Abstract: PURPOSE: Serial dependence refers to an attractive perceptual bias whereby the perception of current visual features can be pulled toward recently exposed features. For example, a direct eye gaze is more likely to be perceived as leftward-facing if a face with extreme leftward eye gaze is recently presented. Serial dependence helps us obtain a stable perception in a noisy environment. Previous research has demonstrated this effect in various aspects of perception, such as orientation, emotion, and facial attractiveness. However, it remains largely unknown whether the serial dependence effects could occur between stimuli from different categories. The current study aimed to investigate whether cross-category serial dependence could exist in social attention by measuring the perception of eye gaze direction and the walking direction of biological motion (BM) stimuli.
METHODS: The discrimination task of eye gaze direction and the “inducer” paradigm were used in all three experiments. Experiment 1 (N = 25) examined whether task-irrelevant inducer stimuli involving information of social attention (i.e., BM stimuli with a leftward or rightward walking direction) could pose an influence on judgments of eye gaze direction (i.e., whether cross-category serial dependence between judgments of BM direction and eye gaze direction could emerge). In each trial, the inducer stimulus (i.e., a BM stimulus) was presented in the center of the screen for 250 msec, followed by a 300 msec mask stimulus. Next, the probe stimulus randomly selected from one of seven levels ranging from 40% leftward to 40% rightward was presented in the center of the screen for 200 msec, followed by a 300 msec mask stimulus. When a question mark appeared in the center of the screen, participants were instructed to press the ‘up’ or ‘down’ keys to indicate which one of the reference stimulus and the probe stimulus was perceived to be more leftward or rightward. Experiment 2 (N = 12) investigated whether the results of Experiment 1 can be replicated when faces with a leftward or rightward eye gaze instead of the BM stimulus were used as inducer stimuli. Experiment 3 (N = 10) investigated whether the cross-category serial dependence could emerge when non-social stimuli (i.e., arrows pointing leftward or rightward) were used as inducer stimuli. The procedure of Experiment 2 and Experiment 3 were identical to Experiment 1, except that different inducer stimuli were used.
RESULTS: The results of Experiment 1 found that after seeing a leftward/rightward walking direction stimulus (i.e., a BM stimulus), participants were more inclined to perceive the gaze direction of the reference stimulus as being more leftward/rightward, and vice versa. There results indicated that the BM stimuli had a cross-category attractive effect on the eye gaze stimuli, making the gaze direction of the reference stimuli becoming closer to the direction of BM stimuli. Similarly, in Experiment 2, where the inducer stimuli were replaced with leftward and rightward eye gaze stimuli, we also observed the attractive effect of the inducer stimuli on the reference stimuli. However, in the Experiment 3 where the inducer stimuli were changed to non-social stimuli (i.e., arrows pointing leftward or rightward), no serial dependence effect was found.
CONCLUSIONS: In sum, different categories of social attention stimuli, rather than non-social attention stimuli, can generate attractive effects. We conclude that cross-category serial dependence can arise when the preceding and current stimuli belong to same social attention stimuli but did not arise between non-social attention stimuli and social attention stimuli.

Key words: serial dependence, social attention, eye gaze, biological motion