Introduction
It is estimated that 80% of individuals with ASD present with motor atypicalities (Green et al., 2009; Weimer et al., 2001), including delayed walking, clumsy gait, and disturbances in reach-to-grasp movements (Fournier et al., 2010; Ghaziuddin and Butler, 1998; Miyahara et al., 1997; Teitelbaum et al., 1998). While there is substantial evidence for atypical motor behavior in autism and movement atypicalities are commonly reported clinically, less is known about the underlying neural substrates. What is more, it is unknown whether these atypicalities are a product of movement planning, sensorimotor integration, alterations in basic motor circuitry, or whether these differences emerge due to dysfunction in downstream processes, such as movement execution or the coordination of movement sequencing.
The majority of work surrounding ASD and the neurophysiology of motor movements focuses on the high-level construct of movement imitation and visual-motor integration, reflecting a tendency in the ASD literature to explore constructs which directly link to impairments in social communication (Bernier et al., 2007; Deschrijver et al., 2017; Ewen et al., 2016; Lidstone and Mostofsky, 2021; Oberman et al., 2005; Sowden et al., 2016; Spengler et al., 2010; Williams et al., 2004; Williams et al., 2001). While existing research provides clear evidence of differences in imitative behavior and neural processes engaged during both the initiation and execution of imitative movements, the role of basic sensorimotor dysfunction in these impaired processes is not well understood. The paucity of data surrounding the neural development of sensorimotor planning in a neurotypical population indicates that a thorough characterization of the ‘normative’ trajectory is needed in order to assess whether children with ASD exhibit significant deviation from this trajectory.
The functional dynamics of spontaneous, voluntary motor movements (such as self-paced finger flexion) are well-characterized in healthy adults using EEG. A common method for studying movement related cortical potentials (MRCP) involves time-locking EEG data to the execution of a motor movement and examining the preceding signal. The sensitivity of EEG on the timescale of milliseconds is a major advantage of this methodology over functional magnetic resonance imaging (fMRI) techniques, which resolve biological signals along a much longer timescale. Using EEG, classical event-related potential (ERP) components of the MRCP can be observed at the scalp. The earliest of these components, known as the early berescheftspotential (BP) or readiness potential (RP), begins 1-3 seconds prior to movement onset and is typically seen as a slow negativity over the central-parietal scalp generated by supplementary motor regions (Deecke, 2007; Ikeda et al., 1992; Kornhuber and Deecke, 1965). The late BP component, beginning 300-500 ms prior to a recorded movement, is visible over the central scalp contralateral to the effector and is caused by a rapid increase in motor excitability thought to be driven primarily by premotor and motor areas (Shibasaki et al., 1980; Shibasaki and Hallett, 2006). Motor excitability peaks at movement onset in a component known as the Motor Potential (MP), generated in primary motor cortices (Böcker et al., 1994). The MP is immediately followed by a rapid positivity known as the reafferent potential (RAP), reflecting sensory-motor integration. The RAP may be generated by the primary somatosensory cortices and peaks between 90 – 130 ms after movement onset (Bötzel et al., 1997; Bötzel et al., 1993). Modulation of movement-related EEG activity has been shown as a function of the complexity and meaning of the intended action, task instructions, and the physical state of the individual performing the action (Bozzacchi et al., 2012; Di Russo et al., 2017; Freude and Ullsperger, 1987; Simonetta et al., 1991). Although not yet explored, it is likely that MRCP components during cued movements may vary as a function of reaction time within an individual; that is, whether presentation of a cue leads to a prompt motor response or a slower response.
However, when motor movements are cued by external stimuli, data indicate that MRCP morphology may differ from the classical models derived from tasks where participants execute spontaneous, self-paced movements (Pereira et al., 2018). For example, the self-initiated BP component has been shown to have an anterior topographic distribution relative to the BP elicited from a cued movement in response to sensory stimuli (Jenkins et al., 2000). Data from fMRI studies suggest that while both spontaneous and auditory-cued motor movements showed strong activation in both motor and premotor areas, auditory-cued movements induced additional bilateral activation in primary auditory cortices, and activation in the basal ganglia was identified for self-initiated movements only (Cunnington et al., 2002). Although sensory processing is a necessary prerequisite for the generation of cued movements, this presents additional complexity to data interpretation. Despite the overlap of sensory and motor preparatory processes in cued motor-response tasks, it is clear that time-locking to a specific task event (either the stimulus or the response) exaggerates some ERP components and reduces the visibility of others (Takeda et al., 2008). It should be noted that both stimulus-locked and response-locked ERP analyses suffer from the issue of temporal smearing of some components—in the case of stimulus-locked analyses, the temporal smearing of motor components, and for motor-locked analyses, the temporal smearing of components related to the sensory stimulus. As such, care should be taken when interpreting data from tasks involving both the processing of a sensory stimulus and the execution of a motor response.
In the current work, we tested the integrity of basic sensorimotor processing in autism in a large sample of participants, in the context of a simple reaction time task. Using response-locked averaging of EEG data, we investigated the neural processes preceding a cued movement response in a large sample of children with ASD (n=84) and without ASD (n=84) with similar age, sex, and IQ. Participants completed an audio-visual speeded reaction time task and were instructed to respond quickly with a button press using the right hand in response to a visual, auditory, or audiovisual bi-sensory cue (Brandwein et al., 2015; Brandwein et al., 2013; Brandwein et al., 2011; Crosse et al., 2022; Molholm et al., 2002). The current work represents a re-analysis of an existing dataset, focusing for the first time on motor-related potentials to assess functional dynamics of sensorimotor processing in Autism. The density of data, both in terms of trials per participant (972 ± 58 trials) and number of participants within group, allowed us to additionally consider how these responses differed with respect to age, speed of response, and sensory cue type, both in Autism and typical development.
Methods
This study involves a secondary analysis of a large dataset collected over a period of 13 years (Brandwein et al., 2015; Brandwein et al., 2013; Crosse et al., 2022). Here, we examined the development of sensorimotor processing and how these processes may differ in Autism. The current work used both a priori and exploratory analytic approaches. A priori analyses were focused on well-defined ERP componentry that have been extensively studied and characterized in adults. Exploratory analysis was undertaken to leverage the richness of the current dataset with regard to the dense electrode array, high temporal resolution, and paucity of existing normative data, the full data matrix (all channels and time-points) were examined across a timewindow of 650 ms, beginning 350 ms prior to the button-press response and ending 300 ms post-response.