Logo: to the web site of Uppsala University

Background Studies of infants with an elevated likelihood of autism spectrum disorder can identify basic developmental processes that are associated with subsequently emerging clinical symptoms. Atypical responsiveness to sounds in infancy is such a potential early marker of autism. Here, we used pupillometry to quantify reactivity to social and nonsocial sounds in infants with a subsequent diagnosis. Previous research suggest that pupil dilation reflects attentional alerting, and link it to the locus coeruleus norepinephrine system. Methods We measured pupil dilation responses to child-directed speech and the sound of running water; sounds infants often hear in their everyday life. The final sample consisted of 99 ten-month-old infants (52 girls), of whom 68 had an elevated likelihood of autism and 31 were typically developing low-likelihood infants. At follow-up (36 months of age), 18 children in the elevated-likelihood group were diagnosed with autism. Results Compared to infants without diagnosis, the infants who were subsequently diagnosed with autism had larger pupil dilation when listening to nonsocial sounds, while reactivity to speech was strikingly similar between groups. In the total sample, more pupil dilation to the nonsocial sound was associated with higher levels of autistic symptoms. We also found that on a trial-by-trial basis, across all conditions and groups, more pupil dilation was associated with making fewer gaze shifts. Conclusions This study did not find evidence of atypical pupillary reactivity to child-directed speech early in life in autism. Instead, the results suggest that certain nonsocial sounds elicit atypically strong alerting responses in infants with a subsequent autism diagnosis. These findings may have important theoretical and clinical implications.

Being looked at is an important communicative signal, and attenuated responses to such direct gaze have been suggested as an early sign of autism. Using live eye tracking, we examined whether direct gaze elicits different gaze responses in infants at ages 10, 14 and 18 months with and without later autism in real-life interaction. The sample consisted of 169 infants: 35 with elevated likelihood of autism and subsequent diagnosis, 94 without subsequent diagnosis and 40 with typical likelihood of autism. Infants in all groups tended to look more towards the adult’s face shortly after the direct gaze occurred. Neither how much nor how quickly the infants responded to the direct gaze differentiated the without elevated likelihood of autism and subsequent diagnosis and with elevated likelihood of autism and subsequent diagnosis groups. Infants in the typical likelihood group looked more at the face after the direct-gaze event than infants in the two elevated likelihood groups, but this result is tentative. In an exploratory analysis, infants in the elevated likelihood of autism and subsequent diagnosis group looked away quicker from faces with direct gaze than infants in the typical likelihood group, but this measure did not correlate with dimensional autism or differentiate between the two elevated likelihood groups. The current results suggest that attenuated behavioural responses to direct gaze in infancy are neither strong nor specific early markers of autism.

Lay abstract

When other people look directly towards us, we often respond by looking back at them, and such direct-gaze responses are important for establishing eye contact. Atypical eye contact is common in autism, but how and when this aspect of autism develops is not well understood. Here, we studied whether how much and how quickly infants respond to others’ direct gaze is associated with autism in toddlerhood. We did this by measuring direct-gaze responses in a playful social interaction using live eye tracking. The study included 169 infants, of whom 129 had an elevated likelihood of developing autism due to having a first-degree family member with the condition, and 40 with typical likelihood of autism. In the elevated likelihood group, 35 were diagnosed with autism spectrum disorder at 3 years of age, and 94 were not. The results showed that infants in all three groups tended to increase their looking towards the adult’s face after the adult looked directly at them. However, neither how much nor how quickly the infants responded to direct gaze by looking back at the adult reliably differentiated the infants with or without subsequent autism. While infants in the elevated likelihood of autism and subsequent diagnosis group tended to look away quicker from faces with direct gaze than infants in the typical likelihood group, this measure did not differentiate between the two elevated likelihood groups. We interpret the results as supporting the view that atypical direct-gaze responses are not early markers of autism.

Background

Pragmatic language is key for adaptive communication, but often compromised in neurodevelopmental conditions such as autism spectrum disorder (ASD). Decontextualized language—to talk about events and things beyond here and now—develops early in childhood and can be seen as a pre-pragmatic ability. Little is known about the factors that contribute to decontextualized language use in toddlers and whether these are different from factors contributing to general language development.

Methods

We studied longitudinal associations between parent-rated core language and non-verbal socio-communicative abilities at 14 months of age, and decontextualized language use at 24 months of age in children with typical and elevated likelihood of ASD (total N = 303). Using twin modelling, we also investigated genetic and environmental contributions on decontextualized language and grammar use in two-year-old twin pairs (total N = 374).

Results

Core language ability was a strong predictor of later decontextualized language use in both children with and without an elevated likelihood of ASD. In contrast, social communication was only a significant predictor of decontextualized language use for children with low levels of core language. This pattern was specific to decontextualized language, and not replicated in prediction of concurrent grammatical ability. Further, there was a large genetic influence on decontextualized language at 2 years of age, which mostly overlapped with the genetic influences on grammatical ability. Shared environment influences were significant for grammatical ability, but not found on decontextualized language. In children with an elevated likelihood of ASD, decontextualized language use was negatively associated with autistic symptoms.

Conclusions

This study suggests that decontextualized language is developmentally associated with, yet dissociable from, more general language development measured as grammatical ability. Already at 2 years of age, parental ratings of decontextualized language is associated to clinician-rated symptoms of ASD.