The main findings in the current study were (1) atypical EDA in the ASD group characterized by a blunted mean EDA reactivity to the anxiety tasks and decreased EDR reactivity in all tasks, (2) negative correlation between EDR frequency and behavioral scores in the anxiety and social domains in the ASD group, and (3) differential mean EDA and behavioral patterns between the low- and high-anxiety subgroups within the ASD group.
Atypical electrodermal activity in ASD
Given the findings in previous literature, we expected that the ASD group would exhibit sympathetic hyperarousal evidenced by atypically increased or decreased EDA level and increased EDR frequency. Although a trend toward decreased mean EDA was observed in the ASD group overall, group differences did not reach statistical significance for any of the EDA measures in our study collapsed across conditions. This may be related to the large variability in autonomic function in this sample. As discussed in the next section, anxiety symptomatology may explain some of this variance and could be used to derive more homogeneous subgroups with respect to electrodermal activity.
Our data also showed that ASD was associated with blunted reactivity to tasks eliciting anxiety, attention, response inhibition (marginally significant), and social cognition. Blunted reactivity to psychosocial challenges in ASD has been previously reported in the literature both in the context of autonomic cardiac measures [
4,
12,
15] and hypothalamic–pituitary–adrenal axis function [
15]. The present study adds to this literature by revealing a dampened response to other mental and cognitive tasks. Such dampened reactivity has also been reported across a number of psychiatric and affective conditions including attention deficit/hyperactivity and conduct disorder [
39], alexithymia [
40], depression [
41], and high trait anxiety [
42,
43]. The inability to regulate physiological responses to environmental stimuli may also be related to emotion regulation difficulties. These difficulties, together with impaired attentional control, may play a role in the early emergence of ASD symptoms [
44] (e.g., by hindering the experience of positive associations from interactions with others). Interestingly, markers of a well-regulated ANS have been associated with improved social function in children with ASD [
45]. Further research is needed to understand the associations between ANS atypicalities and ASD symptomatology.
It is important to note that the literature findings on mean EDA reactivity to anxiogenic stimuli are mixed. Specifically, two previous studies did not find atypical mean EDA reactivity to anxiogenic stimuli in ASD during the Stroop [
27] and phychosocial challenges [
22]. The discrepancies may be related to differences in experimental conditions (e.g., nature of baseline activity) or sample characteristics (e.g., age, IQ, diagnosis status (ASD versus high-functioning autism), presence/exclusion of comorbid conditions) or may indicate a need for larger sampler sizes to capture significant differences in mean EDA, which exhibits high variability in this population.
Increased EDR frequency is generally associated with increased sympathetic activity. Our results therefore suggest decreased sympathetic reactivity in the ASD group. This may be indicative of a deficit in sympathetic modulation to meet task demands and associated with central/peripheral neurobiological differences in ASD. While there is currently no evidence to support differences in peripheral conduction, several neuroimaging/EDA studies suggest central influences on sympathetic atypicalities in ASD. In particular, a widespread network of regions, including the amygdalae, and the prefrontal, anterior cingulate, and insular cortices, has been associated with sympathetic modulation [
46]. ASD has been associated with differences in neuroanatomy, function, and connectivity in these regions [
47‐
49], which may affect sympathetic function. These regions have also been implicated in studies of social deficits in ASD [
47] as well as in neuro-circuitry of anxiety [
50]. It is therefore interesting that our results show a significant correlation between behavioral scores in these domains and decreased EDR frequency in the respective tasks. These results are also consistent with those of [
15] which suggested a negative correlation between anxiety symptom severity and heart rate responsiveness to social stress.
Further supporting atypical central autonomic processing, a study of resting state activity in ASD [
19] found that the EDR signal was positively correlated with activity in several regions involved in autonomic processing (e.g., anterior insular and cingulate cortices) in neurotypical controls, but not in the ASD group. In addition, the results of that paper also suggest that weaker default mode network connectivity in ASD may be partially explained by differences in EDA activity. Future neuroimaging studies are needed to further examine the relation among brain function, autonomic differences, and behavior in these domains.
The blunted task reactivity observed herein may also be related to compensatory down-regulation resulting from chronic exposure to stress [
15]. This would be consistent with the high prevalence of comorbid anxiety in ASD [
51] and previous reports of hyper-arousal in this population [
5‐
8,
27]. In this context, reduced reactivity may reflect inhibitory coping effects [
42].
Finally, other mechanisms may have contributed to decreased arousal during the tasks used in the study. These include deficits in allocation of attentional resources [
39] or other executive functions as well as lower levels of motivation to performance, engagement, or interest in study tasks [
40]. Future studies are needed to further investigate these issues.
ASD subgroups
Our results show that when split based on anxiety symptomatology, two different subgroups emerge within the ASD group, with the high-anxiety group exhibiting significantly decreased mean EDA relative to both the low-anxiety ASD and TD groups. Our results also mirror those reported in [
15] in which a high-anxiety group within ASD had lower heart rate than a low-anxiety ASD group. Overall, these results indicate that anxiety symptomatology may explain some of the variability in EDA findings in existing literature.
Paradoxically, decreased EDA may suggest both hyper- and hypo-arousal as the level of arousal input and physiological output are thought to follow an inverted U-shaped relation. In particular, arousal increases physiological output to a certain point, beyond which the physiological response decreases (a concept similar to Pavlov’s notion of transmarginal inhibition) [
52]. Consistent with this model, individuals with high trait anxiety have been shown to exhibit diminished EDA levels [
52]. Given that our groups were derived based on a measure of trait anxiety, our finding of decreased EDA levels may reflect hyper-arousal in this sample.
The pattern of decreased EDA was not task-specific in our data and was evident even during baseline phases. This may suggest that the observed EDA differences were likely not related to differences in responding to any particular task but to a more global physiological dysfunction or to differences in the overall experience of the experimental setting (e.g., coping with new environment and staff). Decreased levels of EDA have previously been reported in other populations with psychiatric difficulties including depression [
17], anti-social behavior [
53], and externalizing behavior disorders [
54].
In addition to having distinct physiological profiles, the high- and low-anxiety ASD groups differed significantly on a number of behavioral domains. In particular, the high-anxiety group showed more severe symptomatology on measures of affective and psychological difficulties (CBCL anxious/depressed, thought problems, attention problems, rule breaking, and aggressive behavior; RCADS anxiety total (marginally significant), obsessive-compulsive (marginally significant), and depression subscales). These results suggest a different behavioral profile between the two anxiety groups. Our results complement those reported in [
54] where adults with ASD who exhibited low EDA also showed poorer emotion recognition compared to those with higher EDA levels. Given the high rates of comorbidity in the high anxiety group, it remains possible that the findings are driven by the presence of greater comorbid symptoms overall (versus anxiety alone). Future research is needed to further understand the relation between anxiety and other comorbidities in this population.
Overall, our results suggest an interaction between sympathetic function and anxiety and ASD symptomatology. Future research is needed to further clarify the nature of these associations.
Limitations
Some limitations of the present study are noteworthy. First, in this study, sympathetic function as measured by electrodermal activity was examined in isolation. The output of the autonomic nervous system, however, is a result of complex interactions among central and peripheral mechanisms that include both the sympathetic and parasympathetic systems as well as the neuroendocrine system. Further studies are needed to examine these systems simultaneously and to pinpoint system-level differences in this area.
The second limitation of this study was that the specific measures of EDA such as the orienting response and habituation were not examined. This was mainly due to the continuous nature of tasks. Future studies designed specifically for examining these measures (e.g., using discrete stimuli) can further shed light on the nature of EDA atypicalities in this population.
Our sample included participants who were receiving psychopharmacological interventions which may affect autonomic function. While we controlled for the effect of medications on EDA measures in our analyses, future studies with larger sample sizes are needed to quantify these effects.
Finally, given the large variability in EDA measures in this population, our sample size may have contributed to null findings on both the physiological and behavioral domains.