Mechanisms Involved in Risk-Taking During a Contingency Change in the Balloon Analogue Risk Task

Introduction:Research using the Balloon Analogue Risk Task (BART) has primarily interpreted behavior as individual differences in risk preference. However, because participants learn explosion probabilities from experience, performance also reflects latent mechanisms of learning risk contingencies. We studied these mechanisms using a BART in which learned risk contingencies reversed.Methods:In an online sample of healthy adults (n=118, 63 female), we implemented the BART with three balloon colors of different risk levels. After 90 trials, color-risk associations of low and high-risk balloons reversed; afterwards participants completed a single-color low-risk control condition. We fit data from the low-risk balloon, which showed the most behavioral variance, to three candidate hierarchical models separately for each task phase. We then explored whether schizotypy and psychotic-like experiences were linked to BART performance and model parameters.Results:Adjusted pumps were not affected by the reversal, nor were they linked with subclinical traits. The best-performing Exponential Weight Mean Variance (EWMV) model showed a lower inverse temperature (i.e., less systematic responding) after the reversal and a faster shift from prior weighting to accumulated evidence across trials. In exploratory analyses, higher delusional ideation and positive schizotypy were associated with lower inverse temperature before the reversal; positive and overall schizotypy were associated with greater aversion to outcome variance.Discussion:Computational modeling revealed latent processes not visible in pumping behavior alone. Choice stochasticity according to EWMV was higher when multiple contingencies had to be learned and highest after the reversal, though task design limits causal interpretation. Trait associations were modest and should be interpreted cautiously.