The prevailing model of conscious memory holds that a group of loosely related structures, collectively called the medial temporal lobe (MTL), work together as a single functional unit. According to this current orthodoxy, the structures in the MTL are thought to store the memory of specific objects, facts and events. This model has many adherents despite the fact that it is contradicted by a large body of empirical data. We have developed a competing hypothesis, which is consistent with these data. Our model holds that different structures within the MTL have different functions, and that their collective functions extend beyond those assigned to the MTL by the orthodox model. Our work has made use of two measures of memory, both involving objects. One type of memory, known as paired associate learning (PAL), involves arbitrary associations between two different visual stimuli. A second measure of memory involves arbitrary associations between visual stimuli and spatially-directed responses, known as visuomotor learning (VML). Amnesic patients are notoriously poor in acquiring arbitrary associations;by tapping into these kinds of associative memory this work makes a direct link to human amnesia. There is evidence that the hippocampal system is not necessary for PAL. Murray et al. (1993) found that PAL was unaffected by complete removal of the hippocampus. This could mean that the hippocampal system is uninvolved in associative learning when neither component of the association has a relevant spatial attribute. But recent work appears to rule out that account (Brasted et al., 2002, 2003, 2005). Previous work on this project showed that hippocampal-system damage (fornix transection) impaired fast learning of object-response associations even when both the visual stimuli and the responses were nonspatially differentiated. This finding points to another possibility for the results described by Murray et al. (1993). On this explanation of our PAL results, there was no deficit because subjects learned the paired associates slowly. This hypothesis predicts that hippocampal-system damage might cause deficits in the fast learning of paired associates. Accordingly, we are engaged in testing the role of the hippocampus in PAL using a fast-learning procedure. The rationale for this test is the theory that the hippocampal system functions as a rapid acquisition, pattern-associator network, whereas the neocortex acquires similar information, but more slowly. In the context of PAL, this theory implies that the perirhinal cortex, a neocortical area, subserves the slower form of learning and that the hippocampal system subserves the faster form. We have now developed procedures that allow subjects to acquire both VML and PAL associations rapidly, within a single testing session. The results from this project could resolve crucial issues about hippocampal and perirhinal cortex function. A second goal of this project is to determine what part of the hippocampal system (hippocampus proper, subicular complex, or entorhinal cortex) is essential for fast VML. To investigate the role of the hippocampus in spatial and nonspatial associative learning, we trained three subjects on the two behavioral tasks described above. In VML, subjects were presented with one of three images on a touch screen. After the image was touched, it disappeared and three identical target boxes appeared in fixed locations on the screen. If the subject chose the correct location cued by the image for that trial, a reward was delivered. In PAL, subjects were presented with a cue image out of a set of two, and then with two distinct target images appearing at random locations on the screen. Reward was contingent on the subjects touching the target image corresponding to the cue. Each task was run with both familiar sessions, which used well-learned stimuli, and novel sessions, which used new stimuli and therefore required new associative memory formation. To determine the role of hippocampus in these tasks, we assessed the effects on behavior of bilateral infusions into the hippocampus of a neural inhibitor or saline. Preliminary data indicate that there is no effect of hippocampal inactivation on performance of either task relative to saline infusion or no infusion. These findings suggest that regions outside the hippocampus proper, either alone or together with the hippocampus, contribute to performance on these tasks. Accordingly, we plan to assess the effects of inactivation of neighboring structures, with and without simultaneous inactivation of the hippocampus. A third goal of the project, not yet realized, is to test hippocampal-prefrontal interactions in fast associative learning. We predict that both direct (via the fornix) and indirect (via entorhinal cortex) outputs of the hippocampus to the prefrontal cortex are critical to these kinds of rapid learning.
