Attentive brain states in infants with and without later autism

ASD is a neurodevelopmental condition marked by social-communication difficulties and restricted/repetitive behaviors. Although highly heritable, diagnoses typically occur after age 4, limiting early intervention. Prospective infant studies can reveal developmental processes linking diverse risk factors to later ASD. The authors focus on a potential mechanism: engagement of attentive brain states to social stimuli. Infants with a family history of ASD show emerging differences in social attention in the first two years, including reduced attention to eyes and faces and altered responses to gaze. Prior infant EEG studies suggest altered cortical responses to social cues by 6–12 months and differences in face/gaze processing in those later diagnosed with ASD. The research question is whether early neural markers of attentional engagement to faces (direct vs averted gaze) versus non-social stimuli are linked to later ASD diagnosis (categorical outcome) and to dimensional social adaptive skills. The study evaluates the ERP Nc component and whole-brain microstate dynamics, hypothesizing that atypical engagement of attentive brain states to social stimuli is associated with later ASD-related outcomes.

Design and cohort: Prospective longitudinal study within the British Autism Study of Infant Siblings (BASIS). Infants with an older sibling with ASD (FH; elevated likelihood) and controls without family history (noFH) were enrolled in the first year and followed to 3 years for outcomes. - Enrollment: 170 FH and 77 noFH infants enrolled; diagnostic outcome at 3 years determined via best-estimate clinical procedures using ADOS-G, MSEL, and ADI-R. Outcome groups: FH-ASD (met DSM-5 ASD criteria), FH-noASD, and noFH-noASD (none met ASD criteria). - EEG session: EEG and behavioral data collected at 6–11 months (M = 7.92 months, SD = 1.26). Sufficient data obtained from 131 infants: 40 noFH-noASD, 72 FH-noASD, 19 FH-ASD. Task and stimuli: Infants seated 60 cm from a monitor. Continuous blocks presented while attentive (target ~50 blocks). Each block began with a colorful fixation (800–1200 ms), then one of three stimuli: face with direct gaze (FD), face with averted gaze (FA), or a non-social control ('Noise'). Faces displayed various emotions. Trials lasted 800 ms with a 500 ms inter-trial interval. Gaze shifts within blocks (three gaze shifts) were present but not analyzed here. EEG acquisition and preprocessing: 128-channel Hydrocel Sensor Net; sampling rate 500 Hz. Inclusion criteria for trials required central fixation at onset without gaze shifts, blinks, or head movements for 800 ms post-onset, verified offline via video. Data recorded and processed in EGI NetStation 4.0/5 with a consistent pipeline (details in supplementary materials). Standard preprocessing steps per prior protocol (artifact rejection, epoching around stimulus onset, etc.). Analytic strategy: 1) Hypothesis-driven ERP analysis: Focus on the Nc component (~300 ms post-onset over frontal regions), indexing attention engagement. Traditional topographic analyses assessed Nc amplitude and latency across stimuli (FD, FA, Noise) and groups. Power analysis (G*Power 3.1) based on prior effect size for Nc mean amplitude (d = 0.17) indicated ≥11 participants per group for 80% power at alpha 0.05 to detect group-by-stimulus interaction. 2) Data-driven microstate analysis: Whole-scalp spatiotemporal microstates characterized to index transient brain states during attention. Prototypical microstates were derived from the noFH-noASD group during social attention and applied to all infants/stimuli to extract features (e.g., duration, strength/amplitude, timing/latency). 3) Machine learning: Generative algorithm with elastic-net regularization applied to microstate features across stimuli to identify features predictive of (a) categorical ASD diagnosis at 3 years and (b) dimensional variation in social adaptive skills (VABS Socialization standard scores) at 3 years. Regression models also examined associations with family history and outcome groups, and with continuous VABS Socialization to capture variation not limited to diagnostic thresholds.

- Brain state timing vs strength: Measures of brain state timing were associated with categorical ASD outcome, whereas measures of brain state strength related to dimensional variation in social functioning. - Nc component: Infants in the FH-ASD group showed shorter Nc latency compared to FH-noASD and noFH-noASD groups, indicating faster but atypical attentional engagement to faces. - Microstate duration and categorical prediction: The duration of attentive microstate responses to faces contributed information for predicting categorical ASD outcome. - Dimensional social functioning: Reduced Nc amplitude difference between faces without gaze (FA) and non-social stimuli (Noise), together with the strength of the attentive microstate response to faces, contributed to predicting variation in VABS Socialization scores at 3 years. - Overall: Findings suggest atypical cortical activation patterns to social stimuli precede emergent socialization difficulties and can be captured by spatiotemporal EEG features in infancy.

The study addressed whether early neural correlates of attentional engagement to social stimuli relate to later ASD outcomes. The results indicate that infants later diagnosed with ASD exhibit altered timing of attentional engagement (shorter Nc latency) and differences in microstate dynamics when processing faces. Timing-related microstate features were most informative for distinguishing ASD diagnosis (categorical outcome), whereas strength-related features of attentive brain states were linked to continuous variation in social adaptive skills. These findings support the hypothesis that disruptions in social attention processing are part of the developmental pathways leading to ASD. They further demonstrate that characterizing whole-brain spatiotemporal activation patterns (microstates) provides sensitive markers of early brain state engagement to social cues, potentially elucidating neurodevelopmental mechanisms underlying later social difficulties.

The paper shows that atypical cortical activation to social stimuli in infancy precedes later socialization difficulties and ASD diagnosis. Specifically, shorter Nc latency and altered microstate dynamics during face processing are linked to later categorical ASD outcomes, while the strength of attentive brain states relates to dimensional social adaptive skills. Defining infant brain states using whole-brain spatiotemporal EEG features offers a promising approach to understanding mechanisms leading to ASD and to developing early biomarkers. Future work should further validate these markers longitudinally and assess their utility for early identification and targeted intervention strategies.