Rethinking Stereotypies in Autism

Abstract

Stereotyped movements (“stereotypies”) are semi-voluntary repetitive movements that are a prominent clinical feature of autism spectrum disorder. They are described in first-person accounts by people with autism as relaxing and that they help focus the mind and cope in overwhelming sensory environments. Therefore, we generally recommend against techniques that aim to suppress stereotypies in individuals with autism. Further, we hypothesize that understanding the neurobiology of stereotypies could guide development of treatments to produce the benefits of stereotypies without the need to generate repetitive motor movements. Here, we link first-person reports and clinical findings with basic neuroanatomy and physiology to produce a testable model of stereotypies. We hypothesize that stereotypies improve sensory processing and attention by regulating brain rhythms, either directly from the rhythmic motor command, or via rhythmic sensory feedback generated by the movements.

Imagine you’re on your second cup of coffee, sitting in Grand Rounds watching a semi-useful presentation. Your right knee bounces rhythmically, as it often does when watching talks. Your neighbor taps your shoulder gently to get you to stop because it’s distracting them. Later, when reading through some paper charts, you twirl your pen around your thumb or tap it on the page. Meanwhile, at school, your child sits on an exercise ball instead of a chair so they can bounce to get the wiggles out while remaining seated at their desk. They quietly doodle on the corner of their homework while listening to the teacher.

Simultaneously, another child in the class rhythmically rocks their body and periodically flaps their hands. “Alex, please stop that and pay attention,” the teacher says. Alex’s teacher has recommended to Alex’s parents and behavioral therapist that they work on reducing the “stimming” because it keeps Alex from paying attention in class. It’s also part of the reason the other kids at school think Alex is odd. Alex’s parents also wish the stimming would stop because it makes it hard to bring Alex out in public. As soon as they arrive at a store or restaurant, Alex starts body rocking and hand flapping. The parents have tried positive and negative rewards to decrease the stimming, but nothing has worked.

Stereotyped movements or motor “stereotypies” are common. In people deemed “neurotypical”, these behaviors are commonly referred to as “fidgeting.” In people with autism, engaging in these behaviors is often considered problematic and is colloquially referred to as “stimming” (i.e. self-stimulatory behaviors)¹. Stereotypies are often perceived as interfering with normal activity. There is a common belief that if a person is stimming, they are not attending to their environment. Thus, in many contexts, especially educational settings, stereotypies are perceived as counterproductive for learning². They can also be distracting to others. Therefore, behavioral therapies often have a goal of reducing the frequency and intensity of stereotypies.

Here, we propose a new perspective on motor stereotypies. Instead of being a sign that a person with autism is in their own world and isn’t paying attention, we propose that people with autism may perform stereotypies in order to pay attention. What if engaging in these movements helps Alex’s brain process information? What if Alex’s brain is generating this rhythmic movement so they can tune in to what the teacher is saying? What if Alex’s brain generates these stereotypies to help regulate the sensory overload they experience in public places?

We propose that by understanding how and why stereotypies are generated, we can harness those mechanisms to improve sensory processing and attention in people with autism. Here, we propose a testable model of the beneficial purpose of stereotypies. We hypothesize that engaging in stereotyped movements entrains brain rhythms to enhance sensory processing and attention.

Stereotypies are common in people with autism.

Autism is a behaviorally-based diagnosis assigned to people who have disabilities in social communication and who exhibit restricted, repetitive patterns of behavior, interests, and activities. Specifically, the presence of “stereotyped or repetitive motor movements” is a clinical criterion for the diagnosis of autism spectrum disorder³. “Motor stereotypies” can be defined as repetitive, non-goal-directed, rhythmic, patterned movements that stop with distraction and have no accompanying premonitory urge^4–6. Examples include hand flapping, body rocking, spinning, repetitive jumping, and finger flicking. Motor stereotypies are just one type of stereotyped behavior. Other examples of stereotyped behavior include atypical use of language (e.g. echolalia), unusual visual inspection of objects, and more complex seemingly “non-functional” behaviors such as aligning objects⁷. Throughout this paper, we will use the terms “motor stereotypies”, “stereotypies”, and “stimming” interchangeably. We are excluding self-injurious behaviors (SIB) here, as these are beyond the scope of the current discussion.

When do people with autism engage in stereotypies?

Stereotypies can be a visible read-out of a person’s state of being, both negative and positive. The intensity of stereotypies correlates with anxiety⁸ and the severity of core autism symptoms⁹. Observational studies of children with autism show that stereotypies increase following a stressful trigger¹⁰. Based on self-reports from people with autism, stereotypies can help with feelings of anxiety, nervousness, or feeling “wound up”^11–13. People also engage in stereotypies when relaxed and happy¹⁴. Stereotypies are described as comforting, calming, and even enjoyable^{11, 13}. Stimming can help organize thoughts, focus, or get rid of excess energy¹¹. This can be especially helpful in environments where the person feels sensory overload: engaging in stereotypies reportedly modulates overwhelming sensory inputs^{11, 13}.

Stereotypies typically begin unconsciously or involuntarily, but individuals describe intentionally self-perpetuating the movements because of the comfort and control they provide¹³. The cessation of the movements can be internally or externally motivated. Many report a loss of interest in the behavior, or a feeling that the behavior would be inappropriate to continue¹¹. Others report a feeling of emotional fulfillment, or that something has been completed¹¹. Together, these self-reports of people with autism highlight themes of improved sensory processing, emotional regulation, improved focus, concentration, and attention. Despite the prevalence of stereotypies in people with autism, there is scant research quantifying stereotypies for even basic measurements such as the frequency of the rhythmic movements^{15, 16}.

When do people without autism engage in stereotypies?

Repetitive movements are not unique to individuals with autism. People with intellectual disability, severe vision impairment, or institutional care early in life commonly engage in stereotypies^17–19. Neurotypical individuals also engage in repetitive movements, which are typically referred to as fidgeting. Examples include doodling, clicking a pen, rocking in a chair, playing with ones’ hair, humming, chewing gum, or bouncing one’s leg²⁰. These movements are often repetitive or patterned, and are self-initiated. Fidgeting may appear purposeless, but it can have positive effects on attention, concentration, and stress²¹. In fact, therapies based on repetitive manipulation of objects (like small foam “stress” balls) have been found to increase focus and learning performance in some contexts²².

Linking stereotypies to brain rhythms.

Why is engaging in stereotyped movements so common in the general population, and in particular, in people with autism? We hypothesize that the rhythmic brain signal generated to create the movement and/or the rhythmic sensory feedback produced by the movement entrains brain rhythms to enhance information processing (Figure 1).

Figure 1. Hypothesized role of motor stereotypies in the normalization of brain rhythms in sensory processing areas.

A. Case 1: intact efference copy. 1. The motor command center generates a rhythmic signal. 2. Sensory areas are entrained to the command rhythm via the direct pathway. 3. Because the rhythmic signal is also sent to motor effector areas, a rhythmic movement (the stereotypy) is generated as a “side effect”. 4. The stereotyped / rhythmic movement produces a rhythmic sensory experience. 5. The rhythmic sensory experience modulates brain rhythms in sensory areas (“indirect” pathway).

B. Case 2: abnormal efference copy. 1. The motor command center generates a rhythmic signal. 2. Because the branched axons have not formed normally, or are not functioning normally, this motor command signal is not directly relayed to sensory areas. 3. The rhythmic signal is sent to motor effector areas and a rhythmic movement (the stereotypy) is generated. 4. The stereotyped / rhythmic movement produces a rhythmic sensory experience. 5. The rhythmic sensory experience modulates brain rhythms in sensory areas (“indirect” pathway).

Brain rhythms help process sensory information.

Electrical rhythms are found at all scales within the nervous system. The summed postsynaptic potentials of populations of neurons produce rhythmic voltage fluctuations that can be detected microscopically (i.e. in single cell recordings) and macroscopically (i.e. with scalp EEG)^{23, 24}. These oscillations reflect neuronal activity but also influence neuronal signaling. For instance, when an excitatory synaptic input coincides with the peak of a voltage oscillation, the neuron has a higher probability of generating an action potential²⁵.

Brain rhythms influence processing of incoming sensory information^{26, 27}. Sampling of sensory inputs is often timed with brain oscillations to optimize information transfer between the external world and the brain²⁸. Through rhythmic movements, sensory receptors are stimulated at rhythmic intervals that coincide with peaks of brain oscillations. For example, in rodents, olfactory sampling via sniffing and tactile sampling through whisking are phase-locked to internally-generated rhythms in the olfactory bulb and brainstem^{29, 30}. In the visual system of primates, the timing of saccades is aligned with rhythmic field potential activity in visual cortex^{28, 31, 32}. Internally-generated oscillations essentially provide regular “windows of opportunity” for information transfer^{29, 33–35}. Even attention itself has been found to be fundamentally rhythmic, with rhythmic changes in perceptual ability during sustained attention tasks³⁶.

Brain rhythms are disrupted in people with autism.

One prominent hypothesis is that autism is a primary disorder of brain oscillations (i.e. an “oscillopathy”)³⁷. Sensory- and motor-related brain oscillations are abnormal in autism^37–40, both at rest^{41, 42} and during active behaviors^43–45. Differences are reflected in overall power, but also in how different frequencies of oscillations work together to transmit information (“cross-frequency coupling”)^{40, 46}. Differences in oscillations have been observed in autism throughout the brain in the auditory⁴⁷, visual⁴⁰ and motor⁴⁸ systems. We hypothesize that the repetitive motor command signal that produces stereotypies and/or the rhythmic sensory feedback generated by the movements normalize brain rhythms in people with autism to improve sensory processing (Figure 1). To understand how this might work, we next explore how motor and sensory brain regions are directly linked to each other via “efference copy”.

Initiating a movement generates an electrical signal that travels to both motor effectors and sensory areas via efference copy.

To produce movement, a motor command output signal (“efferent”) is transmitted to the appropriate muscle groups via synaptic transmission. Simultaneously, a duplicate (“copy”) of the motor output signal is sent to sensory areas (Figure 1). Through this “efference copy” (also known as “corollary discharge”), the motor area gives the sensory area a “heads up” that there is a movement coming that is self-generated. The brain then suppresses the response to sensory input generated by the movement⁴⁹. Efference copy thus allows us to distinguish between sensations produced by self-generated movements from those generated by external stimuli. For instance, during saccadic eye movements, efference copy of motor commands from the superior colliculus informs visual processing centers that the impending disruption in visual information is due to the internally-generated eye movements rather than a change in the external world^49–53. Thus, we perceive the visual world as stable despite our eyes moving. Efference copy is also potentially why we can’t tickle ourselves⁵⁴.

The anatomical substrate for efference copy is axon branching⁵⁵. By splitting the axon into different branches, single neurons can simultaneously communicate with more than one downstream brain region (Figure 2). Axon branching is an activity-dependent process⁵⁶ and is developmentally regulated⁵⁷. There is some evidence for differences in the axonal branching^{58, 59} and the efference copy system^{60, 61} in autism and schizophrenia⁶², and this is an area in need of further research.

Figure 2. Efference copy throughout the sensorimotor system, a simplified schematic illustrating locations of branching axons based on rodent and primate tracing studies.

A. Incoming sensory information is transmitted from peripheral sensory nerves to lower order” (sensory) thalamus and subcortical motor areas (Sherman, 2017). In the case of visual information, one branch routes to the lateral geniculate nucleus and the other copy is sent to the superior colliculus (Chalupa and Thompson, 1980; Crook et al., 2008).

B. “Higher order” thalamic neurons receive driving input from L5 of cortex (Sherman, 2016). Individual higher order thalamic neurons in mediodorsal thalamus and pulvinar project to multiple cortical regions (Giguere and Goldman-Rakic, 1988; Rockland et al., 1999), and on the way, thalamocortical neurons send axon branches to the thalamic reticular nucleus TRN (Kuramoto et al., 2017). Neurons in the TRN can have axons that branch to distinct thalamic nuclei (Lee et al., 2014).

C. Individual Layer 6 corticothalamic neurons may send recurrent axon branches to cortical Layer 4 (Lee et al., 2012; White and Keller, 1987).

D1. Individual Layer 5 neurons in the prefrontal cortex of mice send axons to both higher order (mediodorsal) thalamus and pons (Collins et al., 2018). Individual Layer 5 corticofugal projection neurons in motor, somatosensory and visual cortices in rats send axon collaterals to the thalamus (Deschênes et al., 1994). Individual Layer 5 corticofugal axons in rat motor cortex send collaterals to multiple subcortical targets including subthalamic nucleus, basal ganglia, and superior colliculus (Kita and Kita, 2012).

D2. Individual Layer 5 pyramidal cells in rat sensorimotor cortex project to the brainstem or spinal cord and send axon collaterals to striatum (Donoghue and Kitai, 1981). This is also seen in the primate primary motor cortex, where corticofugal projections to the brainstem also branch to the striatum (Parent and Parent, 2006).

D3, D4. Individual Layer 5 neurons in rat somatosensory barrel cortex project to pons, thalamus, and superior colliculus (Veinante el al., 2000). Individual Layer 5 neurons in the prefrontal cortex of mice send axons to both higher order (mediodorsal) thalamus and pons (Collins et al-, 2018).

E1. Individual cerebellar fastigial nucleus (cFN) neurons project to both the superior colliculus and thalamus (Katoh et al., 2000).

E2. Additionally, individual cFN neurons branch to both the left and right deep layers of the superior colliculus (Katoh & Benedek, 2003). Of note, within the cerebellum, there is a high degree of axonal branching in mossy fibers, climbing fibers (Mason & Gregory, 1984), and Purkinje cells (Yang et al., 2014).

We hypothesize that stereotypies entrain abnormal brain rhythms to improve sensory processing.

In our proposed model (Figure 1), there are two potential routes by which stereotypies could regulate brain rhythms in sensory areas: “direct” (via efference copy) or “indirect” (via sensory feedback).

By the direct route, a repetitive motor command signal is generated with the goal of entraining abnormal rhythms in the sensory system. The rhythmic motor command is conveyed to sensory areas via efference copy. The motor command signal thus directly influences rhythms in the sensory brain regions through synaptic transmission. Because of efference copy, the motor command signal is also transmitted to motor effectors. Thus, stereotyped movements are produced as a “side effect” of the brain’s attempt to use rhythmic motor commands to regulate sensory area oscillations.

By the indirect pathway, the rhythmic sensory signal generated by stereotyped movements is what entrains sensory areas rhythms. Rhythmic sensory stimuli drive rhythmic voltage responses⁶³ in primary sensory cortices and other brain regions like the hippocampus⁶⁴ and prefrontal cortex⁶⁵. The rhythmic sensory signal generated by stereotyped movements could therefore entrain or modulate⁶⁶ brain rhythms in sensory areas. Through this indirect route, purely externally-generated rhythms (e.g. by watching fans spinning or proprioceptively by swinging), could entrain sensory areas via rhythmic sensory inputs.

In addition to activating sensory regions via the indirect route, externally-generated rhythmic sensory signals may also engage the efference copy system and thus exert influence on brain rhythms via the direct route. For instance, seemingly “passive” activities like being rocked in a swing actually engages core muscles to stabilize body posture by counterbalancing externally-generated forces on the body. Thus, the rhythmic motor command generated to activate core muscles while swinging could rhythmically engage sensory areas via the direct path. Similarly, watching a fan spin likely engages the efference copy system by triggering saccadic eye movements and optokinetic nystagmus. Thus seemingly “passive” stereotypies could engage either the indirect pathway, the direct pathway, or both.

The direct versus indirect pathway framework provides a model to generate testable hypotheses about the role of top-down (direct) versus bottom-up (indirect) signaling driving a person to engage in stereotyped movements. For example, if efference copy is abnormal or underdeveloped in autism⁶⁰, then sensory perception areas may not receive direct copies of motor commands during self-generated movements. As shown in the bottom panel of the Figure 1, we would therefore predict that the indirect pathway would be required to obtain the benefit of stereotypies. If, on the other hand, the efference copy system is functioning normally in autism, then stereotypies could provide benefit through the activation of the direct pathway, the indirect pathway, or both (Figure 1, top panel).

Based on evidence that rhythmic sensory inputs entrain neural oscillations⁶⁶, we hypothesize that the brain rhythms generated by stereotypies (either via the direct or indirect pathway) improve signal processing in sensory areas by providing discrete “windows of opportunity” for incoming sensory signals⁶⁷. Further, stereotyped movements reinforce the cycle of action and perception²⁰, creating a situation where the mismatch is low between expectations and reality. Because stereotypies are repetitive and thus predictable, these movements might provide a stream of sensory input to the brain with low prediction error. This could provide a regular baseline rhythm onto which signals generated by external inputs can be layered.

Outstanding questions.

The literature is scant on how fidgeting and other stereotyped movements influence attention, sensory processing, and brain rhythms. Further, although many people with autism report improved sensory processing during stimming, whether stereotypies enhance sensory signal processing has yet to be directly tested.

We propose that testing our model in individuals with and without autism represents low-hanging fruit for understanding core features of autism spectrum disorder. The experimental paradigms to examine the relationship between stereotypies, brain rhythms, and sensory processing / attention would be straightforward for laboratories already engaged in these lines of research using functional magnetic resonance imaging, EEG, and behavioral assessments. Despite being a well-recognized clinical feature of autism, shockingly little is known about the phenomenology of stereotyped movements. For instance, there are almost no studies on the amplitude, frequency, and regularity of stereotypies. Clinically, they appear rhythmic, but stereotypies are also complex and variable^68–70. Understanding basic parameters like the frequency at which these movements occur^{16, 71} and the regularity of the movements would help predict the patterns of brain rhythms that generate them and the characteristics of the sensory input created by the movements themselves⁷². Through updated methods of wearable sensors in combination with video analysis, measurement and analysis of the physiology of stereotypies as they typically and atypically emerge is now possible⁷³. These technologies could also help quantify similarities and differences in stereotypies across neurodevelopmental disorders^{74, 75} and to establish developmental “growth charts” for typical and atypical stereotyped movements^{76, 77}.

In terms of neurological localization, it still remains unclear or unknown whether the motor command signal for stereotypies is generated subcortically or cortically^{78, 79}. We have therefore been purposely nonspecific in our terminology for the localization for both motor and sensory “areas” in this writing. Thus, it is an open question where in the nervous system (i.e. cortex vs. thalamus vs. basal ganglia vs. cerebellum) modulation of rhythms by motor commands and sensory feedback might be beneficial.

In predicting which frequencies of brain rhythms would be influenced or entrained by stereotypies, it is important to note that the frequency of muscle activation does not necessarily equal the frequency of the brain signal generating the movement (as it is when rhythmic EEG activity is time-locked to clonic movements during seizures). Similarly, rhythmic sensory inputs may generate the same frequency activity in the brain or they may generate harmonics of that frequency (as we commonly see in the photic driving response).

At the microanatomical level, it is unclear how widespread the phenomenon of branched axons is in the brain and the rest of the nervous system⁸⁰. Importantly, most brain connectivity models do not currently include axon branching in the calculations. We expect that including branched axons in functional and structural connectivity models will change predictions of brain connectivity in typically developing individuals and in those with autism⁸¹. Post-mortem brain studies will help determine how early differences in axon branching can be detected. Using animal models, we can test if differences in axon branching in autism are primary or compensatory^{82, 83}.

It is still unclear if and how behaviors that rely on efference copy differ in people with autism. Quantifications of differences in eye^{84, 85} and body movements⁸⁶ offer potential biomarkers for the disorder. Because of efference copy, judging the location of a limb in space is more accurate following active movement than after a passive movement^{87, 88}. If the efference copy system is disrupted in autism, we would predict the difference between perception between active and passive movements to be diminished in individuals with autism. Further, if the efference copy system is disrupted in autism, we would also predict that people with autism would be more likely to be able to tickle themselves than people without autism⁵⁴.

In conclusion, stereotypies are a poorly understood feature of autism spectrum disorder. We hope that by understanding the anatomy and physiology of motor stereotypies, we can make them less stigmatizing and develop ways to harness their benefits to help people with and without autism.