The role of entropy detection in judgments of similarity and difference of visual stimuli

Abstract

The ability to make judgments about degrees of similarity is fundamental to cognition, human or otherwise. Reasoning about the ways in which two things are similar allows people to make logical inferences, create and understand analogy, and organize the world into meaningful categories. Non-human animals are also sensitive to degree of similarity as evidenced by over 100 years of research into associative learning mechanisms. Entropy detection is an appealing explanation for non-human performance on same-different discrimination tasks because it is a zero-parameter model that does not rely on abstract concepts of "same" and "different." The research presented herein indicates that the choices humans make when judging the similarity or difference of visual stimuli, of varying complexity and spatial alignment, are consistent with perceived degree of similarity and of difference being functions of entropy. Additionally, this work demonstrates that one need not invoke the labels of "same" or "different" to prompt humans to sort items based on entropy or to use levels of entropy to construct higher-order relations of sameness and difference. These results offer a highly parsimonious account of human same-different discrimination because they are consistent with the animal literature and, therefore, suggest a conserved mechanism that is potentially involved in many aspects of both human and non-human animal cognition.