Parent Steering: Determine how social partners shape the development of interaction 

The aim of Jasmine's project was to analyse how mothers promote the development of joint-attention skills in infants and to identify the most successful strategies. Joint attention refers to the set of skills infants use to coordinate their attention with that of another person in relation to a mutually interesting property, object, event or third person in the environment. This is a pivotal skill in the early social-communicative development of infants; from the second half of the first year of life onwards, mothers drive social interaction bids with infants and exert influences in promoting joint attention. Traditionally, gaze-following was considered the dominant pathway parents used to establish joint attention with their children. However, evidence shows that joint attention is achieved through a combination of verbal and non-verbal cues interacting in systematic ways. Nonetheless, existing studies examining multimodal pathways to joint attention focus on the co-occurrence of a limited set of domains, using methods like lag-sequential analysis. Yet, alternative computational methods, such as temporal pattern analysis (T-pattern), provide a multivariate approach which may better capture the complexity and temporal dynamics of human behaviour.

The current study therefore aimed to apply T-pattern analysis to examine maternal patterns of behaviour leading up to and sustaining parent-initiated bouts of joint attention, among a preliminary sample of mothers and their infants at elevated likelihood for autism. The project collected mother-child interaction videos of children with an older sibling with ASD and prematurely born children (gestational age < 30 weeks). Videos of children at the ages of 10 and 14 months were collected by the beginning of 2019. Mother-infant interactions were coded for parent-initiated bouts of joint attention, infant-directed speech, touch, deictic and representational gestures. T-pattern analysis was applied to coded interactions to detect recurring patterns of behaviour leading up to and sustaining episodes of joint attention.

Jasmine travelled to Iceland to work with Prof. Jonsson and Prof. Magnusson and learned sequential pattern analysis theory and how to best apply the technique to the data. The secondment lasted 3-weeks which was shorter than initially planned (3-months) due to travelling during the COVID-19 pandemic. Findings demonstrated that maternal behaviour both eliciting and sustaining joint attention exhibits an organised and complex interactive structure of recurring temporal sequences. Linguistic speech represented the highest relative contribution to detected T-patterns – with the most complex patterns involving linguistic speech combined with representational and deictic gestures. Moreover, recurring T-patterns that sustain joint attention were more frequent and showcased a wider range of dimensional pathways and event types, compared to those eliciting joint attention. These findings demonstrate a proof-of-concept that T-pattern analysis is particularly effective in uncovering underlying structures of behaviour during parent-infant interactions. By homing in on subtle differences in interactive behaviour on an individual level, potential challenges in social communication could be highlighted. This may offer parents an individualised approach to fine-tune their interactions to support learning.

The work of Jasmine has put efforts into understanding early social features that may indicate neurodivergent developmental trajectories in children who have an elevated likelihood of developing autism. The collaboration with Arianna Zanatta (ESR 10) has further provide research evidence on the link between brain development and Naturalistic Developmental Behavioural interventions in children with an elevated likelihood of developing autism. Arianna Zanatta and Jasmine Siew have recently submitted a manuscript in June 2023 entitled:

Zanatta, A., Siew, J., van der Paelt, S., Warreyn, P. & Roeyers, H. (2023). Developmental, behavioural and NDBI interventions in autistic children or at elevated likelihood of autism: a systematic review of neural outcomes”.  Submitted for review June 2023: Journal of Autism and Developmental Disorders. 

This work has gathered evidence from previous studies that targeting social-communicative outcomes may have a positive impact on brain development of autistic children and those at elevated likelihood of autism. One example is therapists working in the Flemish home guidance services for autism have been trained to fidelity in Project ImPACT, an evidence-based parent-mediated intervention for improving social and communicative skills in young autistic children. This work has the potential to improve children’s life and their well-being. 

During her PhD, Elena investigated individual differences in brain responses to social cues in infancy. To do this, she first analysed existing datasets of infant neuroimaging data and then collected new data adopting a novel experimental approach. Using data from a large-scale longitudinal study including infants with and without familial likelihood for autism and/or ADHD, she used linear mixed models (using R) to look at associations between electroencephalography (EEG), eye tracking, and parent-report measures of infant behaviour to find out whether the group with elevated likelihood showed altered neural activity at different timepoints in infancy in the context of live social interaction. She found that infants with elevated likelihood for autism significantly differed in their brain response to social vs nonsocial live actions of an experimenter. She also found that looking behaviour to social vs nonsocial live actions did not differ, suggesting that early alterations can be more sensitively captured by neural measures. Finally, she found that responses were not altered in infants with elevated likelihood for ADHD, suggesting that the alteration is not a common signature of neurodevelopmental conditions but specific to autism. 

In the second part of her project, she adopted a novel experimental framework for studying the coupling of early social behaviours and functional brain development on the level of the individual infant. In close cooperation with Pedro (ESR13), she combined for the real-time analysis of infant neuroimaging data with a novel method in machine learning (Neuroadaptive Bayesian Optimisation, NBO). This approach allows efficient identification of the individual’s optimal stimulus from a large stimulus space, by refining its search based on the individual’s response to the presented stimulus. Elena optimised real-time analysis pipelines for two studies using electroencephalography (Enobio by Neuroelectrics) in MATLAB. In a first proof-of-principle study, Elena tested N=62 typically developing infants (aged 5-12 months) presented with morphed faces of their parent and a stranger on a screen. Results revealed lower-than-normal attrition rate and convergence of the NBO to the stimulus reliably eliciting the strongest neural response in a given infant in 85% of the infants completing the study. The proportion of optima did not differ between the parent vs stranger side of the stimulus space. Group-level analyses confirmed significantly stronger responses to mother than stranger and stranger than mother when sub-grouped into infants for whom the algorithm converged at the mother’s vs stranger’s face respectively, providing confirmation that the individual-level preferences identified by the algorithm were robust. Similar group-level responses to familiar and unfamiliar faces in middle infancy relate to heterogeneity in preference, rather than an intermediate preference or no preference. They further revealed that whilst individual infants do show robust preferences for particular faces, these are not systematically related to age or social behaviour in this developmental window.

For a second study, Elena adjusted the real-time analysis pipeline for continuous EEG data allowing to study social attention in a naturalistic context. She tested N=57 typically developing infants (aged 6-12 months) presented with an experimenter exhibiting various social and nonsocial behaviours. In the part of the study where rapid convergence was aimed for, the NBO identified the optimum action for most infants who participated in the paradigm, showing that the method produced a robust signal reliably differentiating between different points in the stimulus space. While no overall enhanced attention for specific actions was observed, attention depended on the infants’ social behaviour skills and parental mood, suggesting that individual differences play a role in which aspect of social interaction is experienced as most attention-capturing. In further pilot studies, Elena has also combined NBO with further measurements including functional near-infrared spectroscopy (fNIRS) and functional magnetic resonance imaging (fMRI). 

Elena has presented her work at different international conferences (including INSAR and ICIS) to further disseminate her research projects across the scientific community. Furthermore, she has actively engaged in several talks to a non-scientific audience hosted by the National Child Trust and at Birkbeck College in London where she interactively presented and discussed her research to a large community of parents. In an outreach paper co-authored by Elena, the BONDS team has emphasized the need to take an individualized approach to studying infants’ development for addressing the fundamental challenges related to reproducibility and generalizability in current infancy research (Gui et al., 2022). In a methodological pre-print co-authored by Elena, the team describes how they applied a Neuroadaptive Bayesian Optimization to studying brain responses in infants (NBO; da Costa et al., 2021). Thanks to a grant from the Birkbeck Wellcome Trust Institutional Strategic Support Fund, Elena has been able to turn the above studies into manuscripts for publication in scientific journals (see Throm et al., 2023 for the live infant NBO study).

The aim of Arianna’s project was to adapt a parent-mediated intervention based on Project ImPACT, suitable for children between 9 and 18 months and to test its effectiveness in children at elevated likelihood (EL) for autism (i.e., preterms, younger infants of children with autism). Arianna spent from October 2019 until September 2023 at Ghent University until finishing the PhD. Due to the COVID-19 pandemic, this project was considerably delayed and could not be completed on time. This in addition to the preliminary results of such project, two studies were conducted as contingency plans focusing on early neural differences in EL children and their neural development. Furthermore, the results of a systematic review on neural outcomes of intervention studies for children with and at elevated likelihood of autism has been carried out. All empirical studies were conducted on EL children. Two studies also included children with typical likelihood of autism (TL) as a control group. 

The first study investigated comparison between EL and TL children in their neural oscillatory activity in response to social vs non-social stimuli. A growing body of research suggests that atypical neural activity may be observed in neurodevelopmental disorders prior to the observation of differences in overt behaviour. In autism in particular, it has been suggested that EL infants may show less pronounced differences in neural responses to social vs. non-social stimuli as compared to TL infants. In research more broadly, the frontal theta frequency band has been correlated to a range of social behaviours. However, little research has investigated whether there are differences between EL and TL children in terms of the modulatory impact of social stimuli on frontal theta responses. Contrary to the expectations, no differences between the two groups were present. More surprisingly, no modulatory effect of social stimuli on either relative or absolute theta power was replicated. 

The second study (conducted on a dataset shared by Birkbeck University of London) was a longitudinal study conducted to investigate whether 1) there was a change over the first 3 years of life in the topographical distribution of neural specialisation to social stimuli; 2) there was a difference in development and topographical distribution of neural specialisation to social stimuli between infants with elevated and typical familial likelihood of autism (i.e., with or without older siblings with autism, EFL and TFL respectively); 3) familial likelihood and sex predicted developmental trajectories of the degree of specialisation of each brain region. The findings seem to indicate that 1) specialisation for social stimuli seems to involve more frontoparietal areas during the first year of life, but then reorganises in temporo-parietal areas at three years; 2) differences in topographical distribution of neural specialisation for social stimuli seem to be present between EFL and TFL; 3) while developmental trajectories of specialisation of each brain region differed between males and females, familial likelihood did not predict these trajectories. 

The third study presents the preliminary results of an RCT investigating the effect of the prodromal adaptation of a parent-mediated intervention for EL infants. As a growing body of literature suggests that early parent-mediated interventions for EL infants might support their neural and behavioural social-communicative development, in this study we aimed 1) to adapt a cost-effective prodromal parent-mediated intervention (Project ImPACT) for EL infants; 2) to assess its effects comparing an intervention group (N = 4) with a no-intervention group (N = 2) in an RCT design; 3) to assess the effect of the intervention after 6 months on the proximal outcomes and its effect on autism traits. The effects of the intervention were investigated on: a) proximal social-communicative abilities b) proximal non-social attentional abilities; c) ‘social’ brain development; d) parent-child interaction. However, due to the impact of the COVID-19 pandemic at the start of this project, results on parent-child interaction, motor imitation, and theta frontal oscillatory power are only preliminary. The findings did not show any effect of the intervention on these measures. However, these results must be interpreted very carefully, especially considering the inherent limitations of such preliminary work and we should await the results of the full RCT to draw final conclusions. 

In the context of neurodevelopmental diversity, Arianna's project has provided new insights on the relation between brain development and, behavioural developmental and Naturalistic Developmental Behavioural interventions in children with and at elevated likelihood of developing autism. In collaboration with Jasmine Siew (ESR 8), Arianna has recently submitted a manuscript in June 2023 entitled Zanatta, A., Siew, J., van der Paelt, S., Warreyn, P. & Roeyers, H. (2023). Developmental, behavioural and NDBI interventions in autistic children or at elevated likelihood of autism: a systematic review of neural outcomes”.  Submitted for review June 2023: Journal of Autism and Developmental Disorders. Their publication has shown that previous studies targeting social-communicative outcomes might have a positive impact on brain development of autistic children and those at elevated likelihood of autism. 

During her PhD, Arianna spent 2 weeks at Kings College London to learn about randomised controlled trials and clinical research with Professor Tony Charman. In November 2022, she spent 1 month at the Donders Centre for Cognitive Neuroimaging under the supervision of Dr. Robert Oostenveld to develop scripts to conduct spectral analyses and pool together large datasets coming from a multisite investigation (the Eurosibs).