Object Discovery and Inverse Physical Reasoning

Abstract

Unobserved objects are typically discovered by making backward inferences from effects to causes. The inverse reasoning account proposes that inferences of this kind are carried out by postulating unobserved causes that best support the forward inference from causes to effects. We evaluated the inverse reasoning account by asking people to reason about hidden attractors and repellers that caused an observed particle to move about an arena. We found that people often evaluated specific hypotheses in a manner consistent with the inverse reasoning account but that hypothesis discovery involved processes that were inconsistent with inverse reasoning.