Schevenels, K. 1 , Vandermosten, M. 1 , De Smedt, B. 2 & Zink, I. . 1
1 Experimental Oto-Rhino-Laryngology, Department of Neurosciences, Faculty of Medicine, KU Leuven, Leuven, Belgium
2 Parenting and Special Education Research Unit, Faculty of Psychology and Educational Sciences, KU Leuven, Leuven, Belgium
Statistical learning enables us to extract regularities from our environment. This is crucial for first language acquisition in children and second language acquisition in young adults. Only few studies have investigated statistical learning in elderly and most of them show a decline in sensitivity to environmental regularities with age. Additionally, current evidence tends to show that statistical learning can appear independently in different modalities and for different kinds of information.
The first aim of this research was exploring the relationship between different statistical learning paradigms. To do this, we measured two types of statistical learning (conditional and distributional) in two modalities (auditory and visual) in 18 adults with a wide age range (aged 28-76). Principal component analysis (PCA) was used to explore the coherence between the different tasks. The second aim was finding a test sensitive and simple enough to tap the implicit learning ability of stroke patients. This population is interesting given the potential prognostic value of statistical learning for language recovery after stroke. Therefore, the test that best tapped the implicit learning ability of the control group was conducted in eight patients (aged 52-96) at four months post-stroke. Their results were related to different language measures and to the results of the control group.
The explorative PCA demonstrated that a person?s learning ability differs between modalities and between different types of statistical learning. This confirms the need to consider statistical learning as a collective term for different learning mechanisms rather than as a unified ability. The stroke patients failed to show significant learning on a visual conditional task, in contrast to the control group. However, their results on this task correlated negatively with language ability, possibly indicating a compensation mechanism. This research is an important step in determining the best-suited task tapping statistical learning in stroke patients.