Full Program »
Modeling without RepresentationalismThe recent philosophical literature on scientific modeling is largely predicated on a twofold representationalist assumption, according to which (R1) a model is a representation of a target, and (R2) the representational relationship between model and target is what secures the epistemic value of modeling—i.e., scientists can learn through modeling insofar as models represent their target. But representationalism is riddled with difficulties. Drawing from cognitive science, I sketch an anti-representationalist view of scientific models as tools or artifacts, and I argue that it can circumvent the problems inherent to representationalism while still preserving some of its attractive features.
Representationalist accounts of scientific modeling differ in how they explicate the nature of representation. Traditional views take representation to be a mind-independent property of the model-target dyad, as is the case of isomorphism- and similarity-based accounts. Other views frame representation as a mind-dependent, conventional feature of modeling, and accordingly describe a triadic (rather than dyadic) representational relationship in which scientists play the central role of determining what in the model represents what in the target, in what way, to what degree of accuracy, and so on. On the one hand, dyadic views of representation support R2 at the expense of R1: that is, they support the claim that the epistemic value of modeling is due to the mind-independent representational relationship between model and target, but they do so by giving unsatisfactory accounts of representation, either as too restrictive (e.g., isomorphism-based views that apply primarily to mathematical models) or too permissive (e.g., similarity applies too broadly: anything is similar to anything else in various ways). On the other hand, triadic views strengthen R1 at the cost of weakening R2: if representation is grounded on scientific practice and is a matter of stipulation, then the representational relationship between model and target cannot be what secures the epistemic value of modeling, for, in this view, anything can represent anything else in some context as long as someone is willing to establish the relevant representational mapping.
These problems motivate pursuing a novel way to make sense of modeling without assuming representationalism—i.e., R1 and R2. This poster presents the problems inherent with representationalism and articulates precisely what an alternative framework can look like. I draw from Gibson's (1979) ecological psychology—an anti-representationalist view of cognition—to account for modeling as a kind of tool-building and tool-using practice. In this view, the epistemic value of modeling is fundamentally tied to the “affordances” or “action possibilities” that models make available to scientists. Affordances are relational properties that encompass both the “abilities of organisms and features of the environment” (Chemero 2003, 189). An affordance-based approach to modeling preserves the objective anchoring characteristic of dyadic views of representation while still including the agential variability of triadic, practice-based views of representation—but it does so without any reference to “representation.”
Chemero, Anthony (2003) An Outline of a Theory of Affordances. Ecological Psychology, 15(2), 181–195.
Gibson, James (1979) The Ecological Approach to Visual Perception. Houghton Mifflin Company Boston.
Guilherme Sanches de Oliveira
University of Cincinnati