A major thread in the cognitive sciences since their inception has been the development of computational models. These models serve (at least) three goals. First, they coerce the precise formulation of specific theories, and their consequences can be understood more reliably than by pen-and-paper speculations. Second, they provide a hypothesis for the "software in the mind", to be corroborated by brain research. Third, they contribute to the development of industrial products ("artificial intelligence"). Mathematical approaches to linguistic theories, assisting and assisted by psycholinguistic and neurolinguistic research and resulting in language technology, offer the best known example. In my paper I argue that we should also develop formal models for other domains of higher cognition, serving the same three goals.
Optimality Theory (OT, Prince and Smolensky 1993/2004, Smolensky and Legendre 2006) has been a successful linguistic framework for two decades, accounting for a huge number of linguistic phenomena. It comes with cognitively plausible computational implementations and learning algorithms. OT might be seen as a general model for higher cognition, and so non-linguistic domains could also be approached with it. In fact, the uniform structure of the brain calls for tackling various domains in a uniform framework. In my talk, I shall illustrate this statement by an OT model of dietary preferences. Thereby I hope to pave the way to a computationally implementable cognitive approach of human culture. OT helps formulate theories in a precise way, leading to hypotheses about the "software of the brain", and possibly even to industrial applications.