RECALIBRATION OF MEC DIRECTIONAL AND SPATIAL REPRESENTATIONS IN AN UNRESTRAINED VIRTUAL ENVIRONMENT

The medial entorhinal cortex (MEC) would continuously recalibrate internal representations to support navigation in changing environments. However, conventional head-fixed virtual reality (VR) paradigms decouple self-motion signals from visual inputs in two-dimensional environments, limiting insight into how this recalibration is achieved under naturalistic navigation. To address this, we aimed to investigate how the same MEC cell ensembles resolve sensory mismatches across varying spatial and temporal scales. Here, we developed a novel unrestrained VR environment that preserves naturalistic self-motion feedback and recorded large-scale MEC activity from freely behaving rats (~500 units in total) using Neuropixels 2.0 probes. We investigated dynamic recalibration of MEC ensembles under two distinct visual challenges: a structural 90-degree Rotation Task involving persistent spatial and directional re-alignment and a repetitive 0.1-s Blink Task involving transient interruption of visual anchor cues. In the Rotation Task, the re-introduction of visual cues after a 10-s dark interval did not lead to an abrupt reset of spatial and directional representations. Instead, decoded maps exhibited a gradual, continuous drift toward the new visual anchor, reflecting a slow spatial recalibration. In contrast, the Blink Task triggered near-instantaneous responses in identified head-direction and spatial cells. Although these responses were heterogeneous, response latencies to the brief flashes progressively shortened across trials. Together, these findings demonstrate that MEC circuits deploy complementary recalibration strategies across distinct timescales: slow continuous drift to resolve structural spatial mismatches and rapid temporal optimization to adapt to momentary sensory events. This dual computational framework enables flexible spatial representations during naturalistic navigation.