This project involves the application of Glymour's "bootstrapping" account of evidential relevance to a methodological controversy in cognitive neuropsychology. The controversy centers on an influential claim by Caramazza that brain-damaged performance can provide relevant evidence for models of cognition only in the form of single-case studies. If this claim is true, then group studies of brain-damaged performance--in particular, all studies based on the classical syndrome categories (such as Broca's aphasia, Wernicke's aphasia, conduction aphasia, etc.)--are worthless as confirmatory or disconfirmatory evidence in constraining rival functional architectures. The primary objective of this project is twofold: firstly, to clarify the different methodological roles of single-case versus group studies, as manifested in dissociations in performance, double dissociations, and associations of deficits (the typical operational elements in cognitive neuropsychology), on the basis of a bootstrapping account of evidential relevance; and secondly, to evaluate bootstrapping relative to other accounts of evidential relevance--for example, the hypothetico-deductive account and the Bayesian account--in terms of the detailed working out of this application. The long-term aim is a complete analysis of the logic of testing models of cognition, as an aspect of the general theory construction in the sciences.