Abstraction in sequence learning
Ferenc Kemény, Ágnes J. Lukács · 2019
The goal of this chapter is to examine the role of abstraction in sequence learning by reviewing and integrating results from different definitions and paradigms, and identify the sources of possible discrepancies. Four different research streams are discussed, all of which test the capacity of the human mind to acquire abstract sequences. Abstract sequences are defined by characteristics that are not (necessarily) directly observable, and not bound to specific physical features. Topics of (1) transfer, (2) learning of category sequences, (3) task complexity related effects and (4) stimulus dependence are covered. The reviewed literature revolves around two important sequence learning tasks: Artificial Grammar Learning and the Serial Reaction Time Task. Based on findings from both tasks, we suggest that previous transfer studies suffer from serious methodological shortcomings, and that the presence or absence of abstraction can be a function of task requirements.