Empirical Constraints on the Concept of Consciousness

Commentary on Crick and Koch’s ‘A Framework for Consciousness’

As other commentators on the target article have pointed out, and as Crick and Koch themselves acknowledge, their hypotheses regarding the neural correlates of consciousness (NCC) have much in common with the work of other researchers. There now seems to be a well-established research consensus that the NCC are distributed, integrated, and semi-hierarchical, extending across many brain systems subserving various cognitive functions. Consciousness seems to involve neural coalitions, not central executives.

Crick and Koch take pains to point out that they aren’t addressing the “hard problem” of consciousness, of how “the redness of red could arise from the actions of the brain,” but only looking to find the NCC. To some such as John G. Taylor (2003), this seems like a patent evasion of the central issue about experience. All you’re going to get from C&K’s framework, he says, is knowledge about how the brain processes information; you’re not going to get anywhere in explaining “inner experience” or the “inner self,” that is, qualitative phenomenal consciousness.

To some extent I share Taylor’s frustration, but C&K are right to be circumspect in approaching the hard problem, since the ultimate explanatory target – phenomenal consciousness – is so ill-define d. What their and others’ work on the NCC will do, I think, is not only to pin down what happens in the brain when we have experience, but to help constrain and define the explanandum itself. That is, what the NCC do should give us clues as to what consciousness is. We should be open to revisions to our very concept of the phenomenal in light of what neuroscience has to say, for there is nothing sacred or immutable about that concept. As Nicholas Humphrey (2000) has suggested, there has to be mutual accommodation between science on the one hand and philosophy on the other, and to do that we have to work both sides of the street.

So what are some of the implications of C&K’s framework for our conception of consciousness? Most generally, their account suggests that the NCC are in the business of representing the world and the organism within in it via what C&K call “feature-detectors,” “explicit neurons,” “essential nodes” and coalitions of such sub-systems in a global workspace (Baars,1988). This implies, although it doesn’t prove, that qualitative phenomenal states serve a representational function. Two neural/functional differences between conscious and non-conscious states, C&K say, are that consciously experienced features require a dedicated set of explicit neurons, and that the essential node formed by this set must participate in a coalition or active network. Consciousness, then, might involve the representation of specific, lower-level features within a wider, integrated representational context, what Thomas Metzinger (2000a) calls the overarching “reality model.” This suggests that discriminable qualitative features in experience (e.g., my sensation of red) are qualitative at least partially by virtue of the participation of the corresponding essential node in the global workspace.

If the NCC involve feature-detectors, this might lead us to suppose that elements of sensory experience, e.g., the red of that apple over there, are representations of regularities in the world our cognitive systems have evolved to detect. That is, consciousness seems to involve informational content about the self and world (Dretske, 1995; Tye, 1995) – highly integrated content that subserves the sorts of flexible and intentional behavior that’s possible only when conscious.

If, as C&K’s framework suggests, consciousness involves representational contents participating in an integrated representational architecture, how might this help address the hard problem? One way it can help is by prompting us to think about what’s specifically qualitative about experience in terms of our representational situation, and to see if this might be explicable on a representationalist account.

For instance, basic elements of visual experience, e.g., hue, contrasts, lines, edges, etc., may correspond to the informational content carried by basic sensory feature-detectors that in effect set the limits of resolution of conscious experience. This can help explain why my experience of red is prototypically qualitative in the sense of being homogeneous, non-decomposable, ineffable, and seemingly intrinsic: there is no more basic feature of the world we’re capable of detecting via our visual color systems; thus there is no further directly represented information available about this content. As a result, there’s nothing to tell or know about the redness of red considered in and of itself; there are only structural relations between it and other hues, plus the various represented objects and events it helps us to characterize. To be qualitative, on this account, involves being a not further directly representable sensory representation.

But C&K’s framework also suggests that for it to be something it’s like for me to experience red, there also has to exist a full-blown representational context, instantiated by neural coalitions forming the global workspace, within which basic representational contents such as hue get bound and integrated into a represented self-in-the-world-of-objects. Although C&K don’t discuss the representational basis of the phenomenal subject in their article, others such as Metzinger and Revonsuo argue, roughly, that to feel oneself a subject in the midst of an experienced world is to cognitively consist of suitably elaborated and integrated representational content that constitutes the behavior-controlling reality model; a model comprised, most basically, of representations of what’s self versus what’s not-self (Metzinger, 2000a, 2003). This model, crucially, cannot directly represent itself as a model, and so becomes a qualitatively experienced virtual world (Revonsuo, 2000) with the represented subject (“inner self”) at its center, for whom it is like to have experience, e.g., of red. The representational situation of basic sensory elements not being further representable, and thus prototypically qualitative, thus in a way repeats at the higher level: the integrated representational content of the self in a world of objects becomes phenomenal partially as a result of our representational limitations.

If consciousness is a certain variety of represented content – that instantiated by the global workspace which dominates in controlling complex behavior – then, as Michael Tye says, we’re not going to see it (e.g., my experienced red) peering inside the brain (1995, p. 151), and of course we don’t. Nor will we find it out there in the world. Conscious content just isn’t the sort of thing that can be seen, since on this account it’s a higher-level property of representational processes instantiated by the brain (or by different physical systems, as long as they embody more or less the same representational architecture). Rather, such content is what conscious seeing, sensing, and understanding consist of and what we consist of as consciously experiencing subjects.

The representational account implicit in their framework suggests, as C&K say, that it’s a mistake to go looking for qualia in “exotic physics” (p. 124), since this is to suppose that the qualitative is a specific sort of entity or physical effect. It seeks to reify an abstract property. Nor, as Dennett (2001) has long argued, should we think of consciousness as being “produced” as a further effect of a mechanism, since it consists of particular sorts of informational content which is a property of the global workspace as it represents the self-in-a-world. So C&K might be somewhat off track to think that we have to “move from correlation to causation” (p. 124) where consciousness is concerned.

But isn’t this talk of content just to replace one mystery, traditional intrinsic qualia, with another, an obscure abstract property of representational states? Not really, since the notion of phenomenal consciousness as a certain variety of highly integrated, behavior controlling informational content is constrained and motivated by what C&K and others are discovering about the NCC. Of course, we don’t have to think about consciousness this way, and many will continue to insist that’s what essentially qualitative necessarily escapes any reductionist account. But what, beyond folk-psychological intuitions enshrined in long-standing varieties of philosophical dualism, motivates this view of the qualitative? Nothing in science, certainly. That the representational account can help explain these intuitions (e.g., about the ineffability and intrinsicality of qualia), plus much else about the uncontroversially informational aspects of experience, counts much in its favor.

Crick and Koch and others exploring the NCC are, I think, helping to solve the hard problem by offering us the empirical basis for reconfiguring our conception of consciousness. We should be open to such conceptual revision under pressure from science, since empirically-motivated change in our pretheoretical notions (e.g., of heat, matter, mass, energy) is how progress gets made in understanding ourselves and the natural world.

© 2003 T.W. Clark

Thomas W. Clark, www.naturalism.org

References

1. Baars, B. (1988), A Cognitive Theory of Consciousness (Cambridge, MA: Cambridge University Press).
2. Crick, F., Koch, K., “A Framework for Consciousness,” Nature Neuroscience, 6 (2), Feb 2003.
3. Dennett, D. (2001), “Are we explaining consciousness yet?,” Cognition 79, pp. 221-237.
4. Dretske, F. (1995), Naturalizing the Mind (Cambridge, MA: MIT Press).
5. Humphrey, N. (2000), “How to solve the mind-body problem,” Journal of Consciousness Studies, 7 (4), pp. 5-20.
6. Metzinger, T. (ed) (2000), Neural Correlates of Consciousness: Empirical and Conceptual Questions (Cambridge, MA: MIT Press).
7. Metzinger, T. (2000a), “The subjectivity of subjective experience: a representationalist analysis of the first-person perspective,” in Metzinger 2000, pp. 285-306.
8. Metzinger, T. (2003), Being No One: The Self-model Theory of Subjectivity (Cambridge, MA: MIT Press).
9. Revonsuo, A. (2000), “Prospects for a scientific research program on consciousness,” in Metzinger 2000, pp. 41-56.
10. Taylor, J.G., (2003), “The missing self, or: 10 ways to be a zombie,” Science and Consciousness Review, http://psych.pomona.edu/scr/editorials/20030203.html.
11. Tye, M. (1995), Ten Problems of Consciousness: A Representational Theory of the Phenomenal Mind (Cambridge, MA: MIT Press).