Evidence supports the contention that humans find it costly to carry out some sorts of strategies. Such costs are unlikely to be observed directly, but various models have been proposed to represent them. We discuss a method for testing the empirical worth of these models.