UCL Logo

Dissociation between spatial and nonspatial information conveyed by parallel input streams into the hippocampus

Jim Knierim

Neurobiology and Anatomy Department, University of Texas Medical School , Houston , Texas , USA

Place cells of the hippocampus form a spatial representation of the environment that plays a critical role in navigation, context-dependent learning, and episodic memory. The precise nature of hippocampal information processing remains unsolved, in large part due to the lack of detailed information about the representations that constitute the input to the hippocampus. One input arises from a set of parahippocampal regions that appear to be segregated into two processing streams that culminate in the lateral entorhinal cortex (LEC) and the medial entorhinal cortex (MEC). Another input arises from a set of cortical and subcortical structures that calculate a representation of the animal’s head direction. To investigate how these representations interact, we recorded from CA1 place cells and cells of the parahippocampal regions, as well as head direction cells of the anterior thalamus. A number of cells in the superficial layers of MEC (in regions that project to the dorsal half of the hippocampus) showed robust spatial selectivity in a pellet-chasing task, although on average the selectivity was less than that of CA1. In contrast, cells in the superficial layers of corresponding regions of LEC showed little spatial selectivity. Cells in the perirhinal cortex, which provide a major input to LEC, also revealed little spatial tuning. Cells of the parasubiculum demonstrated spatial tuning that was similar in quality to cells of MEC. The spatially selective neurons recorded in these regions were controlled by idiothetic cues or by visual landmarks, similar to CA1 place fields and thalamic head direction cells. These results suggest that MEC may convey spatial/contextual information to the hippocampus, LEC may convey individual item/object information, and the hippocampus may create conjunctive object+place (item+context) representations useful for context-dependent learning.