Abstract
Attention can be guided based on previous experience through spatial probability cueing, a selection history effect which is typically considered implicit and inflexible. However, a persistent attentional bias is incompatible with frequently changing goals in our typical daily environment. We recently proposed the goal-specific probability cueing, which suggests explicit attentional goals can unlock learned target-specific spatial attention maps. We train participants on four target items, each with a unique spatial bias, and find participants can learn to flexibly switch between these implicitly learned spatial maps on a trial-by-trial basis following a target cue. Our explanation suggests that these effects are driven by a multidimensional template, which includes both explicit target goal information and implicit spatial biases. One possible alternative explanation for our findings is simple associative learning between the target cue and a spatial location. According to this explanation, there is no multidimensional template because the target cues presented prior to search serve two functions: 1) generate a target template, 2) activate a spatial bias. To test this alternative explanation, we ran a typical training block of goal-specific probability cuing and added a testing block where we dissociated cues and search targets. The previously learned target cues were presented prior to each search trial, but the participants goal was to find a new target item which remained the same throughout the testing block. When participants’ goals changed during the testing block, we no longer found evidence of a spatial bias in RTs- demonstrating the implicit spatial biases are not caused by a simple association between the target cue and space. This provides the strongest evidence to date for goal-specific probability cueing. Multidimensional templates created in the goal-specific probability cueing effect rely on coordination of feature-based and location-based attention, utilizing information that crosses traditional boundaries between top-down control and selection history.