Project 1. How the brain perception multidimensional biological motion (BM) attributes


Wang, R., Lu, X., Jiang, Y. (2023). Distributed and hierarchical neural encoding of multidimensional biological motion attributes in the human brain. Cereb Cortex.

Research background

Since the point-light biological motion (BM), which isolates human body movements, was introduced by Johansson (1973), a large body of research has revealed that the human visual system can efficiently extract and process distinct physical, biological, and social attributes from BM, such as facing direction, gender, and emotional state. Despite that the ability to recognize and evaluate different BM attributes has been well demonstrated at the behavioral level, it remains largely unknown how the brain simultaneously encodes distinct attributes from BM.


Figure 1.1 The stimulus of BM

What analysis did i do

1. performed multiple regression representational similarity analysis (RSA) and multi-voxel pattern analysis (MVPA) to identify the brain network in response to each BM attribute.
2. introduced the scrambled representational dissimilarity matrices (RDMs) in RSA to explore whether the brain areas identified by RSA were specific to the neural encoding of BM attributes.
3. utilized the Dice coefficient (DC) to quantify the dissimilarity between the two RSA results obtained before and after the scrambling procedure.
4. analyzed the correlations between behavioral results and neural responses by performing an RSA with behavior RDMs.
5. tested which regions are primarily involved in the integration process of the BM attributes using psychophysiological interaction (PPI) analysis.


Figure 1.2 Stimuli, task, and RDMs.



These results together demonstrated that the respective neural encoding of the facing direction, gender, and emotional state information embedded in BM stimuli involved considerably overlapping brain regions, which raised the possibility that the neural representations of the three attributes of BM might be shared and possibly interact with each other.


These results demonstrated that the neural encodings of all attributes were influenced by each other to some extent, that is, the neural encodings of the three attributes were recurrently interactive. Among the three attributes, the variations of their neural encodings, inversely related to the DC values, before and after scrambling manipulation were in descending order for facing direction, gender, and emotional state.


In the facing direction representation network, all clusters showed significant functional connections with others. Core nodes included the left lingual gyrus, right FG, right SPL, bilateral insula, and right IFG, which project and receive major functional connectivities. In the gender representation network, the increased connectivity was only found between the left FG and right pSTS, between the right pSTS and left MFG, and between the right lingual gyrus and right pSTS. In the emotional state representation network, most clusters exhibited significant functional connections with others. Core nodes included the left MOG, left precentral, and left MTG.

Co-author: Xiqian Lu, Dr. & Yi Jiang, Prof.
Copyright: Ruidi Wang.
Contact: [email protected] & [email protected]