Jiyeong Ha

Jiyeong Ha

Yonsei University, Korea

Title: The effect of spatiotemporal continuity on face perception


Jiyeong is a second-year graduate student in Cognitive Neuroscience Lab (P.I., Do-Joon Yi) at Yonsei university. Her research focuses on viewpoint-invariant object recognition. Specifically, her work aims to explore the mid-level process between perception and cognition and to find what perceptual features are mostly used to form abstract representations of objects.


The visual system has little trouble recognizing a person even though the retinal image of that person’s face keeps chainging due to expression, illumination, distance, etc. One of the strategies to maintain such stable object representations is considering spatiotemporal continuity of the object with its visual properties. Spatiotemporally connected objects are likely to be considered as the same object despite having different physical appearances. Conversely, if one perceives spatiotemporally separated objects, it is natural to infer them as two different objects even if they look identical. We sought to investigate the effects of spatiotemporal continuity on facial perception.  In each trial, participants were instructed to observe two objects moving in opposite directions across the screen. The objects started as scrambled faces, and turned into intact faces when they reached a fixation. The two faces appeared sequentially. They were the same or different (repeated vs. unrepeated), and they appeared within a single stream or across two different streams (continuous vs. discontinuous). Once the movement ended, participants determined whether the two faces were the same or different. To investigate the effects of spatiotemporal continuity on face perception, we calculated hit and false alarm rates from the repeated and unrepeated conditions, respectively. The results showed that the sensitivity (d’) of the continuous condition was significantly lower than that of the discontinuous condition, but the response criteria (k) for the two conditions were not different. Also, hit rate for repeated faces were lower in the continuous condition than in the discontinuous condition, indicating that repeated faces were more difficult to detect in the same stream than in different streams. Such impaired face identification might be attributed to the failure of token separation under the influence of spatiotemporal continuity. The current findings suggest that spatiotemporal continuity might be utilized as a cue for stable face representations by modulating the process in which perceptual evidence is extracted.