Research Group
in Analytic Philosophy

Abstract

Miscomputation is a deviation from a norm that is set for artificially designed physical systems (e.g., desktops and smartphones), but also for species and organisms. When a system miscomputes, it computes a different mathematical function g, rather than the function, f, which is the norm (i.e., there exists at least one input ix such that g(ix) ≠ f(ix)). Given the importance of miscomputation for any completeaccount of physical computation, surprisingly few works have characterised this phenomenon explicitly (e.g., Colombo, 2021; Fresco & Primiero, 2013; Piccinini, 2015; Tucker, 2018). In contrast to artificial computing systems, there are no design specifications according to which the behaviour of a biological system can be classified as correct or incorrect. ‘Miscomputation’ figures, for example, in computational Bayesian models and computational reinforcement models of mental illnesses. But how should miscomputation be construed in the computational sciences of mind and brain? Is any appeal to ‘miscomputation’ in these sciences merely instrumental? Is this usage aligned with how ‘miscomputation’ is used in computer science and engineering? Most normative accounts of computation face a dilemma: either there is no such thing as miscomputation, or it cannot be explained. In this talk, I will describe the dilemma and outline a possible preliminary solution.