Accounts of Computation in Physical Systems

Attributing computations to physical systems is a commonplace practice in science. A paradigm example can be found in Marr’s (2010) Vision, meanwhile, cognitive science is replete with attributions of computation to the brain. Despite strong intuitions regarding which physical systems do and do not compute, there is an outstanding problem in codifying those intuitions into a criterion or an account of computation. This is a problem insofar as the attribution of computations are taken to be meaningful or explanatory of a given physical system to which they are attributed. A number of philosophical accounts of computation have been pursued, especially recently (Piccinini 2015, Shagrir 2018, Schweizer 2019, Coelho Mollo 2020) aiming to make sense of computation, to rescue such attributions from an explanatory abyss. Both recent and earlier literature on the subject (e.g. Fodor 1975, Pylyshyn 1984, Searle 1980; 1990; 1992; Putnam 1988) seem to disagree about the basic foundations of the question. Some accounts take computation to be essentially abstract, such that the relevant thing to explain is how the abstract relates to the physical. Others take computation to be mechanistic, such that the relevant question is what kinds of mechanisms are computational ones. Still others think computation is essentially representational, a commitment which must be clarified in its own right. In this project, I attempt to organize and chart this terrain. Following convention found in the literature, I organize philosophical accounts of computation into three “families” of accounts: mapping, mechanistic, and semantic accounts. My aim is to provide a clear cartographic guide for understanding both the accounts themselves, and the problems they face in making sense of computing in physical systems (e.g. pancomputationalism). I then analyze these accounts according to three desiderata: individuation, objectivity, and explanation. While I do not attempt to defend these desiderata as desiderata, I frame them as major motivators for the accounts on offer– as goals toward which these accounts are explicitly or implicitly oriented. Although this project is largely cartographic, ultimately I suggest that an underlying assumption of mapping and mechanistic accounts– namely, a commitment to naturalizing computation– inevitably precludes meeting at least one of the three desiderata.

Department dc:contributor.department

History and Philosophy of Science and Technology

Year dc:date.issued

2024

Author dc:creator

• Hall, Jessica Susanne

Advisors dc:contributor.advisor

• Smith, Brian C
• Seager, William

• Computation
• Implementation
• Semantics

Handle dc:identifier.uri

http://hdl.handle.net/1807/140062

OAI identifier oai:identifier

oai:utoronto.scholaris.ca:1807/140062