Expert-Level Understanding of Social Scenes Requires Early Visual Experience. Naveh et al.
Description
We studied 28 late-sighted Ethiopian children who were born with bilateral cataracts and remained nearly-blind for years, recovering pattern-vision only at late childhood. This "natural experiment" offers a rare opportunity to assess the causal effect of early visual experience on later function acquisition. Here, we focus on vision-based understanding of human social-interactions. The late-sighted were poorer than typically-developing peers (albeit better than chance) in categorizing observed social scenes (as friendly or aggressive), irrespective of the display format (i.e. full-body videos, still images, or point-light displays). They were also impaired in recognizing single-person attributes which are useful for human interaction understanding (such as judging heading-direction based on biological-motion cues, or emotional states from body-posture gestures). Thus, comprehension of visually-observed socially-relevant actions and body gestures is impaired in the late-sighted. We conclude that early visual experience is necessary for developing the skills required for utilizing visual cues for social scene understanding. The dataset consists of the stimuli used for the experiments in the paper “Expert-Level Understanding of Social Scenes Requires Early Visual Experience”. The stimuli for experiments 1, 2 & 3 were animated based on motion capture data, obtained from three databases: mocap.cs.cmu.edu, created with funding from NSF EIA-0196217; motekentertainment.com; and the PLAViMoP database (https://plavimop.prd.fr/en/motions). Recorded trajectories were processed and retargeted on avatar models from the Mixamo dataset (https://www.mixamo.com), using the commercial software Autodesk Motion Builder (http://usa.autodesk.com), with different combinations of six avatars, downloaded from the Mixamo dataset (https://www.mixamo.com). The stimuli for experiment 4 are based on images from the BESST body postures pictures set (www.rub.de/neuropsy/BESST.html). The face in each image was obscured by a grey ellipse.