Abstract
Recent behavioral evidence suggests that the influence of semantic relationship between objects on attentional allocation is independent of task-relevance (Malcolm, Rattinger, & Shomstein, 2016). Additionally, neuroimaging data from our most recent work demonstrates that the semantic association between objects modulate spatial attention and strengthens the neural representation of objects in the early visual cortex. These results demonstrate that high-level relationships among objects continuously affect attentional selection. However, these findings are drawn from paradigms examining the influence of semantic relationships between two, maximum three, real-world objects. Since objects are always part of a scene, it is critical to understand whether the semantic properties of a scene have a continuous influence on attention. Here, we investigated the influence of task-irrelevant semantic properties of scenes on attentional allocation and the degree to which semantic relationships between scenes and objects interact. In Experiments 1 and 2, participants were presented with a scene followed by two objects appearing on either side of fixation. Next, two target Gabor patches appeared, one on fixation and one on top of an object, and a checkerboard distractor on the other object. Participants reported whether the orientation of two targets Gabors matched. Critically, only one of the objects was semantically related to the scene and the peripheral target was equally likely to appear on either object, rendering semantic relationships task-irrelevant. Faster RTs were observed for peripheral targets that appeared on the semantically related object. Importantly, RTs were directly predicted by the strength of scene-object relatedness (assessed by a questionnaire). In Experiment 3, the semantic relationship between objects and scene were systematically manipulated and measured using multi-dimensional scaling (MDS) methods. Results suggest that task-irrelevant semantic relationships between scenes and objects continuously influence attention and this influence is directly predicted by each individual’s perceived strength of semantic associations.
Acknowledgement: NSF BCS-1534823 to S. S.