Eye-tracking reveals robust attentional filtering in an object-based attention task

We use object-based attention in our daily lives to process task-relevant objects, and sometimes we must ignore task-irrelevant objects, even if they are salient, dynamic, or situated in front of an attended object. We have previously shown the efficacy of eye-tracking at predicting the focus of object-based attention when participants must attend to one of two naturalistic objects (face, flower) that follow pseudorandom, minimally correlated trajectories while remaining partially overlapping (Pidaparthi & Tong, VSS 2023). Although smooth pursuit eye movements are considered by some to be strongly stimulus-driven, here we asked, how effectively can attention filter out the presence of a task-irrelevant object, as indexed by eye movements? To answer this question, we adapted our paradigm across two experiments. In Experiment 1, subjects were presented with either one or two moving objects and were instructed to respond whenever the task-relevant stimulus underwent brief spatial distortions (2 conditions: attend-face, attend-flower). We then evaluated the selectivity of object-based attentional filtering by using a sliding window correlation analysis to compare gaze trajectories with the attended stimulus trajectories. Notably, even with the overlapping sets of motion signals, pursuit eye movements were not perturbed by the irrelevant motion: observers could follow one attended object in the presence of the distractor object (mean r=0.581) just as accurately as a single object alone (mean r=0.579). In Experiment 2, we replaced the irrelevant object (e.g., flower during attend-face trials) with a moving Gabor stimulus that underwent random bursts of drifting motion (at 4 Hz for 500ms), and measured the extent to which this strong low-level motion signal influenced eye movements. In both instances, we found that observers can selectively attend to the task-relevant object such that gaze-following is unperturbed by extraneous motion signals, demonstrating the robustness of attentional filtering with respect to the eye movement system.