A total of thirty-four children aged 7–11 years participated in the study (18 females, 16 males). The sample included 11 children with developmental coordination disorder (DCD), 11 typically developing children (TD), and 12 typically developing children with extensive musical training (TDM). We defined extensive musical training as a minimum of 2 years of formal piano lessons with regular practice at the time of enrollment. We specifically focused on piano lessons because of their strong emphasis on sensorimotor skills and bilateral movement dexterity. Participant characteristics are detailed in Table 1. Children in the DCD group had been screened for motor difficulties at a local hospital before participating in this study. The diagnostic criteria for DCD were further assessed according to the following DSM-5 standards (3): (A) a score at or below the 15^th percentile on the Movement Assessment Battery for Children, 2^nd Edition (MABC-2) (Henderson et al., 2007); and (B) a parent report confirming the presence of motor coordination difficulties and their effect on the child’s activities of daily living (DCD-Q) (Wilson et al., 2009). The inclusion criteria for typically developing children in both groups (i.e., TD and TDM) included a score above the 30^th percentile on the MABC-2 and a parent report indicating the absence of motor coordination difficulties. All children completed neuropsychological testing and scored above 70 on the Kaufman Brief Intelligence Test (Kaufman and Kaufman, 2013). Right-handedness was assessed using the revised version of the Edinburgh Handedness Inventory (Oldfield, 1971; Williams, 2020). Caregivers reported on the child’s developmental and medical history, musical training (e.g., the duration of musical training and the amount of practice per week, if applicable), and possible symptoms of ADHD and dyslexia. Exclusion criteria were other co-occurring diagnoses and/or medical conditions affecting hearing or motor functioning (e.g., cerebral palsy, hemiplegia, or muscular dystrophy). Considering that DCD and ADHD co-occur in approximately 50% of cases, children with coexisting DCD and ADHD symptomatology were eligible to participate (n = 3). Participants were recruited through the hospital’s research registry, pediatric occupational therapy clinics, local elementary schools, music schools, flyers, and word-of-mouth. All participants provided written assent, and their legal guardians provided written informed consent, in accordance with the Declaration of Helsinki and its later amendments. The study was approved by the Research Ethics Board at the Holland Bloorview Kids Rehabilitation Hospital and the University of Toronto. Participants received monetary compensation after the experiment.

Participants and their caregivers attended two research sessions, each lasting between 60 and 80 min. The first visit involved neuropsychological and motor testing and caregiver questionnaires to ensure the differences in cognitive functioning did not affect the performance on the experimental tasks. The second visit included behavioral tasks and EEG recordings to assess auditory-perceptual and auditory-motor timing abilities across groups.

The perceptual thresholds for duration and rhythm discrimination tasks were calculated by averaging the values obtained across all trials for each task. The score was reported as the percentage of the stimulus IOI (Weber ratio), wherein the smaller number (ms) indicated better performance (i.e., higher perceptual acuity). In synchronization-continuation tapping tasks, only the continuation phase was analyzed. The mean inter-tap interval (ITI) was calculated as a measure of tapping variability (i.e., tapping rate), and its coefficient of variation (CV) reflected motor variability (calculated as the ratio of the SD of the ITIs over the mean ITI). For synchronization tasks, the mean phase synchronization error was computed as the absolute difference between the rhythmic auditory cue (metronome click) and the finger tap, reflecting how accurately the participants could align their taps with the auditory beat (Figure 1B). Synchronization variability was reported as the standard error (SE) (the SD of the synchronization error over the square root of the number of taps). Small synchronization errors (in ms) indicated high accuracy. To compare the performance characteristics of unpaced versus paced tapping, we calculated the absolute tapping period error and its standard deviation by extracting the difference between the absolute mean ITI and the auditory cue.

Statistical analyses were computed using R Studio (version 4.3.0). The alpha level was fixed at p = 0.05 for all data analyses. Data normality was verified with the Shapiro–Wilk test, and the Kruskal-Wallis H was reported if the normality assumption was violated for at least two groups (p < 0.05). Levene’s Test was used to evaluate the homogeneity of variances assumption for each ANOVA. If significant (p < 0.05), we reported Welch’s test. Group differences were assessed using chi-square tests for categorical variables and one-way and mixed-design analyses of variance (ANOVAs) for continuous variables. The lmer function (lme4 package in R) (Bates et al., 2015) was used to fit the ANOVA models. Additional analyses of covariance (ANCOVAs) were performed for perceptual and tapping data with age and working memory (Digit Span Backward) as covariates determined through bivariate correlations of the main study variables. When main or interaction effects were found, post-hoc paired and independent samples t-tests tests were computed using a Bonferroni correction. Two-tailed tests were reported for all analyses, except for the main effect of Task in the EEG analysis, given our a priori hypothesis regarding the direction of effects between paced and unpaced tapping.

Of the 34 participants who completed the study, the tapping data from one participant in the DCD group were excluded as the participant did not follow the task instructions. We further excluded the EEG data from three participants in each group due to an unsatisfactory signal quality caused by excessive motion artifacts, as determined using the HAPPE Processing Report metrics (e.g., if the percentage of good channels was below 65% or if the electrodes overlying the primary cortex were heavily contaminated) (Gabard-Durnam et al., 2018). The groups did not significantly differ in the percentage of bad channels (p = 0.971) and the number of rejected epochs (all p > 0.05). Additionally, the number of excluded taps was comparable across groups, except in the slow unpaced condition (p = 0.002), where more taps were excluded in the DCD group (DCD 10.9 ± 5.8, TD 6.1 ± 3.7, TDM 4.1 ± 2.7). As detailed in Table 1, the three groups did not significantly differ in age, cognitive abilities, immediate attention (Digits Forward), race, and household income (all p > 0.05). For the Digits Backward Test, the data significantly deviated from normality for the DCD (W = 0.794, p = 0.008) and TDM groups (W = 0.852, p = 0.039). A Kruskal-Wallis test revealed significant group differences in working memory abilities [Digits Backward Test; H₍₂₎ = 9.14, p = 0.010], with TD musicians scoring higher than children with DCD. However, this finding should be interpreted with caution as the group differences were only marginally significant after adjusting for age (p = 0.046). As expected, the MABC-2 confirmed the presence of motor difficulties for the DCD group only. Given that three participants with DCD had coexisting ADHD symptoms (caregiver report), we conducted additional analyses and found that participant characteristics and task performance did not significantly differ between the full DCD sample and the subgroup of children who had coexisting ADHD symptoms (DCD + ADHD). Children in the TD and TDM groups had no parent-reported ADHD or dyslexia symptoms.

One-way ANOVAs revealed significant group differences for both perceptual tasks. For the duration discrimination thresholds [F_(2,31) = 3.33, p = 0.049, η_p² = 0.18], the musician group displayed smaller thresholds than both the DCD and TD groups, indicating higher perceptual acuity for interval-based timing. However, the group differences were no longer significant after the Bonferroni adjustment (see Figure 2). The assumption of homogeneity of variances was not met for the rhythm discrimination task [F_(2,31) = 10.56, p < 0.005], therefore the Welch’s test is reported. In the rhythm discrimination task [F_{Welch(2,14.88)} = 18.67, p < 0.001, η_p² = 0.47], musicians had significantly smaller perceptual thresholds than both children with DCD [t₍₂₁₎ = 4.93, p < 0.001, d = 2.06] and their typically developing peers with no musical training [t₍₂₁₎ = 4.50, p < 0.001, d = 1.88]. Although the thresholds for the TD and DCD groups did not significantly differ [t₍₂₁₎ = 1.83, p = 0.082, d = 0.78], children with DCD had larger thresholds (23.10 ± 6.10) than the TD group (19.20 ± 3.60), indicating lower perceptual acuity. The group difference persisted after adjusting for age [F_(2,28) = 3.78, p = 0.035, η_p² = 0.21] (Figure 2).

The current study focused on temporal aspects of auditory processing (i.e., duration and rhythm discrimination) rather than pitch or loudness perception. Our findings suggest that children with DCD have lower perceptual acuity for rhythmic timing than their TD peers with musical training. Although children with musical training also displayed decreased duration discrimination thresholds, group effects were no longer significant after the Bonferroni adjustment. While our findings should be further assessed with larger samples, the results are partially in contrast with earlier studies that reported decreased perceptual acuity on duration-based tasks (i.e., higher thresholds) in children with motor coordination difficulties compared to TD peers (Williams et al., 1992; Lundy-Ekman et al., 1991). A more recent study also found that duration deviations elicited delayed mismatch negativity (MMN) latencies in their sample of 6-7-year-old children at risk for DCD (Chang et al., 2021).

For rhythm-based timing, existing research offers limited and inconclusive results. For example, Chang and colleagues (Chang et al., 2021) found significantly decreased rhythm discrimination sensitivities and delayed P3a latencies in response to rhythm deviants in children at risk for DCD. In contrast, Roche and colleagues (Roche et al., 2016) found no group differences in rhythmic timing between children with DCD and their TD peers. While more research is critically needed, several factors may have contributed to these contrasting results, including differences in psychophysical procedures, inclusion criteria, and levels of musical training across samples. Two studies that reported duration timing difficulties (Williams et al., 1992; Lundy-Ekman et al., 1991) were published before the term DCD was introduced in 1994 (Blank et al., 2012); thus, it is possible that the “clumsy” children from their sample would not meet the current DSM-5 or ICD-10 criteria for DCD (American Psychiatric Association, 2013). Moreover, the lack of group differences in rhythm-based timing reported by Roche et al. (2016) should also be interpreted cautiously since their threshold estimation procedure was incongruous with commonly used adaptive psychophysical measurements (Treutwein, 1995). Most importantly, none of the existing studies controlled for the effects of musical training. Therefore, it is likely that their TD samples consisted of children with varying levels of musical expertise, resulting in heightened differences in perceptual thresholds.

Our findings underscore the positive effects of extensive musical training on rhythmic timing abilities and are congruous with the existing literature showing heightened sensitivity to rhythmic patterns in adult musicians (Chen et al., 2008; Herholz and Zatorre, 2012). Still, there is a continuous debate regarding the impact of individual pre-existing factors on enhanced musical abilities, such as individual cognitive abilities, musical aptitude, and demographic characteristics (Schellenberg, 2016). To control for these factors, we conducted standardized neuropsychological tests and parent questionnaires during the first study visit and found no significant differences in cognitive abilities and socioeconomic status between the groups. Our groups also did not significantly differ in duration thresholds, which suggests that TD musicians in our sample did not have pre-existing auditory acuity. Differences between duration and rhythm discrimination thresholds found in our study may be explained by neuroimaging literature, which indicates that duration and rhythm perception rely on relatively distinct neural circuitries wherein duration discrimination involves cerebellar-cortical pathways and the basal ganglia-cortical pathways support rhythm-based timing (Grube et al., 2010; Teki et al., 2011).

Here, we implemented a simple rhythm discrimination task involving an isochronous sequence of tones provided by the metronome (i.e., anisochrony detection with tones) to ensure the tasks were suitable for children. However, it is possible that more complex stimuli, such as auditory beats embedded in a musical excerpt (i.e., anisochrony detection with music), might be more sensitive in detecting rhythmic timing differences (Sowiński and Dalla, 2013). Indeed, a recent study employed a similar subset of sensorimotor tasks with children with ADHD and found vulnerabilities in rhythm discrimination with complex stimuli but not for simple rhythmic patterns (Puyjarinet et al., 2017). Differences in beat-based timing were also reported across several neurodevelopmental disorders that frequently co-occur with DCD (Lense et al., 1835). Overall, there is a gap in the literature concerning auditory-perceptual timing in DCD, and the existing studies provide conflicting findings and a limited understanding of the role of musical training.

Findings from our tapping tasks are congruous with earlier studies showing that children with DCD are more variable and less accurate than their TD peers both when tapping to the auditory beat (Whitall et al., 2008; Whitall et al., 2006) and when maintaining a tapping pattern without external cues (Williams et al., 1992; Lundy-Ekman et al., 1991). The current study employed unimanual movements as they are largely unaffected by differences in cognitive abilities and cultural exposure (Whitall and Clark, 2018) and can be reliably studied in children. In the synchronization–continuation paradigm, children with DCD exhibited more variability than their peers, especially when coordinating movements at a slow pace. This is not surprising given the inter-tap interval of roughly 500 ms (2 Hz) is reported to be a preferred tapping rate in children (Repp, 2005; McAuley et al., 2006). However, recent literature points to additional factors that influence spontaneous motor tempo contributing to a variability of preferred rates (Desbernats et al., 2023). These factors may be both intrinsic (e.g., age, pathology, expertise) and extrinsic (type of task, physical training, external constraints). Developmental studies further indicate that children have a narrower frequency curve than adults, and their ability to entrain to slower rates broadens with age (Van Noorden and Moelants, 1999). For this reason, the slow condition imposed greater demands for all children in our sample.

When the auditory beat was present, TD musicians consistently displayed higher accuracy and lower variability, followed by TD children without musical training and children with DCD, indicating that long-term musical training can gradually influence how rhythmic periodicities are perceived and internalized, leading to enhanced auditory-motor coupling (Chen et al., 2008; Herholz and Zatorre, 2012). Likewise, it is plausible that children with musical training have developed enhanced beat alignment ability as extensive piano lessons require constant temporal adaptations across different tempi (Scheurich et al., 2020). Interestingly, this ability has been associated with formal musical training rather than current music playing (Spiech et al., 2023). In terms of clinical implications, our results indicate that children with DCD benefit from rhythmic auditory cues when coordinating movements. This finding is in line with existing literature showing positive effects of rhythmic entrainment in adults with movement disorders, such as Parkinson’s disease and stroke (Thaut and Abiru, 2010), and provides compelling evidence in favor of novel auditory/rhythm-based interventions for children with DCD.

To date, no studies have examined the neural dynamics underlying paced versus unpaced rhythmic movements in children with DCD and their TD peers. During paced tapping, the power spectral density in the beta band increased in the cortical motor network. Changes in power spectra closely mirrored the behavioral performance, with more beta power in musicians, followed by the TD and DCD groups. Although neurophysiological studies in DCD are lacking, research with adults showed a similar trend of increased beta power in musicians compared to individuals without musical training (Stegemöller et al., 2018) and decreased beta power in individuals with Parkinson’s disease (Stegemöller et al., 2016; Stegemöller et al., 2017).

We further examined functional connectivity patterns to gain insight into intrahemispheric and interhemispheric communication underlying rhythmic movements (Serrien and Brown, 2002) with high coherence values indicating an increased degree of synchronization (Fries, 2005). Our results showed that unpaced tapping involved a higher degree of neural communication in the ipsilateral intrahemispheric network, particularly in the DCD and TD groups (see Supplementary Table S4). Given their behavioral performance, this may indicate that unpaced tapping posed higher demands as children had to rely on their internal timing mechanisms to maintain the tapping rate. Indeed, improved motor performance has been associated with reduced recruitment of cortical and subcortical networks (Mayville et al., 2002; Pranjić et al., 2024). Studies in adults suggest that, while paced and unpaced performance engage overlapping neural networks, unpaced tapping requires additional processing resources in the medial circuitry (i.e., SMA) (Serrien, 2008; Lewis et al., 2004; Rao et al., 1997). In our study, augmented functional communication occurred in the ipsilateral fronto-central network, an area thought to be engaged with increased motor task complexity (Van Den Berg et al., 2011). Overall, these insights are congruent with developmental studies showing that cortico-cortical connectivity gradually decreases at the onset of locomotion due to synaptic pruning, contributing to the refinement of motor skills (Bell and Fox, 1996). Although age was not included as a covariate in our EEG analyses due to the lack of significant age differences between groups and in task performance, it is important to acknowledge that maturational differences in neural development may significantly influence EEG activity. For instance, prior work has shown that beta-band oscillatory activity develops across childhood and adolescence, with modulations in both spontaneous and task-related beta power reflecting the strengthening of top-down control mechanisms (Trevarrow et al., 2019). These beta dynamics follow distinct trajectories across the lifespan (Heinrichs-Graham et al., 2018). Moreover, developmental changes in functional connectivity also emerge with age and are thought to reflect the maturation of large-scale brain networks involved in executive function, motor planning, and inhibitory control (Smit et al., 2012; Uhlhaas et al., 2010). In younger children, these systems are less functionally integrated, which may result in lower beta coherence or atypical modulation patterns.

Interestingly, more interhemispheric connectivity was found in musicians when auditory cues were present. Similar findings were reported by Blais and colleagues (Blais et al., 2018), who found reduced interhemispheric fronto-central communication in teenagers with DCD who exhibited lower behavioral stability after motor practice. While future work involving larger samples is needed, neuroplastic changes in the corpus callosum have previously been reported in musicians (Schlaug et al., 1995; Vollmann et al., 2014), implying that musical training may enhance communication between hemispheres.

Several limitations should be considered when interpreting our findings. Consistent with common practice in DCD research, we used a cut-off point score of < 15^th percentile on the MABC-2 as our eligibility criterion for children with DCD. Nonetheless, future studies should investigate whether different percentile cut-offs (e.g., 5^th vs. 15^th) can further characterize performance differences within the DCD group. Another limitation of the current study is the modest sample size, which constrains statistical power and may limit the generalizability of findings. This challenge is particularly relevant in EEG data, where high variability and individual differences can impact signal detection. While our multimodal, within-subject design mitigates some concerns, we acknowledge that the observed effects should be interpreted with caution until replicated in larger samples. Given the age of our participants, we employed simple auditory stimuli and unimanual tapping movements. However, more complex auditory-perceptual and motor tasks may offer greater sensitivity in detecting timing vulnerabilities. Taken together, future studies should involve larger samples and employ different levels of task complexity while controlling for the effects of musical training and the presence of co-occurring conditions (Lino and Chieffo, 2022; Pranjić et al., 2023).