A Decidable Confluence Test for Cognitive Models in ACT-R

摘要

Computational cognitive modeling investigates human cognition by buildingdetailed computational models for cognitive processes. Adaptive Control ofThought - Rational (ACT-R) is a rule-based cognitive architecture that offers awidely employed framework to build such models. There is a sound and completeembedding of ACT-R in Constraint Handling Rules (CHR). Therefore analysistechniques from CHR can be used to reason about computational properties ofACT-R models. For example, confluence is the property that a program yields thesame result for the same input regardless of the rules that are applied. In ACT-R models, there are often cognitive processes that should always yieldthe same result while others e.g. implement strategies to solve a problem thatcould yield different results. In this paper, a decidable confluence criterionfor ACT-R is presented. It allows to identify ACT-R rules that are notconfluent. Thereby, the modeler can check if his model has the desiredbehavior. The sound and complete translation of ACT-R to CHR from prior work is used tocome up with a suitable invariant-based confluence criterion from the CHRliterature. Proper invariants for translated ACT-R models are identified andproven to be decidable. The presented method coincides with confluence of theoriginal ACT-R models.