14), except for a marginally significant increase in model-free control in session 3 compared to session 1 (p = 0.04). Model-based control is thought to depend on a number of processes including prefrontal working memory (WM) capacity. Given that studies of WM report lateralized functionality (e.g., Mull and Seyal, 2001), we asked whether the magnitude of a TBS effect might be related to WM

capacity. To examine such interindividual differences, we could not use the population parameter estimates obtained through the regression. Instead, we extracted the numerical magnitude of the main effect of reward, the reward-by-transition interaction, and the difference between the two from each subject’s average stay probability in each of U0126 chemical structure the four reward/transition conditions in each stimulation condition. We first asked whether model-free or model-based control independently correlated with WM in any of the three stimulation conditions. Only the magnitude of the reward-by-transition interaction, inferred as model-based control, correlated with WM after disruption to left dlPFC (r = check details 0.45, p = 0.02; all other p > 0.10). We then correlated the balance between the two systems in all stimulation conditions with WM. Strikingly, only behavior after disruption of left dlPFC was WM dependent (Figure 3; vertex, r = 0.09, p = 0.68; left dlPFC r = 0.53, p = 0.006; right dlPFC, r = −0.05, p = 0.80). Pairwise permutation

tests revealed that the correlation was significantly more positive in left compared to right dlPFC (105 permutations, (-)-p-Bromotetramisole Oxalate p = 0.009), marginally more positive in left dlPFC compared to vertex (p = 0.06), and not significantly different between right dlPFC and

vertex (p = 0.52). Taken together, these data show that the effect of left dlPFC disruption on the balance between model-based and model-free control depends on WM capacity, with high WM participants retaining more model-based control compared to those with low WM. The balance between model-based and model-free control is often framed as a competition between a flexible, forward-looking system and a simpler retrospective stimulus-response-based system (Daw et al., 2005). Our results show that the balance between these two systems can be causally manipulated in the human brain by a disruption to prefrontal cortex. Our data suggest that TBS to right dlPFC impairs a key node in a network that underpins model-based control (cf. Gläscher et al., 2010 and Killcross and Coutureau, 2003). We further show an involvement of left dlPFC in model-based control that is related to individual differences in working memory, suggesting differential roles for left and right dlPFC in the functional architecture underlying deliberative choice. Animal lesion and human imaging work suggest that sectors of prefrontal cortex are involved in high-level cognition and decision making (Miller and Cohen, 2001).