Proceedings of the Annual Meeting of the Cognitive Science Society

We present a novel framework for assessing the fit of cogni-tive models. Using this framework, we highlight limitationsof existing methods of model evaluation, and derive new ap-proaches to validating cognitive models. Tests of topographi-cal consistency emphasize how a model’s structure constrainsbehavior on pairs of coupled stimuli, even when point predic-tions on individual stimuli depend on estimates of the model’sfree parameters. By carefully selecting these coupled stim-uli such that they follow the distinct topography of the model,researchers can overcome some limitations of existing meth-ods. Finally, we provide a proof-of-concept example of how touse our approach to assess a model of multi-alternative, multi-attribute choice.