The Biological Foundations of Cognitive Science
Mark H. Bickhard
One of the founding assumptions that Cognitive Science extracted from the computer model was that mental phenomena can be modeled relatively independently of the brain. Mental phenomena are, in this view, assumed to be akin to the execution of a computer program, and program properties are independent of the specifics of the computer
that is running that program. Similarly, mental properties are assumed to be relatively
independent of the brain in which they are realized. Furthermore, the body is “just” a robot
b o d y h o u s i n g t h a t c o m p u t e r , a n d , p e r h a p s , i n c o r p o r a t i n g t h e s e n s o r y t r a n s d u c e r s t h a t f e e d
inputs into the computer. Recognition that the program/computer distinction might provide a model for the mind/brain relationship was, in fact, one of the early excitements of the field.
The re-emergence of connectionist models in the 1980s was accompanied by claims of greater faithfulness to actual brain organization and process compared to symbol manipulation computationalist models. But the architectural and vector process differences between computationalism and connectionism do not alter the common underlying assumption that, while the brain may realize the relevant processes to constitute mind, the
critical properties of those processes may be still be modeled independently of neural level details in the brain.
More recently, dynamic systems and autonomous agent approaches have argued for the importance not only of specifics about the brain, but of details concerning the body as well. Cognition, in this view, is inherently embodied, and cannot be understood in the disembodied forms provided by either computationalism or connectionism (Beer, 1995;
These dynamic and agent approaches, however, have also reintroduced basic questions about what the aim of modeling in Cognitive Science should be. In particular,
many have argued that mental phenomena such as representation are simply not relevant to either the design or the modeling of embodied autonomous agents. Representation is a vestigial notion from computationalism that should be rejected (Brooks, 1991). Dynamics is everything, so it is held, and getting the dynamics right, whether in design or in modeling, is the only relevant criterion.