Poster de conférence
Résumé : Visual statistical learning refers to the ability to detect and extract regularities from the environment. It allows preverbal infants to organise visual stimulation in a coherent representation. Thus far, statistical learning has been primarily investigated with post-exposure behavioural tasks that can only reveal the outcome of learning. Notably, behavioural tasks may lead to ambiguous interpretations since there is no clear consensus about the directionality of the expected learning outcome in infancy (i.e., novelty vs. familiarity effect). Electrophysiological measures such as steady-state visual evoked potentials (SSVEPs) can be acquired while learning occurs and can shed light onto the temporal course of learning. To date, similar investigations have been limited to the auditory domain (Choi et al., 2020). In this study, we propose an EEG frequency tagging approach to study infants’ SSVEPs in response to visual regularities. Participants were 4- to 6-month-old infants born at term. They were presented with continuous streams of 8 colourful shapes appearing in the centre of the screen at a frequency of 6 Hz. Infants were randomly assigned to one of three conditions: 1) the standard doublet condition, in which shapes were organised in doublets, 2) the control doublet condition, in which only the first element of the pair followed a rule, and 3) the single condition, in which individual shapes were randomly presented. We compared SSVEPs at the frequency of visual stimulation (6 Hz and its higher harmonics) and at the doublet frequency (3 Hz and its higher harmonics) across conditions. If the stream of shapes included visual regularities, we hypothesised not only a strong steady-state response at the base frequency of 6 Hz but also a progressive response at 3 Hz. Results revealed neural entrainment at the base frequency of visual stimulation (6 Hz and its harmonics) that did not differ across experimental conditions. This confirms that infants were similarly attending to the visual stream of stimuli in all conditions. On the other side, activity at the doublet presentation frequency (3 Hz and its harmonics) varied across conditions. Infants assigned to the doublet conditions showed greater responses at the doublet frequency harmonics, especially at 9 Hz, compared to the single condition. Overall, these results suggest that the infant brain can detect visual regularities in a stream of shapes from very early on. These findings are crucial to better understand learning mechanisms during stimulus exposure.