SWIF Philosophy of Mind, 15 May 2002 http://www.swif.uniba.it/lei/mind/forums/bermudez.htm

Forums Forum 2 Carruthers's reply

Commentary on
Carruthers, Peter. Phenomenal Consciousness. Cambridge: Cambridge University Press, 2000.

José Luis Bermúdez

Department of Philosophy
University of Stirling (UK)
homepage

Peter Carruthers’s new book is a welcome addition to the extensive literature on phenomenal consciousness. Developing and expanding the principal themes of his 1996 book Language, Thought and Consciousness, Carruthers offers both an incisive discussion of the principal existing theories of consciousness (including the theories which deny the possibility of explaining consciousness) and a significantly new version of the higher-order thought approach to conscious experience. Unusual among contemporary contributors to the consciousness debate, Carruthers pays far more than lip-service to the interdisciplinary nature of the topic and material from neurophysiology, developmental psychology and the scientific study of consciousness plays a pivotal role in motivating some of Carruthers’s principal conclusions.

The argumentative structure of the book is given by what Carruthers terms the "tree of consciousness" (p. 22). The tree of consciousness is a series of seven choice nodes which effectively delineate the principal theories of consciousness. Carruthers tackles each node and motivates his own response to it until at the bottom of the decision-tree we are left with a single branch – Carruthers’s own non-linguistic dispositionalist higher-order thought theory. Of the seven choice nodes, four are particularly significant. At the first node we have to choose between "no explanation" views (Chalmers, McGinn, Jackson et al) and the possibility of a genuine naturalistic explanation – Carruthers, of course, opts for the latter. Like almost all contributors to the debate, Carruthers thinks that the only hope for a naturalistic theory of phenomenal consciousness is via the representational contents of phenomenally conscious states. Within the general framework of representationalism the key decision (choice node 4) is between first-order and higher-order theories. First-order theories (such as those maintained by Dretske, Tye and others) maintain that phenomenally conscious states are representational states which impact, or are poised to impact, on belief formation and practical reasoning. According to higher-order theories, however, there is a further condition to be met. Phenomenally conscious states must also be the targets of higher-order representational states. Carruthers opts for the higher-order approach. Within the general framework of the higher-order approach, choice node 5 asks us to decide whether the higher-order representational states are thoughts or experiences. Carruthers takes the latter option and (in choice node 7) argues that the relevant higher-order thoughts (HOTs) do not have to be formulated in a natural language.

In many ways the most considerable hurdle for naturalistic theories of phenomenal consciousness comes with the very first choice node – with defending the very possibility of such a theory against the various "mysterian" arguments that have been put forward by Kripke, Jackson, McGinn, Chalmers and others. By seeing how a naturalistic theory confronts these arguments we can appreciate the type of explanation which it takes itself to be offering. In responding to Kripke, Carruthers makes plain the modal strength he thinks appropriate for a naturalistic theory of consciousness. It would be wrong to demand a necessary identity between types of phenomenally conscious states and types of natural properties – for any given phenomenally conscious state type might in a different possible world be differently realised. Nonetheless, a naturalistic theory must entail a logical supervenience claim, so that it cannot even be logically possible for the relevant natural properties to exist in the absence of the appropriate phenomenally conscious state. This means that Carruthers has to defuse the various arguments purporting to show that no such logical supervenience claim could possibly be true.

Carruthers’s strategy here is interesting. He suggests that many of these arguments rest upon a mistaken conception of properties. For Chalmers, for example, a property is simply a function from possible worlds to extensions, such that every coherent concept determines one such mapping function. This is what allows Chalmers to argue that, since the existence of zombie worlds is logically possible, the property having an experience of red cannot logically supervene on any set of natural properties. If we can show that our concepts are such that, for any candidate natural property N which we can conceptualise, it is conceivable that something should be N without having an experience of red, then it seems plausible that the concept having an experience of red cannot determine property N. But, according to Carruthers, if we take properties to be thickly individuated natural entities, then this conclusion does not follow. It might well be the case that our concept having an experience of red picks out a natural property in this world such that, in every world, every individual of whom that property is true will also instantiate the concept having an experience of red – even though there may be worlds in which the concept having an experience of red picks out a different natural property. In such a situation (which, Carruthers suggests, Chalmers et al have not ruled out) there would be logical supervenience without property identity.

Carruthers’s argument here rests on denying the basic intuition underlying many of the arguments in this area, namely, that the concept having an experience of red picks out a phenomenological feel. He suggests, in contrast, that the concept picks out a conceptually individuated natural property – whatever property it is that makes it the case that we have an experience of red. It is hard to see, however, why Chalmers et al. should be convinced by this, since what they are interested in is the phenomenological feel itself, rather than the natural property which might be associated in this world with the experience of red. It is surely open to them to define a new concept, having an experience* of red that is stipulated to pick out the phenomenological feel of the experience of red, in complete independence of any natural property whatsoever. The usual thought experiments will then show that the property which this concept picks out does not supervene logically on any natural property.

Turning now to Carrruthers’s own account of consciousness, his central argument against first-order representationalist theories is that such theories cannot deal with instances of non-conscious experience. That there are non-conscious experiences is, he thinks, shown not simply by exotic neuropsychological phenomena such as blindsight and visual form agnosia, but also by common-or-garden phenomena such as absent-minded perception and non-conscious accommodation to one’s environment as well as by some of the striking dissociations between perception and action in normal subjects which have been taken as evidence for the two visual systems hypothesis. If such non-conscious phenomena have roughly the same functional role as conscious experiences then it follows that simply being poised to impact on belief formation and practical reasoning cannot be sufficient for phenomenal consciousness.

As one would expect, of course, the evidence is far from unequivocal. In all the cases he discusses there is clearly some form of perceptual registration (to use a deliberately neutral term). The question is whether, in each case, the perceptual registration is (a) correctly described as an experience at the personal-level (b) genuinely non-conscious and (c) more or less functionally equivalent to a conscious perception. It is far from clear that all three of these criteria are met by each of the examples which Carruthers adduces. It will be objected to many of the common-or-garden phenomena that they fail to satisfy both (a) and (b). It is hard to see (pace Carruthers’s section 6.1.1) how we might determine whether the absent-minded lorry-driver is non-consciously perceiving the road or simply has an unimpressive short-term memory – here it is tempting to deny (b). As far as subliminal learning is concerned there is a certain plausibility in denying (a). Many of the neuropsychological phenomena, on the other hand, seem to fall foul of (c). The crucial point here is that (c) requires that the non-conscious phenomena do two things – feed into action and feed into belief formation – and it is arguable that the phenomena to which Carruthers draws attention at best do one but not the other. It may well be the case that the perceptual registrations of blindsight and visually agnosic patients do feed into action in more or less the way that conscious perceptions do (although many researchers would deny this), but they certainly do not feed into the processes of belief formation in the right sort of way – which is why such patients describe themselves as merely guessing. Something similar holds for the visual illusions cited to support the two visual systems hypothesis. In the Titchener illusion, for example, normal subjects report that two circles which are in fact equally sized appear differently sized – even though when asked to reach towards the circles the map between finger apperture and target showed that the sizes were being correctly estimated. Again, however, (c) does not appear to be met. It is presumably the non-conscious perceptual registration of the correct size which is supposed to qualify as a counter-example to first-order representationalism – but this perceptual registration does not feed into the processes of belief formation in the right sort of way, which of course is why the Titchener illusion is an illusion.

It looks, therefore, as if the case against first-order representationalism is far from watertight. And one would be forgiven for thinking that this is just as well, since the higher-order version of representationalism which Carruthers offers in its place has some fairly counter-intuitive consequences. If, as higher-order representationalism maintains, higher-order thoughts are essential for phenomenal consciousness, then it follows that creatures incapable of higher-order thought will not be phenomenally consciousness – and, since Carruthers thinks that higher-order thought is the unique preserve of humans and perhaps some of the great apes, he ends up denying sentience to the vast majority of the animal kingdom. In fact, he sees this as an advantage of his theory, since he can find no reason for thinking that non-primates are phenomenally conscious except crude intuitions. We only think that non-primates are conscious because, realising that they have experiences, we attempt to conceptualise those experiences, which are in fact non-conscious, on the model of our own conscious experiences.

Carruthers is no doubt correct that intuitions are not to be trusted in this area. But he neglects the most significant reason for attributing sentience to non-primates. The most plausible model we possess for explaining the vast majority of animal behaviour is that provided by conditioning theory. The basic principle of conditioning theory is that certain patterns of behaviour are reinforced by being associated with primary positive reinforcers, and inhibited by being associated with primary negative reinforcers. But learning through conditioning works because primary reinforcers have qualitative aspects. It is impossible to divorce pain’s status as a negative reinforcer from its feeling the way it does. It is impossible to divorce soothing vocalizations being positive reinforcers from their sounding the way they do. The success of stimulus-reinforcement models of learning therefore provides a powerful motivation for doubting higher-order thought theories of consciousness.

In conclusion, Peter Carruthers has written a rich and rewarding book which both imposes a rigorous framework within which we can evaluate existing theories of consciousness and significantly advances the debate. The level of argumentation is consistently high and a wide range of empirical evidence is brought to bear. Phenomenal Consciousness: A Naturalistic Theory repays careful study and no one working in the philosophy of mind and/or psychology can afford to ignore it.

 

© 2002 José Luis Bermúdez