Some recent studies suggest that it is fairly easy for a human to trust, or even obey, a robot. Maybe a bit too easy. Is the trust and obedience misplaced? Media Equation Theory suggest that we treat the robot in a way we treat humans, since humans are one of the very few social beings around us. If so, a case can be made for using robotics for doing some serious experimental cognitive science: create many different robots with many different cognitive abilities, so that we can create a richer landscape of cognition that will help us better understand these new cognitive agents around us, and socially interact with them accordingly.