Is there a difference between understanding and a simulation of understanding?
Thought experiment: Suppose there is a computer program that "understands" Chinese: it takes as input a stream of Chinese text and produces as output another stream of Chinese text sufficient to pass the Turing test. Now suppose John Searle is placed in a room with an English version of the computer program and enough pencils and paper to work. Chinese text comes through a slot in the door, John applies the rules of the program manually and sends Chinese text out.
Questions: Is there a difference between the computer's execution of the program and John's? Does the computer program "understand" Chinese any more than John does?
This argument is intended as a refutation of functionalist theories of mind.