The Role of Statistical Learning in Attentional Guidance During Search Through Naturalistic Scenes

Associations between objects and locations can guide attention in visual search. Previous research has found that relationships between items and their location (e.g., cup on the counter, or sponge near the sink) could be used to guide the control of attention in naturalistic scenes. These associations are built over an individual’s lifespan based on their own personal experiences, so it is difficult to understand the nature of how these associations have been formed. Previous research has found that different forms of memory, including knowledge concerning semantic relationships between objects and implicit learning, play a role in guiding attention. The present study looks to more closely examine how learning-dependent attentional guidance can occur in naturalistic scenes. Participants were assigned four target shapes, each of which was associated with one of three locations in a scene (floor, counter, or wall). While being eyetracked, on each trial participants were cued with one of the four target shapes and instructed to search for it in the following scene, responding with a keypress based on the orientation of a letter appearing within the target. The target shape appeared in its associated location within the scene on 75% of all trails and in an unassociated location on the remaining 25%, along with three distractor shapes. To require use of the cue, on half of all trials, one of the three non-target distractors was an uncued shape from the target set. Results show that response time was faster on valid trials than on invalid trials, and the first shape fixated was more likely to be the target on valid trials. These results suggest that associations between an arbitrary shape and a location can be used to guide search, providing evidence for the role of statistical learning in attentional guidance during search through naturalistic scenes.