Learning to Reason: The Non-Monotonic Case

We suggest a new approach for the study of the non monotonici ty of human commonsense reasoning The two main premises that underlie this work are that commonsense reasoning is an inductive phe nomenon and that missing information in the in Icraclion of the agent wi th Lhe environment may be as informative for future interactions as observed information This intui t ion is lormahzed and the problem of reasoning f rom incomplete information is presented as a problem ol learning attribute func lions over a generalized domain We consider examples that illustrate various aspects of the non monotonicreasoning phenomena which have been used over the years as bench marks for various formalisms and translate them into Learn ing to Reason problems We demonstrate that these have concise representations over the generalized domain and prove that these representations Lan be learned efficiently The framework developed suggests an opera­ tional approach to studying reasoning that is ne\ ertheless rigorous and amenable to analysis We show that this approach efficiently supports reason­ ing wi th incomplete information and at the same lime matches our expectations of plausible patterns of reasoning in cases where other theories do not This work continues previous works in the Learn ing to Reason framework and supports the thesis that in order to develop a computational account for commonsense reasoning one should study the phenomena of learning and reasoning together