Conceptual constraints on generating explanations
- Zachary Horne, School of Social and Behavioral Sciences, Arizona State University, Phoenix , Arizona, United States
- Sangeet Khemlani, Navy Center for Applied Research in Artificial Intelligence, Naval Research Laboratory, WASHINGTON, District of Columbia, United States
AbstractWhen reasoners explain everyday patterns and observations, they tend to generate explanations based on inherent properties of the observations (Cimpian & Salomon, 2014). Cimpian (2015) and his colleagues hypothesized that inherent properties permit rapid explanation, but the mechanism by which reasoners rapidly build explanations remains unclear. Any given concept may relate to innumerable inherent properties, and no theory explains how reasoners avoid protracted searches through semantic memory. Prasada and colleagues (2013) describe a novel conceptual framework that distinguishes between principled and statistical inherent properties. Here, we argue that the framework can resolve the predicted link between rapid explanation and the inherence bias. Two studies provide evidence that people systematically prefer principled inherent explanations. The finding allows for an integrated, mechanistic account of how reasoners generate explanations in which a preference for inherent explanations emerges from a preference for principled connections.
Return to previous page