Sleep & statistical learning

Date & time

3–4pm 29 April 2013




Dr Joanne Arciuli, Senior Lecturer in the Faculty of Health Sciences, The University of Sydney

Individuals are able to implicitly detect statistical regularities present in their environment. This is known as statistical learning (SL). It is present in young infants and has been shown to operate across a variety of different types of stimuli (auditory, visual, sequential, spatial, adjacent dependencies, nonadjacent dependencies). It is thought that this type of learning may play a role in mental activities such as language processing and object recognition. There is growing interest in SL in relation to sleep.

Arciuli and Simpson (2012) found that adults’ ability to recognise stimuli presented during familiarisation (i.e., recognition of adjacent dependencies within visually presented material) was not affected by a 24 hour delay between familiarisation and testing. Gómez and colleagues (2006) examined nonadjacent dependency learning within auditorily presented stimuli in infants who napped between familiarisation and testing compared with infants who did not sleep. Results showed that the no-nap group preferred listening to familiar over unfamiliar trials, consistent with veridical memory of specific nonadjacent dependencies. Infants in the nap group did not distinguish between familiar and unfamiliar items suggesting that they had abstracted away from particular stimulus items. A study of infants by Hupbach et al. (2009) and of adults by Durrant et al. (2011) provide additional perspectives on the link between sleep and SL. Sleep may contribute to the abstraction of statistical regularities, perhaps by promoting greater flexibility in learning.

I will review this research and discuss the need for further research to untangle effects of age, modality of stimuli, type of sleep, and type of SL paradigm.

Updated:  22 July 2018/Responsible Officer:  Director, RSP/Page Contact:  Web Admin, RSP