A Research On Individual Differences Of Face Recognition Ability Based On Multivariate Analysis Of DTI Brain Network

Face information is one of the most important visual stimuli in human evolution and social life.However,the processing mechanism of facial information in the brain is unknown.The investigation of the brain connectivity network supporting face recognition can help people to understand the processing mechanism of face information in the brain,and can be helpful to clinical researches and treatment of diseases accompanied by face recognition disorders.It can also promote the development of face recognition algorithm in machine vision,and therefore be of great significance to homeland security,public security and traffic security check.In the field of cognitive science,some researches have reported that there are two kinds of extreme cases in face recognition,namely,the participants with extraordinary face recognition ability and ones with severe deficit in face recognition(i.e.,prosopagnosia).Further studies suggested that there was also great individual difference in face recognition ability even among ordinary people.However,such individual difference was frequently neglected in exisiting studies.The present study aimed to explore the mechanism of face processing based on this individual difference in face recognition ability among people.In the field of neuroimaging science,several brain regions associated with face recognition have been identified,such as the fusiform face area(FFA),the occipital face area(OFA)and the temporal lobe.Further,it has been found that face recognition depended on a neural network distributed in the whole brain.According to this network theory,face processing can be decomposed into multiple cognitive processing steps,and therefore was supported by multiple functional regions that were interacted by each other.As a result,face recognition can be thought of a functional integration of multiple cognitive processings.Thus,the investigation of connectivity among brain regions of face processing based on network theory may provide direct neuroimaging evidence for the further studies of neural mechanism of face processing.The present study used Diffusion Tensor Imaging(DTI)to construct the structural network of face processing,and examined the relationship between the connectivity pattern of this network and the individual difference in face recognition ability among healthy participants.In the present study,the face recognition ability of the subjects was first measured through a series of face and object recognition tasks.Then the brain structural connection matrix of the participants was obtained through the DTI.Finally,the relationship between the participants' brain connection matrix and face recognition ability was explored.This final step included four sub-steps.The first was the selection of potential brain regions of face processing according to existing cognitive and neuroimaging evidence.Then the second step was the coarse selection of connection of the brain regions surviving from above step.Third,the elastic network regression was used to establish a face recognition ability prediction model based on brain structure network.Finally,a support vector machine(SVM)classifier was used to further validate effectiveness of face recognition brain network extracted at the third step.In summary,the study obtained an effective brain structure network for face recognition,and demonstrated that the left fusiform gyrus,some areas of the occipital lobe,the right temporal lobe and some areas of the limbic system are closely related to face recognition ability.These findings indicated that face processing brain regions found in traditional functional imaging studies played an important role in face recognition brain structure networks.Additionally,the present study also verified that face recognition information is processed in the brain as a brain network.The present study casted great lights on the transmition pattern of face information in the whole face processing network,and provided the important neuroimaging evidence for the investigation of the neural mechanism of face processing.