Margaret A. Boden
Anti-reductionism is not antiscientific if it offers positive suggestions about empirically-based concepts to explain phenomena not explicable in currently accepted terms. Artificial intelligence (A.I.) is anti-reductionist in a number of respects. Computational concepts can provide explanatory power over and above that of the more basic theories in the life sciences, while being entirely compatible with them. Human and animal intelligence, and psychological phenomena in general, can usefully be thought of in these terms. Behaviourist psychology and neurophsyiology cannot express the phenomena concerned, because their vocabulary has no room for the concepts of representation or intentionality. But A.I. is concerned with symbol-manipulating systems, and these concepts are theoretically central to it. A.I. can be useful in physiology, by clarifying what are the computational tasks which the nervous system is performing, and it may even illuminate some aspects of evolutionary and morphogenetic biology. A computational approach to life and mind is entirely compatible with notions of human freedom. By its emphasis on the subject's representation of the world, it counters the mechanization of the world-picture brought about by the natural sciences.
This paper is not available online