Abstract
Regularities in the environment effectively guide attention. Our previous work has shown that participants quickly learned multiple spatial distributions associated with specific search targets. Importantly, on a trial-by-trial basis these target-specific spatial distributions were activated to bias attention toward likely target locations. This work implies implicit statistical learning is constrained by top-down goals. Within this novel finding, it remains unclear what information underlies these object-location associations. Is it an individual stimulus or the stimulus category that is linked to the spatial distributions? In four experiments, we probed the representations of learned associations with visual and categorical cues. In the learning phase, participants learned spatial distributions of 16 exemplars from four categories through cued visual search tasks. Exemplars from the same category shared a spatial distribution including a high (80%) and low (20%) probability location. In the testing phase, targets appeared at either location with equal likelihood to ensure the results were driven by prior learning. We ensured the spatial biases remained for learned exemplars (LE), while also testing whether the learning transferred to novel exemplars from learned categories (LC). In Experiment 1-3, participants received only visual cues in the learning block, we found the learning from LEs transferred to LCs when participants received a visual cue. However, no evidence of learning or transfer was found when participants received verbal categorical cues in the testing phase. In Experiment 4, participants were trained with categorical cues in the learning block. In the testing block, these categorical cues triggered spatial biases. However, neither visual cues of LEs nor LCs triggered learned biases. This suggests that activation of the associations is strictly constrained by the cued information. Together, our results confirmed that people can acquire object-location associations through search experience, and demonstrated that stimulus categories drove the activation of spatial biases.