People Accurately Predict the Shape of Skill Learning Curves

Abstract

Decades of research have shown that skill learning usually unfolds exponentially — people improve rapidly early on, and then performance levels off. Given how important expectations of learning are for actual learning, we explored whether people accurately intuit this canonical progression in skill learning. Across five pre-registered experiments (n = 375), we find that people correctly predict that skill learning on a novel visuomotor task follows an exponential decay function, both for a naïve player and themselves, before even engaging with the task. Importantly, people are sensitive to conditions that merit exponential learning and rationally only predict exponential learning curves when a player puts in effort and the task is not too difficult. Although people accurately represent the overall shape of their learning in our task, we find that they misestimate specific parameters, which may have negative downstream effects on learning outcomes.