The role of modal and amodal representations in action planning

Abstract

In the context of the Theory of Event Coding (TEC), researchers found evidence for and against abstraction in the sense of generalization in response-effect (R-E) learning. Against the background of the research unit FOR2718, concerned with the overarching topic of modal and amodal cognition, the present work set out to test with a series of experiments whether abstraction can be found in R-E learning experiments 1. in the sense of generalization from exemplars to the corresponding categories (Experiment 1), 2. in the sense of an abstract representation of spatial concepts (Experiment 2), and 3. in the sense of generalization from a picture to the corresponding semantic meaning (Experiment 3). No signs of generalization were found in Experiments 1 and 2, whereas the effect obtained in Experiment 3 could also have occurred because of phonological recoding and thus be confounded. Kernel density plots calculated for Experiment 1 - 3 show a bimodal distribution in all respective control groups. Considering recent literature (Sun et al., 2020), this could point to a propositional nature of the learned knowledge rather than to automatically acquired knowledge as commonly assumed by research in context of TEC. Considering the latter, numerous studies on R-E learning report rather small effect sizes. Experiment 4 was designed to see, if the average effect size can be increased by turning the action effects task-relevant during the acquisition phase task. The results show that this is indeed the case, and the kernel density plot shows that the distribution remains bimodal even with larger effects.