The study by He et al. (1996) is based on a common visual aftereffect (see Figure 3a). If a subject stares for a fraction of a minute at a horizontal grating, and is then tested with a faint grating at the same location to decide whether it is oriented vertically or horizontally, the subject's sensitivity for detecting a horizontal grating will be reduced. This adaptation is orientation specific -- the sensitivity for vertical gratings is almost unchanged -- and disappears quickly. He and colleagues projected a single patch of grating onto a computer screen some 25 degrees from the fixation point. It was clearly visible and their subjects showed the predictable orientation-selective adaptation effect. Adding one or more similar patches of gratings to either side of the original grating -- which remained exactly as before -- removed the lines of the grating from visibility; it was now "masked." Subjectively, one still sees "something" at the location of the original grating, but one is unable to make out its orientation, even when given unlimited viewing time. Yet despite this inability to "see" the adapting stimulus, the aftereffect was as strong and as specific to the orientation of the "invisible" grating as when the grating was visible (see Figure 3b). What this shows, foreshadowed by earlier experiments (Blake and Fox, 1974), is that visual awareness in such cases must occur at a higher stage in the visual hierarchy than orientation-specific adaptation. This aftereffect is thought to be mediated by oriented neurons in V1 and beyond, implying that at least in this case the neurons which mediate visual awareness must be located past this stage.
Figure 3: Psychophysical displays (schematic) and results pertaining to an orientation-dependent after-effect induced by "crowded" grating patches (reproduced with permission by He, Cavanagh and Intriligator).a. Adaptation followed by contrast threshold measurement for a single grating (left) and a crowded grating (right). In each trial, the orientation of the adapting grating was either the same or orthogonal to the orientation of the test grating. Observers fixated at a distance of approximately 25 degrees from the adapting and test gratings. b. Threshold contrast elevation after adaptation relative to baseline threshold contrast before adaptation. Data are averaged across four subjects. The difference between same and different adapt-test orientations reflects the orientation-selective aftereffect of the adapting grating. The data show that this aftereffect is comparable for a crowded grating (whose orientation is not consciously perceived) and for a single grating (whose orientation is readily perceived).
Our ideas regarding the absence of the NCC from V1 are not disproven by PET experiments showing that in at least some people V1 is activated during visual imagery tasks (Kosslyn et al., 1995), though severe damage to V1 is compatible with visual imagery in patients (Goldenberg et al., 1995). There is no obvious reason why such top-down effects should not reach V1. Such V1 activity would not, by itself, prove that we are directly aware of it, any more than the V1activity produced there when our eyes are open proves this. We hope that further neuroanatomical work will make our hypothesis plausible for humans, and that further neurophysiological studies will show it to be true for most primates. If correct, it would narrow the search to areas of the brain farther removed from the sensory periphery.
The Frontal Lobe Hypothesis
As mentioned several times, we hypothesize that the NCC must have access to explicitly encoded visual information and directly project into the planning stages of the brain, associated with the frontal lobes in general and with prefrontal cortex in particular (Fuster, 1997). We would therefore predict that patients unfortunate enough to have lost their entire prefrontal cortex on both sides (including Broca's area) would not be visually conscious, although they might still have well-preserved, but unconscious, visual-motor abilities. No such patient is known to us (not even Brickner's famous patient; for an extensive discussion of this, see Damasio and Anderson, 1993). The visual abilities of any such "frontal lobe" patient need to be carefully evaluated using a battery of appropriate psychophysical tests.
The fMRI study of the blindsight patient G.Y. (Sahraie et al., 1997) provides direct evidence for our view by revealing that prefrontal areas 46 and 47 are active when G.Y. is visually aware of a moving stimulus.
The recent findings of neurons in the inferior prefrontal cortex (IPC) of the macaque that respond selectively to faces--and that receive direct input from regions around the superior temporal sulcus and the inferior temporal gyrus that are well known to contain face-selective neurons--is also very encouraging in this regard (Scalaidhe, Wilson and Goldman-Raleic, 1997). This raises the question of why would face cells be represented in both IT and IPC. Do they differ in some important aspect?
Large-scale lesion experiments carried out in the monkey suggest that the absence of frontal lobes leads to complete blindness (Nakamura and Mishkin, 1980, 1986). One would hope that future monkey experiments reversibly inactivate specific prefrontal areas and demonstrate the specific loss of abilities linked to visual perception while visual-motor behaviors -- mediated by the on-line system -- remain intact.
It will be important to study the pattern of connections between the highest levels of the visual hierarchy -- such as inferotemporal cortex -- and premotor and prefrontal cortex. In particular, does the anatomy reveal any feedback loops that might sustain activity between IT and prefrontal neurons (Crick and Koch, 1997)? There is suggestive evidence (Webster et al., 1994) that projections from prefrontal cortex back into IT might terminate in layer 4, but these need to be studied directly.
Gamma Oscillations
Much has been made of the presence of oscillations in the gamma range (30-70 Hz) in the local-field potential and in multi-unit recordings in the visual and sensory-motor system of cats and primates (Singer and Gray, 1995). The existence of such oscillations remains in doubt in higher visual cortical areas (Young et al., 1992). We remain agnostic with respect to the relevance of these oscillations to conscious perception. It is possible that they subserve figure-ground in early visual processing.
Philosophical Matters
There is, at the moment, no agreed philosophical answer to the problem of consciousness, except that most living philosophers are not Cartesian dualist -- they do not believe in an immaterial soul which is distinct from the body. We suspect that the majority of neuroscientists do not believe in dualism, the most notable exception being the late Sir John Eccles (1994).
We shall not describe here the various opinions of philosophers, except to say that while philosophers have, in the past, raised interesting questions and pointed to possible conceptual confusions, they have had a very poor record, historically, at arriving at valid scientific answers. For this reason, neuroscientists should listen to the questions philosophers raise but should not be intimidated by their discussions. In recent years the amount of discussion about consciousness has reached absurd proportions compared to the amount of relevant experimentation.
The Problem of Qualia
What is it that puzzles philosophers? Broadly speaking, it is qualia --the blueness of blue, the painfulness of pain, and so on. This is also the layman's major puzzle. How can you possibly explain the vivid visual scene you see before you in terms of the firing of neurons? The argument that you cannot explain consciousness by the action of the parts of the brain goes back at least as far as Leibniz (1686; see the translation 1965). But compare an analogous assertion: that you cannot explain the "livingness" of living things (such as bacteria, for example) by the action of "dead" molecules. This assertion sounds extremely hollow now, for an umber of reasons. Scientists understand the enormous power of Natural Selection. They know the chemical nature of genes and that inheritance is particulate, not blending. They understand the great subtlety, sophistication and variety of protein molecules, the elaborate nature of the control mechanisms that turn genes on and off, and the complicated way that proteins interact with, and modify, other proteins. It is entirely possible that the very elaborate nature of neurons and their interactions, far more elaborate than most people imagine, is misleading us, in a similar way, about consciousness.
Some philosophers (Searle, 1984; Dennett, 1996) are rather fond of this analogy between "livingness" and "consciousness," and so are we; but, as Chalmers (1995) has emphasized, an analogy is only an analogy. He has given philosophical reasons why he thinks it is wrong. Neuroscientists know only a few of the basics of neuroscience, such as the nature of the action potential and the chemical nature of most synapses. Most important, there is not a comprehensive, overall theory of the activities of the brain. To be shown to be correct, the analogy must be filled out by many experimental details and powerful general ideas. Much of these are still lacking.
This problem of qualia is what Chalmers (1995) calls "The Hard Problem": a full account of the manner in which subjective experience arises from cerebral processes. As we see it, the hard problem can be broken down into several questions, of which the first is the major problem: How do we experience anything at all? What leads to a particular conscious experience (such as the blueness of blue)? What is the function of conscious experience? Why are some aspects of subjective experience impossible to convey to other people (in other words, why are they private)?
We believe we have answers to the last two questions (Crick and Koch, 1995c). We have already explained, in the section "Why Are We Conscious," what we think consciousness is for. The reason that visual consciousness is largely private is, we consider, an inevitable consequence of the way the brain works. (By "private," we mean that it is inherently impossible to communicate the exact nature of what we are conscious of.) To be conscious, we have argued, there must be an explicit representation of each aspect of visual consciousness. At each successive stage in the visual cortex, what is made explicit is recoded. To produce a motor output, such as speech, the information must be recoded again, so that what is expressed by the motor neurons is related, but not identical, to the explicit representation expressed by the firing of the neurons associated with, for example, the color experience at some level in the visual hierarchy.
It is thus not possible to convey with words the exact nature of a subjective experience. It is possible, however, to convey a difference between subjective experiences -- to distinguish between red and orange, for example. This is possible because a difference in a high-level visual cortical area can still be associated with a difference at the motor stage. The implication is that we can never explain to other people the nature of any conscious experience, only, in some cases, its relation to other ones.
Is there any sense in asking whether the b
