Fig 1. Experimental paradigm and lesion reconstruction.

(a) Participants made a series of choices between one ‘safer’ gamble and one ‘riskier’ gamble. On each trial, participants viewed a new pair of gambles. On some trials, participants made choices alone (Solo trials). On other trials, they were asked to make choices after observing two other players’ choices (Info trials). The Info trials had three different types based on the composition of the two other players’ choices: ‘safe, safe,’ ‘risky, risky,’ and ‘mix.’ Four trial types (Solo, Info: ‘safe’, Info: ‘risky’, and Info: ‘mix’) were intermixed. Reconstructions of (b) dACC (N = 6) and (c) insula (N = 10) lesions are shown. The color bar represents the degree of lesion overlap among patients. To show the lesion overlap across participants, all lesions are shown overlaid on one hemisphere. (d) Power utility model fit for Solo trials as measured by negative log likelihood (−LL; lower values indicate a better fit) shows that individuals with insula or dACC lesions had significantly worse model fit than non-lesion control participants (NCs; P = 0.0072; NC vs insula: P = 0.032; NC vs dACC: P = 0.073), suggesting disrupted utility-based risky decision-making in participants with insula or dACC lesions; see also S1 Fig for model-agnostic data consistent with disrupted sensitivity to risk in lesion participants).