Anxiety Disorders and Sensory Over-Responsivity in Children with Autism Spectrum Disorders: Is There a Causal Relationship?

Studies on hypervigilance and attention in individuals with anxiety disorders can inform our understanding of how anxiety might elicit SOR in children with ASD (see Fig. 1). Evidence indicates that anxiety disorders are characterized by hyperarousal (i.e., heightened baseline levels of autonomic arousal) which supports a state of hypervigilance (i.e., scanning the environment for threat-relevant stimuli, and preparation for potential threat). It is posited that, for anxious individuals, this hypervigilance is part of an overall difficulty in regulating negative affectivity, known as threat-based emotion regulation (Craske 2003). These individuals may have trouble shifting attention and activating or inhibiting behavior in such a way that decreases negative affectivity and arousal.

Studies using paradigms that assess attentional bias support the theory that anxious individuals have higher rates of environmental scanning (i.e., hypervigilance), followed by a narrowing of attention once a threat-relevant stimulus is detected, and then relative difficulty disengaging from that stimulus (Craske et al. 2009; Mobini and Grant 2007). For example, in the dot probe task, individuals with anxiety disorders show slower response latencies to a dot on the opposite side of the screen from a threat-relevant image or word than a dot on the same side of the screen, relative to healthy controls. These findings have been replicated with children, suggesting that relative to healthy controls, children with anxiety disorders show a stronger bias in selectively attending to threat-relevant stimuli, and difficulty disengaging from these stimuli even when they have been asked to attend to other information (Mobini and Grant 2007).

This type of anxious hypervigilance and poor regulation of negative arousal through attentional bias could contribute to SOR. If children are hyperaroused and scanning the environment, looking for threat, they are more likely to notice and react to environmental sensory stimuli. A threat-based style of emotion regulation would exacerbate this reactivity, as they would be more likely to attribute threat to these stimuli, and have difficulty shifting their attention and inhibiting negative affect. A study by Liss et al. (2006) lends some support to the relevance of this model for children with ASD. Liss and colleagues showed that parent ratings of SOR, excellent memory, and overfocused and overselective attention cluster together in a subgroup of children with ASD. The authors suggested that these individuals overreact to aversive sensory stimuli because they are more likely to attend to them and have difficulty disengaging from them. Thus, this study supports the idea that there is a subgroup of children with ASD who are particularly prone to SOR, and that these children show some symptoms similar to individuals with anxiety disorders, including difficulty with rapid disengagement, and selective attention.

Experimental manipulation of anxiety and sensory sensitivity in animal research lends further support to the theory that anxiety may lead to SOR by altering an individual’s ability to regulate his or her reactions to sensory input. In a study by Lepicard et al. (2003) inbred strains of mice that were genetically modified to be anxiety prone showed poorer balance and postural control than non-anxious mice. This may be an indicator of poor modulation of vestibular-proprioceptive input, which is often observed in children with SOR (Miller et al. 2007). Administration of an anxiolytic pharmacological compound to the anxious mice and anxiogenic compounds to the normal mice reversed the group differences so that the originally normal mice had poorer balance and postural control than the originally anxious mice. This causal effect can be found in other sensory modalities as well. For example, Plappert and Pilz (2002) found that mice that were genetically anxiety prone showed a stronger auditory startle reaction to sounds than non-anxious mice. However, such results must be interpreted with caution because the auditory startle reaction is also a symptom of anxiety. More studies are necessary to examine the response of genetically-anxious animals to non-startling sensory stimuli in order to avoid confounding the sensory response with the anxiety response.

Once over-reactivity has been established through hyperarousal, hypervigilance, and threat-based emotion regulation, it may be maintained or exacerbated by classical aversive conditioning, in which an innately aversive event (such as a loud noise) is associated with a previously neutral stimulus (such as an electronic toy). For example, if a child with ASD is constantly scanning the environment for sensory-related threat and preferentially attends to aversive, unexpected noises, he or she is more likely to overreact to these noises. The sensory overreaction feels unpleasant, and the child associates this unpleasant feeling with the aversive noise. This pairing of a conditioned stimulus (the loud noise) with a physiological reaction (e.g., fight-or-flight response) is known as interoceptive conditioning and is present in many different anxiety disorders (Razran 1961). Symptoms associated with anxiety disorders including trait anxiety, physiological arousal, and the perceived uncontrollability of an aversive event all contribute to stronger conditioning (Craske 2003), thus increasing the likelihood that a child with a preexisting anxiety disorder might associate a physiological reaction to a sensory stimulus. Once conditioning occurs, the child is more likely to scan the environment for potential aversive stimuli (e.g., balloons, radios, telephones), more likely to selectively attend to those objects, so more likely to notice a sound and react negatively to it when it does occur. Furthermore, children with anxiety disorders tend to regulate their negative affectivity through avoidance of stimuli that elicit fear (Craske 2003). Avoidance further decreases the possibility that the child will learn to regulate his or her response to the stimuli in an adaptive way. In this way, conditioning serves to maintain or exacerbate the connection between anxious hypervigilance and SOR. Thus, SOR could be a result of hypervigilance, difficulty regulating reactions to aversive sensory stimuli, and then, through conditioning and avoidance, maintaining and strengthening the association of over-reactivity with those stimuli.

SOR as a causal factor for anxiety can be conceptualized in terms of fear and conditioning (see Fig. 2). Unpleasant sensory stimuli (the unconditioned stimuli, such as an aversive noise) are associated with certain objects or situations, such as a balloon or an alarm clock. Consequently, these objects become conditioned stimuli, capable of eliciting a conditional response, such as fear or anxiety. Thus, the balloon may later elicit an anxious response without the presence of an aversive noise. Specific phobia is one of the most common anxiety disorders seen in children with ASD (Gadow et al. 2004; Muris et al. 1998; Sukhodolsky et al. 2008), which is consistent with this fear and conditioning model, as it is easy to imagine how children with SOR might develop phobic responses to stimuli associated with aversive sensory experiences such as balloons, elevators, scratchy clothing, etc. However, conditioning is also a mechanism through which SOR might lead to more generalized anxiety. For example, if the unconditioned stimulus does not always signal the conditioned stimulus (i.e., aversive noises can occur unexpectedly without being predicted by one particular object), the fear is more likely to generalize through what is called context conditioning (Grillon 2008). Through context conditioning, the conditional fear response can shift from being triggered by a specific object to a location or context in which the aversive stimulus has occurred, so the child might refuse to go to a birthday party or into a bedroom because they anticipate there might be aversive sounds at these locations. Animals who have experienced context conditioning become behaviorally avoidant (Bouton and King 1983), and likewise children with SOR often become avoidant of generalized locations such as restaurants, grocery stores, or parties in addition to specific objects (Schoen et al. 2008). The more generalized the avoidance, the more impaired the child is likely to be.

A number of factors have been found to contribute to the “conditionability” of a particular situation, including the intensity of the unconditional response and uncontrollability of the conditioned stimuli (Craske 2003). As children with SOR by definition have an intense unconditional response to sensory stimuli, and as these stimuli are often elicited through factors completely outside the children’s control (Reynolds and Lane 2008), one can see how these children are likely at high risk for fear conditioning. Furthermore, the cognitive and language deficits often associated with ASD may make it more difficult for children to identify relationships of predictability and/or exert control over a stressful situation (e.g. by expressing fear in a way that will cause others to reduce the threat-relevant stimuli). The frequency, uncontrollability, and unpredictability of the conditioned stimuli may then cause the child to become generally hypervigilant to potentially threatening sensory stimuli, and to maintain physiological arousal, resulting in a state of generalized anxiety and worry.

It is possible that attentional bias also plays a role in how SOR contributes to anxiety. SOR may cause attentional bias towards the perceived threatening stimuli, and that attentional bias can cause negative mood states in response to stress (Craske 2003). In particular, MacLeod et al. (2002) demonstrated that training individuals to have an attentional bias towards threat-relevant words subsequently increased their negative response to stress. These results do not specifically address attentional bias as a causal factor for anxiety disorders, but do suggest that attentional bias could contribute to the development of anxiety.

We must consider the possibility that SOR and anxiety are not directly causally related, but are associated through a third variable such as a common risk factor (e.g., amygdala abnormalities) or through overlapping diagnostic criteria (see Fig. 3).

The choice of the most appropriate intervention for children with ASD who show anxiety and SOR symptoms depends on which causal pathway is supported. The two conditions are currently treated very differently, and as if they are entirely separate disorders. The most common interventions for SOR include Sensory Integration therapy (SI), the Sensory Diet, sensory stimulation (SS) techniques, and Auditory Integration Therapy (AIT), but evidence for the efficacy of these interventions is quite mixed and primarily based on case-study designs (see Baranek 2002 for a review). Sensory-based interventions aim to increase participation by reducing negative responses to sensations and/or improving sensory modulation. SI uses somato-sensory and vestibular activities to improve children’s ability to modulate, organize and integrate information from the environment. It is still a controversial therapy, and while some studies show that it can improve children’s play and social-communication skills, there is little evidence that it specifically reduces SOR (Ayres and Tickle 1980; Case-Smith and Bryan 1999; Linderman and Stewart 1999). The Sensory Diet is a program of scheduled sensory activities that meet the child’s sensory needs and involves frequent and systematic application of somatosensory stimulation. Although commonly used, there are no studies of its sensory or emotional outcome for children with ASD (Baranek). Likewise, Auditory Integration Training (AIT), which uses carefully modulated sounds to enhance auditory perception, has not been shown to be consistently efficacious in reducing auditory sensitivity (Baranek). The evidence for SS techniques is more promising. SS is based on the principle that deep pressure can influence the nervous system in a way that helps the child regulate his or her arousal level. For example, weighted vests which provide continuous pressure (Ferterl-Daly et al. 2001) or touch therapy (e.g., massage; Baranek; Field et al. 1997; Silva et al. 2009) have been shown to improve young children with ASD’s responsiveness to sensory input, attention, stereotyped movements, and social-communication symptoms. The only ASD sensory intervention study that used anxiety as an outcome measure was conducted by Edelson et al. (1999). In this study the “hug machine” developed by Temple Grandin, which provides lateral body pressure, has been found to decrease tension and hyperactivity as well as anxiety in children with ASD, especially in children with higher initial levels of arousal or anxiety. This is consistent with work outside the ASD literature showing a decrease in secondary anxiety symptoms in adults with SOR following a sensory-based intervention (Pfeiffer and Kinnealey 2006). Therefore, it appears that certain sensory-based interventions can be successful in reducing anxiety in children with ASD. More rigorous evidence (e.g., from a randomized controlled treatment design) that shows the reduction of anxiety as an outcome of sensory-based interventions for children with ASD would support the Primary SOR model.

Recently, a few studies have shown Cognitive Behavioral Therapy (CBT) to be efficacious in treating anxiety in children with ASD (e.g., Chalfant et al. 2007; Reaven and Hepburn 2003; Sofronoff et al. 2005; Sze and Wood 2007; Wood et al. 2009). The CBT used in these studies was based on CBT for typically developing children, and included psychoeducation, cognitive restructuring, self-talk, relaxation, and exposure to feared stimuli. Chalfant et al. and Sofronoff et al. adapted CBT for children with ASD through methods such as simplifying the cognitive aspects and emphasizing concrete aspects such as relaxation and exposure. Moreover, Wood et al. enhanced a manualized CBT intervention for children with ASD by addressing poor social and adaptive skills in addition to the core anxiety problems. In all of these studies, children in the intervention groups had a significant reduction in anxiety, as compared to wait-list controls.

Psychopharmacological treatments, and specifically SSRIs, have been shown to decrease anxiety in children with ASD (White et al. 2009). SSRIs used to treat anxiety in children may also reduce amygdala volume (Szeszko et al. 2004). It seems possible, therefore, that if an overactive amygdala is implicated in SOR, SSRIs might also reduce SOR by reducing amygdala volume and/or activity.

To our knowledge no one has yet investigated the effect of anxiety interventions such as CBT or exposure therapy on SOR in children with ASD and anxiety. If anxiety is a causal factor for SOR or if SOR is a marker for unrecognized anxiety in young children, then sensory interventions need to take into account the effects of anxiety, and perhaps incorporate cognitive and behavioral strategies from evidence-based anxiety interventions. Likewise, if SOR is a causal factor for anxiety, then cognitive-behavioral programs for children with ASD and anxiety need to assess and treat sensory problems as well. Either way, children might benefit from an integrated intervention approach. For example, exposure therapy might be beneficial to children with primary SOR in addition to those with primary anxiety. Future studies are needed to examine whether young children with SOR who undergo exposure therapy are less likely to develop anxiety later on, which would support the primary SOR model. Exposure therapy might influence SOR through the following mechanism: exposure to fearful objects might provide an opportunity for children to become more aware of the contingencies for when aversive events occur and less likely to experience generalized context conditioning and behavioral avoidance. Exposure therapy could be combined with sensory stimulation intervention, such as body pressure, which would help decrease arousal (Baranek 2002) and make exposure less aversive. Regardless of which causal pathway is supported, pairing sensory-based interventions with CBT or exposure therapy might markedly increase a child’s tolerance for previously aversive sensory stimuli and overwhelming environments.

The discipline of the professional evaluating the child is not only likely to impact the child’s diagnosis but also the type of intervention the child is referred to. This problem was demonstrated by Ben-Sasson et al. (2007a) who found that occupational therapists were more likely to recommend a sensory-based intervention for case studies designed to represent a child with SOR or a child with generalized anxiety disorder whereas psychologists were more likely to recommend family therapy for either case. Professionals from all disciplines would thus benefit from a better understanding of the two conditions through education and training about assessments and interventions available across disciplines. When evaluating a child who presents with symptoms of SOR or anxiety, the professional should consider both for differential diagnosis and accurate referral.

An interdisciplinary framework of care is important but not sufficient given the overlap between assessments for SOR and anxiety. There is a need to develop distinct indicators to better differentiate the two. One way in which the two may be distinguished behaviorally is that SOR should be generalizable to multiple stimuli that elicit the unpleasant sensory response (e.g., several stimuli that create an aversive noise; see Miller et al. 2007) while specific phobia would be to one specific stimulus or situation (e.g. Ollendick et al. 2002). Furthermore, if a child has a general anxiety disorder, by definition he or she should be generally anxious, not only to specific sensory stimuli (e.g., Masi et al. 2004). Developing parent report and observation measures specific to ASD (Davis et al. 2008) that make such distinctions is needed to analyze sources of stress in the frequent case that children are too young or lack the verbal skills to report their own symptoms. Similarly, identifying distinct physiological indicators for clinical use would be advantageous for evaluating SOR and anxiety in ASD, independent of social and cognitive level. These methods can assist professionals across disciplines in considering the possibility of either condition as the primary cause of fear responses in ASD.

To summarize, we have proposed two theoretical, causal models that might explain the association between anxiety and SOR. While these two conditions are related in other populations, we chose to focus on children with ASD, because both conditions are common in and impairing to children with ASD and may be particularly challenging to differentiate in these children. In the Primary Anxiety model, anxiety contributes to SOR as generalized hyperarousal and hypervigilance focuses attention on a specific type of sensory stimulus. In this case, children who are hypervigilant and continuously scanning their environment are more likely to notice and react to aversive sensory stimuli in the environment. The threat-based emotion regulation associated with anxiety also makes it more difficult for children to regulate their emotional and physiological reactions to stimuli. The reaction may then worsen through both classical and interoceptive conditioning. In the Primary SOR model, SOR contributes to anxiety as a specific over-reaction generalizes to an environment or situation through context conditioning. The strength of the unconditional sensory reaction and the uncontrollability of events that elicit sensory stimuli contribute to stronger conditioning. We also addressed the possibility that a common risk factor such as amygdala abnormalities may contribute independently to each condition. The amygdala has long been known to play a role in fear and anxiety, and may also be related to SOR through overestimation of the threat value of a sensory stimulus which triggers an enhanced response to that stimulus.

In reality, these models oversimplify the development of these two conditions in children with ASD, as multiple genetic and environmental factors probably interact to contribute to the etiology and course of both anxiety and SOR. However, the models provide a theoretical basis from which to develop studies that can help us better understand the association between anxiety and SOR. Potentially informative studies include:

Studies such as those listed above would be valuable in furthering our understanding of the association between SOR and anxiety, in all populations, but especially in children with ASD. SOR and anxiety are particularly common and impairing in this population, so the hope is that such research would lead to an improved understanding of the etiology of both conditions, as well as more effective interventions and improved functioning for children with ASD.