Perception and simulation during concept learning.

Abstract

A key component of humans' striking creativity in solving problems is our ability to construct novel descriptions to help us characterize novel concepts. Bongard problems (BPs), which challenge the problem solver to come up with a rule for distinguishing visual scenes that fall into two categories, provide an elegant test of this ability. BPs are challenging for both human and machine category learners because only a handful of example scenes are presented for each category, and they often require the open-ended creation of new descriptions. A new type of BP called physical Bongard problems (PBPs) is introduced, which requires solvers to perceive and predict the physical spatial dynamics implicit in the depicted scenes. The perceiving and testing hypotheses on structures (PATHS) computational model, which can solve many PBPs, is presented and compared to human performance on the same problems. PATHS and humans are similarly affected by the ordering of scenes within a PBP. Spatially or temporally juxtaposing similar (relative to dissimilar) scenes promotes category learning when the scenes belong to different categories but hinders learning when the similar scenes belong to the same category. The core theoretical commitments of PATHS, which we believe to also exemplify open-ended human category learning, are (a) the continual perception of new scene descriptions over the course of category learning; (b) the context-dependent nature of that perceptual process, in which the perceived scenes establish the context for the perception of subsequent scenes; (c) hypothesis construction by combining descriptions into explicit rules; and (d) bidirectional interactions between perceiving new aspects of scenes and constructing hypotheses for the rule that distinguishes categories.