The Information Processing Of Other’s Social Interaction Perception In Human Brain

Man is by nature a social animal.Recognizing and understanding other’s social interactions is an important cognitive function of human beings.In the past two decades,social cognitive neuroscience studies have mainly focused on low-level human recognition and high-level social attribute representation.For mid-level social interaction perception,most studies have only localized relevant brain regions but have not systematically explored how the human brain integrates cues to form social interaction perception from the information processing perspective.Recent technological advances have made it possible to investigate the neural mechanisms of social interaction perception processing.In terms of experimental paradigms,the development of virtual reality technology has enabled the manipulation of social interaction scenarios,providing a quantitative experimental paradigm for investigating social interaction perceptual processing;in terms of neural signal recording means,the recently well-recognized magnetoencephalography technique has high temporal and spatial resolution,making it possible to investigate the neural dynamics of multiple brain regions simultaneously.In this thesis,we explored the neural mechanisms of the human brain’s perception of social interactions of others based on three studies.Study 1 aimed to investigate the way the human brain processes and integrates information about social interactions.We constructed different social interaction scenarios by quantitatively manipulating the interpersonal distances and facing angles of two avatars,and recorded the neural representations of social interaction perception brain regions when human subjects observed these scenarios by functional magnetic resonance imaging technique.The results showed that the early visual cortex represented the interpersonal distance,regions of the ventral pathway represented the facing angle in some situations,and the posterior superior temporal sulcus represented social interaction judgments.The parallel processing hypothesis of social interaction perception was rejected since there were no brain areas represented quantitative changes in facing angle.We further compared the representation of facing angle at different interpersonal distance levels and found that the representation of facing angle in ventral pathway regions was modulated by interpersonal distance information.Besides,the representation of orientation cues in ventral pathway regions was negatively correlated with the representation of distance cues in the early visual cortex.Based on the results,we propose that the human brain’s social interaction perception is based on a hierarchical processing following the least effort principle: the human brain processes more informative interpersonal distance cues first,and then adjusts the representation of facing angle cues based on the information distance cues provided,to obtain sufficient information with least effort.Study 2 aimed to investigate the time course of social interaction perception in the human brain and the information transfer between brain regions.Based on a similar experimental paradigm,we used magnetoencephalography to record the neural dynamics of human subjects when observing social interaction scenes,and explored the representation of social interaction perception brain regions and the information transfer between brain regions using the sourcing algorithm and connectivity analysis.The results showed that interpersonal distances started to be significantly represented in early visual cortex around 50 ms,the modulation of distance information on the representation of orientation cues started to be significantly represented around 200 ms,while facing angles started to be significantly represented in the posterior superior temporal sulcus around 400 ms,supporting the hierarchical processing hypothesis of social interaction perception.Granger causality analysis results indicated that feedback signals were showed from the posterior temporal sulcus to other brain regions starting at about 120 ms,suggesting that hierarchical processing following the least effort principle may rely on top-down feedback modulations.Study 3 aimed to investigate how the posterior superior temporal sulcus,the key region identified in the above study,represented social interaction information based on different stimulus types and to examine whether its representation of social interactions has the ability to generalize across stimulus categories.We recorded the neural representations of the posterior superior temporal sulcus by functional magnetic resonance imaging when human subjects viewed social interaction scenes constructed from different categories of stimuli,then trained classifiers to decode the interaction states of the scenes based on these representations for each category,and applied these classifiers to decode social interaction scenes composed of other categories to test the generalization ability.The results showed that the posterior superior temporal sulcus was able to decode social interaction states for all stimulus types,and the transfer decoding accuracies were all significantly higher than the chance level,indicating that the posterior temporal sulcus represents social interaction information in an abstract way and is not limited to specific stimulus types.Taken together,we systematically reveal the hierarchical processing mechanism and the least effort computational principle for perceiving social interactions of others in the human brain,which makes up for the neglect on the neural mechanism of social interaction perception from the perspective of information processing.We emphasize the key role of the posterior superior temporal sulcus in social perception and cognition,providing a neural marker of people with social abilities deficits.In addition,the hierarchical processing mechanism and the least effort computational principle proposed in this thesis may shed light on constructing neural networks based on human intelligence.