Joe Lau
Department of Philosophy
The University of Hong Kong
Hong Kong
http://www.hku.hk/philodep/joelau
31 March 1999
Michael Tye’s book is a powerful defense of the controversial theory that the phenomenal properties of our conscious mental states are representational in character. The theory is introduced and defended through discussing ten philosophical problems about consciousness. The book is clearly written and arguments are illustrated with interesting thought-experiments and empirical findings. It is one of those delightful occasions where a book is of interest both to professional philosophers and students.
In this review I shall focus on two main theses defended in the book. The first is that all experiences, or conscious mental states, have representational content. The second is that the phenomenal properties of all experiences are determined and explained by their content. Clearly, the latter entails the first but not vice versa. I shall call the first thesis "weak representationalism" and the second one "strong representationalism".
With regard to weak representationalism, I think nowadays most philosophers would agree that perceptual experiences have representational content. But many would deny that this is true of other experiences. Typical counterexamples include bodily sensations such as pain and hunger pangs, and certain conscious emotions and moods. Tye examines these and other examples one by one, and argue in each case that they all have representational content.
The core part of Tye’s response is his causal covariational theory of content for perceptual and sensory representations :
S represents that P = If optimal conditions obtain, S is tokened if and only if P and because P.
Tye then applies this theory to show that even bodily sensations have representational content. So according to him, pain is supposed to correlate with, and hence represent, tissue damage or disturbance. Likewise hunger pangs represent contractions of stomach walls. Having argued for the intentionality of bodily sensations, Tye moves on to discuss emotions and moods. Many conscious emotions obviously have intentional content, like feeling happy that some special event has transpired. In the case of more problematic examples, Tye adapts William James’s argument that felt emotions and moods derive all their phenomenal properties from associated bodily sensations. If this is agreed, then we might say that felt emotions and moods can also have representational content derived from these sensations.
Whether these arguments are convincing or not depends in part on the plausibility of Tye’s causal theory of content. But the theory is inadequate as it stands. First, Tye says that optimal conditions are supposed to be conditions where "there are no distorting factors, no anomalies or abnormalities". But it is not really clear what this means. For example, a visual representation representing an object in front of me is normally tokened only if I open my eyes. So does this mean it is abnormal to blink, or does the visual representation really has the complex content that my eyes are open and there is an object in front of me? A second problem is that it is not clear how the causal theory applies to mental images. There is something it is like to have a mental image, say of a blue square, but such an image is not normally caused by blue squares. Besides, it need not have any single standard cause, so it is not clear how Tye’s theory might apply. Finally, his theory also suffers from the same familiar problems faced by other simple causal theories of content. Consider color experiences, which typically are caused by colored objects, say. But presumably, even under optimal conditions, colored objects can cause color experiences only via light activating the cones in the retina. So why do color experiences represent colored objects instead of stimulation of the cones by photons? These problems are perhaps not insurmountable, but given that they are typical objections to causal theories of content, this is one place in the book which could have been given more detailed discussion.
In any case, I am inclined to agree with Tye that all conscious mental states do have representational content. In the case of bodily sensations, as Tye points out, assigning representational content to them can help explain the phenomenon of phantom limbs, since having sensations in amputated limbs is then a matter of having sensations that represent states of body parts that actually do not exist. Furthermore, weak representationalism fits with a plausible informational account of the biological function of phenomenal consciousness : when a mental state is conscious, the information it contains is made directly available for reasoning and rational control of action.
Still, Tye intends to go further. The main objective of his book is to defend what I have called strong representationalism. It is not just that conscious mental states have content. Tye thinks that the phenomenal properties of experiences are in fact identical to their representational content. In other words, what it is like to have an experience is to be explained by what the experience represents. Tye’s theory is closely related to that of Dretske (1995) and Lycan (1996). In the rest of this book review I hope to point out some of their common strengths and weaknesses.
On the face of it, what is plausible about such theories is that they explain why what it is like to have an experience varies systematically with its representational content. What is it like to hear a sound, for example, varies with the direction the sound is represented as coming from. What it is like to have a visual experience also varies with what the experience represents the external world as being. A visual experience representing a red object to the left has a different phenomenal quality compared with a visual experience representing a blue object to the right. These observations are only to be expected if the phenomenal properties of an experience are to be identified with its representational content.
There is however a serious problem with such a view. Since phenomenal properties are representational, and representational content depends on causal covariation with states of the body or the environment, Tye’s theory ( and also that of Lycan and Dretske ) entails the externalist conclusion that the phenomenal properties of our experiences are not determined by the internal states of the brain. In other words, if there is a creature with a brain that is in exactly the same physical state as I am, it can still possess different experiences, or even no experience at all, provided it is embedded in a suitably different environment.
With cognitive states such as belief, there is the strong intuition that in a lot of cases their content depends on having the right historical causal connections to the social or physical environment. Thus we can readily accept that a döppelganger with the same brain state might nonetheless have beliefs with different content. However, there is no comparable intuition that my döppelganger could have different or no experiences. If anything the intuition runs in the opposite direction, so the burden of proof is surely upon the externalist to justify their claim.
To his credit Tye acknowledges this problem and tries to deal with it. Take Davidson’s famous example of the swampman, a complete physical duplicate of Davidson produced by a random cosmic accident. While it seems plausible that the swampman lacks many of Davidson’s beliefs, most people would grant that there is something it is like to be the swampman. Unlike Dretske, Tye agrees that it is implausible to deny phenomenal consciousness to swampman, but argues that this can be accommodated within his theory. His response is that swampman is a living creature which can function well or badly depending on the environment. Optimal conditions for this creature would be conditions under which it would flourish, and we can discover the representational content of its internal states by finding out which features are tracked by those states under optimal conditions so understood. There is no difficulty then in admitting that swampman has phenomenal consciousness.
But what if swampman is in an environment which is rapidly changing, or perhaps an unstable one for which there is no clear answer to the question as to whether it would flourish or not? Or what if swampman has been kidnapped by an evil scientist and is about to be immersed in a virtual reality world? Which environment is the one we should appeal to in determining the representational content of its internal states? If we adopt Tye’s theory, we need to answer such questions before we know whether swampman has experiences or what those experiences are like. But an internalist position is I think simpler and remains more intuitive in that we only need to consider the internal states : swampman has exactly the same experiences if the internal states remain the same across all these situations.
Of course, the fact that internalism is more intuitive does not mean it is true. But internalism with regard to phenomenal consciousness is also an implicit assumption underlying a lot of recent empirical research on consciousness. Many cognitive scientists believe that if consciousness can be explained, it has to do with the neurophysiological or computational properties internal to the brain. [footnote 1 : For neurophysiological approaches, see Crick and Koch (1998). For cognitive and computational approaches, see Johnson-Laird (1983), Baars (1997), Jackendoff (1987).] But all these theories can be dismissed if Tye is right. The question therefore is what special reasons can Tye offer in support of externalism with regard to phenomenal consciousness.
One such argument is that if we open up the brain and look inside, phenomenal properties are nowhere to be found. A person might visually experience a blue square, but there is nothing square or blue in that person’s brain. This observation is of course quite correct, but all it shows is that representational content need not be in the head. We can agree that a visual representation can represent properties of external or non-existent objects. But this fails to show that what it is like for the subject to have that visual experience is not something that is wholly determined by what goes on inside the brain.
It might be useful at this point to introduce an alternative proposal, one which we might call a "vehicle" theory of phenomenal consciousness. It endorses weak representationalism in that it also takes normal conscious mental states as states with representational content. But it denies that the content determines what it is like to be in a conscious state. Rather, phenomenology is determined by the nature of those mental representations that have the content. In other words, the vehicle theory says that there is a special level of representations in the brain which generates conscious experience. Let us call them "phenomenal vehicles". They normally represent certain bodily or environmental states which is why all experiences have content. But it is the nature of the phenomenal vehicles that determine what our experiences are like, not their content.
Like Tye’s theory, the vehicle theory can also accommodate the fact that phenomenology varies systematically with the representational content of experiences. This is simply because differences in content corresponds to differences in phenomenal vehicles, and differences in vehicles corresponds to differences in phenomenal properties. But notice that the vehicle theory is still an internalist theory. Thus it avoids the counter-intuitive consequences of strong representationalism while retaining its advantages, and at the same time it is more consonant with empirical approaches to consciousness.
Returning to Tye’s argument, the vehicle theory can accept that when we have a visual experience of a blue square, we have certain phenomenal vehicles that represent blueness and squareness. The latter properties are of course nowhere to be found in the brain, but the vehicles themselves are still internal to the brain, and these vehicles determine what it is like for the subject to have the experience. It is true that when we observe these vehicles in the subject’s brain, we ourselves do not end up seeing blueness and squareness, but that is simply because observing these vehicles normally do not cause the tokening of the same vehicles in our own brains, since they are normally caused by blueness and squareness instead. Tye’s observation therefore lends no support to his theory over the vehicle theory. What is surprising is that Tye does not discuss the vehicle theory anywhere in his book. This is in my opinion the weakest part of his defense of strong externalism. I shall now look at some of his other arguments and show that they also fail to rule out the vehicle theory.
One line of argument has to do with intensionality in our description of our experience. Tye notes that in describing what our experiences are like we might talk about what it is like to have them, or how things look. But such discourse might be intensional : the tiles might look square, but they do not look the shape of the picture, even though the shape of the picture is square. From this and other examples Tye again draws the conclusion that phenomenal properties are identical to representational content. This is of course one possibility, but it is also possible that when we describe our experiences, we use representational content only as the means to pick out their phenomenal properties. This does not show that the contents themselves are essential to or constitutive of phenomenal properties.
I think the most interesting argument that Tye offers for strong representationalism is the argument from transparency. As suggested by G. E. Moore and others, in introspecting the character of one’s experiences our attention seems to focus inevitably on aspects of what we experience, such as the colors and shapes of external objects or the conditions of our body. Introspection does not seem to reveal any properties of experiences other than what they represent. Tye claims that this is because the representational contents of experiences are the same as their phenomenal properties.
But such an argument might be resisted in various ways. First, the assumption that experiences are transparent in the sense described is controversial. But even if it is true, it need not undermine the vehicle theory. For it is possible that phenomenal vehicles determine what it is like to be conscious of those external properties accessible through introspection. When I look at a red apple I am conscious of its red color, and the representational content of my experience is not internally determined. But how that red color appears to me can still be completely determined by the internal representational vehicles that the apple causes to instantiate in me. In other words, an internalist can reject the assumption that those intrinsic properties of the brain which give rise to conscious experience must themselves be available to introspection.
In conclusion then, none of Tye’s arguments for strong externalism is successful. In my opinion the vehicle theory remains a more attractive internalist approach to phenomenal consciousness, and it is unfortunate that Tye fails to consider this competing proposal in his book. Having said that, whether you agree with Tye or not, Ten Problems of Consciousness is a clearly written and enjoyable book, one which should be read by anyone interested in the state of the art philosophical discussion about the nature of phenomenal consciousness.
References
Baars (1997) In the Theater of Consciousness : the Workspace of the
Mind, New York : Oxford University Press.
Crick and Koch (1998) "Consciousness and Neuroscience" Cerebral Cortex 8,
pp.97-107.
Dretske, F. (1995) Naturalizing the Mind, Cambridge, Mass: The MIT Press, Bradford
Books.
Johnson-Laird (1983) Mental Models, Cambridge, UK: Cambridge University Press.
Jackendoff, R. (1987) Consciousness and the Computational Mind, Cambridge, Mass :
The MIT Press, Bradford Books.
Lycan, W. (1996) Consciousness and Experience, Cambridge, Mass: The MIT Press,
Bradford Books.