‘Eye Know’: Gaze reflects confidence in explicit predictions while relying on a distinct computational mechanism

Abstract

AbstractWe learn by continuously forming associations to predict future events. This learning is manifested in both explicit decisions and implicit sensorimotor behaviors. Despite significant advances in understanding each of these learning systems, their computational interplay remains unclear. We explored the relationship between explicit predictions and oculomotor expectations during associative learning in virtual reality, across an exploratory and two additional pre- registered experiments (Total N = 115). Participants’ explicit predictions about target location and their subsequent anticipatory gaze patterns both showed learning patterns. Moreover, gaze exhibited computational hallmarks of confidence in the explicit prediction, possibly reflecting an oculomotor confidence-like assessment. However, ocular and explicit learning also diverged significantly. Oculomotor learning exhibited reduced accuracy and metacognitive sensitivity relative to explicit responses. Oculomotor’s computational learning mechanism was characterized by more exploratory behavior, increased rule changes, and reduced perseverance. These findings suggest complementary learning processes for explicit and oculomotor systems that enable adaptive behavior.