It is necessary now to explain how the optimal risk bonus scaling itself was calculated. We simulated, for every trial, all unique decision sequences, each associated RG7420 purchase with a different risk bonus scale by calculating their modified values and using the aforementioned decision rule (Figure S3). For every unique decision sequence, generated with our value modification model, we could compute an end of block expected value. We defined the optimal risk bonus scaling as the risk bonus scale, which led to the decision sequence with the highest end of block value. It is important to note that,

when doing so, we took into account that all net outcomes that fell short of the target value had a value of 0. Although we do not assume that participants

were able to track the exact optimal risk bonus scaling, it served as an approximation of how the values of specific choices should be modified as a result of the context on a given trial. Task parameters were chosen to maximize its parametric range. It is, furthermore, possible to calculate the risk bonus scale that leads to the point of equivalence for a given pair of options. In other words, at an optimal risk bonus scaling equal or above this value for an option pair, the riskier option should be preferred: equation(5) equivalenceriskbonus_scale=(MS×PS−MR×PR)/(MR×(1−PR)−MS×(1−PS))orequivalenceriskbonus_scale=(MS×PS−MR×PR)/((MR−MS)−(MR×PR−MS×PS)),where Electron transport chain MR, MS, PR, and PS refer to OSI-906 cell line the reward magnitudes associated with the riskier and safer options and reward probabilities

associated with the riskier and safer options, respectively. By computing this value for all remaining decisions and rank-ordering decisions from the least to the most risky, we could estimate the value of all unique decision sequences and select the one that led to the highest end of block value. In all neural and behavioral analyses, the risk bonus scale used is, therefore, equal to the optimal risk bonus scaling in a given trial, i.e., the risk bonus scale that generates a sequence of future decisions that would lead to the highest expected value at the end of the block, taking into account the current context (risk pressure) and future prospects (set of options left and the pair presented). The optimal risk bonus scaling is, therefore, a contextual parameter reflecting the degree of bias toward riskier choices that is optimal for a given context and applies to both options in a trial in the same way. The option bonus becomes larger for riskier choices, compared to safer choices, as the optimal risk bonus scaling increases, reflecting the riskier choices’ increased utility for reaching the target.