THE CASE AGAINST COGNITIVE REDUCTIONISM
So far in this book, I have argued two main points. The first was that all formal models, whether mathematical, computational, quasi-neural, or verbal, are neutral with regard to the inner workings of the brain--mind complex. By neutral, I mean that although both the mental and neural mechanisms can be functionally described by models, they cannot be reductively explained by them any more than can any other physical system. I argued that the reasons for this are fundamental and are not dependent on the current state or future possibilities of our technology.
The second main point was that, for many reasons, relations drawn between psychophysical data and neural mechanisms, although often quite imaginative and sometimes based on functional analogies, are almost always unconvincing and fanciful, if not downright wrong. The main reason for this conclusion was that psychophysical methods were prototypical input--output comparisons for which the "black box" constraint had to hold. Therefore, the top-down (from psychophysical findings to neurophysiological events) approach could never achieve its goals.
I also stressed the point that neurophysiological methods, even when they were successful in opening the "black box," could never, for equally compelling reasons, unravel the tangled web of interactions among the enormous number of neurons that make up our brains as well as our thoughts. The extreme levels of complexity that characterize realistic neural nets prohibit the bottom-up (from neurophysiological findings to mental events) approach from explaining psychological phenomena. The complexity of the brain, in the formal mathematical sense of the word, is far greater than most neuroreductionists seem willing to accept.