| Summary: | The rodent hippocampus and associated structures are implicated in spatial memory and navigation. A necessary requirement of these roles is the ability to integrate incoming sensory information with pre-existing knowledge about an environment. With this dichotomy in mind, sensory control over place cell firing in the hippocampus, and grid cell firing in the entorhinal cortex were investigated. In the first experimental chapter a computational model of hippocampal place cell firing is presented. The model, a two layer feed-forwards network, describes place fields as a function of the distance and direction to boundaries surrounding an animal. It is shown that by incorporating the idea of boundaries with distinct sensory qualities, and by allowing synaptic weights to be updated by application of the BCM learning rule, the model is able to capture: (1) Experiential changes in place fields resulting from prolonged exposure to a static environment, and (2) Changes in place field position and firing rate induced by movement of cues and boundaries in a familiar environment. The model is shown to compare favourably with novel electrophysiological data collected for this purpose and with experimental findings published by other authors. The second experimental chapter investigates the affects geometric manipulations of a familiar environment have on the firing of medial entorhinal grid cells. Novel data is presented that shows grid cell firing represents an experience-dependent interaction between sensory input and learnt expectation about the size of an animal's enclosure. It is also shown that this interaction evolves with time to resolve conflict between the two sources of information. Finally it is noted that grids from a single animal are aligned and have fixed relative sizes. The final chapter discusses the data in terms of how the brain perceives and represents the world.
|
|---|
