On the Matter of Memory: Neural Computation and the Mechanisms of Intentional Agency
Humans and other animals are intentional agents; they are capable of acting in ways that are caused and explained by their reasons. Reasons are widely held to be mediated by mental representations, but it is notoriously difficult to understand how the intentional content of mental representations could causally explain action. Thus there is a puzzle about how to ‘naturalize’ intentional agency. The present work is motivated by the conviction that this puzzle will be solved by elucidating the neural mechanisms that mediate the cognitive capacities that are distinctive of intentional agency.
Two main obstacles stand in the way of developing such a project, which are both manifestations of a widespread sentiment that, as Jerry Fodor once put it, “notions like computational state and representation aren’t accessible in the language of neuroscience”. First, C. Randy Gallistel has argued extensively that the mechanisms posited by neuroscientists cannot function as representations in an engineering sense, since they allegedly cannot be manipulated by the computational operations required to generate structurally complex representations. Second, William Ramsey has argued that neural mechanisms in fact do not function as representations in an explanatory sense, since the explanatory role they play is more like that of a causal relay rather than an internal ‘stand-in’ for external entities.
I argue that the criticisms developed by Gallistel and Ramsey rest on a misapplication of relevant theoretical notions from computer science, and an impoverished appreciation of the rich variety of mechanisms posited in contemporary neuroscience. My central argument, though, is that the conception of representation presupposed by those researchers, according to which representations have the same abstract structure as the external systems they represent, encompasses states in all sorts of systems, including mindless systems such as plants. Thus these ‘structural representations’ are not distinctively cognitive representations, and although they play an important explanatory role in neuroscience and other disciplines, they cannot on their own elucidate the nature of intentional agency. I conclude by sketching some ways in which neuroscientists are beginning to elucidate the representational mechanisms that do mediate distinctively cognitive capacities. Thus, by removing the obstacles put in place by Gallistel and Ramsey, this work clears a path toward the ‘naturalization’ of agency in terms of neural mechanisms.