The functional Magnetic Resonance Imaging (fMRI) data of both the ventral pathway and the dorsal pathway on the visual cortex in a classification task was analyzed. We found that the classification performance improved hierarchically from lower-level regions to higher-level regions in both pathways, which partly verified the visual pathway theory proposed in cognitive neuroscience. Moreover, the LO (Lateral Occipital), V3a and V3b fMRI data were good classification basis no worse than the widely-used features such as GIST, HOG and LBP. It indicated that imitating the activity patterns of visual cortex to design new feature-extraction algorithms might be favorable. Finally, the performance of V3a and V3b voxels were very close to that of LO voxels. Consequently, in the design of brain-like intelligence systems, we should consider the coordination mechanism between the two pathways rather than focusing on the ventral pathway alone. The relationship of human visual pathway and deep learning structure was also discussed tersely. Keywords