a, While entropy is negatively related to value difference, there was no difference between the ACC-TUS and SHAM conditions (n = 18 sessions). A linear model was used and a one-sample t-test was performed on the resulting coefficients. b, Similarly, the negative relationship between cumulative stay and value difference was not different between the ACC-TUS and SHAM conditions. c, Maintain model’s parameters (left panel). There was no difference in temperature parameter between the fMRI, SHAM TUS and ACC-TUS sessions (right panel). A similar picture emerged for the learning rates. While the mean learning rate in the ACC-TUS session was higher than in the fMRI and SHAM sessions, this result was not significant. A linear regression model was used to determine the difference between conditions. d, Decision accuracy (selecting the option with the highest subjective value) plotted as a function of difficulty (difference in subjective value between the best (HV) and worst (LV) presented options). Subjective values were estimated using the RL model (Methods). Each bin contains data binned according to percentile, with each point corresponding to the [0–5%], [5–10%], [10–15%], …, [90–95%], [95–100%] of the value difference amplitudes. Accuracy is the rate at which the participant picked the subjectively better option. Because subjective value estimates are obtained by fitting a model to the data and reflect the animals’ actual patterns of decisions, the overall difference in performance levels is no longer apparent between the TUS and sham conditions. If the animal is poor at learning or choosing the best option to take, then this pattern of behavior will mean that the RL model will estimate that the animal’s subjective valuations of the choices are closer together than they objectively are. The RL model therefore ‘explains away’ differences in performances as differences in subjective valuations. However, once again it is clear that there was no evidence of a TUS-induced impairment in performance that increased with difficulty (smaller HV – LV value differences on the left of the figure). Solid lines are linear fits to the data and the shaded area is the 95% confidence interval. Two-sample t-tests were performed between the ACC-TUS and control sessions (n = 18 sessions for each group, non-significant).