Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Sep 1.
Published in final edited form as: Clin Neurophysiol. 2024 Jun 8;165:55–63. doi: 10.1016/j.clinph.2024.05.015

Parsing Evoked and Induced Gamma Response Differences in Autism: A Visual Evoked Potential Study

Abigail Dickinson a, Declan Ryan a, Gabrielle McNaughton a, April Levin b, Adam Naples c, Heather Borland d, Raphael Bernier e, Katarzyna Chawarska c, Geraldine Dawson f, James Dziura g, Susan Faja h, Natalia Kleinhans i, Catherine Sugar j, Damla Senturk j, Frederick Shic k, Sara Jane Webb k, James C McPartland c, Shafali Jeste l; Autism Biomarkers Consortium for Clinical Trials
PMCID: PMC11684857  NIHMSID: NIHMS2041167  PMID: 38959536

Abstract

Objective:

Electroencephalography (EEG) measures of visual evoked potentials (VEPs) provide a targeted approach for investigating neural circuit dynamics. This study separately analyses phase-locked (evoked) and non-phase-locked (induced) gamma responses within the VEP to comprehensively investigate circuit differences in autism.

Methods:

We analyzed VEP data from 237 autistic and 114 typically developing (TD) children aged 6–11, collected through the Autism Biomarkers Consortium for Clinical Trials (ABC-CT). Evoked and induced gamma (30–90Hz) responses were separately quantified using a wavelet-based time-frequency analysis, and group differences were evaluated using a permutation-based clustering procedure.

Results:

Autistic children exhibited reduced evoked gamma power but increased induced gamma power compared to TD peers. Group differences in induced responses showed the most prominent effect size and remained statistically significant after excluding outliers.

Conclusions:

Our study corroborates recent research indicating diminished evoked gamma responses in children with autism. Additionally, we observed a pronounced increase in induced power. Building upon existing ABC-CT findings, these results highlight the potential to detect variations in gamma-related neural activity, despite the absence of significant group differences in time-domain VEP components.

Significance:

The contrasting patterns of decreased evoked and increased induced gamma activity in autistic children suggest that a combination of different EEG metrics may provide a clearer characterization of autism-related circuitry than individual markers alone.

Keywords: Autism, Electroencephalography, Visual Processing, Gamma Oscillations, Visual Evoked Potentials

1. Introduction

Autism spectrum disorder (“autism”) is a complex neurodevelopmental condition characterized by social, communicative, and behavioral differences that emerge in early childhood (American Psychiatric Association., 2013). As a non-invasive measure of neural circuit function, EEG provides a powerful technique for developing objective biomarkers to augment traditional behavior-based diagnostic and monitoring methods in autism (Jeste et al., 2015; Levin et al., 2020; McPartland et al., 2020; Port et al., 2014). EEG measures of event-related visual evoked potentials (VEPs) are particularly valuable, as these stereotyped responses reflect highly organized neural mechanisms involved in visual processing and provide a robust framework for investigating circuit differences in autism (Jeste & Nelson, 2009; Modi & Sahin, 2017).

Although VEPs have been studied in autism using a range of visual stimuli (see Farashi et al., 2023 for a review), pattern-reversal VEPs are particularly promising in the context of biomarker development (McPartland et al., 2020; Siper et al., 2016). These simple checkerboard stimuli can be precisely controlled and elicit stereotypical waveforms with highly consistent morphology and temporal characteristics, providing precise insights into specific neural mechanisms that shape the VEP (Zemon & Gordon, 2018). Moreover, using geometric stimuli minimizes confounding factors related to social and emotional cognition, which complex stimuli like faces could introduce. Therefore, identifying differences in pattern-reversal VEPs provides a targeted lens for understanding specific visual circuit disruptions in autism and may reveal broader neural mechanisms implicated in the disorder (Beker et al., 2021).

However, no consensus exists regarding autism-related differences in traditional time-domain VEP components. Studies characterizing stereotypical deflections in the averaged pattern-reversal VEP waveform describe a range of findings in autism compared to typically developing comparison groups, including reduced P100 amplitude (Brittenham et al., 2022; Kovarski et al., 2016, 2019; Siper et al., 2016) and delays in P100 (Hamdan et al., 2022) and P145 latency (Sayorwan et al., 2018). A recent meta-analysis highlights this considerable variability and reports that reduced N100 amplitude response to non-face stimuli in autism is one of the only consistent VEP findings observed across studies (Farashi et al., 2023). Moreover, in the multi-center Autism Biomarkers Consortium for Clinical Trials (ABC-CT), N100 and P100 amplitude VEP components did not differ in a large group of autistic children compared to their typically developing peers (Webb et al., 2023).

VEP-derived event-related spectral perturbations (ERSPs) offer a more specific method to investigate neural circuit variations in autism than average time-domain responses. ERSPs provide in-depth analysis of oscillatory activity changes in response to stimuli, shedding light on transient circuit dynamics that shape the VEP. Within this framework, high-frequency gamma oscillations (>30Hz) play a pivotal role. Originating from the interaction between inhibitory GABA-ergic interneurons and excitatory pyramidal cells (Bartos et al., 2007; Cardin et al., 2009; Traub et al., 1998; Whittington et al., 2000; Whittington & Traub, 2003), this high frequency activity reflects the excitatory and inhibitory interactions that drive rapid circuit processing. Gamma oscillations exhibit two distinct types of functional responses to sensory input: evoked and induced (Tallon-Baudry et al., 1996). Evoked gamma activity is phase-locked to the stimulus onset and represents the brain’s immediate and predictable responses. Induced gamma activity, which is not phase-locked, is associated with higher-order cognitive processes in visual perception, such as attention and integration (David et al., 2006).

Recent evidence reinforces the potential of utilizing VEP-derived gamma oscillations to detect circuit disruptions linked to autism. Autistic children exhibit weaker evoked gamma responses (30–48Hz) to pattern-reversing checkerboard stimuli than their TD peers (Brittenham et al., 2022; Siper et al., 2016). Expanding on previous studies of evoked activity to include measures of induced gamma oscillations is particularly significant, given the heightened neural response variability reported in autism (Dinstein et al., 2012; Haigh et al., 2015; Milne, 2011). For instance, higher neural response variability may weaken phase-locked evoked responses without affecting non-phase-locked induced activity similarly.

Investigating both evoked and induced activities within the VEP could provide a more comprehensive and nuanced perspective on neural circuit disparities in autism (Rojas et al., 2008). This study leverages VEP data from the extensive ABC-CT cohort data to compare evoked and induced gamma responses between autistic and typically developing children. We utilize a data-driven clustering method to explore potential group differences in gamma responses (30–90Hz), avoiding the limitations of analyses confined to predetermined time or frequency ranges. Considering the lack of amplitude- and latency-based VEP component group differences within the ABC-CT sample (Webb et al., 2023), our study presents a unique opportunity to determine if gamma oscillations reveal nuanced neural circuit variations not captured through conventional time-domain VEP analysis.

2. Methods

2.1. Participants.

Participants in the present analysis are part of the ABC-CT cohort, a multi-site initiative dedicated to uncovering biomarkers of autism to inform clinical trials. The ABC-CT protocols (McPartland et al., 2020; Webb et al., 2020) and sample (Faja et al., 2023) are comprehensively described in previous publications. Inclusion criteria for the autism group included a diagnosis confirmed with the ADOS-2. TD children had no autism diagnosis or first-degree relative (parents or siblings) with autism. Exclusionary criteria for both groups included neurological conditions, epilepsy or use of anti-epilepsy medications, clinically significant auditory or visual impairment, sensory-motor challenges that would impact participation, prematurity or pre/perinatal birth injury or brain damage, or severe environmental difficulties that would impact neurological development. A central institutional review board reviewed and approved all procedures, and guardians or participants provided informed consent or assent in accordance with the declaration of Helsinki.

The ABC-CT study includes measures of social function and a battery of EEG and eye-tracking tasks administered across five sites. These assessments take place during two-day visits at three time points. This study focuses on the first (Time 1) study visit, at which point 280 children with a confirmed autism spectrum disorder diagnosis (mean age=8.6 years; SD=51.6, range 56.0–11.5) and 119 typically developing children (mean age=8.5 years, SD=51.6 months; range 56.0–11.5) were enrolled. Our analysis utilizes EEG data collected during a binocular VEP paradigm, as previously described in a report of time-domain VEP components (Webb et al., 2023). As outlined in the initial paper, 237 children with ASD and 114 TD children from the Time 1 ABC-CT cohort provided usable VEP data. Among those who did not contribute usable data (n=48), n=20 did not complete the VEP paradigm, n=11 recordings experienced equipment failures or experimenter error, and n=17 had inadequate trial numbers/VEP data morphology. All participants had normal or corrected-to-normal vision. Demographic details for this subsample of participants are reported in Table 1.

Table 1.

Demographic Characteristics of Participants in the ABC-CT VEP Subsample.

Autism VEP (N=237) TD VEP (N=114)
Sex
n girls (%) 55 (23%) 35 (31%)
n boys (%) 182 (77%) 79 (69%)
Age in years at EEG 8.7 (1.6) 8.5 (1.6)
Income <$75,000 (n %) 76 (32.9%) 22 (19.8%)
Race
n (%)
African American or Black 17 (7%) 4 (4%)
American Indian and Alaska Native 2 (1%) 0 (0%)
Asian 13 (5%) 1 (1%)
Multi-racial 36 (15%) 13 (11%)
Other 6 (3%) 1 (1%)
White 163 (69%) 95 (83%)
Ethnicity
n (%)
Hispanic 38 (16%) 5 (4.4%)
Maternal/Paternal Education
n (%)
Maternal <Bachelors 82 (34.6%) 14 (12.4%)
Paternal <Bachelors 88 (38.4%) 14 (12.5%)
Full Scale IQ 98.6 (17.9) 115.5 (12.5)
Verbal IQ 98.1 (20.3) 116.5 (11.3)
Nonverbal IQ 99.0 (16.8) 112.6 (14.0)
ADOS Calibrated Severity Score 7.6 (1.8) 1.6 (0.9)
Open in a new tab

2.2. EEG Collection.

EEG data were collected using an EGI 128-channel acquisition system using the Net Amps 300 series (three sites) or 400 series (two sites) amplifier and 128-electrode EGI HydroCel Geodesic Sensor Nets. A standard acquisition setup implemented a 1000-Hz sampling rate with a 0.1- to 200-Hz filter. EPrime (version 2.0) was used to present visual stimuli on a 23-inch monitor, with stimulus presentation timing monitored in real-time using a Cedrus StimTracker. Two checkerboard stimuli (20×20 checks; 26cm x 26cm) featuring a central red fixation point underwent phase alternation every 500ms, creating a standard pattern-reversal stimulus for 200 trials, divided into four blocks. Checkerboards had an average luminance of 80cd/m^2 and a contrast of 99%, set against a mean luminance background. Participants were positioned 66.5 cm away from the screen, with the stimuli occupying 22.15 degrees of visual angle at this viewing distance. The experimenter monitored and recorded the participant’s attention to the screen throughout acquisition. EEG Data Processing: EEGLAB (Delorme & Makeig, 2004), FieldTrip (Oostenveld et al., 2011), and custom MATLAB scripts (The MathWorks, Inc., Natick, MA) implemented data processing procedures. Continuous EEG data were filtered to remove frequencies below 1 Hz and above 110Hz, using a finite impulse response filter. Data was segmented into epochs based on stimulus onset (0ms) to 300ms, with a 100ms baseline from −100ms to 0ms. Digital markers indicating the timing of phase reversals were corrected for latency delays before epoching. Trials associated with periods of inattention, as identified by the examiner, were excluded, ensuring that only epochs during which participants had confirmed visual attention to the stimuli remained.

2.3. EEG Processing.

We used the ERPLab function pop_artexval implemented in EEGLab to detect and remove channels and trials with voltage deviations exceeding +/−150mv. On average, 163.27 (SD=32.24) artifact free, attended trials were retained. The number of artifact free, attended trials available was significantly higher for the TD group (M=176.79; SD=19.36, range: 115 −200), as compared to children with autism (M=156.76; SD=35.09; range: 65–200; t=−6.87, df=342.62, p=<0.001). There was no significant difference in the number of channels rejected between autistic (M= 21.70; SD=7.74) and TD children (M=20.18; SD=6.71; t=1.89, df=254.15, p=0.06). After removing unattended and artifactual epochs, we converted the data to continuous form and interpolated it to align with the international 10–20 system’s 25-channel montage (Jasper, 1958). We then implemented an additional cleaning stage using independent component analysis (ICA) (Onton et al., 2006). Electromyogram (EMG), saccadic spike potentials and line noise represent non-neural sources that can interfere with gamma activity due to their high-frequency nature (Yuval-Greenberg et al., 2008). While pervasive, gamma artifacts can manifest at amplitudes too subtle to be identified by an amplitude-based cleaning algorithm. As such, we used ICA to decompose continuous EEG data into maximally independent components (ICs). The EEGLAB function iclabel was then used to categorize components automatically, facilitating the identification and removal of artifacts and emphasizing primarily neural data. For further analysis, we retained only those ICs classified as more than 50% likely to originate from neural activity. Finally, data were re-segmented into their original epochs for time-frequency analysis. All participants had >60 trials, consistent with prior analyses (Webb et al., 2023).

2.4. Time Frequency Analysis.

We extracted measures of evoked and induced gamma activity (30–90Hz) from the averaged data of two occipital channels (O1 and O2). To accurately represent dynamic time and frequency changes in our data, we implemented wavelet analysis using the newtimef function in EEGLAB. Wavelet cycle parameters increased linearly with frequency from 1 to 7, ensuring a balance between time and frequency precision for our frequency range (30–90Hz). While conventional trial averaging is suitable for analyzing tightly synchronized evoked responses, induced activity, which is not strictly phase-locked to stimulus onset, necessitates more specialized signal processing techniques (Cohen, 2014). After subtracting the averaged VEP from each trial, we calculated induced activity as the average time-frequency decomposition of these adjusted individual trials (Cohen, 2014). We then subtracted the induced signal from the total signal (average time-frequency decomposition of each trial) to calculate evoked power (Cohen, 2014).

2.5. Statistical Analysis.

We used cluster-based permutation testing in Field Trip to examine group differences in evoked and induced time-frequency representations. Permutation clustering is a non-parametric statistical technique that retains the temporal and spectral structure of ERSP data while accommodating multiple comparisons (Maris, 2012; Maris & Oostenveld, 2007). This method allows for the unbiased detection of potential group differces distributed in time and frequency in high-dimensional EEG data, avoiding the need for pre-defined summary metrics. Permutation clustering analysis initially calculates uncorrected t-tests for every time-frequency point between groups. Adjacent data points (in both time and frequency) exceeding a pre-defined significance threshold (0.025) are grouped into clusters, with the sum of the t-values forming a cluster-level statistic. This initial map of potentially significant effects is then refined using a permutation procedure. Specifically, test statistics are recalculated after random re-shuffling of group labels to obtain a null distribution against which the observed clusters can be compared. Repeating this process 1000 times derived a distribution of cluster-level statistics under the null hypothesis of no group differences. Only clusters exceeding a significance threshold compared to the null distribution (0.0125) were retained. By comparing observed clusters to this permuted distribution, we controlled the family-wise error rate across the multiple comparisons conducted in our time-frequency analyses, providing a reliable indication of significant group differences. Our alpha thresholds (0.025 & 0.0125) were adjusted to reflect separate analysis of evoked and induced gamma, mitigating potential Type I errors (Pernet et al., 2015). Age, sex, full scale IQ, and the number of trials were included as covariates in permutation clustering.

In follow-up analyses, we examined individual power values and frequency profiles within each significant cluster. While cluster-based permutation tests provide valuable insights into group differences in multi-dimensional data, it is advised to interpret specific time and frequency boundaries with caution, as these may not be definitive (Sassenhagen & Draschkow, 2019). Given these considerations, we extracted values from spectro-temporal clusters of interest to assess the distribution of gamma power values across groups. First, we explored the frequency distribution of power values within the indicated significant time intervals of each cluster. We then utilized the precise time and frequency coordinates indicated for each cluster to calculate individual power values (which were defined as the mean power within the time and frequency ranges of interest for each cluster). To determine if permutation outcomes were disproportionately affected by outliers, we removed outlier values using Tukey’s 1.5 Interquartile Range (IQR) method and employed ANCOVAs to assess group differences. By examining individual differences and reevaluating group differences without outliers, we aimed to further investigate the reliability and broader applicability of our findings.

3. Results

3.1. Evoked and Induced Time Frequency Representations.

Examining time-frequency plots showed an expected distribution of evoked and induced gamma power within the ERSP signal. Phase-locked evoked gamma manifested as a consistent response, most prominent ~ 50 to 80 milliseconds in lower gamma frequencies. Induced activity was distributed more diffusely across frequencies and time, reflecting the inherent variability in induced responses.

3.2. Group Differences.

Permutation cluster analysis revealed significant group differences in evoked and induced gamma signals. Table 2 describes the characteristics of identified clusters. The summary t statistic for each cluster reflects the magnitude of the observed group contrast, with larger absolute values indicating more pronounced differences. The direction of the t statistic (positive or negative) shows the direction of the group difference. Clusters with negative t statistics denote a power increase in TD compared to autistic children, whereas positive values indicate a reversed trend, highlighting elevated power in the autism group. In addition to summary t values, Figure 2 illustrates the magnitude of t values within each cluster to show each cluster’s distribution of t statistics, highlighting regions with the most pronounced group differences. The clusters’ time and frequency ranges provide insights into the spatial-temporal dimensions of the significant regions, revealing how diffuse or localized each group difference is and the specific frequencies and times implicated.

Table 2.

Characteristics of Significant Induced and Evoked Power Clusters.

Cluster Time Range (ms) Frequency Range (Hz) Cluster Statistic Permutation-Based Probability Effect Size
Induced 1 218.80 – 243.86 38.74 – 83.97 6123.54 <0.001 0.4543
Induced 2 49.37 – 72.43 38.44 – 62.26 2272.46 <0.001 0.3816
Induced 3 149.62 – 174.69 70.10 – 90.00 2253.48 <0.001 0.3514
Induced 4 124.56 – 141.60 30.00 – 47.19 1354.37 <0.001 0.3085
Evoked 1 72.43 – 85.46 44.47 – 63.77 −1292.09 0.0009 −0.3599
Open in a new tab

Figure 2. Group-Averaged Time-Frequency Representations of Gamma Power.

Figure 2.

Open in a new tab

Plots display the mean time-frequency representations for evoked power for A) autistic and B) TD participants, and induced power for C) autistic and D) TD participants. Power (in dB) is color-coded as indicated by the color bar.

3.2.1. Evoked Power.

Permutation clustering highlighted increased evoked power in TD compared to autistic children across one significant cluster. This effect was observed from 72.43ms to 85.46ms following stimulus onset, with the frequency range primarily spanning 44–64Hz. The temporal range of this cluster suggests that it reflects the early stages of visual processing, potentially indicating differences in the highly time-locked and phase-locked responses characteristic of the early stages of the visual response. Comprehensive statistics, including the summary t-value, p-value, and effect size related to this cluster, are presented in Table 2.

3.2.2. Induced Power.

Permutation clustering revealed four significant clusters representing increased induced power in the autism group compared to the TD group. The clusters represent distinct areas of significance across various time and frequency intervals, as detailed in Table 2. Notably, the primary cluster (Cluster 1) was observed from 218.80ms to 243.86ms post-stimulus onset, covering frequencies ranging from 38.74Hz to 83.97Hz. This cluster falls within a time window associated with later stages of visual processing, suggesting that it may reflect higher-order cognitive functions such as object recognition or attentional modulation.

In the same time range, while evoked gamma responses were decreased in autism, induced responses were stronger, which may suggest differences in the phase-locking of responses. Specifically, this pattern could indicate less consistent phase-locked responses in autism, reflecting potential variations in the synchronization of neural activity during visual processing, as discussed below.

To gain a more comprehensive understanding of the power differences represented by significant clusters, we extracted individual summary power measures. Notably, ANCOVAs revealed that group differences for Induced Clusters 1 and 2 remained significant after outlier removal. Figure 3 displays the individual values post-outlier removal, offering a rigorous assessment of power value disparities both within and between the two groups.

Figure 3. Visualization of group differences in Induced and evoked gamma power.

Figure 3.

Open in a new tab

A) Evoked: Areas of significant evoked power differences reflect increased evoked power in TD children (TD power-autism power). The one significant evoked cluster is outlined in white. B) Induced: There were four areas of significant group differences in induced power, all characterized by increase power in autistic as compared to TD children (autism power-TD power). The four distinct induced clusters are marked in a gradient from white to black, sequentially representing clusters 1–4 based on their prominence. The lower panel illustrates heat maps of t statistics for C) Evoked and D) Induced clusters.

4. Discussion

This study examines evoked and induced gamma oscillatory responses within the VEP to understand visual processing dynamics in children with autism. Although the timing and patterns of evoked and induced responses aligned with well-established neurophysiological norms, there were significant group differences in their strength. Specifically, we observed weaker evoked but stronger induced gamma activity in autistic children. This finding is particularly significant given the previously reported absence of group differences in traditional time-domain VEP components for this sample (Webb et al., 2023). Together, these data suggest that gamma oscillations provide a sensitive metric for detecting visual processing differences in autism. In turn, gamma variations could guide the development of objective neural markers to help us better understand the circuit dynamics that contribute to the distinct sensory experiences commonly reported in autism.

Our study found that autistic children showed significant reductions in evoked gamma power early in the visual response (~72 to 85ms) in lower gamma frequencies (~44–64Hz). These group differences align with previous reports of reduced gamma coherence in autism during a similar visual paradigm, measured using magnitude square coherence (MSC) (Brittenham et al., 2022; Siper et al., 2016). While MSC does not measure evoked oscillations directly, it inherently emphasizes phase-locked evoked responses over induced activity. As such, the present findings provide further evidence of decreased phase-locked activity in individuals with autism.

Evoked gamma activity is generally associated with canonical neural processing pathways and is often considered to reflect bottom-up circuit processing (Arieli et al., 1996). Reduced evoked power in autism indicates that there may be inherent differences in the initial circuit mechanisms that process visual stimuli in autism. One possible mechanism contributing to differences in low-level circuit processing could be an increased intrinsic variability in visual responses to sensory stimuli. Studies have consistently highlighted an enhanced variability in neural responses among autistic individuals (Dinstein et al., 2012; Haigh et al., 2015; Milne, 2011). This heightened variability could disrupt the coherent synchronization of gamma oscillations, potentially leading to the diminished evoked activity observed in autism.

Along with reductions in evoked power, our data reveal stronger induced gamma responses in autistic compared to TD children. While evoked differences were characterized by localized power variations, induced differences were more temporally and spectrally diffuse. Significant induced differences spanned four clusters reflecting various time and frequency intervals, as described in Table 1. Cluster 1 represents an autism-related power increase later in the visual response (218–243ms post-stimulus onset) across a broad frequency range (38–83Hz). Moreover, induced differences exhibited greater effect sizes than evoked differences and remained significant after outlier removal. These findings underscore the necessity of including induced activity in gamma analysis, which traditional averaging-based methods neglect. The dispersed pattern of group differences reflects the non-phase-locked nature of induced oscillations, aligning with the idea that induced activity represents brief neural assembly synchronization in response to stimuli. Thus, the group differences reported here might indicate changes in ongoing processing associated with integrating stimuli and other cognitive or sensory functions. Increases in induced power may not necessarily indicate dysfunctional processing mechanisms; instead, they could reveal differences in how the visual system communicates with other brain areas in autism, highlighting unique aspects of neural circuitry. Future studies are essential to understanding how simultaneous shifts in evoked and induced gamma may translate to altered sensory experiences in autism.

More broadly, the present findings align with prior evidence of reduced evoked but increased induced gamma responses in autism. Specifically, Rojas and colleagues (2008) reported a similar evoked-induced power distinction in autism, demonstrated by lower evoked but higher induced gamma power to a simple auditory tone in children with autism. This consistent pattern suggests that the differences in evoked and induced gamma activity reported here are not unique to visual processing but may represent a broader characteristic of neural circuit differences in autistic individuals, affecting basic low-level sensory processing across multiple sensory modalities. Furthermore, the contrasting gamma patterns reported here highlight the importance of studying both evoked and induced activities to gain a more nuanced understanding of neural circuits. For instance, composite metrics that can quantify the relative strength of induced and evoked gamma activity within individual participants could provide novel avenues to advance biomarker development.

4.1. Limitations.

While our study benefits from a large sample size, a key strength in autism research, it is relatively restricted in other sample characteristics. Importantly, children in the present sample do not reflect the diverse range of abilities and demographic backgrounds that are present in autism, including individuals with significant language or cognitive challenges. Therefore, future research is needed to expand the sample diversity to test the applicability of our findings across a broader autism sample. The checkerboard-reversal VEP paradigm utilized in this study is passive, brief (lasting only three minutes), and does not require explicit task comprehension, making it a widely used method across samples accommodating a range of ages and abilities. However, it is important to note that in the present analyses, autistic children had fewer trials available, indicating a higher amount of unattended and artifactual data. This potential limitation should be considered when including a broader range of cognitive and language abilities in future studies. However, subsequent research employing this paradigm could help determine if variability in gamma metrics maps onto wider phenotypic heterogeneity in autism and other neurodevelopmental conditions.

It is also important to note that while informative, the gamma differences identified in this study are not immediately applicable as individual-level biomarkers. For example, substantial inter-individual variability within each group complicates the use of gamma measures for clear group differentiation. To advance the search for precise biomarkers, future research should focus on how gamma activity patterns relate to individual traits or behaviors, which may help to characterize specific subgroups of children with autism. Additionally, while the present study does not explicitly focus on measures of trial-to-trial variability, examining such variability could offer key insights. Specifically, leveraging individual trial responses may provide important opportunities to capture detailed insights from VEP data and could yield a more in-depth understanding of autism-specific visual processing. In our ongoing work, we plan to utilize sophisticated signal processing techniques to gain a clearer understanding of this variability, thereby enriching our perspective on the complex neural dynamics in autistic individuals.

4.2. Conclusions.

In conclusion, our study highlights the significant role of gamma oscillations in understanding neural differences underlying autism. Importantly, our findings indicate that differences in gamma power dynamics in children with autism are not confined to specific time points or frequency bands. Instead, they span a range of gamma frequencies and time periods within the visual response, affecting both evoked and induced activities in distinct ways. This complex pattern suggests that neural processing deviations in autism are not characterized by one simple circuit alteration. Instead, the time-frequency differences reported here might indicate both perturbations and possible compensatory mechanisms, providing a more nuanced perspective for understanding the variability inherent in autism.

Figure 1. Example Time-Frequency Representations of Evoked and Induced Gamma Power for Individual Participants.

Figure 1.

Open in a new tab

Individual plots depict both evoked (phase-locked) and induced (non-phase-locked) gamma power for A) 5 participants with autism and B) 5 TD participants (randomly selected). Each row represents a participant, and power (in dB) is color-coded as indicated by the color bar.

Figure 4. Frequency Profiles and Individual Power Values within Significant Evoked and Induced Clusters.

Figure 4.

Open in a new tab

Top Panel: Frequency Profiles for A) Evoked and B) Induced clusters. Average power values for both groups (Autism shown in red; TD shown in blue) are plotted across frequencies, capturing the distinct frequency profile for each significant cluster within its designated time range. Background gray-shaded regions indicate the specific frequencies where significant group differences were observed within each cluster. Bottom Panel: Average power values for every participant are shown for A) Evoked and B Induced clusters. Average power values were computed based on the mean power values within the time and frequency boundaries of each cluster.

Acknowledgement:

Support for the Autism Biomarkers Consortium for Clinical Trials was provided by NIMH U19 MH108206. NIH scientific partners and members of the FNIH Biomarkers Consortium served on the Steering Committee and Biomarkers Consortium Project Team and provided consultation on study design and analysis. No company contributed to funding of this study. Additional important contributions were provided by members of the ABC-CT consortium including: Madeline Aubertine, Jessica Benton, Cynthia Brandt, Carter Carlos, Shou-An A. Chang, Kelsey Dommer, Alyssa Gateman, Simone Hasselmo, Julie Holub, Toni Howell, Ann Harris, Alexander Hoslet, Kathryn Hutchins, Kelsey Jackson, Scott Johnson, Lily Katsovitch, Minah Kim, Beibin Li, Kelsey MacDonald, Samantha Major, Samuel Marsan, Adriana S. Méndez Leal, Takumi McAllister, Lisa Nanamaker, Leon Rozenblit, Megha Santhosh, Helen Seow, Laura Simone, Dylan Stahl, Cindy Voghell, Andrew Yuan. Consultation for the ABC-CT EEG was provided by the EU Aims LEAP team, including Emily J.H. Jones and Luke Mason.

Footnotes

Conflict of Interest. Geraldine Dawson is on the Scientific Advisory Boards of Akili, Inc, Nonverbal Learning Disability Project, and Tris Pharma, Inc., received speaker fees from WebMD and book royalties from Guilford Press, Oxford University Press, Springer Nature Press. Dawson reports grant funding from NICHD, NIMH, and the Simons Foundation. Dawson has developed technology, data, and/or products that have been licensed to Apple, Inc. and Dawson and Duke University have benefited financially. James C. McPartland consults with Customer Value Partners, Bridgebio, Determined Health, Apple, and BlackThorn Therapeutics, has received research funding from Janssen Research and Development, serves on the Scientific Advisory Boards of Pastorus and Modern Clinics, and receives royalties from Guilford Press, Lambert, Oxford, and Springer. Dr. Shic has served as a consultant for BlackThorn Therapeutics, Janssen Research and Development, and Hoffmann–La Roche. April Levin has served as a consultant for Jaguar Gene Therapies and Lab 1636, LLC. All other authors (AD, DR, GM, HL, RB, KC, JD, SF, NK, CS, SJW, SJ) declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Ethical statement

The research was conducted in accordance with the principles embodied in the Declaration of Helsinki and in accordance with local statutory requirements. Human subjects oversight was provided by the Yale University Human Research Protection Program. To participate in the study, the parent or legal guardian provided written informed consent and the child provided assent.

Data Availability

We refer to a number of experiments as well as support documents detailing our standard operation procedures and manuals of operation for the ABC-CT; these documents can be accessed by request from the Principal Investigator (james.mcpartland@yale.edu) and data are available via NIMH NDA (#2288).

References

  1. American Psychiatric Association. (2013). Diagnostic and Statistical Manual of Mental Disorders (5th ed.). American Psychiatric Association. 10.1176/appi.books.9780890425596 [DOI] [Google Scholar]
  2. Arieli A, Sterkin A, Grinvald A, & Aertsen A (1996). Dynamics of Ongoing Activity: Explanation of the Large Variability in Evoked Cortical Responses. Science, 273(5283), 1868–1871. 10.1126/science.273.5283.1868 [DOI] [PubMed] [Google Scholar]
  3. Bartos M, Vida I, & Jonas P (2007). Synaptic mechanisms of synchronized gamma oscillations in inhibitory interneuron networks. Nature Reviews. Neuroscience, 8(1), Article 1. 10.1038/nrn2044 [DOI] [PubMed] [Google Scholar]
  4. Beker S, Foxe JJ, Venticinque J, Bates J, Ridgeway EM, Schaaf RC, & Molholm S (2021). Looking for consistency in an uncertain world: Test-retest reliability of neurophysiological and behavioral readouts in autism. Journal of Neurodevelopmental Disorders, 13(1), 43. 10.1186/s11689-021-09383-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Brittenham C, Gordon J, Zemon VM, & Siper PM (2022). Objective frequency analysis of transient visual evoked potentials in autistic children. Autism Research, 15(3), 464–480. [DOI] [PubMed] [Google Scholar]
  6. Cardin JA, Carlén M, Meletis K, Knoblich U, Zhang F, Deisseroth K, Tsai L-H, & Moore CI (2009). Driving fast-spiking cells induces gamma rhythm and controls sensory responses. Nature, 459(7247), Article 7247. 10.1038/nature08002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cohen MX (2014). Analyzing Neural Time Series Data: Theory and Practice. MIT Press. [Google Scholar]
  8. David O, Kilner JM, & Friston KJ (2006). Mechanisms of evoked and induced responses in MEG/EEG. NeuroImage, 31(4), 1580–1591. 10.1016/j.neuroimage.2006.02.034 [DOI] [PubMed] [Google Scholar]
  9. Delorme A, & Makeig S (2004). EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. Journal of Neuroscience Methods, 134, 9–21. 10.1016/j.jneumeth.2003.10.009 [DOI] [PubMed] [Google Scholar]
  10. Dinstein I, Heeger DJ, Lorenzi L, Minshew NJ, Malach R, & Behrmann M (2012). Unreliable evoked responses in autism. Neuron, 75(6), Article 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Faja S, Sabatos-DeVito M, Sridhar A, Kuhn JL, Nikolaeva JI, Sugar CA, Webb SJ, Bernier RA, Sikich L, Hellemann G, Senturk D, Naples AJ, Shic F, Levin AR, Seow HA, Dziura JD, Jeste SS, Chawarska K, Nelson CA III, … Trials A. B. C. for C. (2023). Evaluation of clinical assessments of social abilities for use in autism clinical trials by the autism biomarkers consortium for clinical trials. Autism Research, 16(5), 981–996. 10.1002/aur.2905 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Farashi S, Jenabi E, Bashirian S, Fayyazi A, Rezaei M, & Razjouyan K (2023). Differences Between Autism Spectrum Disorder and Typically Developing Individuals During Visual Information Processing—A Systematic Review and Meta-analysis Focusing on Visual Event-Related Potentials. Review Journal of Autism and Developmental Disorders. 10.1007/s40489-023-00383-5 [DOI] [Google Scholar]
  13. Haigh SM, Heeger DJ, Dinstein I, Minshew N, & Behrmann M (2015). Cortical Variability in the Sensory-Evoked Response in Autism. Journal of Autism and Developmental Disorders, 45(5), 1176–1190. 10.1007/s10803-014-2276-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hamdan F, Shareef H, & Jasim H (2022). Visual evoked potential findings and correlation between visual evoked potential and clinical severity in children with autism spectrum disorder. Medical Journal of Babylon, 19(2), 133. 10.4103/MJBL.MJBL_88_21 [DOI] [Google Scholar]
  15. Jasper H (1958). The ten twenty electrode system of the international federation. Electroencephalography and Clinical Neuroph Siology, 10, 371–375. [PubMed] [Google Scholar]
  16. Jeste SS, Frohlich J, & Loo SK (2015). Electrophysiological biomarkers of diagnosis and outcome in neurodevelopmental disorders. Current Opinion in Neurology, 28(2), Article 2. 10.1097/WCO.0000000000000181 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Jeste SS, & Nelson CA (2009). Event Related Potentials in the Understanding of Autism Spectrum Disorders: An Analytical Review. Journal of Autism and Developmental Disorders, 39(3), 495–510. 10.1007/s10803-008-0652-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Kovarski K, Malvy J, Khanna RK, Arsène S, Batty M, & Latinus M (2019). Reduced visual evoked potential amplitude in autism spectrum disorder, a variability effect? Translational Psychiatry, 9(1), 341. 10.1038/s41398-019-0672-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kovarski K, Thillay A, Houy-Durand E, Roux S, Bidet-Caulet A, Bonnet-Brilhault F, & Batty M (2016). Brief Report: Early VEPs to Pattern-Reversal in Adolescents and Adults with Autism. Journal of Autism and Developmental Disorders, 46(10), 3377–3386. 10.1007/s10803-016-2880-8 [DOI] [PubMed] [Google Scholar]
  20. Levin AR, Naples AJ, Scheffler AW, Webb SJ, Shic F, Sugar CA, Murias M, Bernier RA, Chawarska K, Dawson G, Faja S, Jeste S, Nelson CA, McPartland JC, Şentürk D, & and the Autism Biomarkers Consortium for Clinical Trials. (2020). Day-to-Day Test-Retest Reliability of EEG Profiles in Children With Autism Spectrum Disorder and Typical Development. Frontiers in Integrative Neuroscience, 14. https://www.frontiersin.org/articles/10.3389/fnint.2020.00021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Maris E (2012). Statistical testing in electrophysiological studies. Psychophysiology, 49(4), 549–565. 10.1111/j.1469-8986.2011.01320.x [DOI] [PubMed] [Google Scholar]
  22. Maris E, & Oostenveld R (2007). Nonparametric statistical testing of EEG- and MEG-data. Journal of Neuroscience Methods, 164(1), 177–190. 10.1016/j.jneumeth.2007.03.024 [DOI] [PubMed] [Google Scholar]
  23. McPartland JC, Bernier RA, Jeste SS, Dawson G, Nelson CA, Chawarska K, Earl R, Faja S, Johnson SP, Sikich L, Brandt CA, Dziura JD, Rozenblit L, Hellemann G, Levin AR, Murias M, Naples AJ, Platt ML, Sabatos-DeVito M, … the Autism Biomarkers Consortium for Clinical Trials. (2020). The Autism Biomarkers Consortium for Clinical Trials (ABC-CT): Scientific Context, Study Design, and Progress Toward Biomarker Qualification. Frontiers in Integrative Neuroscience, 14, 16. 10.3389/fnint.2020.00016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Milne E (2011). Increased intra-participant variability in children with autistic spectrum disorders: Evidence from single-trial analysis of evoked EEG. Frontiers in Psychology, 2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Modi ME, & Sahin M (2017). Translational use of event-related potentials to assess circuit integrity in ASD. Nature Reviews Neurology, 13(3), Article 3. 10.1038/nrneurol.2017.15 [DOI] [PubMed] [Google Scholar]
  26. Onton J, Westerfield M, Townsend J, & Makeig S (2006). Imaging human EEG dynamics using independent component analysis. Neuroscience & Biobehavioral Reviews, 30(6), Article 6. [DOI] [PubMed] [Google Scholar]
  27. Oostenveld R, Fries P, Maris E, & Schoffelen J-M (2011). FieldTrip: Open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data. Computational Intelligence and Neuroscience, 2011, 1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Pernet CR, Latinus M, Nichols TE, & Rousselet GA (2015). Cluster-based computational methods for mass univariate analyses of event-related brain potentials/fields: A simulation study. Journal of Neuroscience Methods, 250, 85–93. 10.1016/j.jneumeth.2014.08.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Port RG, Gandal MJ, Roberts TPL, Siegel SJ, & Carlson GC (2014). Convergence of circuit dysfunction in ASD: a common bridge between diverse genetic and environmental risk factors and common clinical electrophysiology. Frontiers in Cellular Neuroscience, 8, 414. 10.3389/fncel.2014.00414 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Rojas DC, Maharajh K, Teale P, & Rogers SJ (2008). Reduced neural synchronization of gamma-band MEG oscillations in first-degree relatives of children with autism. BMC Psychiatry, 8(1), Article 1. 10.1186/1471-244X-8-66 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Sassenhagen J, & Draschkow D (2019). Cluster-based permutation tests of MEG/EEG data do not establish significance of effect latency or location. Psychophysiology, 56(6), e13335. 10.1111/psyp.13335 [DOI] [PubMed] [Google Scholar]
  32. Sayorwan W, Phianchana N, Permpoonputtana K, & Siripornpanich V (2018). A Study of the Correlation between VEP and Clinical Severity in Children with Autism Spectrum Disorder. Autism Research and Treatment, 2018, e5093016. 10.1155/2018/5093016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Siper PM, Zemon V, Gordon J, George-Jones J, Lurie S, Zweifach J, Tavassoli T, Wang AT, Jamison J, & Buxbaum JD (2016). Rapid and objective assessment of neural function in autism spectrum disorder using transient visual evoked potentials. PLoS One, 11(10), e0164422. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Tallon-Baudry C, Bertrand O, Delpuech C, & Pernier J (1996). Stimulus Specificity of Phase-Locked and Non-Phase-Locked 40 Hz Visual Responses in Human. Journal of Neuroscience, 16(13), 4240–4249. 10.1523/JNEUROSCI.16-13-04240.1996 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Traub RD, Spruston N, Soltesz I, Konnerth A, Whittington MA, & Jefferys JGR (1998). GAMMA-FREQUENCY OSCILLATIONS : A NEURONAL POPULATION PHENOMENON, REGULATED BY SYNAPTIC AND INTRINSIC CELLULAR PROCESSES, AND INDUCING SYNAPTIC PLASTICITY. Progress In, 55(98), Article 98. http://www.sciencedirect.com/science/article/pii/S0301008298000203 [DOI] [PubMed] [Google Scholar]
  36. Webb SJ, Naples AJ, Levin AR, Hellemann G, Borland H, Benton J, Carlos C, McAllister T, Santhosh M, Seow H, Atyabi A, Bernier R, Chawarska K, Dawson G, Dziura J, Faja S, Jeste S, Murias M, Nelson CA, … McPartland JC (2023). The Autism Biomarkers Consortium for Clinical Trials: Initial Evaluation of a Battery of Candidate EEG Biomarkers. American Journal of Psychiatry, 180(1), Article 1. 10.1176/appi.ajp.21050485 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Webb SJ, Shic F, Murias M, Sugar CA, Naples AJ, Barney E, Borland H, Hellemann G, Johnson S, Kim M, Levin AR, Sabatos-DeVito M, Santhosh M, Senturk D, Dziura J, Bernier RA, Chawarska K, Dawson G, Faja S, … the Autism Biomarkers Consortium for Clinical Trials. (2020). Biomarker Acquisition and Quality Control for Multi-Site Studies: The Autism Biomarkers Consortium for Clinical Trials. Frontiers in Integrative Neuroscience, 13, 71. 10.3389/fnint.2019.00071 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Whittington MA, & Traub RD (2003). Interneuron Diversity series: Inhibitory interneurons and network oscillations in vitro. http://www.sciencedirect.com/science/article/pii/S0166223603003369 [DOI] [PubMed]
  39. Whittington MA, Traub RD, Kopell N, Ermentrout B, & Buhl EH (2000). Inhibition-based rhythms: Experimental and mathematical observations on network dynamics. 38, 315–336. 10.1016/S0167-8760(00)00173-2 [DOI] [PubMed] [Google Scholar]
  40. Yuval-Greenberg S, Tomer O, Keren AS, Nelken I, & Deouell LY (2008). Transient Induced Gamma-Band Response in EEG as a Manifestation of Miniature Saccades. Neuron, 58(3), Article 3. 10.1016/j.neuron.2008.03.027 [DOI] [PubMed] [Google Scholar]
  41. Zemon VM, & Gordon J (2018). Quantification and statistical analysis of the transient visual evoked potential to a contrast-reversing pattern: A frequency-domain approach. European Journal of Neuroscience, 48(2), 1765–1788. 10.1111/ejn.14049 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

We refer to a number of experiments as well as support documents detailing our standard operation procedures and manuals of operation for the ABC-CT; these documents can be accessed by request from the Principal Investigator (james.mcpartland@yale.edu) and data are available via NIMH NDA (#2288).

RESOURCES