Using the Mcgurk Effect to Investigate the Neural Indices of Audiovisual Speech Processing in Infants at Risk for ASD

Poster Presentation
Saturday, May 4, 2019: 11:30 AM-1:30 PM
Room: 710 (Palais des congres de Montreal)
K. H. Finch1, C. A. Nelson2 and H. Tager-Flusberg3, (1)Boston University, Boston, MA, (2)Boston Children's Hospital, Boston, MA, (3)Psychological and Brain Sciences, Boston University, Boston, MA

The McGurk effect is an audiovisual (AV) illusion such that when the listener hears one syllable, /ba/, while seeing a speaker articulate /ga/, people report hearing /da/ (McGurk & MacDonald, 1976). Individuals with ASD are thought to have impairments in AV integration since they have difficulty perceiving the McGurk effect (Irwin et al., 2011) and show atypical neural processing to AV speech as measured by event-related potentials (ERPs; Magnee et al., 2011). Little is known about how ERPs to AV speech differ earlier in development.

Objectives: Our current study investigated ERPs in response to AV speech in infants at risk for developing ASD. We included infants at familial risk, infants who fail a 12-month screener, and an age-matched group of low-risk infants.


ParticipantsSeventy 12- to 14-month-olds (M=13.47) were divided into two groups: 1)low-risk typically developing controls (LR; N=41) or 2)high-risk for ASD (HR; N=32) defined as having an older sibling with ASD (N=18) or failing the Communication and Symbolic Behavior Scales Checklist (Wetherby & Prizand, 2002; N=14).

ProcedureERPs were recorded while infants watched 4 video types up to 120 times: 1)AV-congruent-ba (Visual-ba, Audio-ba or VbaAba), 2)AV-congruent-ga (VgaAga), 3)AV-incongruent-impossible (VbaAga), and 4)AV-incongruent-McGurk (VgaAba).

AnalysisAnalyses focused on the mean amplitude of the AV mismatch response (AVMMR; 290-390ms post-audio-onset) as it is modulated by incongruent AV speech in infants (Kushenerenko et al., 2013). We combined the AV-congruent conditions (conditions 1 and 2). Controlling for age, we analyzed AVMMR differences using an ANOVA across 3 conditions (AV-congruent, AV-impossible, AV-McGurk), two regions of interest (central, temporoparietal), and two hemispheres (left, right; see Figure1). Also, given that the AVMMR relates to age specifically in the AV-impossible condition, we conducted Pearson correlations between age and AVMMR.


We found no condition or group differences at central sites (all p’s>0.200). At temporoparietal sites, there was a significant condition by hemisphere response (F(2, 140)=4.36, p=.015; see Figure2). Follow-up analyses revealed that AV-impossible showed a more negative response in the left temporoparietal sites compared to AV-Congruent (p=.034) and AV-McGurk (p=.050). There were no other main or interaction effects, and no effects involving group. We found a significant association between age and the AVMMR response to AV-impossible in the left temporoparietal sites (r=0.288, p=.013) which was driven by the LR infants (r=0.377, p=.015). No association was found in the HR infants (p=0.355).


The AVMMR was modulated by incongruent AV stimuli. Infants showed a more negative response to speech that could not be fused together into a single percept (VbaAga) compared to congruent (VbaAba, VgaAga) or fusible (VgaAba) speech percepts. There were no group differences as LR and HR groups showed the same pattern. However, older LR infants showed a more positive response to AV-impossible speech, similar to past findings (Kushnerenko et al., 2013), while HR infants did not show this age difference. Future work will follow these infants longitudinally to investigate potential maturational delays in the AVMMR for HR infants, including HR infants who are later diagnosed with ASD.