Heal (1986) and Goldman (1989) promptly replied to Dennett. Fair enough, if a system S tries to simulate the state of a radically different system Q (e.g., if a human being tries to simulate the state of a bridge), then S’s simulation must be guided by a theory. However, if a system S tries to simulate the state of a relevantly similar system S*, then S’s simulation can be entirely process-driven: to simulate the state which S* is in, S simply has to run in itself a process similar to the one S*