Anatomic model of hippocampal encoding of spatial information.

Place cells were recorded simultaneously from identified locations along the longitudinal axis of the CA3 and CA1 subregions of hippocampus with a sixteen site electrode array while rats performed a simple pellet chasing task (Deadwyler et al., J Neurosci 1996;16:354-372; Hampson et al., Hippocampus 1996;6:281-293). Cells in CA3 or CA1, separated by 100-300 microm (two electrode locations), exhibited high cross-correlations with respect to place field firing in a given location in the chamber. This pattern of co-activity changed abruptly to low cross-correlations when the longitudinal distance between recording sites increased to 400-1,000 microm. Surprisingly, cells located 1,200-1,400 microm apart again exhibited similar place fields, suggesting a repeating pattern of place field representation within hippocampus. These features were used to construct a model of hippocampal place cell activation using known anatomic connections and projections between CA3 as well as CA1 pyramidal cells. The model provides a topographic representation in hippocampus of the animals' movements around the chamber as different place cells become activated. The model utilizes key landmarks (i.e., corners and walls) to define the animals' movement trajectories through successive place fields and to construct corresponding patterns of place cell firing in hippocampus. This is accomplished via a topological transformation of the chamber's key landmarks projected onto the anatomy of the CA3 and CA1 subregions. The primary feature of the model is that it can, within limited capacity, accurately encode where the animal has been, rather than where it is going. The model, therefore, appears to be more appropriate for memory required to return to particular locations than for initial guidance into those locations.