We present a new approach for revising reference theories, based on the well-known idea by Quine and others that even a simple scientific hypothesis, test, or observation rests on numerous assumptions. In our proposal, by examining some of the key underlying assumptions of existing theories, IS scholars can revise these theories and add more accurate explanations. We use Rational Choice Theory (RCT) and a cybersecurity example to illustrate how the same general RCT explanation can be further refined into two distinct RCT theories, each with specific explanations and predictions. The objection that no one knows or tests all assumptions, while true, is mitigated by focusing on those assumptions that are explanatory (E) and predictively (P) necessary for the theory. Such sacrosanct assumptions, which are E or P necessary, play a crucial role in determining whether something qualifies as a new variant of that theory.