit of the topic, you omit the most important feature of consciousness. This, I think, is a cognitive illusion of the most pernicious sort -- but of course just badmouthing it doesn't make it go away, sad to say.

Daniel Dennett, the author of Darwin's Dangerous Idea, is Distinguished Arts and Sciences Professor, Professor of Philosophy, and Director of the Center for Cognitive Studies at Tufts University.