Exploring the neuroanatomical model of visual word reading is one of the essential theoretical issue in cognitive neuroscience of language. Converging findings reveal that the dorsal and ventral routes are involved in reading neural network, but how they overlap and interact remains unclear and controversial. Using the Dynamic Causal Modeling (DCM) approach, the current project proposes two fMRI experiments to investigate the ventral and dorsal routes and their cooperation in Chinese character reading. Experiment 1, taking advantage of fMRI rapid adaption techniques, is designed to identify brain regions for cognitive components of reading processing. Both the GLM and DCM analysis are conducted to investigate the neural network, as well as two routes (the dorsal and ventral routes) for Chinese character reading. Experiment 2 further manipulates the stimulus properties (semantic and phonological cues) and task demands (symbolic, semantic, phonological detection task), to test the brain activation and connectivity among regions of reading neural network. By further comparing the DCMs including the same brain structures across three tasks, the different connectivity patterns will reveal how our brain regions in the same reading network interact for different types of stimuli or different processing demands. The combined results will test the idea of universal brain neural basis for cross languages reading. Also, it will provide theoretical guidance for empirical studies and applications, such as language teaching, treatment of reading disorders and clinical diagnoses.