Recently it has been proposed that the hippocampus in rats and non-human primates does not encode only spatial information, but rather is responsible for the encoding of associations between the multiple cues present in the environment, whether spatial, temporal, visual, auditory, etc. (Eichenbaum et al., 1990; Sutherland & Rudy, 1989; Rolls et al., 1989). The goal of the proposed research is to examine whether a """"""""relational"""""""" approach can also explain the amnesic deficit seen in humans following hippocampal damage. Amnesics generally show impaired performance on explicit memory tests of recognition, cued recall, and recall, but are intact on implicit memory tests such as priming and skill learning. While describing the amnesic deficit as one of explicit access to declarative information accurately characterizes the behavioral consequences of hippocampal damage in humans, it does not explain the processing mechanism that leads to this deficit. It has been suggested that a deficit in relational processing may underlie the behavioral deficit found in amnesia, that is difficulty with explicit memory tasks (Eichenbaum, 1992; Squire, 1992). The present proposal defines the deficit in relational processing as an impairment in forming long lasting, multiple relations between cues present during encoding. In the proposed research amnesics and controls will be given several variations of three types of learning tasks; conditioning tasks, implicit memory tasks, and explicit memory tasks. It is expected that increasing the relational demands on all types of tasks assessing long term memory will detrimentally effect amnesics' performance more than normal controls. In addition, it is expected that amnesics will encode fewer relations with the original learning context than controls and, therefore, will: 1) show less interference than controls when tested in a misleading context, and 2) benefit relatively more than controls from having aspects the original learning context reinstated in the testing environment. Finally, it is hypothesized that the memory representation for amnesics is less complex and complete than controls, and, therefore, will be less flexible in novel situations. Results from the proposed experiments will help determine if a relational approach could accurately describe the deficit seen in human amnesia.
