Fig. 2: A subpopulation of CA1 cells remaps relative to reward.
From: A flexible hippocampal population code for experience relative to reward
a, Five example cells (mouse m4, ENV 2, day 14). Left panel: spatial activity per trial normalized (norm.) to the cell’s mean activity. White lines, reward zone starts. Right panel: trial-by-trial correlation (corr.) matrix. b, Example RR cells (orange labels) versus non-RR remapping cells (black labels) (four mice, replicated in n = 11 mice; see Extended Data Figs. 2 and 4). Top: mean-normalized spatial activity, original track axis. Bottom: trials after the switch (blue line) circularly shifted to align reward locations within (day 14) and across environments (day 8). c, Left: example peak activity per place cell (translucent dots, jittered for visualization; m12, day 14) before versus after the switch in original track coordinates. Middle-left: data rotated in periodic coordinates to align the starts of reward zones at zero. Middle-right: TR cells removed. Right: data compared to a random-remapping shuffle (red). d, Peak activity per remapping cell relative to reward (day 3, n = 1,082 cells, 11 mice). Orange shading, ≤50 cm (~0.698 radians) between relative peaks. e, Top: histogram of the circular difference (diff). between relative (rel.) peaks of remapping cells after minus before the switch (orthogonal to the unity line in d). Solid and dashed red lines, mean and upper 95% of the shuffle distribution. Bottom: distribution of mean RR firing position for cells with ≤50 cm between relative peaks (orange shading in d; n = 453 cells, 11 mice, non-zero mean, 0.441 radians; 95% confidence interval (lower, upper), (0.306, 0.575); circular mean test). Black dashed line, mean; blue shading, reward zone. f, Same as d but for the last switch (day 14, n = 1,153 cells, 11 mice). g, Same as e but for the last switch (day 14). Bottom: n = 645 cells with ≤50 cm between relative peaks (non-zero mean, 0.292 radians; 95% confidence interval, (0.205, 0.380); circular mean test). h, Above-chance fraction of RR remapping cells grows linearly with task experience. Dots, combined fraction across n = 11 mice; gray line and shading, best fit ± s.d. of linear regression. i, Above-chance fraction of RR remapping cells after exclusion of cells within each distance (±) from both reward zone starts.
