Figure 2.
Remapping of single cells in CA1 and mPFC across rules. A, Illustrative examples of spatial representations in CA1 and mPFC showing remapping on the common trajectories performed across the two rules, left to center or center to left. i, 2D place fields at stable performance of Rule 1 (top) and Rule 2 (bottom) in interleaved blocks (first block for each rule showed on the left and second block shown on the right; numbers in circles denote the order of blocks), with peak firing rate on top. ii, Linearized firing fields of the cells shown above (mean ± standard error). B, Illustrative examples showing firing rate change of neurons at the same well locations across the two rules, aligned to reward onset for the same common trajectory across rules (left to center or center to left). C, Correlation of single-cell firing patterns during individual trials to average firing patterns during stable performance of a given rule (>60th percentile of the performance of the day) during rule-switching blocks. Note the change in correlations that occur after rule switch. D, Proportion of remapped cells for each region. During running: CA1 27.3% (39/143, 32 Pyr, 7 Int), mPFC 25.8% (33/128, 26 Pyr, 7 Int), χ2 = 0.077, p = 0.78. At reward location: CA1 14.7% (21/143, 13 Pyr, 8 Int), mPFC 14.8% (19/128, 18 Pyr, 1 Int), χ2 = 0.0013, p = 0.97. Fisher's exact test for proportions of remapped cells between regions. E, Distribution of spatial information encoded by individual cells for each brain region. CA1, median 2.3 bits/s, n = 143; mPFC, median 1.3 bits/s, n = 128; VTA, median 0.23 bits/s, n = 53. CA1–mPFC p = 0.0010, mPFC–VTA p = 6.2 × 10−8, CA1–VTA p = 7.7 × 10−17, Kruskal–Wallis test with multiple-comparisons correction.