Reasoning with partial knowledge
Abstract
We investigate how sociological argumentation differs from classical first-order logic. We focus on theories about age dependence of organizational mortality. The overall pattern of argument does not comply with the classical monotonicity principle: Adding premises overturns conclusions in an argument. The cause of nonmonotonicity is the need to derive conclusions from partial knowledge. We identify metaprinciples that appear to guide the observed sociological argumentation patterns, and we formalize a semantics to represent them. This semantics yields a new kind of logical consequence relation. We demonstrate that this new logic can reproduce the results of informal sociological theorizing and lead to new insights. It allows us to unify existing theory fragments, and it paves the way toward a complete classical theory. Observed inferential patterns which seem “wrong” according to one notion of inference might just as well signal that the speaker is engaged in correct execution of another style of reasoning. —Johan van Benthem (1996)