Availability: The Cognitive Basis of Experience?

Availability: The Cognitive Basis of Experience?

David J. Chalmers

Department of Philosophy
University of Arizona
Tucson, AZ 85721

chalmers@arizona.edu

[[This was written as a commentary on Ned Block's paper "On A Confusion about a Function of Consciousness". It appeared in Behavioral and Brain Sciences 20:148-9, 1997, and also in the collection The Nature of Consciousness: Philosophical Debates (MIT Press, 1997) edited by Block, Flanagan, and Guzeldere.]]

Introduction

Block's distinction between access consciousness and phenomenal consciousness (or experience) is very useful. There is clearly a conceptual distinction here, as illustrated by the facts that: (1) one can imagine access without experience and vice versa; (2) access can be observed straightforwardly, whereas experience cannot; and, most important, (3) access consciousness seems clearly amenable to cognitive explanation, whereas phenomenal consciousness is quite perplexing in this regard. But the tight empirical link between the two phenomena deserves attention.

Bringing access and experience closer together.

Block himself notes that P-consciousness and A-consciousness often occur together. This is no accident, as one can see by noting that a P-conscious experience is usually reportable, and that reportability implies accessibility of the corresponding information. Block does not think they always occur together, but I think that with appropriate modifications they might. One of the most interesting projects in this area is that of modifying the concept of A-consciousness in such a way as to make it plausible that A-consciousness (in the modified sense) and P-consciousness are perfect correlates.

A good start is the modified notion of direct availability for global control. That is, a content is A-conscious in the modified sense when it is directly available for use in directing a wide range of behaviors, especially deliberate behaviors. I am not sure how different this is from Block's definition: it plays down the role of rationality and reasoning (after all, impairments of rationality probably do not diminish phenomenal consciousness), it relegates verbal report to the status of a heuristic (as Block himself suggests), and there is another important difference that I will come to shortly. The restriction to direct availability works to eliminate contents that can be retrieved with some work but that are not conscious.

To see how well this modified notion of A-consciousness correlates with P-consciousness, we need to see how it handles Block's examples in which one sort of consciousness occurs without the other. Block's examples of A-consciousness without P-consciousness are all mere conceptual possibilities (zombies and super-blindsight, for example), so they are not relevant here, but to illustrate P-consciousness with A-consciousness he gives some real-world examples. One is Sperling's example in which all nine letters in a square array are experienced, but only three can be reported at a time. In this case, only three letter-representations are accessed, but it is nevertheless plausible that each of the nine was available, until the process of access destroyed their availability. This works because the modified notion of A-consciousness is dispositional - not access but accessibility is required. And it is plausible that all nine letter-representations are A-conscious in the modified sense. So even in this case, P-consciousness and modified A-consciousness occur together.

The case of the drilling noise in the background can be handled similarly. Here it seems reasonable to say that the information was directly available all along; it simply wasn't accessed. The case of experience under anesthesia (if this is actual) is trickier, but we might handle it by saying that in these cases the corresponding contents are available for global control; it is just that the control mechanisms themselves are mostly shut down. We might say that the information makes it to a location where it could have been used to direct behavior, had the motor cortex and other processes been functioning normally.

Other cases could be considered and further refinements could be made. A fuller account might flesh out the kind of availability required (perhaps a kind of high-bandwidth availability is required for experience, or at least for experience of any intensity) and might specify the relevant kind of control role more fully. Counterexamples are not threatening but helpful; they allow us to refine the definition further. The details can be left aside here; the point is that this project will lead to a functionally characterized property that might correlate perfectly with P-consciousness, at least in the cases with which we are familiar.

This property - something in the vicinity of direct availability for global control - could then be thought of as the information-processing correlate of P-consciousness, or as the cognitive basis of experience. There are some interesting consequences for the issues that Block discusses.

Empirical work on consciousness.

Block notes that researchers on consciousness often start with an invocation of phenomenal consciousness but end up offering an explanation of A-consciousness and leaving P-consciousness to the side. The tight link between the two suggests that a somewhat more charitable interpretation is possible. If experience correlates with availability for global control, much of this work can be interpreted as seeking to explain A-consciousness, but trying to find a basis for P-consciousness. For example, Crick and Koch's oscillations are put forward because of a potential role in binding and working memory; that is, in integrating contents and making them available for control (working memory is itself an availability system, after all). If both the empirical hypothesis (oscillations subserve availability) and the bridging principle (availability goes along with experience) are correct, then the oscillations are a neural correlate of experience, which is just what Crick and Koch claim.

The same holds elsewhere. Shallice's "selector inputs" for "action systems" (1972) and his "Supervisory System" (1988a; 1988b) are clearly supposed to play a central role in availability and control; if the empirical hypothesis is correct, these could reasonably be regarded as part of the basis for conscious experience. Similarly, the "global workspace" of Baars (1988), the "high-quality representations" of Farah (1994), the "temporally-extended neural activity" of Libet (1993), and many other proposals can be all be seen as offering mechanisms in the process whereby some contents are made available for global control. The common element is striking. Of course, it is an empirical question which of these proposals is correct (although more than one might be, if they offer accounts of different parts of the process or descriptions at different levels). But insofar as these mechanisms play a role in the availability/control process, they are candidates to be neural or cognitive correlates of experience, which is often what the authors suggest (correlation is all that Farah and Libet claim; Shallice and Baars oscillate between "correspondence" and explanation).

The picture is this: (1) we know that availability goes along with experience; (2) we empirically discover that some mechanism plays the central role in the availability process. We may then conclude that the mechanism is part of the explanation of A-consciousness and part of the basis of P-consciousness. Of course, the story about the mechanism does not alone explain P-consciousness, as we still have not explained why availability always goes along with experience; we have simply taken for granted that it does. But if we are prepared to take the link between availability and experience as a kind of background assumption (perhaps for later explanation), this can provide a useful partial explanation of the contents of experience.

A phenomenal consciousness module?

Interestingly, this analysis allows us to make some sense of the idea of a phenomenal consciousness module. If it turns out that there is a single system responsible for mediating the availability of certain contents for global control - something like Baars' global workspace or Shallice's supervisory system - then it might be plausible that the contents of that system correspond precisely to the contents of experience, and maybe we could call it a P-consciousness module. I do not think it is probable that there is such a module - more likely there are many different mechanisms by which contents become available for a control role - but at least the idea makes sense. But the only way there could be a "P-consciousness" module would be for it to be an availability/control module. If a module were dissociable from the relevant role in availability and control, the considerations above suggest that it would be dissociable from P-consciousness too.

In particular, there is something very strange about the idea of an "epiphenomenal" P-consciousness module (Block's Figure 3). The main motivation for epiphenomenalism is surely that experience seems superfluous to any information-processing; but Block's idea suggests an implausible epiphenomenalism within the information-processing story. Indeed, if the module has no effect on other processes, then we could lesion it with no external change (same reports, even), and no empirical evidence could support the hypothesis. Perhaps Block means to allow that the module has the very limited function of causing phenomenal reports, so that lesioning it eliminates remarks such as "I am having a blue sensation." But now either (1) remarks such as "There is a blue object", confident blue-directed behavior, and so on are all eliminated too - in which case the module had an important function after all - or (2) they are preserved (a kind of ultra-superblindsight), implying an extraordinary independence between the pathways responsible for phenomenal report and those responsible for visual descriptions and normal visual processing. Given the remarkable coherence between visual descriptions and reports of visual experience, one presumes that they are tied more closely than this.

The function of consciousness?

The link between P-consciousness and (modified) A-consciousness makes the search for a function for P-consciousness even more hopeless. Given the correlation, any purported function for P-consciousness can be attributed to A-consciousness instead.

Only those who implausibly identify the concept of P-consciousness with that of (modified) A-consciousness have a way out. If one accepts the conceptual distinction, one will accept the conceivability of zombie functional isomorphs (made of silicon, say). To be consistent, one must then accept the conceivability of zombie physical isomorphs, as there is no more of a conceptual entailment from neural stuff to consciousness than there is from silicon stuff. From here, it is easy to see that P-consciousness gives me no functional advantage. After all, I am different from my zombie twin only in that I have P-consciousness and he does not, but we are functionally identical.

Block suggests that P-consciousness might "grease the wheels" of A-consciousness, but this cannot work. P-consciousness is redundant to the explanation of the physical mechanisms of A-consciousness, as the conceivability of the zombie shows: same physical mechanisms, same explanation of A-consciousness, no P-consciousness. The remaining option is to "identify" P-consciousness with modified A-consciousness (empirically but not conceptually), solving the problem by fiat. I think this sort of identification without explanation misunderstands the way that scientific identification works (see Chalmers 1995; Jackson 1994), but in any case it still leaves the concept of P-consciousness with no explanatory role in cognitive functioning. The independent concept of A-consciousness does all the work. I think it best to accept instead that phenomenal consciousness is distinct from any physical or functional property, and that it does not need to have a function to be central to our mental lives.

References

Baars, B.J. 1988. A Cognitive Theory of Consciousness. Cambridge University Press.

Chalmers, D.J. 1996. The Conscious Mind: In Search of a Fundamental Theory. Oxford University Press.

Crick, F. and Koch, C. 1990. Towards a neurobiological theory of consciousness. Seminars in the Neurosciences 2: 263-275.

Farah, M.J. 1994. Visual perception and visual awareness after brain damage: A tutorial overview. In (C. Umilta and M. Moscovitch, eds.) Consciousness and Unconscious Information Processing: Attention and Performance 15. MIT Press.

Jackson, F. 1994. Finding the mind in the natural world. In (R. Casati, B. Smith, and S. White, eds.) Philosophy and the Cognitive Sciences.

Shallice, T. 1972. Dual functions of consciousness. Psychological Review 79:383-93.

Shallice, T. 1988a. From Neuropsychology to Mental Structure. Cambridge University Press.

Shallice, T. 1988b. Information-processing models of consciousness: possibilities and problems. In (A. Marcel and E. Bisiach, eds.) Consciousness in Contemporary Science. Oxford University Press.