Patients with Alzheimer's disease (AD) perform poorly on tasks dependent on access to, and utilization of, previously acquired knowledge and skills. It has been commonly assumed that impaired knowledge in AD, as well as in other patients with cortical lesions, is due to an actual loss or disorganization of a specific knowledge base or system. This hypothesis has, however, recently been called into question by data from tasks that purport to tap knowledge on a more automatic and implicit level. For example, although AD patients are impaired on object naming and verbal fluency tasks, they show a normal pattern of semantic facilitation on reaction time based priming tasks. In fact, the level of facilitation or activation on these tasks has often been reported to be greater in AD patients than in normal individuals. These and similar data have been used to support arguments that performance decrements in AD patients are due to deficits in attentional and/or retrieval processes rather than a degradation of knowledge stores. The central focus of this project will be to test a model of the semantic representations of object that predicts increased facilitation or hyperpriming in AD patients as a result of degraded representations. The relationship between performance on on-line priming tasks, visual attention and spatial processes, and explicit and implicit measures of memory also will be examined. In addition to normal controls, patients with cognitive and memory impairments, but without semantically-based naming difficulties (elderly depressed, Huntington's disease, Korsakoff's disease) will serve as controls for overall slowness of response and degree of explicit memory deficit.