Electrophysiological evidence for memory schemas in the rat hippocampus
MetadataShow full item record
According to Piaget and Bartlett, learning involves both assimilation of new memories into networks of preexisting knowledge and alteration of existing networks to accommodate new information into existing schemas. Recent evidence suggests that the hippocampus integrates related memories into schemas that link representations of separately acquired experiences. In this thesis, I first review models for how memories of individual experiences become consolidated into the structure of world knowledge. Disruption of consolidated memories can occur during related learning, which suggests that consolidation of new information is the reconsolidation of related memories. The accepted role of the hippocampus during memory consolidation and reconsolidation suggests that it is also involved in modifying appropriate schemas during learning. To study schema development, I trained rats to retrieve rewards at different loci on a maze while recording hippocampal calls. About a quarter of cells were active at multiple goal sites, though the ensemble as a whole distinguished goal loci from one another. When new goals were introduced, cells that had been active at old goal locations began firing at the new locations. This initial generalization decreased in the days after learning. Learning also caused changes in firing patterns at well-learned goal locations. These results suggest that learning was supported by modification of an active schema of spatially related reward loci. In another experiment, I extended these findings to explore a schema of object and place associations. Ensemble activity was influenced by a hierarchy of task dimensions which included: experimental context, rat's spatial location, the reward potential and the identity of sampled objects. As rats learned about new objects, the cells that had previously fired for particular object-place conjunctions generalized their firing patterns to new conjunctions that similarly predicted reward. In both experiments, I observed highly structured representations for a set of related experiences. This organization of hippocampal activity counters key assumptions in standard models of hippocampal function that predict relative independence between memory traces. Instead, these findings reveal neural mechanisms for how the hippocampus develops a relational organization of memories that could support novel, inferential judgments between indirectly related events.