Neural Mechanism Of The Binocular Disparity Processing Based On The Encoding And Decoding Method

Vision is a main way for human to understand the physical world.Binocular disparity is one of the important visual cues for human brain to estimate the depth structure of the environment.Therefore,exploring the neural mechanism of disparity processing is essential for understanding human visual system.Functional magnetic resonance imaging(f MRI)technology provides an effective method for human brain studies.Generalized linear model(GLM)and multi-voxel pattern analysis(MVPA)are conventional encoding and decoding analysis methods for f MRI data.The two methods have been widely used in some f MRI-based human brain studies and achieved good results.With the development of f MRI study and computer technology,a voxel-wise encoding model is proposed.The model can describe the mapping between specific features of stimuli and voxel responses and be used for f MRI data encoding analysis and decoding analysis.Most previous f MRI studies of binocular disparity used conventional encoding and decoding analysis methods to reveal brain areas that were involed in disparity processing and had discriminative power to disparity.However,the previous studies still exist the following problems.(1)The neural mechanism of the human visual areas in the disparity sign(crossed/uncrossed disparity)processing is still not clear.(2)The disparity-tuning characteristics in human brain have not been revealed yet.(3)The neural mechanism of the human brain processing stereo image with disparity is still not clear.For the above three problems,three innovative studies are conducted.1.The first study explores the neural mechanism of disparity signs processing in human brain.For this purpose,a f MRI experiment is designed.GLM and MVPA are used to analyze experimental data and the brain areas that are engaged in disparity sign were located.The results indicate that the higher dorsal and ventral areas are closely related to the disparity sign processing.2.The second study explores the neural mechanism of disparity-tuning in human brain.For this purpose,a f MRI experiment is designed.According to the characteristics of experiment,the fine-coarse voxel-wise encoding model based on Gabor function(FCM)is constructed.Compared with uncrossed-crossed and single voxel-wise encoding model based on Gabor function(UCM and SM),FCM can better describe the disparity-tuning characteristics of voxel.The results indicate that the higher dorsal visual area V3 A complex show the higher identification power for disparity levels by FCM.3.The third study explores the neural mechanism of stereo image processing in human brain.For this purpose,a f MRI experiment is designed.According to the characteristics of experiment,a voxel-wise encoding model based on lasso regression is constructed and 3D features can be used as input to the model.The results indicate that the primary visual area V1 and higher visual areas(h MT+/V5 and LO)show the higher identification power for stereo image by the model based on lasso regression and 3D features with disparity information can help these visual areas identify stereo images.