Extended Data Fig. 4: Simultaneous diversity within session.
From: Distributional reinforcement learning in prefrontal cortex
Four simultaneously recorded cells from the session with most reward-sensitive cells (9 in total), demonstrates there is diversity in optimism even within a session. Across cells, responses to middle value levels are both above and below the linear interpolation between lowest and highest values’ responses. Mean normalised firing is plotted for each of the 4 value levels. Error bars denote SEM. Firing rates are normalised such that responses to value 1 and 4 have mean firing rate 0 and 1, respectively. Normalisation allows comparison across cells of responses to middle value levels. Responses to value 2 across the 9 simultaneously recorded cells were significantly diverse; ANOVA rejected the null hypothesis that across cells the value 2 responses were drawn from the same mean (F(8,405) = 3.56, P = 0.0005). The same was true for responses to value 3 (F(8,441) = 2.16, P = 0.0291). This diversity was also present when including all cells in the analysis (value 2: F(40,1658) = 3.82, P = 2.74 × 10−14, and value 3: F(40,1842) = 4.73, P = 4.99 × 10−20), and in individual subjects (first animal: value 2; F(11,516) = 3.18, P = 0.0006, value 3; F(10,520) = 3.61, P = 0.0001; second animal: value 2; F(29,1142) = 3.92, P = 2.6 × 10−11 value 3; F(29,1322) = 5.47, P = 1.7 × 10−18).
