Abstract for: Experts learn to disagree—and that’s not a bad thing

The different causes of expert disagreement offered in the literature have one thing in common: they are static explanations. While we do not refute any of the commonly known causes, we seek to offer an alternative dynamic explanation of how the process of decision making and learning from experience can cause within-cohort disagreement, as well as between-cohort disagreement in the form of growing bias towards the options with less feedback availability. We build upon the experiential learning theory to form a dynamical system model and investigate how asymmetrical (conditional upon decision) availability of feedback and sensitivity of outcome to skill interact to create path dependent and diverging opinions among the learning agents. In cases where objects of the experts’ decisions have agency over screening and choosing among experts, their choices contribute to path dependency, yet improves the total outcome of the practice. This improvement in outcome is not a result of diverse opinions contributing to innovation in open-ended situations. Instead, the customers’ active screening aligns the experts’ accumulating skill with the preferred process for the cases they experience. In contrast, top-down interventions to address disagreement end in decreasing the cumulative outcome of experts’ decisions, even though they successfully reduce disagreement.