Modeling Cognition: How Fiction Relates to Fact

Modeling Cognition: How Fiction Relates to Fact Anna-Mari Rusanen (anna-mari.rusanen@helsinki.fi) Philosophy of Science Group/ Department of Philosophy, History, Art and Culture Studies, PO BOX 24 00014 University of Helsinki, FINLAND Otto Lappi (otto.lappi@helsinki.fi) Cognitive Science Institute of Behavioural Sciences, PO BOX 9 00014 University of Helsinki, FINLAND properties and their causal relations - gives rise to or produces the phenomenon. Constructing an explanatory mechanistic model involves mapping elements of a mechanistic model to the system of interest, so that the elements of the model correspond to identifiable constituent parts with the appropriate organization and causal powers to sustain that organization. The mechanistic account of explanation is a typical example of the realist interpretation of scientific models. According to realism, a model explains the behavior of a target system, if and only if it is a correct account of the target’s behavior underlying observed phenomena – i.e. the model must correspond to, depict or represent the target system in a sufficiently correct way. In addition, many current realist accounts require that the target systems are actual or real – i.e. have causal power to generate observable phenomena and data. However, models are always more or less abstract, simplified and idealized descriptions of their real world target systems. Target systems are just too complicated to be studied in a full fidelity, and thus all kinds of assumptions are made to reduce the complexity of a model. Thus most (if not all) models used in science are unrealistic. Often models are nevertheless considered useful, even if they are known to be false, and they are known to contain assumptions that are not even approximately true, but highly idealized. For this reason, it has been argued that this feature of modeling seriously undermines the realist interpretation of models. If all models involve unrealistic elements, how is it possible that they could correspond, depict or describe the real world target system in a correct or truthful way? If they do not, where does their explanatory force come from? Sometimes models involve assumptions about fictional entities and processes that are known not to exist in the real world. These fictional models describe systems that (i) do not exist in the real world or (ii) have elements that do not exist in the real world. Obvious examples of fictional models in cognitive science are for instance the models of Abstract The increasing use of computational modeling and simulation methods offers interesting epistemic and theoretical challenges for the philosophy of science. One of the main questions discussed in the philosophical literature relates to the explanatory role of false, unrealistic and sometimes even fictional models. In this paper we argue that (i) some fictional models can offer explanations known as structural model explanations, and (ii) at least some variants of realism, such as the information semantic account of scientific models, can consistently hold that this subset of fictional models are explanatory. Keywords: Models; information semantics fictional models; explanation; Introduction For a philosopher of science interested in the philosophical issues of modeling, cognitive science is a wonderful source of case studies. Cognitive science utilizes modeling in a unique way, both methodologically and theoretically. The increasing use of computational modeling and simulation methods offers interesting methodological challenges for scientists, but also philosophers of science find many things of interest in the theoretical and epistemic status of modeling methods. One of the main questions discussed in the philosophical literature relates to the explanatory role of models. A growing number of philosophers have proposed that explanation of the behavior and capacities of complex systems (such as those found in the cognitive, biological and neurosciences) does not typically involve natural laws, but specific models of particular mechanisms (Bechtel and Richardson, 1993; Craver, 2006, 2007; Machamer, Darden, and Craver, 2000). It has also been argued that this mechanistic account of explanation could be extended to cover explanations in cognitive science (Kaplan & Craver, 2011, Sun, 2008) and computer sciences, as well as computational neuroscience (for instance, Piccinini, 2007). According to this account, to explain a phenomenon is to construct a model of how a causal mechanism - a hierarchical system composed of component parts, their