Generating Saccades for Reducing Uncertainty: Cognitive and Sensorimotor Trade-Offs

In uncertain environments, eye movements inspect relevant stimuli that reduce uncertainty, but the mechanisms of this process are not well understood. How do we trade off the benefits of reducing uncertainty against the costs of planning and executing saccades? To examine this question, we designed a task in which participants attempted to estimate the relative probabilities of two possible world-states, and used saccades to inspect a set of visual stimuli that had different predictive accuracy (diagnosticity) regarding those states. Diagnosticity was explained to participants in advance and indicated by stimulus color. The time available for inspection was limited, so that participants could only inspect a subset of the available stimuli. Preliminary results together with computational modelling suggest that participants traded off stimulus diagnosticity against eccentricity (saccade size) and planned saccade sequences rather than individual saccades. Future task versions and computational models will examine how these trade-offs depend on uncertainty, shedding light on the sensorimotor mechanisms that the brain uses to sample visual stimuli to reduce uncertainty.