Extended Data Fig. 7: Task difficulty and behaviour variability do not drive the reduction in reward representation over sessions (days). | Nature

Extended Data Fig. 7: Task difficulty and behaviour variability do not drive the reduction in reward representation over sessions (days).

From: Predictive coding of reward in the hippocampus

Extended Data Fig. 7

a, The percentage of reward cells across sessions is shown. Grey points represent all sessions, while black points indicate sessions with a 2-second delay. Both groups exhibit a negative correlation over time. b, The percentage of reward cells is plotted against task performance. In both groups, the correlation is weak. c, Correlation values for both groups, across all mice (n = 7 mice), are shown. The data reveal a strong negative correlation between the percentage of reward cells and session number, and a weaker correlation with performance in both conditions. df, The same analysis as in ac is applied to reward information content (f: n = 7). The results similarly show a negative correlation between reward MI and session number, and no significant correlation with performance. These patterns are consistent across both groups: the full dataset and the subset of sessions with a 2-second delay. gn show that the slight variability of mice’s behaviour across sessions does not drive the reduction in reward representation over time. g, The percentage of reward cells is negatively correlated with session number. h, There is a weak correlation between the percentage of reward cells and the time interval between choice and reward. The time between choice and reward is used as a behavioural measure that quantifies how fast the mouse approaches the reward after poking the screen and selecting the correct choice. i, Correlation values across all mice (n = 7 mice) confirm a strong negative correlation of percentage of reward cells with session number and a weak correlation with behaviour. j, Modelling results indicate that session number explains a greater proportion of variance than latency time (n = 7 mice). kn, The same analysis as in gj is repeated for reward MI instead of the percentage of reward cells. These results similarly show that reward MI decreases over sessions and is weakly influenced by behaviour (m,n: n = 7 mouse). Bar graphs and error bars in inset of c,f,i,j,m,n show mean ± s.e.m.

Back to article page