Evaluating Goodness-of-Fit in Comparison of Models to Data

While the representational theory of mind is without doubt one of the central theoretical underpinnings of Cognitive Science, the use of cognitive simulation models to explain human behavior is one of its main methodological tools. Not only does cognitive modeling exemplify a view of human thinking as symbol processing, it also provides us with rigorous qualitative and quantitative predictions that can be used to discriminate between alternative models and uncover which aspects of a given theoretical framework require further elaboration. In his scientific life, Werner Tack moved from (numerical) mathematical theories (Tack, 1969) to computational (symbolic) models of human cognition (Tack, 2004) — and both authors of the present paper had the luck to experience his vivid enthusiasm for formal approaches to the study of mind. The formal rigor of theorizing in terms of such computational approaches has, however, not yet been met by equally rigor procedures for evaluating the goodness of fit of cognitive models to data — and we remember many fruitful discussions with Werner Tack on this topic. In this paper we aim at addressing the gap between the rigor in model construction and model evaluation, and look forward to the next round of exciting discussions with Werner Tack.