On a Philosophy Bites podcast I heard Daniel Dennett mention the following thought experiment, which he attributed to Galileo. Suppose that heavier objects fall faster than light ones. Take two objects, A and B, where A is lighter than B. Connect A and B with a string and drop them. Since A is lighter than B, A will act as a drag on B, and B will fall more slowly than it would have alone. Yet since A and B are jointly heavier than B, and heavier objects fall faster than light ones, B will fall faster than it would have alone. We have a contradiction. Therefore, heavier objects do not fall faster than light ones. I thought that this was really marvelous and also very surprising. I had been under the impression that one could not arrive at ostensibly substantive empirical claims like the one in question just by considering thought experiments. I was hoping that one of the panelists could explain exactly how Galileo's thought experiment works here.

One way of thinking of this kind of thing is that it uncovers a subtle contradiction between views we already hold. But it doesn't, by itself, prove anything positive. It's conceivable that one could deny, for example, that A and B are "connected" in the right way for them to fall faster together.

The question how thought experiments work in science is an important one, and philosophers and historians of science have written on it extensively. One good place to start might be Thomas Kuhn's paper, "A Function for Thought Experiments", which discusses this one, and also Einstein's famous thought experiment involving the moving train. There's a nice You Tube version of that one here.

Check out
the SEP article on thought experiments, as well.

Read another response by Richard Heck
Read another response about Physics