Distinct selection mechanisms for when predictions and rewards guide visual selective attention

Abstract

How does the brain combine information predictive of the value of a visually guided task (incentive value) with information predictive of where task relevant stimuli may occur (spatial certainty)? Human behavioural evidence indicates that these two predictions may be combined additively to bias visual selection (additive hypothesis), whereas neuroeconomic studies posit that they may be multiplicatively combined (expected value hypothesis). We sought to adjudicate between these two alternatives. Participants viewed two coloured placeholders that specified the potential value of correctly identifying an imminent letter target if it appeared in that placeholder. Then, prior to the targets presentation, an endogenous spatial cue was presented indicating the targets more likely location. Spatial cues were parametrically manipulated with regards to the information gained (in bits). Across two experiments, performance was better for targets appearing in high versus low value placeholders and better when targets appeared in validly cued locations. Interestingly, as shown with a Bayesian model selection approach, these effects did not interact, clearly supporting the additive hypothesis. Even when conditions were adjusted to increase the optimality of a multiplicative operation, support for it remained. These findings refute recent theories that expected value computations are the singular mechanism driving the deployment of endogenous spatial attention. Instead, incentive value and spatial certainty seem to act independently to influence visual selection.