Predictive modeling of optimism bias using gray matter cortical thickness

Abstract

Abstract Optimism bias (OB) is an expectancy bias, where people expect irrationally good future outcomes for themselves. Predictive modeling for OB would open new opportunities for estimating an overall state of well-being and understanding clinical conditions such as depression. To our knowledge, this is the first study attempting to address OB implementing a dedicated machine-learning based predictive modeling. We calculate people’s OB via a soccer paradigm, where participants rate their comparative chances for a successful outcome against their rival (i.e., personal OB) and a rival team (social OB). Later, using gray matter cortical thickness (CT) in a machine-learning framework, we predict both POB and SOB. Our results reveal a significant brain structure-based predictive model for experimentally assessed POB (Pearson’s r = 0.41, p = 0.006). Strongest predictors include left rostral and caudal ACC, right pars orbitalis and entorhinal cortex, all shown to have a role in OB before. Our confounder analysis suggests that the predictions are predominantly driven by CT measures and are not corrupted by demographic data (e.g., age and sex). There were no predictors recognized for estimating SOB. More of such predictive models on a large-scale data platform are needed, to help us understand positive psychology and individual well-being.