In this chapter, I reconsider the standard models of “rationality” in economics, asking whether subjects' decisions, often judged in various contexts to be irrational from the standard perspective, might not be judged differently when viewed from a different perspective, including that of the subjects and the environment to which they have become adapted. A divergence between the predictions of a theory and the results of tests of the theory implies that either the theory or the experimental test is inadequate. Just as the experimentalist can get it wrong in designing a test, and returns to the drawing board when testing problems or suspected design flaws show up in the first experiments, so the theorist can get it wrong in deciding what assumptions, criterion of action, and constraints are most relevant for the decision problem faced by the individual. Science is about reducing error – deviations between observation and prediction – and the two instruments of science are experiment and theory (a third instrument is the technology, or machines, of experiment, which I discuss in Chapter 13); reducing error necessarily involves being prepared to adjust either or both of these instruments. If the data are persistent and robust to procedural and contextual issues and imply that the behavior stems from ecologically rational actions, we are well advised to reevaluate our constructivist model of rational action.