Some Theories are Unfalsifiable

Abstract

Trafimow (2009) claims that there are no unfalsifiable theories: To test any theory, one must make auxiliary assumptions, and with sufficient creativity about auxiliary assumptions one could always arrive at reasonably risky predictions. We argue that a prediction from a theory plus a given set of auxiliary assumptions will not be risky for the theory when the initial level of confidence in the theory is greater than the initial level of confidence in one or more of the auxiliary assumptions. Some theories, we claim, are so basic that initial confidence in these theories will be greater than initial confidence in any set of auxiliary assumptions with which they could be tested, and such theories are unfalsifiable. We illustrate this with a principle that forms part of the theory of reasoned action.