Abstract
While much research has shown that attention enhances perceptual processing, it is unclear how unattended and irrelevant information influences perceptual representations of attended objects. Here we asked how ensemble processing of a set of attended faces is affected by other, irrelevant face ensembles presented at unattended locations. Participants monitored a stream of images that contained faces and noise patches on one side of the display (e.g., right visual field) and reported the average emotion of the faces at the end of each trial. In Experiment 1 (N=20) we used a set of faces that continuously varied in emotional units from happy to sad (Haberman & Whitney, 2007), and presented a set of four faces in each stream. Another stream of images was presented on the unattended, task-irrelevant side (e.g., left visual field) and participants were instructed to ignore it. Critically, the average emotion of these irrelevant faces was systematically varied such that it would either match the average emotion of the attended faces, or be on average happier or sadder than the attended set. We found that the perceived ensemble was biased towards the average emotion of the irrelevant face set, such that if happier faces were presented in the unattended stream, participants were biased to perceive the attended faces as happier, and vice versa (p=0.007). Similar biases towards the irrelevant face set were observed in Experiment 2 (N=25; p=0.005), in which we used a circular face space with 360 faces that varied in emotional units from happy to sad to angry. Together, these data indicate that percepts of objects within the spatial focus of attention are not isolated from the rest of the visual scene, but are instead influenced by irrelevant inputs, at least those that match the current attentional template.