The two-part activity outlined in this paper reveals to undergraduate students that assumptions made in theory building remain unquestioned until one steps outside the initial realm of expectations, and that theories adopted have a demonstrable impact on behaviors. Part I defines a theory, describes the roles of assumptions and knowledge in theories, and provides a test to reveal that biases are involved whenever tests are made or taken. The test demonstrates that knowledge is needed to make useful theories and that sometimes the knowledge is based on assumptions that turn out not to be useful. Part II shows how a theory which is not perfect may be better than no theory at all, even if such a model leads to predictable and consistent errors. For example, when human behavior is predicted, it is useful to be right even if the theory is right for the wrong reasons. Further, even if there are gaps in the model, on the whole it may lead to better predictions than random guesses. Part II shows this by presenting a memory task twice. The first time students simply try to memorize a set of numbers in 20 seconds, while the second time they are guided by information that will lead to a "theory"; that is, they use a mnemonic device to aid their recall. (JDD)