Origin and role of path integration in the cognitive representations of the hippocampus: computational insights into open questions

Abstract

Path integration is a straightforward concept with varied connotations  that are important to different disciplines concerned with navigation,  such as ethology, cognitive science, robotics and neuroscience. In  studying the hippocampal formation, it is fruitful to think of path  integration as a computation that transforms a sense of motion into a  sense of location, continuously integrated with landmark perception.  Here, we review experimental evidence that path integration is  intimately involved in fundamental properties of place cells and other  spatial cells that are thought to support a cognitive abstraction of  space in this brain system. We discuss hypotheses about the  anatomical and computational origin of path integration in the wellcharacterized circuits of the rodent limbic system. We highlight how  computational frameworks for map-building in robotics and cognitive  science alike suggest an essential role for path integration in the  creation of a new map in unfamiliar territory, and how this very role can  help us make sense of differences in neurophysiological data from  novel versus familiar and small versus large environments. Similar  computational principles could be at work when the hippocampus  builds certain non-spatial representations, such as time intervals or  trajectories defined in a sensory stimulus space.