I’m pleased to announce that our paper on the CAL model has been accepted for publication, at Psychological Review (lead author: René Schlegelmilch). You can read the preprint here.

CAL is a new formal theory of category learning, which assumes both rule and exemplar-like processes. In this regard, it has some similarities to previous accounts, such as RULEX, ATRIUM, and COVIS. The main thing that is new about CAL is its explanation of how rules are learned from scratch based on three central assumptions:

  1. Category rules emerge from two processes of stimulus generalization (similarity) and its inverse (contrast), operating on independent dimensions.

  2. Two attention mechanisms guide learning by focussing on rules, or on the contexts in which they produce errors.

  3. Knowing about these contexts inhibits executing the rule, without correcting it, and consequently leads to applying partial rules in different situations.

Among the category-learning phenomena explained by CAL are:

  • the Six Problems (Shepard et al., 1961), the effect of rule instructions thereon (Kurtz et al., 2013), and the observation that attention (as measured by eye gaze) shifts only after errors cease.

  • A2 vs. A1 training advantage on the 5-4 task (Medin & Schaffer, 1978)

  • the non-linearly-separable advantage (Medin & Schwanenflugel, 1981), and individual differences therein (Levering et al., 2019).

  • extrapolation of incomplete XOR (Conaway and Kurtz, 2017). Here’s an animated graphic showing CAL operating on this task.

  • peak shift, and individual differences therein (Lee et al., 2018).