Could this problem be solved by connecting two brains together in some elaborate way? It is impossible to do this at the moment, or in the easily foreseeable future. One is therefore tempted to use the philosopher's favorite tool, the thought experiment. Unfortunately, this enterprise is fraught with hazards, since it inevitably makes assumptions about how brains behave, and most of these assumptions have so little experimental support that conclusions based on them are valueless. For example, how much is a person's percept of the blue of the sky due to early visual experiences?
The Problem of Meaning
An important problem neglected by neuroscientists is the problem of meaning. Neuroscientists are apt to assume that if they can see that a neuron's firing is roughly correlated with some aspect of the visual scene, such as an oriented line, then that firing must be part of the neural correlate of the seen line. They assume that because they, as outside observers, are conscious of the correlation, the firing must be part of the NCC. This by no means follows, as we have argued for neurons in V1.
But this is not the major problem, which is: How do other parts of the brain know that the firing of a neuron (or of a set of similar neurons) produces the conscious percept of, say, a face? How does the brain know what the firing of those neurons represents? Put in other words, how is meaning generated by the brain?
This problem has two aspects. How is meaning expressed in neural terms? And how does this expression of meaning arise? We suspect (Crick and Koch, 1995c) that meaning derives both from the correlated firing described above and from the linkages to related representations. For example, neurons related to a certain face might be connected to ones expressing the name of the person whose face it is, and to others for her voice, memories involving her and so on, in a vast associational network, similar to a dictionary or a relational database. Exactly how this works in detail is unclear.
But how are these useful associations derived? The obvious idea is that they depend very largely on the consistency of the interactions with the environment, especially during early development. Meaning can also be acquired later in life. The usual example is a blind man with a stick. He comes to feel what the stick is touching, not merely the stick itself. For an ingenious recent demonstration along similar lines, see Ramachandran and Hirstein (1997).
Future Experiments
Although experiments on attention, short-term and working memory, the correlated firing of neurons and related topics may make finding the NCC easier, at the moment the most promising experiments are those on bistable percepts. These experiments should be continued in numerous cortical and thalamic areas and need extending to cover other such percepts. It is also important to discover which neurons express the NCC in each case (e.g., which neuronal subtype, in what layer, and so on), how they fire (e.g., do they fire in bursts) and, especially, to where they project. To assist this, more detailed neuroanatomy of the connectivity will be needed. This is relatively easy to do in the macaque but difficult in humans (Crick and Jones, 1993). It is also important to discover how the various on-line systems work, so that one can contrast their (unconscious) neuronal activity with the NCC.
To discover the exact role (if any) of the frontal cortex in visual perception, it would be useful to inactivate it reversibly by cooling and/or the injection of GABA agonists, perhaps using the relatively smooth cortex of an owl monkey.
Inevitably, it will be necessary to compare the studies on monkeys with similar studies on humans, using both psychophysical experiments as well as functional imaging methods such as PET or fMRI. Conversely, functional imaging experiments on normal subjects or patients, showing for instance the involvement of prefrontal areas in visual perception (Weiskrantz, 1997; Sahraie et al., 1997), can provide a rationale for appropriate electrophysiological studies in monkeys. It would help considerably if there were more detailed architectonic studies of cortex and thalamus, since these can be done post-mortem on monkeys, apes and humans. The extremely rapid pace of molecular biology should soon provide a wealth of new markers to help in this endeavor.
To understand a very complex nonlinear system, it is essential to be able to interfere with it both specifically and delicately. The major impact of molecular biology is likely to be the provisions of methods for the inactivation of all neurons of a particular type. Ideally, this should be done reversibly on the mature animal (see, for example, No et al., 1996; Nirenberg and Meister, 1997). At the moment this is only practical on mice, but in future one may hope for methods that can be used on mature monkeys (perhaps using a viral vector), as such methods are also needed for the medical treatment of humans.
As an example, consider the question of whether the cortical feedback pathways -- originating in higher visual area (in the sense of Felleman and Van Essen, 1991) and projecting into a lower area -- are essential for normal visual consciousness. There are at least two distinct types of back pathways (Salin and Bullier, 1995): one, from the upper cortical layers, goes back only a few steps in the visual hierarchy; the other, from the lower cortical layers, can also go back over longer distances. We would like to be able to selectively inactivate these pathways, both singly and collectively, in the mature macaque. Present methods are not specific enough to do this, but new methods in molecular biology should, in time, make this possible.
It will not be enough to show that certain neurons embody the NCC in certain -- limited -- visual situations. Rather, we need to locate the NCC for all types of visual inputs, or at least for a sufficiently large and representative sample of them. For example, when one blinks, the eyelids briefly (30-50 msec) cover the eyes, yet the visual percept is scarcely interrupted (blink suppression; Volkmann et al., 1980). We would therefore expect the NCC to be also unaffected by eye blinks (e.g., the firing activity should not drop noticeably during the blink) but not to blanking out of the visual scene for a similar duration due to artificial means. Another example is the large number of visual illusions. For instance, humans clearly perceive, under appropriate circumstances, a transient motion aftereffect. On the basis of fMRI imaging it has been found that the human equivalent of cortical area MT is activated by the motion aftereffect (in the absence of any moving stimuli; Tootell et al., 1995). The time course of this illusion parallels the time course of activity as assayed using fMRI. In order to really pinpoint the NCC, one would need to identify individual cells expressing this, and similar, visual aftereffects. We have assumed that the visual NCC in humans is very similar to the NCC in the macaque, mainly because of the similarity of their visual systems. Ultimately, the link between neurons and perception will need to be made in humans.
The problem of meaning and how it arises is more difficult, since there is, as yet, not even an outline formulation of this problem in neural terms. For example, do multiple associations depend on transient priming effects?Whatever the explanation, it would be necessary to study the developing animal to show how meaning arises; in particular, how much is built in epigenetically and how much is due to experience.
In the long run, finding the NCC will not be enough. A complete theory of consciousness is required, including its functional role. With luck this might illuminate the hard problem of qualia. It is likely that scientists will then stop using the term consciousness except in a very loose way. After all, biologists no longer worry whether a seed or a virus is "alive." They just want to know how it evolved, how it develops, and what it can do.
Finale
We hope we have convinced the reader that the problem of the neural correlate of consciousness (the NCC) is now ripe for direct experimental attack. We have suggested a possible framework for thinking about the problem, but others may prefer a different approach; and, of course, our own ideas are likely to change with time. We have outlined the few experiments that directly address the problem and mentioned briefly other types of experiments that might be done in the future. We hope that some of the younger neuroscientists will seriously consider working on this fascinating problem. After all, it is rather peculiar to work on the visual system and not worry about exactly what happens in our brains when we "see" something. The explanation of consciousness is one of the major unsolved problems of modern science. After several thousand years of speculation, it would be very gratifying to find an answer to it.
Notes
We thank the J.W. Kieckhefer Foundation, the National Institute of Mental Health, the Office of Naval Research and the National Science Foundation. For helpful comments we thank David Chalmers, Leslie Orgel, John Searle and Larry Weiskrantz.
Correspondence should be addressed to Dr. Francis Crick, The Salk Institute, 10010 North Torrey Pines Road, La Jolla, California 92037, U.S.A.
References
Adolphs R, Tranel D, Damasio H, Damasio A (1994) Impaired recognition of emotion in facial expressions following bilateral damage to the human amygdala. Nature 372: 669-672.
Baddeley A (1992) Working memory. Science 255: 556-559.
Ballard DH, Hinton GE, Sejnowski TJ (1983) Parallel visual computation. Nature 306: 21-26.
Blackwell KT, Buchsbaum G (1988) Quantitative studies of color constancy. J Opt Soc Amer A5:1772-1780.
Blake R, Fox R (1974) Adaptation to invisible gratings and the site of binocular rivalry suppression. Nature 249:488-490.
Boussaoud D, di Pellegrino G, Wise SP (1996) Frontal lobe mechanisms subserving vision-for-action versus vision-for-perception. Behav Brain Res 72:1-15.
Bradley DC, Chang GC, Andersen RA (1997) Activities of motion-sensitive neurons in primate visual area MT reflect the perception of depth. Nature, in press.
Braun J, Julesz B (1997) Dividing attention at little cost. Percep & Psychophys, in press.
Britten KH, Shadlen MN, Newsome WT, Movshon JA (1992) The analysis of visual motion: a comparison of neuronal and psychophysical performance. J Neurosci 12:4745-4765.
Chalmers D (1995) The Conscious Mind: In Search of a Fundamental Theory. Oxford:Oxford University Press.
Coltheart M (1983) Iconic memory. Phil Trans R Soc Lond B 302:283-294.
Cowey A, Stoerig P (1995) Blindsight in monkeys. Nature 373:247-249.
Crick F (1994) The Astonishing Hypothesis. New York:Scribner's.
Crick, F (1996) Visual perception: rivalry and consciousness. Nature 379:485-486.
Crick F, Jones E (1993
