Plausible Generalization: Extending a Model of Human Plausible Reasoning
Mark H. Burstein, Allan Collins, Michelle Baker · Journal of the Learning Sciences · 1991
Transcripts of people answering questions or carrying on dialogues about everyday matters are filled with plausible inferences - inferences that are not certain, but that make sense. The same patterns of inferences occur in many different contexts. Often, in forming these inferences, people make generalizations that are equally uncertain but nevertheless are useful guides to reasoning. This article describes some important extensions to our earlier description of a core theory of plausible reasoning, based in large part on a new protocol study. The extensions are both data driven and theory driven. The primary focus here is on the inductive inference patterns people use to form plausible generalizations, that is, weakly held beliefs based on few examples but annotated with the same forms of certainty and similarity information that supported the inferential patterns described in our earlier work. We also provide examples of qualitative reasoning with inequalities and extend our formalism to cover that type of reasoning.