This article describes recent work I am doing with Prof. Emily Jones and our colleagues at the Centre for Brain and Cognitive Development (CBCD, Birkbeck College, University of London). We are aiming to understand whether we can leverage individual differences in infant neural profiles to create personalised, inclusive social contexts for autistic people from as early as possible. In this article I discuss my PhD study looking at neural responses to social and non-social images in infants with and without autism, and present my current post-doc project to develop a method to observe infant brain responses in social contexts.
In our paper, which has been accepted for publication in a scientific journal (Translational Psychiatry) and will be out soon, we looked at 131 infants aged between 6 and 10 months (average 8 months) and measured their brain activity using a non-invasive neuroimaging technique called electroencephalography, or EEG. In the experiment, infants saw images of faces with a direct or averted gaze or a non-face image (created by scrambling the pixels of a face images) appearing on a screen. We recorded the subtlest changes in electrical activity of the underlying cortex from the scalp using net of small “channels” on the infant’s head (like in the figure below). We then measured Event-Related Potentials (ERPs), these are changes in the electrical field that occurred in specific scalp regions in response to the presentation of the each of the images.
For our analysis we were interested in the Negative Central (Nc) ERP. Nc typically occurs around 300 milliseconds after an image is presented and describes an increased negativity in the electric field over the fronto-central area of the scalp. Previous research has found that the more negative is the Nc, the higher the attention engagement in the infant. Most interestingly, a preliminary study showed that the Nc in response to faces is reduced in 6-month-old infants who later received a diagnosis of autism compared with infants who did not received a diagnosis of autism. Our first goal for the study was to see if we found the same in our larger study sample, because meaningful scientific findings are those that can be replicated. Second, we wanted to understand whether the infant neural activity in response to faces and non-faces recorded with EEG was predictive of 1) categorical autism (whether or not an infant received a diagnosis of autism at the age of three years) and 2) a score representing how well the child coped with social situations in everyday life at the age of three (socialization score), as reported by parents using the Vineland Adaptive Behaviour Scales questionnaire when the child was three. A low score indicates that children find social situations challenging. In our study the group of children with autism had a socialization score between 61 and 110, while neurotypical children had a socialization score between 94 and 122.
The infants took part in our experiment as part of the British Autism Study of Infant Siblings (BASIS, www.basisnetwork.org), a longitudinal study following up infants who had an older sibling with or without a community diagnosis of autism over multiple visits, including a diagnostic assessment at the age of three by research clinicians at King’s College London. Because the EEG experiment data was collected some years ago and these infants are now older than three, we could compare the ERPs of infants without a family history of autism with those of infants with a family history of autism who did or did not meet criteria for autism at the three-years visit, to see if there were differences between the different groups.
We found that: 1) when faces with a direct gaze are presented, infants later diagnosed with autism showed a slower Nc response, 2) the amplitude of the Nc was reduced in response to faces and enhanced in response to the non-face stimulus in infants with a family history of autism. Enhanced Nc amplitude (possibly indicating higher attention engagement) to the non-face compared to the face image was found in infants with lower socialisation score (i.e., parent-report difficulties in social contexts at the age of three).
We validated these findings using machine learning to test whether the individual infant’s pattern neural activity over the entire brain in response to the face/non-face images was predictive of autism. We found that shorter duration of the brain states reflecting attention engagement to faces with direct gaze predicted autism diagnosis. Moreover, a reduced strength of brain activation when looking at faces with direct gaze and an increased strength when looking at non-face image were predictive of difficulties in coping with social situations in everyday life (indicated by lower parent-report socialisation score).
Our results are in line with other studies suggesting that attention engagement for social and non-social stimuli might differ between autistic and neurotypical individuals and that these attentional profiles might be observed since infancy. To further investigate this, we started a new research study; the Behaviour and Online Neuroimaging to study the Development of Socialization (BONDS study). With this project we aim to understand the origins of neurodiversity by developing a machine-learning method to know, looking at the infant’s brain activity, what aspects of the outside world are more interesting to each individual infant. We use a wireless EEG cap to obtain the Nc or other neural signatures as measures of the individual’s engagement with a variety images, sounds, even characteristics of the other’s actions during social interaction. This way we can try to understand what are each unique infant’s “preferred” stimuli and use this information to create personalised motivating and positive contexts to facilitate the development of socialization.
For more about this study and/or updates, feel free to follow us: http://www.cbcd.bbk.ac.uk/research/bonds, twitter: @annagui86, @bondcbcd.