BEHAVIORAL AND BRAIN SCIENCES (1999) 22, 127–196 Printed in the United States of America A connectionist theory of phenomenal experience Gerard O’Brien Jonathan Opie Department of Philosophy, The University of Adelaide, Adelaide, South Australia 5005, Australia. [email protected] chomsky.arts.adelaide.edu.au/Philosophy/gobrien.htm [email protected] chomsky.arts.adelaide.edu.au/Philosophy/jopie.htm Abstract: When cognitive scientists apply computational theory to the problem of phenomenal consciousness, as many have been do- ing recently, there are two fundamentally distinct approaches available. Consciousness is to be explained either in terms of the nature of the representational vehicles the brain deploys or in terms of the computational processes defined over these vehicles. We call versions of these two approaches vehicle and process theories of consciousness, respectively. However, although there may be space for vehicle theories of consciousness in cognitive science, they are relatively rare. This is because of the influence exerted, on the one hand, by a large body of research that purports to show that the explicit representation of information in the brain and conscious experience are dissociable, and on the other, by the classical computational theory of mind – the theory that takes human cognition to be a species of symbol manipulation. Two recent developments in cognitive science combine to suggest that a reappraisal of this situation is in order. First, a number of theorists have recently been highly critical of the experimental methodologies used in the dissociation studies – so critical, in fact, that it is no longer reasonable to assume that the dissociability of conscious experience and explicit representation has been adequately demonstrated. Second, classicism, as a theory of human cognition, is no longer as dominant in cognitive science as it once was. It now has a lively competitor in the form of connectionism; and connectionism, unlike classicism, does have the computa- tional resources to support a robust vehicle theory of consciousness. In this target article we develop and defend this connectionist ve- hicle theory of consciousness. It takes the form of the following simple empirical hypothesis: phenomenal experience consists of the ex- plicit representation of information in neurally realized parallel distributed processing (PDP) networks. This hypothesis leads us to reassess some common wisdom about consciousness, but, we argue, in fruitful and ultimately plausible ways. Keywords: classicism; computation; connectionism; consciousness; dissociation; mental representation; phenomenal experience 1. Computational theories of consciousness: Vehicle versus process Gerard O’Brien completed his There is something it is like to be you. Right now, for exam- D.Phil. degree at the University of ple, there is something it is like for you to see the shapes, tex- Oxford before taking a position in the tures, and colors of these words, to hear distant sounds fil- Department of Philosophy at the tering into the room where you sit, to feel the chair pressing University of Adelaide, Australia, against your body, and to understand what these sentences where currently he is a senior lecturer mean. In other words, to say that there is something it is like and the director of the Graduate Program in Cognitive Science. He to be you is to say that you are phenomenally conscious: a lo- has published widely in the fields of philosophy and cus of phenomenal experiences. You are not alone in this re- mind and the foundations of cognitive science, with a spect, of course, because the vast majority of human beings particular focus on exploring the connectionist have such experiences. Furthermore, there is probably alternative to the classical computational conception of something it is like to be a dog, and perhaps even fish have human cognition. phenomenal experiences, however minimal and fleeting they may be. On the other hand, there is surely absolutely Jonathan Opie recently com- menced an Australian postdoctoral nothing it is like to be a cappuccino, or a planet, or even an 1 research fellowship in the Depart- oak tree. These, at least, are the standard intuitions. ment of Philosophy at the University It is clearly incumbent on any complete theory of the of Adelaide, after completing his mind to explain phenomenal experience. And given that Ph.D. there. He has published arti- our best theory of the mind will likely issue from cognitive cles on computational approaches to science, it seems incumbent on this discipline, in particu- consciousness in a number of major lar, to provide such an explanation. What is special about journals. cognitive science is its commitment to the computational © 1999 Cambridge University Press 0140-525X/99 $12.50 127 O’Brien & Opie: Connectionism and phenomenal experience theory of mind: the theory that treats human cognitive pro- crete object is typically said to be represented explicitly. In- cesses as disciplined operations defined over neurally real- formation that is stored in a dispositional fashion, or em- ized representations.2 From this perspective, the brain bodied in a device’s primitive computational operations, on is essentially a very sophisticated information-processing the other hand, is said to be represented nonexplicitly.3 It device; or better, given what we know about brain archi- is reasonable to conjecture that the brain uses these differ- tecture, an elaborate network of semi-independent ent styles of representation. Hence the obvious emendation information-processing devices. to the original suggestion is that consciousness is identical The computational vision of mind and cognition is by now to the explicit coding of information in the brain, rather very familiar. The question we want to consider here is how than the representation of information simpliciter. we might exploit the resources of this paradigm to explain Let us call any theory that takes this conjecture seriously the facts of phenomenal consciousness. Given that compu- a vehicle theory of consciousness. Such a theory holds that tation is information processing, and given that information our phenomenal experience is identical to the vehicles of must be represented to be processed, an obvious first sug- explicit representation in the brain. An examination of the gestion is that phenomenal consciousness is somehow inti- literature reveals, however, that vehicle theories of con- mately connected with the brain’s representation of infor- sciousness are exceedingly rare. Far more popular in cog- mation. The intuition here is that phenomenal experience nitive science are theories that take phenomenal con- typically involves consciousness “of something,” and in being sciousness to emerge from the computational activities in conscious of something we are privy to information, either which these representational vehicles engage.4 These typi- about our bodies or the environment. Thus, perhaps phe- cally take the form of executive models of consciousness, nomenal experience is the mechanism whereby the brain according to which our conscious experience is the result of represents information processed in the course of cognition. a superordinate computational process or system that priv- However, to identify consciousness with the mental rep- ileges certain mental representations over others. Baars’s resentation of information is to assert two things: that all “Global Workspace” model of consciousness (1988) is a phenomenal experience is representational, and that all the representative example. Baars’s approach begins with the information encoded in the brain is phenomenally experi- premise that the brain contains a multitude of distributed, enced. Theorists have difficulties with both aspects of this unconscious processors, all operating in parallel, each identification. On the one hand, it is commonplace for highly specialized, and all competing for access to a global philosophers to argue that certain kinds of phenomenal ex- workspace – a kind of central information exchange for the perience are not representational (Searle [1983, pp. 1–2], interaction, coordination, and control of the specialists. e.g., cites pains and undirected emotional experiences in this Such coordination and control is partly a result of restric- regard); and on the other, it is sheer orthodoxy in cognitive tions on access to the global workspace. At any one time, science to hold that our brains represent far more informa- only a limited number of specialists can broadcast global tion than we are capable of experiencing at any one moment messages (via the workspace) because different messages in time. So sensations, undirected emotions, and memories may often be contradictory. Those contents are conscious immediately pose problems for any account that baldly iden- whose representational vehicles gain access to the global tifies phenomenal consciousness with mental representation. workspace (perhaps as a result of a number of specialists The advocate of a such an account of consciousness is not forming a coalition and ousting their rivals) and are subse- completely without resources here, however. With regard quently broadcast throughout the brain (pp. 73–118). The to the first difficulty, for example, there are some philoso- nature of the vehicles here is secondary; what counts, as far phers who, contrary to the traditional line, defend the po- as consciousness is concerned, is access to the global work- sition that all