Abstract & Overview
This thesis focuses on mechanistic computationalism or a mechanistic variant of the computational theory of cognition in philosophy of cognitive science. In the four research articles comprising this dissertation, I address some of the questions related to mechanistic computationalism discussed in recent years and raise some new ones. In the first two, co-written articles, we analyze arguments against computationalism that concern the role of representations in computationalism. Employing Gualtiero Piccinini’s mechanistic account of physical computation and the computational framework of reinforcement learning, we argue that the analyzed arguments miss their target. In the remaining two, solo-written articles, I analyze problems that concern the relationship between computational and implementational properties in Piccinini’s mechanistic account of physical computation. Focusing on the role of abstraction in the mechanistic account, I introduce a theoretical distinction between vertical and horizontal abstraction as a potential solution to the analyzed problems. However, a deeper look into the provided suggestion raises some new questions and challenges for the mechanistic account concerning the nature of computational properties.