Despite ever-growing interest, and some empirical progress, in understanding the neural basis of consciousness in the last decade, its evolutionary function has been relatively neglected. I will briefly review existing hypotheses, and advance my own action-based hypothesis concerning the evolutionary and computational functions of consciousness that I term “consciousness as credit allocation” (CCA). This holds that the function of consciousness, from a computational viewpoint, is to allocate credit and blame to the multiple parallel hypotheses that typically underlie (and precede) action. Coherent updating in a parallel system demands a system for credit and blame allocation, so that each of the semi-independent processing units (small assemblages of neurons) be informed about the final "decision" of the system as a whole. The complementary evolutionary function follows from the Helmholtzian view of the brain as an unconscious inference machine, but as applied to action: namely that action is typically preceded by multiple unconsciously simulated actions, competing in parallel for final execution. In order to learn from the outcome, a “tagging” system is required that broadcasts the final action chosen and executed to all relevant brain regions, allowing appropriate allocation of credit (for positive outcomes) or blame (for mistakes). This model implies the existence of first-order consciousness in most vertebrates and in any brain capable of simultaneously entertaining competing hypotheses (e.g. including cephalopods).
Fitch (in press) Why evolve consciousness? Neural credit and blame allocation as a core function of consciousness. Behavioral and Brain Sciences.
Fitch (2008). Nano-intentionality: A defense of intrinsic intentionality. Biology and Philosophy, 23(2), 157-177