Abstract
Visual attention can be guided by goal-directed and stimulus-driven attentional control. An important yet unresolved question is how these two sources of guidance interact to determine attentional priority. We hypothesized four models. First, goal-directed and stimulus-driven control are simultaneously integrated on a common priority map to guide attention (coactivation model). Second, the two forms of guidance are independent mechanisms of control and race toward a solution (separate-activations model). Third, the two forms of guidance have different time courses of influence, with stimulus-driven control implemented before goal-directed control (sequential model). Finally, stimulus-driven and goal-directed control compete before the computation of priority, with only the stronger influencing the allocation of attention (competition model). A redundancy-gains paradigm was used. On each trial, participants searched a circular array of colored shapes for the presence of either a goal-directed feature, a stimulus-driven feature, or both (redundant). The goal-directed feature was an item matching a shape cue presented immediately before the search array. The stimulus-driven feature was a color singleton. The redundant target matched the shape cue and was also a color singleton. We assessed whether the RT distribution in the redundant condition contained a substantial proportion of trials that were faster than could have been generated by the faster of two individual guidance processes operating independently in parallel: i.e., violation of the race model inequality (RMI). Only the coactivation model predicts this effect. Robust violations of the RMI were observed, consistent with the coactivation model but inconsistent with the other three models. In addition, violations of the RMI were eliminated when redundant features appeared at different locations, indicating that coactive guidance is implemented on a spatially organized priority map. In summary, the present research demonstrates that goal-directed and stimulus-driven modes of guidance converge to jointly influence a common, spatially organized representation of attentional priority.