Who's Afraid of Multiple Realizability?: Functionalism, Reductionism, and Connectionism
I argue that this view is radically mistaken. If valid, the argument proves too much. A functional state of a system is one intrinsically characterized by its causal or computational role in mediating between inputs (e.g., sensory stimula tion) and outputs (e.g., behavior), as well as by the internal relations among the functional states themselves. Multiple realizability is a property of any func tionally characterized state, so the argument would establish the irreducibility of any theory that appeals essentially to such states, such as transmission genetics, which adverts to the states and properties of genes. Gene is a functional term. To say x is a gene is to characterize it in terms of its functional role in transmitting hereditary information. Genes are therefore multiply realizable. But if the multi ple realizability argument for autonomy denies the reducibility of genetics, this is a reductio: transmission genetics is in part reduced to molecular genetics.