A Frame-Theoretic Model of Bayesian Category Learning

Abstract

Bayesian models of category learning typically assume that the most probable categories are those that group input stimuli together around a maximally optimal number of shared features. One potential weakness of such feature list approaches, however, is that it is unclear how to weight observed features to be more or less diagnostic for a given category. In this theoretically oriented paper, we develop a frame-theoretic model of Bayesian category learning that weights the diagnosticity of observed attribute values in terms of their position within the structure of a frame (formalised as distance from the frame’s central node). We argue that there are good grounds to further develop and empirically test frame-based learning models, because they have theoretical advantages over unweighted feature list models, and because frame structures provide a principled means of assigning weights to attribute values without appealing to supervised training data.

A Functional Contextual Account of Background Knowledge in Categorization: Implications for Artificial General Intelligence and Cognitive Accounts of General Knowledge

“…Here, a cascade is a combinations of frames in a tree, and category prototypes are structured within these knowledge trees. The frames mediate the input information and output behavior through a Bayesian inference model of category learning ( Taylor and Sutton, 2021 ). Frame-theoretic representations in the form of recursive attribute-value structures organized around a central node is clearly an improvement compared to simple feature list models ( Anderson, 1991 ; Sanborn et al, 2006 ; Goodman et al, 2008 ; Shafto et al, 2011 ).…”