|
BEING THERE: PUTTING BRAIN, BODY AND WORLD TOGETHER AGAIN, by ANDY CLARK. Cambridge MA: MIT Press, 1997, 280pp. A great deal of philosophy of mind in the modern era has been driven by an intense aversion to Cartesian dualism. In the 1950s, materialists claimed to have succeeded once and for all in exorcising the Cartesian ghost by identifying the mind with the brain. In subsequent decades cognitive science put scientific meat on this metaphysical skeleton by explicating mental processes as digital computation implemented in the brain's hardware. The fundamental message of Andy Clark's latest book is that this orthodox materialist-computationalist framework seriously misrepresents the nature of mind. Ironically, this is because it has failed to make a sufficiently clean break with Cartesianism. The old mind-body opposition, of ontological substances, is recapitulated in a new opposition between the mental on one hand and the body and world on the other. The mind is the inner realm of pure reason, while the body and world are just all that stuff on the far side of occasional symbolic inputs and outputs. This implicit dualism in nature is reflected in the methodological supposition that the cognition can be studied quite independently of any detailed consideration of the brain, the body, and the physical and social environment. In contrast, Clark proceeds from the premise that the brain/mind is fundamentally a controller of embodied action, and must be understood as such. The brain's primary task is to assist the organism survive in its particular niche. Evolution stumbled across neural machinery just adequate for the purpose. This machinery works by fast parallel pattern completion, of roughly the kind found in standard layered connectionist models. These "associative engines" manage remarkable feats of coordination by opportunistically exploiting sources of order and coherence already contained in the body and environment. The system is decentralized: there is no central executive or planner directing operations, nor is there any general-purpose, action-independent inner code or world model. Control is an emergent property of a self-organizing, distributed system embracing brain, body and aspects of the world. How do these simple associative engines manage to achieve anything more interesting than motor control? Clark answers that advanced cognition results from the interaction of associative engines with highly structured environments. For example, doing long-division by hand involves repeatedly responding to marks on paper; the computation is thus really a collective achievement of brain and body with pencil and paper. This kind of scaffolding can involve complex social institutions. The design and construction of jumbo-jets involves thousands of associative engines in myriad simple interactions with each other and with special artifacts. Of these artifacts, the most important is language, whose essence is to be a "computational transformer that allows pattern-completing brains to tackle otherwise intractable computational problems" (194). Of course, these external structures are usually themselves products of associative engines; an important task for cognitive science is to understand the bootstrapping process through which organisms create and maintain the scaffolding which makes advanced cognition possible. One important consequence is that mind "leaks" out of its supposed home inside the skull. It is not just that certain cognitive achievements, such as navigating a ship, are collective phenomena involving various people and artifacts interacting within special social structures. Rather, the point is that our own cognitive performances-and to that extent, our minds-are distributed over shifting assemblies which include not only our bodies but also aspects of our physical and social contexts. The boundary between mind and world is both vague and plastic. Running counterpoint to Clark's unfolding account of the nature of mind and cognition is a concern with the methodology of cognitive science. What kind of investigatory approaches and conceptual tools are appropriate to "active, embodied cognition"? Clark argues for an "explanatory liberalism" in which any framework or strategy is embraced as long as it provides some kind of genuine grip on the subject matter. He quarrels only with extremists who insist that insight into cognition will be derived exclusively from their favored approach. In particular, he argues against what he calls the thesis of Radical Embodied Cognition, according to which the notions of representation, computation and decomposition-so central to traditional accounts-are to be discarded in favor of pure dynamical systems stories. To illustrate and support this vision, Clark takes us on an eclectic, whirlwind tour of contemporary cognitive science and related ideas. We visit can-collecting robots, simulated cockroaches, infant walking, computer games, ship navigation, termite mounds, steam engines, monkey finger movements, slime moulds and mangrove swamps, as well as numerous more standard destinations. In every case Clark extracts a clear and useful lesson for understanding human cognition. With this book, Clark consolidates his position as a (perhaps the) leading exponent of contemporary anti-Cartesian cognitive science. The book synthesizes a remarkably broad range of material to arrive at a comprehensive, coherent and enticing vision of the nature of cognition and the ways it can be studied. Chapters bubble with provocative insights, some culled from the scientific or philosophical literature, but many others original with the author. The book is written in the engaging and entertaining style for which Clark is increasingly renowned. Most importantly, the book is commendable for its high quotient of truth. Few books as adventurous as this one are so consistently correct in their basic outlook. For reasons such as these the book deserves to be widely read-by philosophers wanting to be brought up to date with cognitive science; by scientists hoping to grasp more of the philosphical context in which they work; and by any intelligent person seeking better to understand how minds really work. Nevertheless, placing the book under an analytical microscope does reveal some flaws. To begin with, it is surprisingly hard to pin down the major claims with any precision. Like an impressionist painting, the book is best viewed from a distance. One reason is that, more often than not, Clark qualifies his claims with what informal logicians sometimes call 'weaselers'. For example: "Much of what we commonly identify as our mental capacities may likewise, I suspect, turn out to be properties of the wider, environmentally extended systems of which human brains are just one (important) part." (214; my italics). Another problem is that critical concepts remain vague and underdeveloped. For example, one theme of the book is that radical dynamicists are misguided in claiming that natural cognition does not involve computation. Clark points out that the concept of computation is unclear and so a "watertight case" is hard to make. He provisionally defines computation as "mechanistically governed transition between representations." But this "gloss" is too loose. Randomly selecting books from a sale bin would count as computation in this sense. With an overly broad notion of computation, Clark can sustain his position but only by robbing it of interest. Third, the book's central arguments are diffuse and weak. Its modus operandi is much more suggestion, repetition and evocation than it is logical implication. For example in Chapter 10 Clark defends his interesting position on the role of language mainly by outlining and illustrating the idea, rather than compelling our assent with strong and explicitly formulated arguments. (Note that Clark has in fact taken up much of this argumentative burden in subsequent journal articles.) Content-wise, the major weakness of the book is that it is founded on an 80's-era connectionist conception of neural processing as a kind of pattern-completing. The idea that brains are basically associative engines is fast losing plausibility as we develop more sophisticated understanding of the dynamics of cognitive processes at all levels. Clark acknowledges that the vision he presents is but "one more step" in a long intellectual journey. The next genuine advance will be one that consigns associative engines to the same theoretical scrapheap as telephone switchboards and digital computers. On balance, however, the book's virtues strongly outweigh its problems. Allan Newell used to say "Don't bite my finger-look where it's pointing!" Being There is pointing in the right direction, and does so in a delightful way. Tim van Gelder University of Melbourne |
|
This page, its contents and style, are the responsibility of the author and do
not represent the views, policies or opinions of The University of Melbourne.
|