In cognitive psychology, theoretical models of the mental processes underlying cognitive abilities are proposed, tested and revised on the basis of patterns of performance over a range of cognitive tasks in single-case and group studies involving subjects with varying degrees of brain damage. A theory of cognition is a modular functional architecture that provides a computationally explicit, information-processing account of how the mental representations required to perform these cognitive tasks are produced. Professor Bub is continuing his research applying Dr. Clark Glymour's `bootstrapping' account of relevant evidence to a current methodological controversy in cognitive neuropsychology, concerning the relation between theory and evidence from single-case and group studies on brain-damaged and normal subjects. Professor Bub has reformulated the original project as an aspect of the general problem of how to reliably infer causal relations (representing a model of cognition) from probability relations (representing performance data from normal and brain-damaged subjects). This new investigation concerns the status of some of the assumptions that guarantee the existence of a reliable methodology for cognitive neuropsychology, as well as the single-case versus group studies problem that motivated the original inquiry.