Culture and Computation: Steps to a Probably Approximately Correct Theory of Culture

• In Event: 230 - Theory Section. New Developments in Contemporary Theory

Abstract

This paper outlines some provisional steps toward a theory of culture grounded in computational thinking. This theory provides computational micro-foundations for sociological claims about perception, classification, learning, and habit. These micro-foundations address the computational problems faced by social actors, and the potential algorithmic and representational solutions consistent with what we know about the bodies, brains, and material environments in which they are implemented. I begin by describing computational thinking, drawing on “Marr’s hierarchy” for the analysis of information processing systems. I then address the definition of culture and social action, arguing with Sperber that culture is a property of causal chains, rather than a thing-in-the world. I briefly address contemporary debates over the nature of culture—embodied versus embedded—and argue for an ecological approach in which culture-in-action unfolds as embodied schemas encounter (and produce) “handles” in the environment. When schemas are “objectively adapted” to the handles, this generates action that is ecologically rational—what Bourdieu would call “ontological complicity.” To explain ontological complicity, I outline a formal approach to cultural learning based on Probably Approximately Correct (PAC)-learning theory. I then describe a concrete model of enculturated action, drawing on the classifier system approach to distributed computation. Taken together, these two approaches provide a new angle on many of the unanswered questions raised in DiMaggio’s foundational article on cognition and culture. To illustrate the synthetic value of a computational theory of culture, I conclude by applying this approach to the sociology of science.

Author

• Jacob Gates Foster, University of California-Los Angeles