I assume one would try to "prove" Church's thesis by arguing that any physically realizable machine (under laws of physics that have some form) can be simulated by a Turing machine to arbitrary precision, and hence any computation it can do, a Turing machine could do as well. Alex