Neural Correlates of Early Language Processing in 9-Month-Old Infants at Risk for ASD

Background:  Language delays are one of the earliest autism-related concerns that parents report. Infants who develop autism spectrum disorders (ASD) often exhibit atypicalities in their early language profiles (e.g., Hudry et al., 2014). Moreover, toddlers with ASD and significant verbal delays show hypoactivity in canonical temporal language-processing regions during speech processing (e.g., Lombardo et al., 2015). Since language learning appears to be gated by the social brain (Kuhl, 2007), early deviations in language processing may index altered social development prior to overt deficits in social or verbal skills. Examining language processing during foundational stages of language acquisition may thus inform risk for ASD and later language difficulties.

Objectives:  We examined the neural circuitry subserving language processing in 9-month-old infants at high (HR) and low risk (LR) for ASD to identify patterns of brain activity that may predict altered trajectories in language development and ASD risk. This will further substantiate the role of language in the early manifestation of ASD.

Methods:  19 HR and 13 LR infants underwent fMRI during natural sleep and were presented with speech samples produced by different female native speakers of English and Japanese. Both English and Japanese stimuli were matched for duration, intensity, peak amplitude, pitch, and pitch range. English and Japanese speech segments (8 segments per language, each lasting18s) alternated between periods of silence (12s each) in a traditional fMRI block-design. Individual data were registered to an infant atlas; preprocessing (including motion censoring) and statistical analyses were conducted in FSL. Behavioral measures of language, cognitive, and social functioning were assessed at 12 months [i.e., Mullen Scales of Early Learning (MSEL), Vineland Adaptive Behavior Scale (VABS), and Autism Observation Scale for Infants (AOSI)].

Results:  Relative to LR infants, HR infants showed hypoactivity for both English and Japanese in temporal and frontal language regions, the cerebellum, and regions associated with attention (precuneus) and reward (amygdala, and caudate nucleus). In particular, LR infants showed greater activity in left temporal pole and caudate nucleus than HR infants for English, the infants’ native language. The relation between neural activity for English and 12-month Verbal T Scores from the MSEL were also examined. For LR infants, greater activity in temporal language areas at 9 months predicted better verbal scores at 12 months. In contrast, for HR infants, better verbal scores were inversely related to activity in the precuneus and medial prefrontal cortex, the hubs of the default mode network (DMN), a resting-state network that is most active during passive rest and deactivated during information processing. Mean activation in these DMN hubs during language exposure was negatively correlated with 12-monthVineland Socialization Scores (r =-0.61, p= 0.01).

Conclusions: HR infants already show attenuated neural responses to language at 9 months, prior to overt signs of atypical development. The distinct patterns of brain activity during native language processing and its association with later verbal skills suggest that HR infants may be less attuned to the language inputs of their immediate environments, with negative downstream effects for language and social development.