Systematic Review and Meta-Analysis of Mismatch Negativity in Autism: Insights Into Predictive Mechanisms.

Researchers studied brain responses to unexpected sounds in autistic people. They found that autistic children have weaker brain responses to complex sound changes, but autistic adults have stronger responses. Autistic people of all ages showed weaker responses specifically to speech sounds. The timing of brain responses was normal. This suggests that how the autistic brain processes and predicts sounds changes with age.

This systematic review and meta-analysis examined mismatch negativity (MMN), a brain response measuring auditory processing and change detection, in autism spectrum disorder. Analyzing 76 studies for amplitude and 62 for latency, researchers found age-dependent differences. Autistic children and adolescents showed reduced MMN amplitudes in complex auditory tasks, while autistic adults demonstrated increased amplitudes. Autistic individuals had smaller MMN responses specifically to speech sound changes.

No significant differences in response timing were found. Results suggest developmental changes in predictive brain mechanisms in autism, with heightened prediction errors potentially emerging in adulthood. Findings support theories about altered predictive coding in autism.

Findings suggest age-dependent patterns in auditory processing that may inform developmental interventions. Reduced speech sound processing across all ages supports targeting phonological awareness. The developmental trajectory from reduced to heightened responses may guide timing of sensory interventions and explain changing sensory profiles with age.

Residual heterogeneity remained across studies despite identifying contributing factors. The authors note that larger samples and improved data reporting are needed. Individual study quality and methodological variations may have contributed to inconsistent findings across the literature.

Mismatch negativity (MMN) has been frequently used to assess auditory processing and change detection in autism spectrum disorder (ASD), but findings have been fairly inconsistent. To address this issue, we conducted a systematic review and meta-analysis of MMN amplitude (76 effect sizes) and latency (62 effect sizes) in ASD to identify factors contributing to this heterogeneity and to interpret findings within the predictive coding framework. While residual heterogeneity remained, significant effects of the interaction between age group and design type (unifeature vs. multifeature, i.e., one or several types of deviants) and deviant type were found for MMN amplitude. In multifeature designs, autistic children and adolescents exhibited reduced MMN amplitudes compared to neurotypical peers (g = 0.25, p = 0.01), whereas autistic adults showed increased MMN amplitudes (g = -0.26, p = 0.02).

In addition, autistic individuals had significantly smaller MMN amplitudes than neurotypical individuals in paradigms using phoneme deviants (g = 0.41, p < 0.001). Across designs, no significant MMN latency differences were observed between neurotypical and autistic individuals. These results are discussed within the predictive coding framework, as MMN responses are thought to reflect prediction errors, aligning with theories suggesting heightened prediction errors in autistic adults. Future studies with larger samples and improved data reporting are needed to further clarify the developmental trajectory and variability of MMN responses in ASD.

Additionally, computational modeling approaches can help characterize learning dynamics and disentangle predictive coding accounts among autistic individuals.