Abstract
How does selective attention navigate us to people-related information in the social world, for example, to find the same solitary person as a partner in a grand dancing party? According to the prevalent object-based attention theory, single person can be selected as a unit and guide attention, and this theory has received numerous supports in the past thirty years including both behavior and neurophysiological evidence. However, in real-world scenes, we need to detect more social information, such as whether people are involved in social interactions. We thus proposed a new concept of event-based attention that interacting social events can be the elementary unit of selective attention. We examined these two different theories using a paradigm modified from classic two-rectangle cuing paradigm. Observers looked at the four upright biological motions interacting in pairs (paired condition) or not (unpaired condition), and discriminate whether the probe was “T” or “L” after a cue on one agent (Experiment 1). Results indicated a “same-event advantage” only in paired condition: performance was better for probes presented on the same-event agent, compared to probes presented on the equidistant agent but involved in a different event. There was no such difference in inverted biological motions (Experiment 2), excluding any possible influence of low-level features. We thus concluded that interacting social events can serve as the elementary units of selective attention.