Abstract Subjects selected data in order to decide from which of two ‘islands’ an ‘archeological find’ had come. The results replicated two established phenomena in cognitive psychology: (1) the tendency to ignore base rate data given individuating information, and (2) the tendency to seek confirmatory evidence. The major outcome of the study was, however, to reveal a new phenomenon in information search. Subjects displayed a surprising and strong tendency to seek diagnostically worthless information. They then altered their conclusion based on that information. For example, subjects who had already obtained P( D 1/ H 1) selected P( D 2/ H 1) when P( D 1/ H 2) was equally easily available, and when they had no relevant experience to bring to bear on the estimation of P( D 1/ H 2). This phenomenon, which appears to be a wholly dysfunctional cognitive tendency, was labeled pseudodiagnosticity.