On shared structure of dynamics, computations and mechanisms

Theoretical neuroscience deals with dynamical and computational models of cognition. Such models explain their target phenomena by providing dynamical and computational explanations. In contrast, proponents of mechanistic explanations have proclaimed that all explanations in the neurosciences appeal to mechanisms: since both dynamical and computational models refer to abstract formal properties of physico-neural systems, they are in and of themselves not explanatory, while only the realizing mechanisms are. I shall argue that this conclusion is overhasty. Both dynamical and computational models pick out relevant spatiotemporal-cum-causal relations of the underlying mechanisms. This shared structure can be Nagel-identified with the physical organization structure of the mechanisms. The structural grounding of dynamical and computational explanations accounts for their powerful generalizability and the multiple realizability of their target systems. I shall sketch a general and integrative account of “vertical” and “horizontal” types of scientific explanations illuminating the different perspectives as well as “intersections” of mechanistic explanations versus dynamical and computational explanations.

Preceded by a pre-talk for graduate students, 1:30PM Benjamin Library.

3:30-5:30PM Thursday 1 Nov 2018

Sir Roland Wilson Building 1.02, ANU

If you would like to come to dinner following the talk, please let me know.