Empirical test of a formal framework of forgetting

Forgetting is a fundamental cognitive process that allows to efficiently discard outdated and irrelevant knowledge and manage information for humans and artificial systems alike. Originating from the latter, Beierle et al. (2019) propose a formal framework for modeling forgetting operators from a common-sense point of view. In this article, we aim for bridging the gap by investigating the formal forgetting operators with respect to their effects on human cognition utilizing an abstract experimental paradigm – the Counting Game. Using participants’ accuracy and response time in rule-based tasks, we examine if and how the forgetting operators affect human performance in response to rule manipulations. We found that the addition, contraction and revision of a rule have the most remarkable effect on performance. Therefore, we modeled them using features describing the rule system and current scenario to predict the change in accuracy they prompt. By analyzing the effect that the operators have on humans when used to manipulate rule systems, we make a step towards further modeling the formal aspect of changes to rule systems in applications in order to understand the potential effect they have on the users.