Causal networks or causal islands? The representation of mechanisms and the transitivity of causal judgment