Summary: ASSERTIONAL AND BEHAVIORAL APPROACHES TO
Uri Abraham, Ben-Gurion University, Beer-Sheva, Israel
We compare two proofs of the mutual-exclusion property of the well known critical
section algorithm of Peterson: an assertional proof and a behavioral one. The accepted
view is that behavioral proofs are informal and are, for some intrinsic reason, error prone.
We try to present a different view and to outline a framework within which the behavioral
approach can be formalized in a way that keeps the intuitive content of the behavioral
Behavioral reasoning is useful because it gives insight, but it is prone to errors.
A. U. Shankar, An Introduction to Assertional Reasoning, 1993.
Whereas the invariantassertional method is well defined, and its formal development is a
well established field of pure and applied information science, the term behavioral reasoning
(or operational reasoning) is only loosely used to denote a certain pre-formal activity which
belongs, some would say, more to psychology than to computer science. Without denying the
proven industrial and theoretical value of the assertional reasoning, I argue that this strong and
successful scientific activity has neglected the need to develop the behavioral approach, not as
a substitute to the assertional method, but as a way to explain algorithmic ideas and to develop