Whether touch can substitute for light in MEC maps
The medial entorhinal cortex (MEC) holds several kinds of spatially modulated cells whose firing tracks something about the animal's position or orientation: grid cells fire in a repeating hexagonal pattern, head-direction cells fire according to where the head points, border cells fire near walls, and non-grid spatial cells fire in localized fields. Spatial maps must stay anchored to external cues, and the study asks how visual and tactile cues jointly support these cells.
The authors used miniature two-photon calcium imaging in freely moving male mice, recording neural activity while animals explored open arenas. Calcium events were converted into firing-rate maps binned at 2.5 × 2.5 cm for position and 3° for head direction, and cells were classified from a baseline session. This design lets the same cells be compared across sensory conditions rather than across different animals.
Darkness degraded grid and spatial coding first
In Environment 1, an open field with three visual cue cards on the walls, mice explored under light (L), then in complete darkness (D), then in darkness after whisker trimming (DW). Darkness reduced grid scores and spatial information, and it increased the variability of grid spacing and grid ellipticity, meaning grid patterns became less regular. Among 215 grid cells, 181 (84.2%) changed significantly in at least one metric after light deprivation, while 34 (15.8%) were classified as resistant.
Non-grid spatial cells followed a similar pattern: of 501 identified spatial cells, 405 (80.8%) were affected by light deprivation and 96 (19.2%) were resistant, and their mean field size increased significantly in darkness. Population-level position decoding from all functional cell types also worsened: mean absolute error rose from 13.35 cm in light to 19.88 cm in darkness. Even so, decoding error stayed below shuffled controls in every session, which the authors interpret as residual map construction, potentially through path integration.
Whisker trimming added little after darkness
When whiskers were trimmed in the dark, grid scores and spatial information did not decline further at the population level, and decoding error was unchanged (16.84 cm in DW). The authors caution that this sequence was not independent: light deprivation came first, so the DW session reflects combined visual and whisker deprivation, and grid cells may have become less sensitive to further sensory loss once visual input was gone.
Individual cells tell a different story from the population average. After whisker trimming, 133 of 215 grid cells (61.9%) were still classified as affected and 82 (38.1%) as resistant, and for non-grid spatial cells 275 of 501 (54.9%) were affected while 226 (45.1%) remained resistant. Head-direction and border cells were more clearly altered by whisker trimming than grid cells were, which is why the paper describes the effects as cell-type specific rather than uniform.
Tactile landmarks anchored every cell type
To test whether touch alone can anchor spatial representations, the authors built Environment 2: an arena with no visual cue cards, two sandpaper strips of different grit on two walls, and a square of sandpaper on the floor in one corner. Mice were habituated in complete darkness for about one week, so any spatial tuning had to be built from tactile and self-motion information.
In this tactile cue–dominated setting, all spatially modulated cell types formed cue-anchored representations. The result matters because it separates two claims that are easy to conflate: darkness alone does not abolish MEC spatial coding, and the failure in Environment 1 was about losing the dominant visual anchor, not about losing light per se. The same cells can lock onto a different sensory reference when that reference is the salient one.
Whisker loss mattered more with touch anchors
Once tactile landmarks were the anchoring cue, whisker trimming impaired grid and spatial coding more strongly than it had in the visually dominated arena. This is the paper's central comparison: the same manipulation produced a small additional effect in Environment 1 and a larger one in Environment 2, so the impact of losing whisker input depends on which cue the map was built around.
The authors also removed the arena walls in darkness after whisker trimming, eliminating boundary-related tactile feedback. That manipulation severely degraded representations across all spatially modulated cell types, with border-cell fields nearly completely abolished. This supports the interpretation that boundary-related tactile signals help sustain some neurons even after combined visual and whisker deprivation, which is why some cells remained stable in the DW session.
Some MEC neurons tracked whisker movement
In a separate head-fixed analysis, the authors identified 191 MEC neurons whose calcium activity correlated with whisker movement. Most of these, 67.0%, showed no spatial tuning during free exploration. Of the remaining 63 cells (33.0%), head-direction cells were the largest group at 17.3%, followed by border cells at 12.0%, spatial cells at 2.1%, and grid cells at 1.6%.
Compared with chance overlap computed from a hypergeometric distribution, border cells were significantly enriched among whisker-responsive neurons, while grid and spatial cells were significantly depleted. The authors note that whisker movement covaries with orofacial movements, arousal, and other behavioral states, so these recordings cannot establish that the signal is specifically tactile. The finding is presented as a possible route for whisker-related information to reach MEC, not as proof of one.
What the deprivation sequence can and cannot show
The study used male mice only, and the whisker trimming always followed light deprivation in Environment 1, so the two manipulations were not independent there. The authors also report that whisker trimming reduced head angular speed, which means part of the head-direction effect may reflect altered self-motion or vestibular input rather than touch alone. Investigators were not blinded to conditions, though all imaging and cell classification used the same predefined automated pipelines.
These limits qualify how far the results travel. The paper supports a context-dependent account in which MEC cells weight whichever sensory reference is most informative in a given arena, not a fixed hierarchy in which vision always dominates touch. The most decision-relevant split is between environments: whisker input matters little for grid coding when visual landmarks are present and dominant, and matters substantially when tactile landmarks are the only anchor available.
Sources used in this answer
Differential effects of visual and tactile deprivation on spatially modulated cells in the medial entorhinal cortex
Darkness disrupted all spatially modulated MEC cell types in a visually dominated arena, while in a dark tactile-landmark arena all cell types anchored to touch cues and whisker trimming more strongly impaired grid and spatial coding.