meadowreader 2007-05-24
10 of 12 people found the following review helpful:
Daniel Dennett once characterized Darwin's theory of natural selection as the best single idea anyone ever had. I generally agree with that, so I am naturally well-disposed toward selectionist accounts, of which Edelman's Neural Darwinism is an example. I also have run out of patience with the clever word games that unfortunately constitute far too much of epistemology as it exists in academic philosophy, and that makes me a receptive audience for this kind of selection-based approach. I also agree with Edelman's rejection of computer-based models of human cognition and of Chomksky's mythical language organ. So, yes, I liked the book, found many insights in it, and I recommend it as a stimulating read. All that said, and given the extremely informative review provided below by L. Guzman, I will focus on what I found less than satisfactory.
When it comes to psychology, Edelman's view of the field seems to be bounded by Piaget and Freud. He gives the occasional, semi-perfunctory nod to the environment, but never in serious detail considers the importance for both brain and behavior of the history of interaction between organism and environment. Nor does he show any knowledge of the huge literature describing detailed analyses of environmental effects on behavior, analyses that specifically emphasize the selective effects of environmental consequences. Edelman's account reminds me of how the genome-phenome distinction is sometimes treated in relation to selection, as if genes themselves are directly selected for or against, and then go forth and do things in the world. In this case, it's always the brain doing this or that, with little or no acknowledgment that it is behavior that makes contact with the environment and is subject to selection, with resultant effects on the brain. We know from sensory deprivation experiments that, absent a reasonably normal environment, brain activity quickly drifts into disorder and incoherence. One suspects that the well-known behavior-specific effects of certain brain injuries, which Edelman describes in some detail, have been overgeneralized, resulting in an overly brain-centric view of behavior. What's needed here is an explicit input-output model, where the inputs are the initial state of the organism and the functional characteristics of the environment in which it operates; the fact that computer-based models have used input-output terminology should be considered of no relevance. At some points in Edelman's presentation, I found it difficult to decide just what constituted the output side -- and whether we were ever finally emerging from the neurological realm to the world outside.
A key notion for Edelman is the idea that the organism's inherited neurological structure incorporates biases that will determine something about the way certain stimuli are responded to. This "value system" is considered to be a product of evolutionary history, but it's not entirely clear exactly what the particular functions of this value system are supposed to be. Is it a kind of pre-wiring that makes certain stimuli more salient? For example, the visual appearance and smells associated with a particular species' preferred food might as a matter of inherited tendency trigger consumatory behavior, or at least make it more probable. Besides in-born sensitivity to releasor stimuli, would the value system include inherited behavior itself, ranging from reflexes to simple fixed-action patterns to very complex response sequences? One thinks of the elaborate behavior observed in courtship displays, nest-building, or nurturing the young. Is the notion of value system meant to stand as a neurological-level explanation for the ordinary behavioral effects of reinforcing and punishing consequences? Or is it meant specifically to account for some built-in extreme susceptibility of certain behavior to its consequences, thus amplifying normal reinforcement-punishment effects? Edelman includes the inherited value system as a hypothetical entity or process in his theoretical system, but it's not clear from this book exactly what its functions are, or how they interact with behavior or with the environmental events that precede, accompany, or follow behavior.
In Chapter 12, on Brain-Based Devices, we find, as is typical, that when it comes to actually making something happen the environment suddenly becomes important. Edelman makes much of this extremely interesting work using robotic devices with simulated brains that allow them to learn through trial and error, rather than being driven by pre-programed instructions. It is noteworthy, and completely predictable, that the descriptions of these experiments turn out to be descriptions of (1) the initial structure and behavioral capabilities of the simulated organism, and (2) operations involving the provision of specific environmental stimuli and environmental feedback. Results indicate that interaction with the environment produces changes in the device's behavior and in the organization of its simulated brain. If results didn't turn out that way, the researchers presumably would tinker with (1) and (2) above until they did. One long-term potential here seems to be the rediscovery of behaviorism, but with much better illumination of its neurological underpinnings. That would be an outcome devoutly to be wished, but getting there will require a broader, less brain-centric view than Edelman's alone.