Chunk decomposition is defined as a cognitive process which breaks up familiar items into several parts to reorganize them in an alternative approach. The present study investigated the effective connectivity of visual streams in chunk decomposition through dynamic causal modeling (DCM). The results revealed that chunk familiarity and perceptual tightness made a combined contribution to highlight not only the "what" and the "where" streams, but also the effective connectivity from the left inferior temporal gyrus to the left superior parietal lobule.