Home, head direction stability, and grid cell distortion

Abstract

The home is a unique location in the life of humans and animals. In rats, home presents itself as a multicompartmental space that involves integrating navigation through subspaces. Here we embedded the laboratory rat's home cage in the arena, while recording neurons in the animal's parasubiculum and medial entorhinal cortex, two brain areas encoding the animal's location and head direction. We found that head direction signals were unaffected by home cage presence or translocation. Head direction cells remain globally stable and have similar properties inside and outside the embedded home. We did not observe egocentric bearing encoding of the home cage. However, grid cells were distorted in the presence of the home cage. While they did not globally remap, single firing fields were translocated toward the home. These effects appeared to be geometrical in nature rather than a home-specific distortion and were not dependent on explicit behavioral use of the home cage during a hoarding task. Our work suggests that medial entorhinal cortex and parasubiculum do not remap after embedding the home, but local changes in grid cell activity overrepresent the embedded space location and might contribute to navigation in complex environments.NEW & NOTEWORTHY Neural findings in the field of spatial navigation come mostly from an abstract approach that separates the animal from even a minimally biological context. In this article we embed the home cage of the rat in the environment to address some of the complexities of natural navigation. We find no explicit home cage representation. While both head direction cells and grid cells remain globally stable, we find that embedded spaces locally distort grid cells.

Keywords: grid cell; head direction; home; homing; navigation.

Fig. 1.

Home paradigm and tetrode recordings. A, top: schematic of the test environment with the home cage in the center; the animal’s home cage was modified such that not only could the lid be removed, but also two gaps in the sides of the cage could be opened. Walls are covered in complex cues. Bottom: we recorded 12- to 25-min sessions without removing the rat from the arena. B: histology of tetrode recordings in the medial entorhinal cortex (MEC) and the parasubiculum (PaS). Fluorescence microscopy of a tangential section of layer 2/3 border of the PaS and MEC. Red: calbindin (Cb); green: DAPI staining. The calbindin stripe clearly demarcates the end of the PaS and beginning of MEC. Tetrode tracks are observed as either holes in the section or highly DAPI-fluorescent tracks. Tracks of 4 tetrodes in the PaS are demarcated with asterisks, while a single track in the MEC is demarcated with a hashtag. D, dorsal; L, lateral; M, medial; V, ventral.

Fig. 2.

Head direction (HD) discharge is not affected by the embedded home. A: schematic of the conditions depicted for a head direction cell, open field (blue), home cage center (green), home cage moved (red). B: head direction rate polar plots (rate in corresponding color, angular occupancy in black). C: spikes of the head direction cell are clearly preferring a stable head direction. Black and gray, head direction of the animal in time (duplicated for visualization). Spikes are plotted on top in the corresponding RGB color scheme. D: head direction Rayleigh vector lengths are not affected by the presence of the home cage. Gray lines show individual cells. Means are depicted in black. Kolmogorov–Smirnov (KS) normality test, P = 0.34, t test, P = 0.96, N = 68. E: cumulative frequency function of the distributions of differences between preferred angle in the open field and in each home cage condition. Note that the inflection point is at 0 and the very steep slope. KS normality test for angle difference, P = 0.6993. F: correlation between the angular rates of both home conditions in relation to the open field. Cumulative frequency graph shows that correlations are distributed close to 1. G: head direction polar plots of a conjunctive grid cell show a consistent head directionality between sessions in open field (blue) and home center (green) conditions. H: cumulative distribution of preferred angle differences in conjunctive grid cells. KS normality test, P = 0.26, N = 39. I: head directional representation inside and outside of the home cage. Left: separation of spikes from a head direction cell into inside spikes (dark green) and outside spikes (bright green). Right: polar plot of head direction rates inside (dark green) and outside (bright green), and their corresponding head directional occupancy for outside (gray) and inside (black). J: cumulative frequency function of the distributions of differences between preferred angle inside and outside of the embedded space. Dark green corresponds to the embedded space in the center. Dark red corresponds to the embedded space rotated and on the side (KS normality test for angle difference, P = 0.013, N = 62).

Fig. 3.

The home cage does not induce egocentric home bearing discharges. A, top: representation of head direction angle. Bottom: representation of home direction angle. Note that in this case corresponds to the angle between the head direction vector and the vector pointing from the head of the animal to the center of the home (x). Hence, the animal facing the home retrieves a value of 0 and the animal running away from the home results in a value of 180 or −180. B: schematic of the conditions depicted for a nongrid and non-head direction cell; open field (blue), home center (green). C: polar plots of home direction rates (blue and green according to B, showing no clear home direction vectors). D: plot of the home directionality of the spikes for this cell shows no clear home direction preference. E: distribution of fictive home bearing vector lengths in the open field (calculated to the center of the arena) and the corresponding vector lengths once the home cage is placed at the center of the arena. Mean and SE are depicted in black showing no significant increase in vector length. Kolmogorov–Smirnov (KS) normality, P = 0.0037, signed-rank test, P = 0.21, N = 220. All cells not classified as grid cells or head direction cells). NS, not significant. F: the head direction vector (HDVector) of one head direction cell. G: note the lack of clear home bearing tuning curve for the same cell as in F. H: the distribution for all pure head direction (HD) cells of home bearing vector lengths is much smaller than for head direction. I: home bearing vector lengths of pure head direction cells did not change with the presence of the home. KS normality, P = 0.367, t test, P = 0.560, N = 68. Gray lines show individual cells. Means are depicted in black. J: home direction polar plot for a grid cell, showing lack of home bearing encoding. K: home bearing vector lengths for all grid cells are very low and did not change with the presence of the home. KS normality, P = 0.780, t test, P = 0.250, N = 54. Gray lines show individual cells. Means are depicted in black.

Fig. 4.

Weak egocentric bearing representation. A: distribution of egocentric bearing vector lengths with respect to the center of the arena for cells with significantly nonuniform egocentric home direction tuning curves. Only 4 significantly nonuniformly tuned cells are above the 0.3 cutoff for the open field condition (blue), 0 cells for the home center condition (green). B: multi-reference-point analysis for egocentric bearing direction. Egocentric bearing direction vector length maps for the cell in Fig. 3, in two conditions. We calculated egocentric bearing tuning vector length for each bin of space and computed the maximum vector length (MaxVL). C: distribution of maximum vector length values for all Rest cells [cells not classified as head direction cells, pure grid cells, or conjunctive grid cells (see text); n = 220]. Distributions show very few cells beyond cutoff of 0.3 vector length, even considering it is maximizing its module. D: position of reference point for MaxVL for all Rest cells (n = 220) in both conditions (blue: open field, green: home center). It can be noted that there is no apparent systematic bias in the location of the reference point, or a systematic bias for the center of the arena induced by the home. E: same plot as D for cells with MaxVL > 0.3 cutoff. The only cell with an egocentric bearing direction bias toward the position of the home is our example cell in B and in Fig. 3.

Fig. 5.

Single firing fields of grid cells shift toward home cage location. A: grid cell under 3 conditions: open field, home center, and home moved (blue, green, red respectively). Left: normalized rate maps. Right: spikes (RGB) superimposed on the rat trajectory (gray). B: composite plot of the spike positions for the three conditions of the cell in A. Note the change in positions for home center spikes. Bottom: composite rate map. Each rate map is normalized and assigned the corresponding channel in an RGB image. Side panel demonstrates the color compositions for the different RGB mixtures [cell recorded in dorsal medial the medial entorhinal cortex (MEC)]. C: two parasubicular grid cells, under the same 3 conditions as in B. Note how single fields move toward the position of the home. D: normalized firing rate increases in a region of the arena corresponding to the location of the home in comparison to the same area in the open field condition. ARB, arbitrary units. Kolmogorov–Smirnov normality, P = 0.280, t test, P = 0.0029, N = 51. Gray lines are individual cells. Mean and SE are depicted in black. E: the increase in normalized rate is evident in the Euclidian profile to the center of the home. Solid line corresponds to mean normalized rate, and shaded area correspond to the SE. Green: home center; blue: open field (t test for shaded area). F: the same effect is evident in the spatial averages of the peak normalized rates. Spatial averages of peak normalized rate maps for all cells in the open field (left) and home center (middle). On the right the difference between these two average maps is shown. G: brute force split between cells upmodulating their normalized firing rate in the home area (red, n = 32) and not-upmodulating (black, n = 19). H, top: same scheme as F for upmodulating cells from G, with the inclusion of the Euclidian profile to the left. Bottom: same scheme for downmodulating cells from G, with the inclusion of the Euclidian profile to the left (t test for shaded area). I: speed matching control of grid cell shifts. Two exemplary cells comparing unmatched and speed matched between home cage and open field conditions. Speed matching did not alter substantially the changes in grid pattern. J: lack of increase in normalized rate of Rest cells [cells not classified as head direction cells, pure grid cells, or conjunctive grid cells (see text)] in the Euclidian profile to the center of the home. Solid line corresponds to mean normalized rate, and shaded area correspond to the SE. Green: home center (HC); blue: open field (OF) (signed-rank test for shaded area).

Fig. 6.

Effect of home on grid cell population is driven by geometry. A: sliding window correlation analysis between conditions. Left: example of peak normalized rate maps of a grid cell in the open field (top row) and home north (middle) condition, and the corresponding sliding window correlation between these two (bottom). Right: same as left but for another cell compared with the home center condition. B: sliding window correlation analysis for all grid cells in both conditions. Note the lower average local correlation in the position of the home for both cases. C: inside embedded space correlation for the home cage center condition, and for an equivalent area (white dashed box in B). The presence of the home produces a lower correlation in comparison to an equivalent area that remains unchanged (signed-rank test, P < 0.001, N = 69). Black box corresponds to interquartile range, and the horizontal red line corresponds to the median. D: inside embedded space correlation for the home cage moved condition, and for an equivalent area (white dashed box in B). The presence of the home produces a lower correlation in comparison to an equivalent area that remains unchanged (signed-rank test, P < 0.001, N = 42). E: sliding window correlation analysis comparing the home versus a cardboard box with similar dimensions, showing that the effect is related to geometry not valence of the home. Left: the comparison between the open field and the cardboard box is shown. F: high correlation comparison between the home and the cardboard box. G: sliding window correlation analysis between the open field and a tall object in the center of the arena did not produce low correlations near the object. H: the presence of a linear corridor in the arena in different angles produced shifts of grid fields in some of the conditions further demonstrating the strong effect of internal geometry. The moving field is highlighted in pink.

Fig. 7.

Head direction discharge and grid cell activity is not strongly altered by a pellet hoarding task. A, left: behavior of the rat during pellet hoarding task, presenting high speeds and centralized home basing behavior. Middle: behavior of the rat in the open field while foraging for chocolate treats. Right: behavior of the rat with the embedded home cage, but while foraging for chocolate treats, behavior closely resembles open field behavior. B: home cage and pellets produces homing behavior in the rat. Rats spend time in the home and have outgoing trips to find pellets and incoming trips to hoard them in the home. Incoming trips are faster than outgoing trips. C: head direction rate polar plots (rate in corresponding color, angular occupancy in black). D: cumulative frequency function of the distributions of differences between preferred angle in the open field and in the pellet hoarding conditions. Note that the inflection point is at 0 and the very steep slope. Kolmogorov–Smirnov (KS) normality test for angle difference, P = 0.6993). E: spikes of the head direction cell are clearly preferring a stable head direction. Black and gray, head direction of the animal in time (duplicated for visualization). Spikes are plotted on top in the corresponding color scheme. F: head direction Rayleigh vector lengths are not systematically affected by the presence of the home cage during the execution of the hoarding task. KS normality test, P = 0.0.851, signed-rank test, P = 0.426, N = 20). Gray lines are individual cells. Means are depicted in black. G: inside home cage correlation comparison. The hoarding/home cage condition is compared on one had to the no task/open field condition and on the other hand to the no task/home cage condition. Example correlation analysis shows a higher correlation inside the home cage between the two conditions with the embedded space present. H: inside home cage area correlation for both comparisons in F. Showing high correlation between both conditions including the embedded space and lower correlation between the hoarding session and the open field (signed-rank test, P = 0.0039, N = 9).