We found that the vmPFC was modulated by signals related to the subject’s own reward probability in the Other task. Whole-brain analysis during the Other task identified BOLD signals in several brain regions, including the vmPFC (p < 0.05, corrected; Figure 4A), that were significantly modulated by the subject's reward probability (for the stimulus chosen by the subject) at the time of decision (Table 1). The subject's reward probability is the decision variable closest to their choices, as it is the farthest downstream in the hypothesized computational processes for
generating their choices, but it is also based on simulating the other’s decision-making processes, in particular, the simulated-other’s reward probability (Figure S1A). To determine whether the activation of the vmPFC that was significantly modulated by the subject’s reward
probability was compounded by, or possibly rather due to, the simulated-other’s Galunisertib reward probability, we conducted two additional whole-brain analyses: when the simulated-other’s reward probability (for the stimulus chosen by the subject) was added to the regression analysis as a potential confounder and when the regressor variable of the subject’s buy PLX3397 probability was first orthogonalized to the simulated-other’s reward probability and then included in the regression analysis together with the simulated-other’s reward probability. In both cases, vmPFC activation remained significantly modulated by the subject’s reward probability (p < 0.05, corrected). These results indicate that at the time of decision during the Other task, vmPFC activation was significantly modulated by the subject's reward probability. For comparison, the significant vmPFC signals related to the sRPE are also shown in Figure 4A. Here, we emphasize that the sRPE was not the subject's own reward prediction error (the difference
between the subject’s own outcome and his/her own reward probability) during the Other task. Indeed, no region was significantly activated by the subject’s own reward prediction error during the Other task. This observation was confirmed by an additional whole-brain analysis that was conducted in the same way as the original analysis, except that we added the regressor variable for the subject’s own reward prediction error and removed the regressors Astemizole for the sRPE and sAPE. Whole-brain analysis during the Control task revealed significant modulation of vmPFC activity (p < 0.05, corrected) by the reward probability (for the stimulus chosen by the subject) at the time of the decision and the reward prediction error at the time of the outcome (Figure 4B; Table 2). These activities remained significant (p < 0.05, corrected) when the following potential confounders were added to the analysis: the reward magnitude of the chosen stimulus with the reward probability and the value and reward probabilities of the chosen stimulus with the reward prediction error.