01whole.pdf (6.56 MB)
Download file

Decoding hidden language: towards a reliable neural assessment of language comprehension in minimally-verbal autistic children

Download (6.56 MB)
posted on 28.03.2022, 19:25 by Selene Petit
Language is fundamental for cognition and social functioning, and something we often take for granted. In cases where people cannot use language to communicate, it can be difficult to evaluate cognitive abilities and the extent of spoken language understanding. In the case of minimally-verbal autistic people, accumulating evidence suggests a discrepancy between their receptive and productive language skills. In particular, it is becoming evident that at least some minimally-verbal children may understand more language than they can demonstrate. Because of the nature of their condition, including behavioural challenges and difficulties in social situations, autistic children often perform poorly on traditional measures of language such as standardised tests. Therefore, passive tests that do not require verbal answers may be more suitable. The aim of this thesis was to develop neural tests of language comprehension for minimally-verbal autistic children that do not require overt behavioural answers. Chapters 2 to 4 cover the development and validation of four paradigms in neurotypical children, and Chapter 5 presents the results of one of the paradigms applied to three minimally-verbal autistic children. In Chapter 1, I present the scope and the challenges of this work, and I review the relevant literature. In Chapter 2, I developed two child-friendly auditory paradigms that use spoken words and sentences to detect lexico-semantic processing in neurotypical children (N = 31), using electroencephalography (EEG). In this work, I also evaluated the data recording quality of a low-cost portable EEG system, Emotiv EPOC+, which provides an easy-to-setup and affordable EEG option. I compared two data analyses approaches: univariate and multivariate pattern analyses (MVPA). Three main findings emerged: first, there was large inter-individual variability in neural signals, with only around 50% of individuals showing a statically significant univariate effect. Second, the EPOC+ recorded similar signal to the research-grade system, when analysing the data with a univariate approach. Third, MVPA, which is robust to intra-individual differences in the time course and topology of effect, improved our reliability at the individual level to a maximum of 88% of children. This provided a promising avenue for a covert assessment of lexico-semantic processing in children. In Chapter 3, I extended this work and designed an updated paradigm in which I better controlled the stimuli, and added visual animations. I again found large inter-individual variability in the neural responses to semantic processing in children (N = 20), and the detection rate did not improve from Chapter 2. In Chapter 4, I turned to a novel paradigm that used functional transcranial Doppler ultrasound (fTCD) to assess mental task-following from the brain activity of neurotypical children (N = 20). I found that, for identical visual stimuli, children showed distinct lateralisation patterns when performing language and visuo-spatial memory tasks, with language being left-lateralised, and visuo-spatial memory being more bilateral. However, at the individual-subject level, I again found only half of participants showing statistically reliable task-following neural patterns. In Chapter 5, I selected the most promising paradigm (one of the EEG paradigms from Chapter 2) and tested three minimally-verbal autistic children (aged 5 to 15 years). I demonstrated that my techniques can be used with this population and I found evidence of language comprehension from one child's neural activity without requiring behavioural answers from him. Finally, in Chapter 6, I discuss the implications of these results for assessing language comprehension in minimally-verbal populations, and propose some future directions for this work. Together, this research provides a rigorous exploration of the methodological issues in using neuroimaging to assess cognition at the individual subject level, and an initial proof-of-concept that we can measure intact lexico-semantic processing in populations that may otherwise struggle to communicate -- summary.


Table of Contents

Chapter 1. Introduction -- Chapter 2. Towards an individualised neural assessment of receptive language in children -- Chapter 3. Even neurotypical children are heterogeneous: using multivariate decoding to improve individual subject analysis of lexico-semantic EEG data -- Chapter 4. Finding hidden treasures: a child-friendly neural test of task-following in individuals using functional transcranial Doppler ultrasound -- Chapter 5. Neural assessment of lexico-semantic processing in minimally-verbal autism: a pilot case-series -- Chapter 6. Discussion -- Appendix


Includes bibliographic references Thesis by publication.

Awarding Institution

Macquarie University

Degree Type

Thesis PhD


PhD, Macquarie University, Faculty of Human Sciences, Department of Cognitive Science

Department, Centre or School

Department of Cognitive Science

Year of Award


Principal Supervisor

Alexandra Woolgar

Additional Supervisor 1

Nicholas Badcock

Additional Supervisor 2

Liz Pellicano


Copyright Selene Petit 2019 Copyright disclaimer: http://mq.edu.au/library/copyright




1 online resource (xiii, 247 pages)

Former Identifiers

mq:72270 http://hdl.handle.net/1959.14/1283108