Proceedings of the Annual Meeting of the Cognitive Science Society

Spatial cognition relies on an internal map-like representationof space provided by hippocampal place cells, which in turnare thought to rely on grid cells as a basis. Spatial Seman-tic Pointers (SSP) have been introduced as a way to representcontinuous spaces and positions via the activity of a spikingneural network. In this work, we further develop SSP rep-resentation to replicate the firing patterns of grid cells. Thisadds biological realism to the SSP representation and links bi-ological findings with a larger theoretical framework for rep-resenting concepts. Furthermore, replicating grid cell activitywith SSPs results in greater accuracy when constructing placecells.Improved accuracy is a result of grid cells forming the op-timal basis for decoding positions and place cell output. Ourresults have implications for modelling spatial cognition andmore general cognitive representations over continuous vari-ables.