Fig. 5. Area 47/12o TUS does not alter GRS impact on behavior or neural activity.

(A) Hypotheses: 47/12o TUS (red) should not disrupt GRS representations (illustrated as in Fig. 1E). In addition, no effect of aPFC TUS (green) is expected in comparison to the sham control (blue). (B) The influence of the past reward history (abscissa, highlighted) on which choice the animal will take next in the sham condition (blue), after 47/12o TUS, and after aPFC TUS. (C) The same analysis is visualized as a histogram. (D) The overall influence that GRS has on learning was not different across experimental conditions. This was demonstrated by examining W_RT [the weight exerted by GRS when prediction error (PE) learning occurred]. When W_RT is multiplied by the learning rate (how much animals learn on each trial), then this reveals the overall influence that GRS has on learning and this is what is plotted here. (E) In sham, parametric variation in GRS was associated with variation in neural activity amygdala, operculum but most prominently IA. There was no change in this activity after either 47/12o TUS or aPFC TUS. (F) Effects extracted from IA region shown in inset are shown for comparison of sham and 47/12o TUS and comparison of sham and aPFC TUS. As in Fig. 3, separate panels illustrate contrast of sham and 47/12o TUS (i) and sham and aPFC TUS (ii) because small differences in model fitting resulted in small differences in effect size estimates in the sham group depending on the comparison group when fitting was done (however, careful fitting to either sham versus 47/12o or sham versus aPFC guarded against the possibility that any changes in activity that might have been found could have been a consequence of a poor model fit when TUS was applied).