ABSTRACT
This chapter explores high-level connectionist models for everyday commonsense reasoning from a broad perspective, investigating their connections to existing paradigms. It discusses rules in general, and the formal logic accounts of reasoning as a special case of the rule-based approach. The chapter looks at some logics for dealing with non-monotonicity in reasoning. It also discusses fuzzy logic. The chapter also looks at probabilistic views of rule-based reasoning. Among the paradigms used in the existing accounts of commonsense reasoning, the rule-based paradigm is by far the most prominent. In this paradigm, generally, some generic, common syntactic forms that consist of antecedents and consequents are used, as a basic representational means. The probabilistic approach is a mixed symbolic/numeric approach for modelling reasoning. It treats beliefs as probabilistic events, and utilizes probabilistic laws for belief combinations. Therefore, in contrast to logics, it deals with inexact information.
