Combining real-time analysis with machine learning to study neural correlates of social attention in infancy

Throm, Elena Veronika (2024) Combining real-time analysis with machine learning to study neural correlates of social attention in infancy. PhD thesis, Birkbeck, University of London.

Abstract

Autism is characterised by social behaviour difficulties which might be rooted in altered processing of social cues in early infancy. Enhanced synchronised oscillations in the theta band of the electroencephalography (EEG) associated with attention and learning are stronger during social compared to nonsocial live interaction in infancy, while this differential response is altered in older children with autism. The present thesis aimed to test whether theta power responses during live social interaction are altered in infants with elevated likelihood for autism, and to introduce and apply a novel experimental approach allowing to identify what are the aspects within social interaction that maximally trigger an elevated attentional state in the individual infant. Chapter 1 describes theory and empirical signs of social attention in infancy, with a focus on live social interaction. Chapter 2 argues how Neuroadaptive Bayesian Optimisation can be used with infant EEG data to reliably map an individual infant’s attention response across a wide range of social stimuli. Chapter 3 shows that theta power during live social compared to nonsocial behaviour of a social partner is already altered in infants with familial likelihood of autism, and Chapter 4 suggests that this difference emerges in the second half of the first year of life. Chapter 5 presents a proof-of-principle study testing the feasibility of Neuroadaptive Bayesian Optimisation with infant EEG data to study which among a range of familiar and unfamiliar faces presented on a screen maximally triggers attention responses in the individual infant. Chapter 6 applies this method to a live interaction setting to identify which aspect of live social interaction the individual infant maximally engages with, suggesting that individual differences in social behaviour and environment might play a role in what type of interaction an infant prefers. Implications of the present findings and future research directions are discussed.

Metadata

Statistics

DownloadsShow export options

Export as XMLJSONCSV

Activity Overview

6 month trend

169Downloads

6 month trend

35Hits

Additional statistics are available via IRStats2.

Archive Staff Only (login required)

Edit/View Item