Since rationality is a normative ideal, it is difficult to see how a theory of rationality might be subjected to empirical evaluation. This paper explores various aspects of this problem in relation to the work of L. J. Cohen, Amos Tversky and Daviel Kahneman, Ellery EeUs, Isaac Levi, and Henry Kyburg. Special consideration is given to its significance for testing systems of inductive logic.
If rationality is only a normative ideal, then it is difficult to imagine how theories of rationality can be empirically tested so as to be confirmed or disconfirmed. How we should think and act is constrained only in the loosest sense by the ways in which we do act and think. On the other hand it is entirely possible to describe the patterns of thought and action manifested by human populations which these populations characterize as rational, and even to give an explanatory account of why these patterns of thought and action are manifested. Hence it is possible to test the empirical adequacy of theories of this second sort.
Two kinds of theories of rationality are distinguishable. The first kind provides an answer to the question:
(1) How should we act or think?
while the second kind provides an answer to the question:
(2) How do we act or think?
Received doctrine states that while theories of the first kind are not accepted or rejected on the basis of experimentally determined data, theories of the second kind can be so accepted or rejected. Is there any way in which normative theories of rationality are subject to empirical testing? I claim not only that this is possible but desirable. I claim that this is the case when normative theories are conceived of as answers to the question:
(3) How should we think about how we think?