Physical Computation and Cognitive Science
There are currently considerable confusion and disarray about just how we should view computationalism, connectionism and dynamicism as explanatory frameworks in cognitive science. A key source of this ongoing conflict among the central paradigms in cognitive science is an equivocation on the notion of computation simpliciter. 'Computation' is construed differently by computationalism, connectionism, dynamicism and computational neuroscience. I claim that these central paradigms, properly understood, can contribute to an integrative cognitive science. Yet, before this claim can be defended, a better understanding of 'computation' is required. 'Digital computation' is an ambiguous concept. It is not just the classical dichotomy between analogue and digital computation that is the basis for the equivocation on 'computation' simpliciter in cognitive science, but also the diversity of extant accounts of digital computation. What does it take for a physical system to perform digital computation? There are many answers to this question ranging from Turing machine computation, through the formal manipulation of symbols, the execution of algorithms and others, to strong pancomputationalism, according to which every physical system computes every Turing-computable function. Despite some overlap among them, extant accounts of concrete digital computation are intensionally and extensionally nonequivalent, thereby rendering 'digital computation' ambiguous. The objective of this book is twofold. First, it is to promote a clearer understanding of concrete digital computation. Accordingly, the main underlying thesis of the book is that not only are extant accounts of concrete digital computation non-equivalent, but most of them are inadequate. In the course of examining several key accounts of concrete digital computation, I also propose the instructional information processing account, according to which nontrivial digital computation is the processing of discrete data in accordance with finite instructional information. The second objective is to establish the foundational role of computation in cognitive science whilst rejecting the extrinsically representational nature of computation proper.