The chain structure has not been easily inferred when three events are causally related like a chain (A causes B, and B in turn causes C). We hypothesized that focusing on the intermediate event make the chain structure more learnable, if the causal structure is locally computed (Fernbach & Sloman, 2009). In Experiment 1, focusing is manipulated by informing which event is important. Participants who were informed that the intermediate event (B) is important inferred the chain structure (61%) more than participants who were informed the terminal event (C) is important (20%) and participants who were not informed (21%). In Experiment 2, causal mechanism between B and C was provided to focus on the intermediate event, because cause is overestimated than the effect. Participants given BC mechanism inferred the chain structure (64%) more than the participants given AB mechanism (48%). Results of two experiments corroborated our hypothesis.I