Recently, it has been proposed that cortical regions in the inferior temporal lobe support recognition based on discriminations of stimulus familiarity, and that this function is independent of the hippocampal formation. The framing of recognition as a familiarity discrimination process has a long cognitive history, and was formalized with the application of signal detection theory to recognition data. However, many researchers have argued that this single process theory is insufficient because the conscious recollection of qualitative, contextual information can also be used during recognition judgments. Both animal and patient evidence suggest that this ability critically depends on the hippocampal formation. A dual process model, consistent with reports of preserved recognition in patients and animals, suggests that both the hippocampal formation and outlying regions contribute to recognition; the former critical for recollection, and the latter for familiarity discriminations. If correct, the preserved recognition in amnesics with circumscribed hippocampal damage, will be driven largely by item familiarity, and therefore will closely conform to the predictions of signal detection theory. For normal subjects, this is typically not the case, presumably because recollection also contributes to their performance. The current project will contrast controls and patients in four experiments in which the predicted trend of performance is different depending on whether recognition is a signal detection process, a discrete recollective process, or the result of both. If the hippocampal formation and outlying cortical regions perform a single, integrated memory process, then performance of amnesics should simply resemble a degraded version of controls. Such findings would be consistent with the declarative memory model, and recent single process accounts of recognition. However, if the hippocampal formation primarily serves recollection, then the pattern of preserved recognition in patients should qualitatively differ from controls. Such a finding would provide a neurological footing for dual process theory, help clarify the nature of preserved recognition in amnesics, and highlight the inadequacy of single factor memory estimates (e.g., d', percent correct, hits minus false alarms) for identifying functional deficits in amnesia.