Introduction
Selective mutism (SM) is a mental health disorder in which affected children fail to speak in certain social situations where they are expected to, but their speech remains unaffected in other situations [
1]. The disorder is associated with a chronic course of increased psychopathological symptoms persisting into adulthood and resulting in impairments in academic and socio-emotional development [
2‐
4]. Previous research suggests a prevalence of approximately 1% [
5], although the occurrence of SM is probably underestimated [
6]. With the introduction of the DSM-5 [
1], SM was classified among anxiety disorders for the first time. The reason for this was evidence that SM shares numerous similarities with other anxiety disorders, particularly social anxiety disorder (SAD) [
5]. SAD is characterized by a marked fear of being evaluated by others in social situations as well as physiological symptoms [
1]. The overlap between SM and SAD is, for example, reflected in high comorbidity rates up to 100% of SAD in children with SM [
7] and findings that social fear is a central phenomenon in children with SM [
8]. Additionally, etiological similarities between both disorders exist such as the increased risk of the temperament of Behavioral Inhibition (BI) during early childhood [
9].
In light of this extreme overlap, hitherto an unresolved question is why children with SM fail to speak in certain social situations, whereas children with SAD do not, despite both clinical groups experiencing marked social fear. In this context, SM has been conceptualized as an extreme variant of SAD with an extreme fear and overarousal in social situations, in which failure to speak functions as an avoidance mechanism leading to a reduction in fear arousal [
10‐
15]. While the assumption of SM as an extreme version was based on findings that children with SM are rated by clinicians as more anxious in social situations than children with SAD [
11,
14], comparable levels of self-reported social trait anxiety between both groups speaks against this notion [
11,
14,
16]. However, recent studies on children with SM suggest that it is important to differentiate between social situations in which children are expected to speak (verbal) and situations which do not require speech (nonverbal) [
10,
13,
17]. During nonverbal social situations, children with SM and SAD report comparable levels of state anxiety [
13] and are rated as comparably anxious by parents [
17], teachers [
17] and clinicians [
10], but still show increased levels of fear compared to children with typical development (TD) [
10,
13,
17]. In verbal social situations, children with SM are rated as more anxious by clinicians [
10] and report higher levels of fear than children with SAD [
13]. Therefore, research supports the assumption that children with SM show an extreme level of fear compared to children with SAD, even if this only applies to verbal situations. This suggests that failure to speak is related to extreme fear in verbal situations, which implies that it is part of a dysfunctional fear response. Consequently, differences in fear responses between SM and SAD may provide insights into mechanisms of failure to speak. However, support mainly comes from studies using data based on self-report or behavioral observations, and few quasi-experimental studies exist, which use objective measures of fear such as physiological processes. When faced with a fear-inducing situation, a fear response occurs in order to cope with the threat. Immediate fear responses to threat in the form of fight, flight or freeze are mediated by the autonomic nervous system [
18], which consists of a sympathetic and a parasympathetic part. The presence of threat leads to the activation of the autonomic nervous system, which results in an increase in sympathetic and decrease of parasympathetic activity [
18]. This change of arousal in response to a fear-related situation already initiates in anticipation of the threat [
19,
20] and is called reactivity (in experimental designs defined as change from baseline to confrontation with threat). Reactivity is associated with a higher sympathetic activity indicated by an increase in skin conductance level (SCL) and heart rate (HR) and a reduced parasympathetic activity indicated by a decrease of the respiratory sinus arrhythmia (RSA) [
18,
21]. Recovery from threat leads to decrease of SCL and HR and increase of RSA. Aside from the phasic change in physiological arousal in a fear-inducing situation (reactivity or recovery), the basic level of arousal in the absence of stress is called tonic autonomic arousal [
22]. Disorder-specific models of other anxiety disorders [
23], as well as psychophysiological theories of anxiety in general, propose physiological mechanisms to be involved in the symptomatology of anxious individuals [
22,
24,
25]. In contrast, no empirically based model exists in SM and mechanisms of failure to speak are largely unclear, as psychophysiological research in children with SM is scarce. The assumption that failure to speak occurs due to more intense fear and associated overarousal in children with SM compared to children with SAD implies that children with SM have
higher autonomic reactivity during an expectation to speak. At the same time, disorder-specific models of SAD suggest that individuals with SAD exhibit increased reactivity compared to non-anxious individuals [
23]. However, the few existing studies on autonomic activity in SM [
14,
26] do not support the notion of a higher reactivity in SM compared to SAD as a mechanisms of failure to speak. In this context, Heilman et al. [
26], who did not include a group of children with SAD, found no difference in reactivity during a non-standardized verbal social stress task between children with SM and typical development (TD). Young et al. [
14] did not examine the change of autonomic arousal from the baseline to a non-standardized speech demanding task, but only investigated the level of arousal during both phases. Therefore, no conclusions about reactivity of the autonomic arousal can be drawn from this study. Also against the assumption of SAD models, most psychophysiological studies do not show a higher reactivity in children with SAD compared to children with typical development (TD) during social situations [
17,
18,
21‐
24].
In contrast to the higher reactivity suggested by disorder-specific theoretical assumptions, psychophysiological theories of anxiety propose a
lower reactivity to be indicative of a pathological fear response [
18,
22]. In line with the idea that a fear response is an adaptation to the fear-inducing situation, it is assumed that strong and flexible reactivity is functional, as long as the arousal recovers quickly after the situation. Instead, they propose the combination of a
tonic elevated autonomic arousal (in the absence of stress) and a blunted response of the autonomic nervous system to stress, consisting of a
lower reactivity (lower increase of arousal) and
slower recovery to be indicative of pathological anxiety [
18,
22]. Tonically increased autonomic arousal is associated with reduced responsiveness of the autonomic nervous system [
22]. The combination of high tonic arousal and blunted response is also known as restrictive autonomic flexibility, which can transdiagnostically be found in anxious individuals and has been interpreted as chronic dysregulation of the autonomic nervous system [
27,
28]. While autonomic flexibility has not directly been studied in children with SM, research in children with SAD consistently indicates a restricted autonomic flexibility during social stress. Both an increased tonic arousal [
27,
28] as well as a blunted response to social stress, with lower physiological reactivity and a subsequent slow reduction of arousal can be found in children with SAD [
28‐
30], while the restriction of autonomic flexibility is positively related to the level of state anxiety [
31]. However, physiological studies in socially anxious children have been criticized for being limited in indicating whether arousal is actually tonically increased or whether arousal is already influenced by the unfamiliar laboratory setting (e.g., presence of strangers) that induces stress in this clinical group [
32]. For this reason, Asbrand et al. [
32] examined resting arousal in a familiar environment and the absence of social stress, and found that children with SAD showed higher autonomic resting arousal here as well. This tonically increased arousal, even in non-stress inducing situations, can already be observed in toddlers with a high level of the temperamental BI feature [
33‐
35], which is a precursor for both SM and SAD.
Given that inflexibility of the fear response, rather than high reactivity, is indicative of pathological anxiety, and that children with SM experience an extreme fear in verbal social situations compared with SAD, SM might be associated with an even stronger inflexibility of the fear response. However, psychophysiological studies focusing on autonomic activity in children with SM have not yet examined tonic autonomic arousal in the absence of stress, nor do they include direct analyses of both reactivity and recovery over the course of a stressful situation [
14,
26]. Young et al. [
14] found no significant differences in level of autonomic arousal between children with SM, SAD, and TD during a baseline in a laboratory setting, whereas statistical power was limited due to the very small sample size. The authors did not examine arousal after the verbal stress task and did not analyze differences between phases, so no conclusions about reactivity or recovery can be drawn from this study. Heilman et al. [
26] examined baseline arousal in a laboratory setting in combination with physiological responses during and after a verbal social stress task. Although the authors examined reactivity to the verbal stress task, they did not examine recovery. The authors found a higher HR as well as a lower RSA in children with SM compared to children with TD during a baseline condition. Children with SM showed lower RSA after the task, indirectly suggesting a slower recovery from the stressor. However, the authors did not correct for differences in level of arousal between groups (baseline correction), did not use a standardized social stress task and did not control for speech production, which is known to influence physiological responses [
36]. Taken together, these findings also seem to indicate a restrictive autonomic flexibility in children with SM, although not all components in SM have yet been investigated. Furthermore, baseline measurements of both studies were performed at an unfamiliar place and in the presence of strangers, both of which induce symptoms in children with SM. In this respect, it is questionable whether the findings of Heilman et al. [
26] indicate a chronically increased arousal in children with SM or are a consequence of the factors inducing symptoms in SM. Complementary to the assumption that SM is associated with extreme fear, it has been proposed that failure to speak is an avoidance mechanism [
5,
12,
14,
37]. Here, the assumption is that the intense fear experienced by children with SM during a verbal social situation is reduced by the failure to speak [
14]. While this is supported by evidence from a qualitative study in which individuals with SM report in some cases that failure to speak is associated with the experience of safety [
8], there is also initial evidence from the above-mentioned psychophysiological studies. Children with SM showed a comparable level of autonomic arousal as children with TD during verbal tasks of both above-mentioned physiological studies and a lower arousal than children with SAD in the study of Young et al. [
14]. The lower arousal compared to children with SAD has especially been interpreted in favor of considering failure to speak as avoidance mechanism. However, the design of both studies [
14,
26] is not completely suitable for testing this assumption due to two reasons: (1) they did not examine a detailed course of the fear response. While an increase in autonomic arousal already starts in anticipation of a fear-inducing situation, it is implicit in the assumption of the avoidance mechanism that arousal decreases again as soon as the expectation to speak sets in and the child fails to speak. Therefore, it is important to capture the course of anticipation of the fear-inducing situation and the situation in which the child does not speak. (2) Both studies did not control for all important confounding variables. In this context, it is especially important to control speech production during the verbal stress task, as it has an impact on the physiological response. Additionally, in order to investigate a possible avoidance mechanism, it would be important to consider a nonverbal social stress task, which also induces fear in children with SM but cannot be avoided by failure to speak. Given that previous studies have not done this, a comparison of fear responses between these situations could provide insight into whether reduction happens only in the verbal situation that can be avoided by failure to speak.
Therefore, the current study aims to examine the responses of the autonomic nervous systems of children with SM compared to children with SAD and TD. In addition to tonic autonomic arousal during rest, the fear response (reactivity and recovery) is investigated with regard to both verbal and non-verbal social standardized stress situations. Additional to autonomic flexibility, we examine the course of the fear response comparing both types of situations in order to capture a potential reduction of arousal through failure to speak in SM. In order to gain detailed insight into the course of the fear response, we analyze arousal in different phases of the fear-inducing situations (baseline, anticipation, performance, recovery). While a story is to be retold in the verbal task, the nonverbal task consists of drawing a previously heard story, so that it cannot be avoided by failure to speak. To induce social stress, both tasks take place in front of an unfamiliar experimenter and children had been told that other children will later evaluate the recorded tasks. Furthermore, we analyze and control for possible baseline differences as well as confounding variables such as age, gender, and the number of spoken words for the analysis of physiological response. Furthermore, we take into account the level of BI retrospectively assessed for toddlerhood, as this has to be shown to have an influence on the tonic arousal level.
We aim to address the following hypotheses:
1.
Autonomic flexibility: We aim to shed light on the fear response according to the concept of restricted autonomic flexibility in children with SM:
1.1.
Resting Arousal:
a.
We expect children with SM and SAD will show higher tonic autonomic arousal (indicated by elevated HR and SCL and reduced RSA) during a rest period in a non-threatening environment compared to TD children.
b.
To investigate whether the higher arousal is rooted in the temperament BI, which is a precursor to both disorders, we test whether the level of BI can predict arousal beyond the symptomatology of SM and SAD.
1.2.
Reactivity and recovery during nonverbal social stress: We expect that children in the clinical groups will show a lower reactivity and a slower recovery on sympathetic and parasympathetic markers during nonverbal social stress compared to TD children (SM = SAD > TD).
1.3.
Reactivity and recovery during verbal situations:
We further expect the children with SM will show a lower reactivity and slower recovery during verbal stress compared to both children with SAD and children with TD and we expect children with SAD will show lower reactivity and slower recovery compared to children with TD (SM > SAD > TD).
2.
Avoidance: We aim to investigate the failure to speak as a potential disorder-specific mechanism in SM. In order to address limitations of previous physiological studies in SM [
14,
26], we aim to investigate the course of the fear response in children with SM and compare it between a verbal and a nonverbal social stress situation. Hereby, we want to identify whether a reduction of the physiological arousal occurs with the onset of an expectation to speak after it increases in anticipation of the fear-inducing situation.
2.1.
Reduction of arousal: We hypothesize that autonomic arousal in children with SM will increase (increase of HR and SCL and decrease of RSA) in anticipation of the speech demanding situation and will decrease (decrease of HR and SCL and increase of RSA) during the speech demand (performance of verbal task).
2.2.
Comparison of paradigms: We hypothesize that a reduction from anticipation to performance of the stress task in children with SM will only occur in a verbal task with the expectation to speak, but not in a nonverbal social stress task that cannot be avoided by failure to speak. We therefore assume that the difference in autonomic arousal between anticipation and performance of the verbal social stress task (performance–anticipation) in children with SM will be larger than in the nonverbal situation.
Limitations
There are methodological limitations, which confine the conclusions of our study. First, the stress paradigm used in the study is standardized but not validated. However, instructions and procedures were closely aligned with the TSST-C and there was a main effect phase with respect to physiological responses. Therefore, it can be assumed that the paradigms successfully induced social stress. Secondly, the recovery phases in our paradigm were shorter than in previous studies in children with SAD, so that we cannot make any assumption regarding the long-term recovery of the arousal. Thirdly, we have some heterogeneity in our SM group, as we included children with SM and comorbid SAD as well as non-socially anxious children with SM. Although the rate of SAD found in our SM group (80%) is consistent with rates reported in the literature, this heterogeneity did not allow for a clear differentiation between physiological processes in children with both SM and SAD, children with SAD only, and children with SM only and thus disorder-specific conclusions are limited. However, because we performed additional analyses within the SM group and correlational analyses based on symptom scores, we nonetheless also obtained disorder-specific insights based on our results. To fully disentangle disorder-specific mechanisms, future studies should include larger samples with the aforementioned subgroups. Fourthly, it is important to note that we were not able to diagnose autism-spectrum-disorder, speech or language disorders or assess information regarding developmental history based on the clinical interview used in the current study. Furthermore, we also did not capture cognitive abilities of the participating children. Fifthly, given that we chose a rather narrow age range of 8–12 years old due to the influence of age on physiological responses reported in the literature, our results cannot be generalized to other age groups. Therefore, the physiological findings found here and derived explanations for the function of failure to speak cannot be applied to other age groups (e.g. children at the preschool age or adolescents with SM). In order to be able to draw valid conclusions regarding etiological mechanisms of SM, studies with younger children and a longitudinal design would be of great relevance. Sixthly, although there was no experimenter in the room during the measurement at rest, the children may still have experienced participation in the study itself as a socially evaluative setting. However, because the groups did not differ in the subjective level of anxiety during the measurement at rest, it can be assumed that a possibly induced stress level was reduced to a minimum. Seventhly, it is important to emphasize that although we had sufficiently large statistical power for the main analyses, the further analyses of children with SM who spoke and who did not speak were based on only very small sample sizes. Future studies should more closely examine the mechanisms identified in the present study using larger sample sizes.