Abstract
Selected visual information receives preferential processing when top-down attention is applied to sensory input. Multiple selection mechanisms can be deployed simultaneously, but the extent to which the unique influences of each combine to facilitate behavior remains unclear. We utilized an integrated pe-cue containing both location and color information that reliably predicted a target in a sparse display to investigate possible interactions between space-based (SBA) and feature-based (FBA) attention effects. Previously, we observed little additive perceptual benefit when both were cued simultaneously in a sparse display. The contribution from FBA was restricted to non-perceptual decision-making processes where it depended on the presence of SBA, while SBA influences additionally facilitated target enhancement (Liang & Scolari, VSS 2018). Here, we examined whether differences in the reliability of each pre-cue component modulates the presence and magnitude of attention effects and their interactions across perceptual decision-making processes. Across two experiments, we independently manipulated the proportion of trial-by-trial pre-cues that contained valid space and/or feature information. Regardless, SBA and FBA effects in accuracy were additive, indicative of independent mechanisms within target enhancement, consistent with previous work. Estimated drift rates produced by a simple drift diffusion model exhibited a similar pattern, such that the speed with which target information was accumulated in response to one dimension of the cue did not depend on the other. However, we also observed super-additive relationships between SBA and FBA within other decision-making components, again replicating our previous findings, and these depended on cue reliability. Non-decision time (proportion of RT outside of evidence accumulation) was reduced and responses were more conservative for the most likely pre-cue combination—even in scenarios where the most likely combination included an invalid component. These data indicate a higher-order dependency between selection mechanisms outside of signal enhancement, where decision-making processes may track cue reliability.