UCLA

The hippocampal formation has long been implicated in the formation of episodic memories. Equally, the activity pattern of neurons in these brain regions is thought to serve as a neural substrate for spatial representation thus constructing a cognitive map of space. This spatial aspect of hippocampal activity has been extensively studied over the past decades, particularly in rodents. In addition, hippocampal neurons exhibit a robust temporal code, termed phase precession, which is hypothesized to facilitate sequential learning and complement the hippocampal spatial code.

Despite decades-long research, the underlying mechanisms governing the hippocampal spatial and temporal codes remain poorly understood. It has been shown that multiple sensory and motor inputs reach the hippocampus and can modulate its activity. In real world (RW) environments however, the contributions of these inputs are confounded. Thus, to dissociate these contributions and thereby elucidate the mechanisms of spatial selectivity, we used a virtual reality (VR) setup.

We found comparable levels of hippocampal spatiotemporal selectivity on linear tracks in RW and VR. In contrast, during random foraging in two-dimensions, spatial selectivity was severely diminished in VR. Nevertheless, most spikes occurred within ~2-s-long hippocampal motifs—with similar structure to that in RW—within which the hippocampal temporal code was intact, demonstrating a decoupling between the spatial and the temporal codes. These observations also impose constraints on the validity of phase precession models. Further, additional experiments and analyses revealed significant directional selectivity in the hippocampus in RW and VR, challenging commonly held beliefs about hippocampal directionality. Notably, contrary to the impairment of spatial selectivity in VR, the degree of directional selectivity was identical in both worlds and was determined by the angular information contained in the visual cues. Similar to place cells, grid cells in the medial entorhinal cortex (MEC) lost their spatial selectivity and periodic firing pattern in VR. Finally, most MEC head-direction cells did not maintain their selectivity, though there was a small but significant subset of head-direction cells that displayed selectivity with respect to the visual cues in VR.

These results suggest that visual cues alone are insufficient to generate a stable localized representation in the spatial domain in place cells and grid cells. Conversely, they are sufficient to elicit—and play a causal role in—hippocampal directional selectivity and drive directional responses in a subset of head-direction cells. Additionally, these findings are reminiscent of the reports in human and non-human primates and can thus bridge the gap between these studies. Taken together, the investigation of the influence of multisensory inputs on hippocampal responses can provide insight into the underlying neural mechanisms of spatial representation in the brain.