The University of Arizona
NSMA Neural Systems, Memory and Aging Division of ARL

Society for Neuroscience Research Abstracts

Path integration-based attractor dynamics in the entorhinal cortex.

*S. LEUTGEB1, L. L. COLGIN1, K. JEZEK1, J. K. LEUTGEB1, M. FYHN1, B. L. MCNAUGHTON1,2, E. I. MOSER1, M.-B. MOSER1

1. Ctr. Biol. Memory, NTNU, Trondheim, Norway;
2. ARL Div. Neural Systems, Memory and Aging, Univ. Arizona, Tucson, AZ

Hippocampal cell ensembles can accommodate changing inputs in an additive manner, forming a continuity of increasingly more different representations for progressively dissimilar stimuli or, alternatively, can switch sharply between mutually exclusive attractor states. The existence of discrete, non-overlapping representations in the hippocampus raises the question of whether the discreteness is generated intrinsically in the hippocampus, possibly as a consequence of attractor dynamics in the recurrent CA3 network, or whether it is imposed from outside, possibly as a result of aligning grid cells in the medial entorhinal cortex to different sets of external landmarks. We hypothesized that attractor dynamics in hippocampus depend on the presence of distinct, previously stored spatial representations in the entorhinal cortex. In order to associate two sets of path integrator coordinates with different landmarks, rats explored and became well familiarized with a square and circular enclosure connected by a corridor during several days of pretraining. As expected for different portions of a continuous space, grid cell maps were aligned in an unrelated way to the walls of each enclosure. During the second phase of training, the square and the circle were moved to a common intermediate location and presented on alternating trials. Entorhinal grid cells retained their map alignment with respect to the walls of the enclosures, as indicated by shifted center peaks from the origin of the crosscorrelation maps between the square and the circle, while simultaneously recorded hippocampal place cells exhibited strong remapping. Once coordinate shifts and remapping had been verified at the common location, the boxes were morphed on a final test day by presenting intermediate shapes interspersed between the original square and circular shapes in a predetermined sequence. The grid cell alignment changed abruptly and coherently at an intermediate shape between the square and circle. The transition was not only coherent among simultaneously recorded entorhinal cells, but it also occurred along with global remapping in hippocampal place cells. These results suggest that, when the representations in entorhinal cortex correspond to two different path integrator coordinates, both the entorhinal and hippocampal networks will switch sharply and without any intermediates from one representation to the other. Sharp and coherent transitions between place-cell ensembles in the hippocampus may reflect the retrieval of different sets of spatial coordinates in the entorhinal cortex rather than the mere activation of discrete attractor states within the hippocampus.