Pavlidou, E. . 1, 2 , Pugh, K. 2 , Gumkowski, N. . 2 & Williams, J. 1
1 The University of Edinburgh
2 Haskins Laboratories
Children were trained and tested on a visual (i.e. unfamiliar shapes) artificial grammar learning (AGL) task, which pitted abstract learning against stimulus-specific learning: we measured children's sensitivity to novel sequences to test whether they would preferentially learn the specific features of the itemsor the rules that generated those items. Analysis (e.g.repeated measures ANOVAs,tests&correlations) indicated that typically developing children show learning of both abstract as well as stimulus specific information. Subsequently, to test whether this type of learning is subject to modality constrains,children were tested on an auditory and tactile analogues of the visual task. Analysis showed that children are sensitive to abstract (verylikely in the form of rules) as well as specific (in the form of chunks) information in both auditory and tactile tasks. There was no correlation of performance across modalities: one would expect that if implicit learning is a domain general process served by a single mechanism then performance should be correlated. However, new perspectives on implicit learning(see Frost et al., 2015) embodymodality-specific constraints within a domain general approach of implicit statistical learning. We will discuss the way our data fit current theories ofimplicit learning and enhance our understanding on developmental effects of this type of learning. The European Commission via the Marie Sklodowska-Curie Outgoing Fellowships scheme funds this work.