The effects of Parkinson's disease, music training, and dance training on beat perception and production abilities.

Humans naturally perceive and move to a musical beat, entraining body movements to auditory rhythms through clapping, tapping, and dancing. Yet the accuracy of this seemingly effortless behavior varies widely across individuals. Beat perception and production abilities can be improved by experience, such as music and dance training, and impaired by progressive neurological changes, such as in Parkinson's disease. In this study, we assessed the effects of music and dance experience on beat processing in young and older adults, as well as individuals with early-stage Parkinson's disease. We used the Beat Alignment Test (BAT) to assess beat perception and production in a convenience sample of 458 participants (278 healthy young adults, 139 healthy older adults, and 41 people with early-stage Parkinson's disease), with varying levels of music and dance training. In general, we found that participants with over three years of music training had more accurate beat perception than those with less training (p < .001). Interestingly, Parkinson's disease patients with music training had beat production abilities comparable to healthy adults while Parkinson's disease patients with minimal to no music training performed significantly worse. No effects were found in healthy adults for dance training, and too few Parkinson's disease patients had dance training to reliably assess its effects. The finding that musically trained Parkinson's disease patients performed similarly to healthy adults during a beat production task, while untrained patients did not, suggests music training may preserve certain rhythmic motor timing abilities in early-stage Parkinson's disease.

Introduction

Most humans naturally perceive the underlying temporal regularity in music termed the beat. Humans often spontaneously synchronize their body movements to music through tapping or clapping. The process of synchronizing, or entraining, movement to the beat engages motor areas of the brain. In particular, the basal ganglia have been shown to play a key role in perceiving the beat [1], and Parkinson’s disease patients, who have dysfunctional inputs to the basal ganglia, show specific beat perception impairments [2]. Beyond Parkinson’s disease, however, even neurotypical individuals show a striking range in how accurately they both perceive and synchronize to a beat. Some of this variability is related to past experiences, such as music and dance training [3, 4]. It is therefore possible that perception or production deficits in Parkinson’s disease may be offset by a music or dance background.

Neural timing networks involve both cortical and subcortical motor control areas [5, 6]. These cortical structures include the premotor cortex and supplementary motor area (SMA), and the subcortical structures include the basal ganglia and the cerebellum [5]. The basal ganglia are affected in Parkinson’s disease, a neurodegenerative disease characterized by progressive cell death of dopaminergic neurons in the substantia nigra, resulting in loss of excitatory stimulation of a part of the basal ganglia called the putamen [7]. The disruption of dopamine projection within these networks appears to result in beat processing deficits [2, 8, 9]. Patients are impaired on tapping tasks involving finger tapping to a metronome followed by paced tapping without a metronome [8] and more complex rhythm discrimination tasks that required participants to decipher whether two beat-based rhythms were identical [2]. These deficits may be related to dopamine levels in the basal ganglia, as Parkinson’s disease patients improve on rhythm discrimination tasks after taking dopaminergic medication [9]. Similar trends of impaired temporal discrimination performance are observed among healthy adults when dopamine uptake is disrupted by dopamine receptor antagonists [10]. Furthermore, beat perception and production deficits are correlated with idiopathic REM sleep disorder which commonly occurs prior to Parkinson’s disease onset and is often considered a prodromal-Parkinson’s disease symptom [11]. These studies confirm the crucial role of dopamine in timing and rhythm processing.

Music training

The effects of music training have been studied previously by comparing musicians and non-musicians. Musicians can distinguish changes in beat-based and nonbeat-based rhythms better than non-musicians [12]. Similarly, musicians are better able to extract metrical structures from music [13]. Aside from perceptual advantages, music training relates to better motor and timing abilities. On a simple tapping task, musicians demonstrated lower tapping variability and more accurate synchronization to the external rhythmic stimulus [14]. In addition, musicians were more sensitive to tempo changes in the stimulus and exhibited faster phase correction compared to non-musicians [14]. Recently developed behavioral batteries of rhythm tasks such as the Battery for the Assessment of Auditory Sensorimotor and Timing Abilities (BAASTA) and the Harvard Beat Assessment Test (H-BAT) corroborate other work showing that music training is associated with better beat processing abilities [15, 16].

Dance training

Similar to musicians, dancers also have superior timing perception abilities compared to non-dancers [17]. However, unlike musicians, dancers are especially skilled at entraining their movements to visual events. Dancers must synchronize their movements to the music and with the movements of other dancers, resulting in elevated motor entrainment abilities with both auditory and visual stimuli [18]. Dancers learn choreography by watching others perform and later fine-tuning their movements to match with other dancers [19]. Dancers engage both visual and motor networks during their training and are found to be better at extracting a beat from visual stimuli compared to musicians [20]. In addition, viewing dance movements enhances auditory meter perception in dancers suggesting visual-auditory entrainment abilities [21]. Dancers are experts in whole-body coordination to auditory cues. A study comparing dancer and non-dancer muscle contractions at the onset of salient metronome beats found that dancers had more accurate movements compared to non-dancers [22]. Overall, dancers have exceptional whole-body sensorimotor entrainment and motor coordination, both dynamic properties that are thought to contribute to their superior visual-motor and auditory-motor entrainment abilities [23].

Study rationale

Though beat perception and production are thought to be related, there is evidence that the abilities dissociate [24, 25], thus many assessments include both perceptual and production tasks [16, 26, 27]. The Beat Alignment Test (BAT) is one such assessment [26]. The beat perception task of the BAT has been extensively used in various studies [15, 24] and has successfully identified people with impaired beat perception but intact beat production, and vice versa [24, 25]. It is simple and brief, therefore especially useful in testing clinical populations in which complex or fatiguing tasks are less feasible. Recent research has used the BAT to assess sensorimotor integration as well as beat perception and production in Parkinson’s disease patients [9, 28].

Beat perception and production accuracy are positively impacted by music and dance training but negatively impacted by neurological changes in Parkinson’s disease. Previous studies have used the BAT to assess rhythmic abilities in healthy and clinical groups. We expect training to relate to better rhythmic abilities in healthy adults, but whether this advantage is preserved in Parkinson’s disease is unknown. In this study, the BAT was used to measure beat perception and production in the three participant groups: healthy young adults, healthy older adults, and people with early-stage Parkinson’s disease, all groups with varying levels of music and dance training. The data were aggregated from several studies conducted over a period of several years, in which all individuals completed the BAT and a demographics questionnaire in the context of other studies. We hypothesized that music and dance training would correlate with better beat perception and production abilities, while Parkinson’s disease would reduce these abilities relative to controls. The behavioral benefits of music and dance training may still be preserved in Parkinson’s disease. Therefore, Parkinson’s disease patients with previous training may have better abilities than patients without training.

Methods

Participants

278 healthy young adults (M = 20.41, SD = 3.01), 139 healthy older adults (M = 64.63, SD = 9.27) and 41 people with early-stage Parkinson’s disease (M = 68.28, SD = 7.73) were recruited for various music and gait studies conducted in the Music and Neuroscience Lab. People with early-stage Parkinson’s Disease (Hoehn & Yahr stages 2–3) were recruited from the community of Southwestern Ontario through community outreach and flyers. Given the exploratory nature of the study, Parkinson’s disease patients were not excluded based on medication regimen, years since diagnosis, or having received deep brain stimulation. Six participants who did not complete both beat perception and production tasks of the BAT and eight participants who did not indicate the years of previous music and dance training experience were excluded from the analyses. The final N = 458 only includes participants who completed the BAT and indicated the years of previous music and dance training experience. Participants of each group varied in level of music and dance training (Table 1). Based on years of training, participants were divided into two categories: 0–2 years and 3+ years. This threshold was chosen to balance sample sizes across participants with different degrees of music and dance training while maintaining the distinction between minimal and more extensive training. Informed consent was obtained from all participants and approval for all studies was obtained from the Western Medical Research Ethics Board or the Western Nonmedical Research Ethics Board (104487,106385).

Table 1

Participant demographics.

N = 458		Age	Music training (years)		Dance training (years)
		Years (SD)	0–2	3+	0–2	3+
Young adults	278	20.41 (3.01)	111	167	208	70
Older adults	139	64.63 (9.27)	71	68	111	28
Parkinson’s disease patients	41	68.28 (7.73)	25	16	40	1

Stimuli

Musical stimuli were taken from the Beat Alignment Test of the Goldsmiths Musical Sophistication Index (Gold-MSI) v1.0 [26] downloaded from https://www.gold.ac.uk/music-mind-brain/gold-msi/download/. Version 1.0 of the Gold-MSI is optimized relative to the version reported in Müllensiefen et al., (2014) [26], with 17 items, selected as described in documentation available at https://www.gold.ac.uk/music-mind-brain/gold-msi/download/ (S1 Fig). Musical excerpts were chosen from a variety of music genres and ranged from 10 to 16 seconds in duration. In the beat perception task, beeps were superimposed on the music excerpts 5 seconds into the music. The BAT was administered on a PC laptop using E-Prime (2.0) software (Psychology Software Tools, 2002). Auditory stimuli were delivered through Sennheiser HD 280 headphones. All participants completed both beat perception and production tasks in one session.

Beat perception task

Participants listened to musical excerpts (3 practice trials, 17 test trials) with superimposed metronome beeps either on or off the beat. Off-beat excerpts could either result from beeps coming in too early or too late relative to the actual beat (phase error), or from beeps too fast or too slow relative to the tempo of the actual beats (period error). Phase shifts of the superimposed beeps were adjusted 10% or 17.5% ahead relative to the musical beat. Period shifts of the superimposed beeps were adjusted 2% slower or faster relative to the musical tempo (S1 Fig). Participants were tasked to identify whether the superimposed beeps were “on the beat” or not, without using body movement to assist the judgement. The trial order was randomized, and participants rated how confident they were of their judgment after each excerpt on a 7-point Likert scale.

Beat production task

Participants heard the same musical excerpts as the beat perception task with the superimposed beeps removed. Each participant synchronized their finger tapping as soon as they perceived the beat of the music. Each excerpt was presented twice consecutively. The order of the musical excerpts was randomized, and participants were asked to rate their familiarity with each musical excerpt on a 7-point Likert scale. The extent to which participants matched their tapping to the actual beats was measured based on phase and tempo accuracy, as well as tapping variability.

Phase matching accuracy was represented by asynchrony (Eq 1), which measured the absolute difference between tap time and nearest beat position. The asynchrony score was obtained by taking the mean of the absolute difference between each tap and its nearest beat divided by the mean inter-beat interval (IBI). IBI was calculated by subtracting consecutive beat onsets. High asynchrony scores reflect high tapping phase error, indicating that participants tapped too early or too late relative to the actual beat. In contrast, low asynchrony scores reflected taps that were more aligned with the musical beat. Asynchrony scores were averaged across the 17 trials to obtain an average asynchrony score for each participant.

Tempo matching accuracy was represented by the coefficient of deviation (CDEV) score (Eq 2), which measured the absolute deviation between inter-response interval (IRI) and inter-beat interval (IBI). The inter-response interval was determined by subtracting consecutive tap onset times. High CDEV scores reflect high tapping period error, meaning that participants either tapped too fast or too slow relative to the actual beat tempo. In contrast, low CDEV scores reflected more accurate tempo matching abilities. CDEV scores were averaged across the 17 trials to obtain an average CDEV score for each participant.

Tapping variability was represented by the coefficient of variation (CoV) score (Eq 3), which measured motor response variability independent of the stimuli. High CoV scores reflect less consistent tapping, whereas low CoV scores reflect more consistently paced tapping. CoV scores were averaged across the 17 trials to obtain an average CoV score for each participant.

Demographic questionnaire

After the beat perception and production tasks, participants filled out a demographic questionnaire to describe their age and years of music and dance training experience. This exploratory study employed convenience sampling of participants that partook in different music and walking studies. Consequently, healthy young and older adults were given a general demographic questionnaire while Parkinson’s disease patients were given the Gold-MSI Questionnaire [26].

Statistical analyses

Beat perception ability was quantified by the percent of correct responses. Beat production was quantified by asynchrony and coefficient of deviation scores reflecting phase matching and tempo matching accuracy, respectively, as well as coefficient of variation scores reflecting tapping variability. To investigate the effects of music training, 3 (healthy young adults, healthy older adults, Parkinson’s disease patients) x 2 (0–2 years, 3+ years of music training) ANOVAs were conducted on beat perception and production measures. We excluded the Parkinson’s disease group from the dance training analysis because only one Parkinson’s disease patient reported 3+ years of dance experience. For this reason, a separate 2 (healthy young adults, healthy older adults) x 2 (0–2 years, 3+ years of dance training) x 2 (0–2 years, 3+ years of music training) ANOVA was conducted to investigate the combined effects of dance training and music training on beat perception and production in the healthy adult groups, excluding the Parkinson’s disease group. For this ANOVA, only main effects of or interactions with dance are reported, as music training effects are covered in the 3 x 2 ANOVA that enabled the inclusion of the Parkinson’s disease group. Follow-up ANCOVAs using age, music training and participant groups as covariates were conducted to investigate the dance training effects. Main effects and interactions were confirmed by follow-up simple main effect and post-hoc pairwise comparisons using Bonferroni correction. In addition to traditional frequentist statistical approaches, we conducted two Bayesian ANOVAs for each dependent variable similar to the traditional ANOVAs. Bayes hypothesis testing allows for the distinction between “absence of evidence” (i.e., the data is not informative, design underpowered) or “evidence of absence” (i.e., the data supports the null hypothesis), allowing for a more informed understanding of the results. One Bayesian ANOVA included music training levels and all three groups, and the other included dance training levels and only the young/older adult groups, to further quantify support for the null versus experimental hypotheses. A 2 x 2 x 2 Bayesian ANOVA with music training, dance training, and young/older groups produced similar outcomes to the individual music and dance training 2x2 Bayesian ANOVAs, so is not reported. Data were analyzed and visualized using JASP and R software.

Results

The mean age for older adults (64.5) significantly differed from that of Parkinson’s disease patients (68.2), as shown by a Welch’s unequal variance two-sample t-test (t = -2.58, p = .012). However, linear models fitting age and beat perception, asynchrony, coefficient of variation and coefficient of deviation indicated age did not predict performance on any of the dependent variables in these groups. Therefore, age differences in the range found in the older adults and Parkinson’s disease group do not appear to reliably affect beat perception or production (all p’s > .05).

Beat alignment test perception scores

Participants with 0–2 years and 3+ years of music training averaged 61% (SD = 0.16) and 70% (SD = 0.17) correct responses, respectively. The 3 (group) x 2 (music training) ANOVA revealed a main effect of music training [F(1, 457) = 20.42, p < .001, ηp2 = 0.043]. Participants with greater music training, regardless of participant group, demonstrated more accurate beat perception compared to those with minimal music training. The main effect of music training was qualified by a Music training x Group interaction [F(1, 457) = 3.49, p = .031, ηp2 = 0.015]. Simple main effects revealed that for both young adults and Parkinson’s patients, those with more extensive music training differed from those with minimal music training [young adults: F(1, 277) = 30.06, p < .001, ηp2 = 0.098; Parkinson’s: F(1,40) = 7.62, p = .006, ηp2 = 0.16], but there was no reliable effect of music training for older adults [F(1, 138) = 1.04, p = .31, ηp2 = 0.0075] (Fig 1A). No main effect of or interactions with dance training were noted (all p’s>.05) in the 2 (healthy adult groups) x 2 (music training) x 2 (dance training) ANOVA (Fig 2A).

Fig 1

Music training effects on beat perception and production.

Performance broken down by group (young, older, Parkinson’s disease) and music training (0–2 years, 3+ years) for beat perception (A), beat production phase matching (B), beat production tempo matching (C), and beat production tapping variability (D). For beat perception, young adults and Parkinson’s disease patients with more extensive music training were significantly better than those without. For asynchrony (phase matching), Parkinson’s disease patients with minimal music training were significantly worse than all other groups. No significant differences were present for coefficient of deviation (tempo matching). For coefficient of variation (tapping variability), older adults and Parkinson’s patients were more variable than younger adults, and participants with more extensive music training (regardless of group) were less variable than those with little training. Error bars indicate the standard error of the mean. ** = p < .01, *** = p < .001.

Fig 2

Dance training effects on beat perception and production.

Performance broken down by group (young, older, Parkinson’s disease) and dance training (0–2 years, 3+ years) for beat perception (A), beat production phase matching (B), beat production tempo matching (C), and beat production tapping variability (D). No significant differences were present for beat perception, phase matching (asynchrony) and tempo matching (coefficient of deviation). Tapping variability did differ between groups. Error bars indicate the standard error of the mean.

Beat alignment test production scores

Beat production: Phase matching

The 3 (group) x 2 (music training) ANOVA revealed a significant main effect of music training [F(1, 457) = 8.86, p = .003, ηp2 = 0.019] and a Music training x Group interaction [F(1, 457) = 4.19, p = .016, ηp2 = 0.018]. Follow-up simple main effects indicated that Parkinson’s patients with minimal music training had lower phase matching accuracy than Parkinson’s patients with more extensive training [F(1, 41) = 10.07, p = .002, ηp2 = 0.20]. Parkinson’s patients with minimal training also had lower phase matching accuracy than young adults and older adults in both music training groups (all p-values < .05) (Fig 1B). Interestingly, Parkinson’s patients with music training performed similarly to healthy adults with training, suggesting that music training was associated with retained beat production abilities for Parkinson’s disease patients. The 2 (healthy adult groups) x 2 (music training) x 2 (dance training) ANOVA revealed no significant dance training effects or interactions in healthy young and older adults (Fig 2B).

Beat production: Tempo matching

There were no significant effects in the 3 (group) x 2 (music training) ANOVA (Fig 1C). The 2 (healthy adult groups) x 2 (music training) x 2 (dance training) ANOVA revealed no main effects of or interactions with dance training (all p’s >.05) (Fig 2C).

Beat production: Tapping variability

The 3 (group) x 2 (music training) ANOVA revealed a main effect of music training [F(1,457) = 33.28, p < .001, ηp2 = 0.069] and a main effect of group [F(1, 457) = 6.00, p = 003, ηp2 = 0.026] (Fig 1D). The 2 (healthy adult groups) x 2 (music training) x 2 (dance training) ANOVA revealed no significant dance training effects or interactions in healthy young and older adults (Fig 2D).

Bayesian analyses

The Bayesian ANOVA compares the predictive performance of each model with and without each independent variable and their interactions. The P(M) column is the prior model probability which assumes that all rival models are equally likely to represent the data. The P(M|data) column indicates the probability of each model given the actual data. The BFm column indicates the relative likelihood of each model compared to the average of all other models, and the BF10 column indicates the relative likelihood of each model compared to the null model. As BF10 deviates from 1, support for the null or alternative hypothesis increases. Generally, 0.33 < BF10 < 3 indicates that the data is insufficient to support either null or alternative hypothesis. 0.1 < BF10 < 0.33 provides moderate support for the null hypothesis and 3 < BF10 < 10 provides moderate support for the alternative hypothesis. Finally, a BF10 < 0.1 provides strong support for the null hypothesis and BF10 > 10 provides strong support for the alternative hypothesis [29, 30]. Strength of the evidence can be quantified based on the Bayes Factor (e.g., BF10 = 8 is twice as strong as BF10 = 4 in supporting the alternative hypothesis).

Beat perception

When assessing music training and group differences on beat perception, the model including only the music training factor was more supported than the null model by a Bayes Factor of 685845, which is strong evidence (Table 2). Likewise, models including music and group main effects, as well as music and group main effects plus the music*group interaction had large BF10 values of 46220 and 36339, respectively (Table 2).

Table 2

Comparison of Bayes models: Music & group.

Models	P(M)	P(M|data)	BF M	BF 10	error %
Null model	0.20	0.00	0.00	1.00
music	0.20	0.89	33.23	685845	0.00
music + group + music*group	0.20	0.06	0.26	46220	1.85
music + group	0.20	0.05	0.20	36339	1.48
group	0.20	0.00	0.00	0.12	0.02

Note: P(M) = prior model probability; P(M|data) = posterior model probability; BFm = change from prior to posterior model odds; BF10 = Bayes Factor in favor of each model compared with the null model. Music = music training level, group = young/older/Parkinson’s group.

When assessing the effects of dance training on beat perception, the dance model had a BF10 value of 0.33 (Table 3), indicating insufficient support for either the null hypothesis or the alternative hypothesis [29, 30].

Table 3

Comparison of Bayes models: Dance & group.

Models	P(M)	P(M|data)	BF M	BF 10	error %
Null model	0.20	0.57	0.57	1.00
dance	0.20	0.19	0.19	0.33	0.00
group	0.20	0.17	0.17	0.29	0.00
dance + group	0.20	0.06	0.06	0.10	1.08
dance + group + dance*group	0.20	0.01	0.01	0.02	2.54

Note: P(M) = prior model probability; P(M|data) = posterior model probability; BFm = change from prior to posterior model odds; BF10 = Bayes Factor in favor of each model compared with the null model. Dance = dance training level, group = young/older group.

For music training, music and group models revealed a BF10 value between 0.33 and 3 (Table 4), suggesting that the data was not well represented by any of the provided models [29]. While the traditional 3 x 2 ANOVA above revealed a music*group interaction, the effect size was small, driven by Parkinson’s disease non-musicians differing from Parkinson’s disease musicians and healthy adults. Thus, both analyses indicate that the Music training x Group interaction was small, and, in the Bayesian analysis, insufficient to power the selection of music training and group as the best overall model. However, the Bayesian analysis provides additional information that the null hypothesis was also not strongly supported by the data.

Table 4

Comparison of Bayes models: Music & group.

Models	P(M)	P(M|data)	BF M	BF 10	error %
Null model	0.20	0.40	2.72	1.00
music	0.20	0.25	1.30	0.61	0.03
music + group + music*group	0.20	0.15	0.73	0.38	0.00
music + group	0.20	0.12	0.55	0.30	1.80
group	0.20	0.08	0.33	0.19	8.34

For dance training effects on phase matching, the dance model had a BF10 of 0.17 (Table 5), indicating moderate support for the null hypothesis.

Table 5

Comparison of Bayes models: Dance & group.

Models	P(M)	P(M|data)	BF M	BF 10	error %
Null model	0.20	0.76	12.86	1.00
dance	0.20	0.13	0.59	0.17	0.00
group	0.20	0.09	0.40	0.12	0.00
dance + group	0.20	0.02	0.06	0.02	2.15
dance + group + dance*group	0.20	0.00	0.02	0.01	5.91

For tempo matching, the music model had a BF10 of 0.50 (Table 6), indicating insufficient support for either the null hypothesis or alternative hypothesis. For dance, BF10 = 0.22 (Table 7), indicating moderate support for the null hypothesis.

Table 6

Comparison of Bayes models: Music & group.

Models	P(M)	P(M|data)	BF M	BF 10	error %
Null model	0.20	0.60	5.97	1.00
music	0.20	0.30	1.72	0.50	0.00
music + group + music*group	0.20	0.07	0.31	0.12	0.02
music + group	0.20	0.03	0.11	0.04	2.69
group	0.20	0.00	0.01	0.00	2.09

Table 7

Comparison of Bayes models: Dance & group.

Models	P(M)	P(M|data)	BF M	BF 10	error %
Null model	0.20	0.69	8.73	1.00
dance	0.20	0.15	0.71	0.22	0.00
group	0.20	0.12	0.56	0.18	0.00
dance + group	0.20	0.03	0.11	0.04	1.12
dance + group + dance*group	0.20	0.01	0.05	0.02	1.44

Beat production: Tapping variability. When assessing music training and group differences on tapping variability, the music and group model was more supported than the null model by a Bayes Factor of 5.30e+9, followed by the full model with interaction (music and group plus music*group interaction) with a Bayes Factor of 1.19e+9 and the music model with a Bayes Factor of 2.65e+8 (Table 8). The music and group model has the largest Bayes Factor and thus suggests that it is most likely to accurately reflect the variance seen in the data. When assessing dance training and group differences on tapping variability, the group and dance model was more supported than the null model by a Bayes Factor of 2.63, which is weak evidence. However, unlike other measures of beat perception, the group model was best supported with a Bayes Factor of 16.78 (Table 9).

Table 8

Comparison of Bayes models: Music & group.

Models	P(M)	P(M|data)	BF M	BF 10	error %
Null model	0.20	0.00	0.00	1.00
music + group	0.20	0.79	14.90	5.30e+9	7.12
music + group + music*group	0.20	0.17	0.84	1.19e+9	1.75
music	0.20	0.04	0.16	2.65e+8	0.00
group	0.20	0.00	0.00	301.27	0.02

Table 9

Comparison of Bayes models: Dance & group.

Models	P(M)	P(M|data)	BF M	BF 10	error %
Null model	0.20	0.05	0.20	1.00
group	0.20	0.79	15.02	16.78	0.00
dance	0.20	0.12	0.56	2.62	1.68
dance + group	0.20	0.03	0.13	0.68	1.37
dance + group + dance*group	0.20	0.01	0.03	0.16	0.00

Discussion

This study examined the effects of music and dance training on beat perception and production abilities across the life span and in the context of Parkinson’s disease. We predicted that music and dance training would improve beat perception and production skills, while the neurological deficits associated with Parkinson’s disease would negatively affect these skills. We further predicted that some positive impacts of music and dance training would be preserved despite disease state. Indeed, on the beat perception task, young adults and Parkinson’s disease patients who were musically trained did better than those who weren’t, but this difference was not significant for older adults. For beat production, only the Parkinson’s groups differed as a function of musical training: Parkinson’s patients with music training performed comparably to healthy adults, whereas patients with minimal training showed significantly worse phase matching accuracy (high asynchrony scores). However, tempo matching accuracy was not affected by music training in any group. Furthermore, no dance training effects were found, although the dance training analysis was restricted to older and younger adults, as only one Parkinson’s patient had 3+ years of dance training. Finally, interpretation of the reported null results from the traditional frequentist statistical analyses is aided by Bayesian analyses indicating that true null results are unlikely, suggesting that greater power may be necessary to detect effects of dance training or music.

We observed that music training was associated with better beat perception in young adults and Parkinson’s disease patients, but not older adults. While it is widely documented in the literature that music training is associated with better beat perception, the results obtained in the current study may suggest that music training effects on beat perception could decay over time since training, as evidenced by the significant differences seen in healthy young adults, but not healthy older adults. However, this decay either does not apply to the Parkinson’s patients, or other factors are at play, as music training was associated with better performance for patients.

Contrary to our hypothesis, Parkinson’s disease patients were not significantly impaired on the beat perception task. These results were consistent with Cameron and colleagues’ findings that beat perception tested on the BAT did not differ across Parkinson’s disease and healthy control groups [9]. In contrast, our results differed from Benoit et al.’s findings that Parkinson’s patients showed worse timing perception than healthy adults [31]. However, in their study, Parkinson’s patients were tasked to detect misaligned beats in a two-measure music excerpt, as opposed to several seconds of tones overlaid on music in the BAT. This task is less taxing on memory and attention, and thus may not be able to differentiate between healthy adult groups and Parkinson’s patients. Rhythm perception was also examined in Cameron et al.’s study using a rhythm discrimination task (not part of the BAT). In their work, the rhythm discrimination task was more sensitive to timing perception deficits in Parkinson’s disease than BAT. While both BAT and rhythm discrimination tasks measure beat perception abilities, there are important differences. Beat perception in the discrimination task relies solely on temporal information, without the additional beat cues afforded by music in the BAT (e.g., pitch, harmony, timbral, and amplitude cues). The rhythm discrimination task also involves working memory to compare consecutively presented rhythms. In contrast, the BAT relies on a comparison between simultaneous temporal sequences (musical stimuli and overlaid tones), and it does not require attending to the stimuli for the entire duration nor remembering them. Cochen de Cock et al. corroborated these findings by suggesting that cognitive abilities such as attention, executive function and cognitive flexibility could influence beat perception abilities [32]. Thus, the mechanisms required to perform the beat perception task in the BAT could be intact in Parkinson’s disease while mechanisms for strictly temporally-based rhythm perception and comparison could be impaired.

We predicted that rhythm-intensive training, such as through dance or music, would improve beat perception skills. Surprisingly, only music training elicited positive effects on beat perception. Although dancers generally performed better on the beat perception test than non-dancers, the differences were not significant, consistent with other studies that tested young adults on the BAT [33]. Importantly, it appears unlikely that years of training differed between musicians and dancers. The Gold-MSI groups music training into multi-year levels, and the average level for the musicians was 6–9 years. The dance questionnaire requested specific numbers, and the 3+ years group averaged 8.1 years. Thus, the lack of dance training effect relative to musical training effect seems unlikely to be caused by differences in years of training. Of course, years of training is an imprecise quantification of true training effects, as the rigor of different training programs and hours of deliberate practice varies across individuals.

We found that Parkinson’s disease patients performed worse than healthy adults on the beat production task, however, this appeared to be influenced by music training. Parkinson’s disease patients with minimal music training exhibited lower phase matching accuracy than healthy adults, as reflected by their increased asynchrony scores. However, decreased accuracy on the motor tasks overall may reflect nothing more than the generalized effects of Parkinson’s disease, consistent with previous findings that patients tap more variably than healthy controls [34]. Interestingly, Parkinson’s disease patients with more extensive music training exhibited phase matching similar to that of healthy adults, unlike patients with minimal training. This suggests that music training may have influenced motor control or temporal accuracy. Though the literature on music training effects in the Parkinson’s disease population is limited, these results were consistent with the trends seen in healthy adult populations [35, 36]. Older adults in our study (regardless of whether they have Parkinson’s disease) displayed higher tapping variability compared to younger adults, and musicians (regardless of group) displayed more consistent tapping. These results were consistent with trends seen in Thompson et al.’s cross-sectional study [37]. However, an additional music*group interaction is plausible provided that the model representing music and group and their interaction is well supported by the 3x2 Bayesian ANOVA. Contrary to previous findings that Parkinson’s disease patients tapped either faster or slower on synchronization and self-paced timing tasks [38], patients in this study did not demonstrate significantly different tapping tempos compared to healthy adults.

People with minimal (0–2 years) dance training did not differ from those with more extensive (3+ years) dance training on beat production. We performed another analysis using a stricter cutoff for dance training (0–5 vs. 6+ years) but still did not find any significant differences. These results contrast previous findings on rhythm entrainment which found that dancers were better at synchronizing whole-body movements to a recurring beat compared to non-dancers [23]. However, the beat production task may contain components that are emphasized more by music training than dance training. For example, dancers often use visual cues, observing other dancers, to fine-tune their entrainment [19]. Furthermore, dancers use more whole-body movements when synchronizing with music rather than finger-tapping [14, 22]. Future studies could investigate both auditory-motor and visual-motor entrainment to better understand the effects of dance training on motor entrainment skills.

Parkinson’s disease-related motor symptoms are most commonly treated using pharmaceutical therapies, such as levodopa, MAO-B inhibitors and dopamine agonists [39]. Medications help manage some motor symptoms but don’t necessarily improve gait symptoms, such as shuffling and freezing [40]. Therefore, Parkinson’s disease patients are frequently treated with rehabilitative therapies; rhythmic auditory stimulation (RAS) is a common therapy aimed at improving gait patterns. RAS provides temporal cues, such as a metronome, to which a person can entrain their walking pattern. Better rhythm processing abilities correlate with better RAS outcomes in both healthy and Parkinson’s disease populations [32, 41]. Therefore, Parkinson’s disease patients with music training may be better candidates for music- and rhythm-based therapies. The current results suggest it may be possible to adapt the BAT to be an effective screening for Parkinson’s disease patients who might benefit from rhythm-based interventions by identifying those with intact beat processing abilities.

The current study employed a convenience sample from multiple studies. Parkinson’s disease patients completed the Gold-MSI questionnaire, which groups music training into categories, (e.g., 0, 1, 2, 3, 4–5, 6–9 years) whereas healthy adults completed a survey that simply asked for the number of years in which they engaged in regular music practice. Music training in healthy adults was grouped to match Parkinson’s disease patients to analyze the entire dataset. To minimize sample size disparity, music training categories were grouped once more to create two reasonably sized groups (0–2 vs. 3+ years) while maintaining the distinction between minimal and more extensive training. Few participants reported dance training experience, so two sizable groups (0–2 vs. 3+ years) were created for statistical analysis purposes while excluding Parkinson’s disease patients from this analysis due to low sample size. In addition, most Parkinson’s disease participants were receiving dopaminergic therapy, which improves beat perception and production abilities [9, 38]. Therefore, greater group differences may be found in an off-medication paradigm. Hoehn and Yahr stage is also the only indicator of disease severity collected in the study, and future work addressing the impact of these factors across disease stages would benefit from including more specific disease severity information such as disease duration or the levodopa equivalent daily dose and by including participants with moderate to severe stages of Parkinson’s.

Conclusion

Our findings indicate significantly better beat perception and production skills among participants with more extensive music training. Parkinson’s disease patients with more extensive music training exhibited better beat perception and production skills than patients with minimal training. These results contribute to the growing knowledge of the long-term effects of music training and suggest that music training may preserve certain motor timing functions related to beat processing in early-stage Parkinson’s disease.

Stimuli used in beat alignment perception test v1.0.

Note: The order of the musical excerpts was randomized.

(PDF)

Click here for additional data file.

23 Nov 2021

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Sonja Kotz

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match.

When you resubmit, please ensure that you provide the correct grant numbers for the awards you received for your study in the ‘Funding Information’ section.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thank you for giving me the opportunity to review this article. This study provides an insightful view of the role of music and dance training on the change in rhythmic skills with age. The authors tested rhythm perception and tapping synchronization with musical material in three different groups (healthy adults, healthy older adults, and older adults with Parkinson’s disease) with or without music or dance training. The main finding is that Parkinsonian participants with music training performed as well as healthy adults in synchronization task.

The article is well written, and the methodology is appropriate. My main concern is that the authors did not use an index of tapping variability per se (e.g., coefficient of variation of the asynchronies). They claim that the asynchrony score reflects variability and consistency, which is not exact. I suppose that additional group differences would be highlighted by tapping variability, and this might change the overall findings pattern. Please see below for additional suggestions about this point.

Here are the specific comments that, I think, should be addressed by the authors to improve the manuscript:

Introduction

- “simple timing tasks [8] and more complex beat-based rhythm discrimination tasks [2].” (l65). What are the simple tasks that the authors refer to? Interval-based? It would be helpful to describe briefly the tasks typically used in these experiments here. In the same paragraph, two other types of tasks are introduced (“rhythm discrimination tasks”, l67, and “temporal discrimination task”, l69). Please clarify by giving a short description of the tasks, as very few readers will be familiar with that.

- Note that in some studies, the BAT is only the perceptual task ([11], [13], [16] in the article reference list), whereas in others it includes a variety of other tasks ([14] in the article reference list). It would be worth mentioning it to avoid possible confusion.

- The second paragraph of the “Rhythm disturbance in Parkinson’s disease” section does not seem to fit in well with the section. The authors can consider moving it to another subsection.

- Because the article does not include any brain imaging measures, I suggest not to start by describing the brain areas involved in PD, music training, and dance training in the different sections. Focusing on the behavior is more relevant. Neuronal underpinnings can be briefly mentioned at the end of each paragraph, probably in one sentence, especially because they are not discussed in the rest of the article.

- “As the basal ganglia have not been implicated in the neural changes associated with music and dance training” (l126). I would tone down this statement. It may be true that no changes have been directly observed in the basal ganglia, and I trust the authors who have certainly carefully reviewed the literature on this topic. Nevertheless, this absence of observed change in the basal ganglia may have different causes, and the basal ganglia are part of networks that are associated with music and dance training.

Methods

- Is there a significant difference in age between the older adults and the PD groups?

- “Six participants who did not complete both beat perception and production tasks of the BAT and participants who did not indicate the years of previous music and dance training experience were excluded from the analyses” (l140-142). Does the final N of participants presented in table 1 include those participants? If so, I suggest to rather present only the participants that were included in the analyses.

- The initial battery developed by Müllensiefen, Gingras, Musil, & Stewart (2014) that is used in this study contains 18 stimuli. Why did the authors use only 17? I would find it useful to give some precision about the stimuli (e.g., N of trials with phase and period changes, percentage of change etc.) so that the reader does not have to read the article by Müllensiefen et al. to understand the methods. This is particularly important if the number of stimuli differs.

- “low asynchrony scores reflected less variable and more consistent tap times” (l180). I do not think the asynchrony score reflect variability and consistency. It tells how far from the beat the participants tapped; a high score can reflect a very consistent performance in antiphase, for example. None of the two indexes reflect variability; it could be worth using a coefficient of variation of the asynchronies [SD(asynch)/Mean(asynch)], where asynch = Mean(Response - Beat). This index might reveal additional effects that the tempo matching index failed to capture. Notably, I expect that participants with dance and music training are less variable in their tapping performance. For example, a recent study on children with cerebellar anomalies showed that a dance training protocol reduced their variability in a synchronization task (Bégel et al., 2021).

Results

- “When assessing the effects of dance training on beat perception, the null model had a BF10 of 1 (Table 2b), suggesting that the data could not discriminate between the alternative hypothesis from the null hypothesis” (l288-290). This is also the case for the music & group model. Please explain a bit more how you come to this conclusion and what is the difference between the two Null models. This remark applies to the phase matching analyses as well.

- Figure 2. Because the PD group was not included in the analyses with dance training, I think their results should not appear in the figure.

- Figure 3. It is not clear what the three lines (user, wide, ultrawide) are. Most readers are not familiar with Bayesian statistics. It is important to give more information.

Discussion

- The absence of difference between the PD and healthy adults groups in beat perception is inconsistent with the results of Benoit et al. (2014). There are interindividual difference in Parkinson’s patients that may explain the contradictory results (Cochen de Cock et al., 2018). I think this should be mentioned in the discussion.

- The “Implications” and “Limitations” sections are quite long. Condensing them into one section seems appropriate.

References

Bégel, V., Bachrach, A., Dalla Bella, S., Laroche, J., Delval, A., Riquet, A., & Dellacherie, D. (2021). Dance improves motor, cognitive and social skills in children with developmental cerebellar anomalies. The Cerebellum, Advance Online Publication

Benoit, C. E., Dalla Bella, S., Farrugia, N., Obrig, H., Mainka, S., & Kotz, S. A. (2014). Musically cued gait-training improves both perceptual and motor timing in Parkinson’s disease. Frontiers in Human Neuroscience, 8, 494.

De Cock, V. C., Dotov, D. G., Ihalainen, P., Bégel, V., Galtier, F., Lebrun, C., ... & Dalla Bella, S. (2018). Rhythmic abilities and musical training in Parkinson’s disease: do they help?. NPJ Parkinson's disease, 4(1), 1-8.

Reviewer #2: Hsu et al. explored in a large population, the effects of music and dance training on beat perception and production abilities across the life span and in the context of Parkinson’s disease, predicting that music and dance training would improve beat perception and production skills, while Parkinson’s disease would negatively affect these skills.

They observed that young adults and Parkinson’s disease patients who were musically trained did better than those who weren’t in beat perception tasks. This effect was not significant in older adults.

For beat production, only Parkinson’s patients with music training performed comparably to healthy adults, whereas patients with minimal training had significantly higher asynchrony scores.

Tempo matching was not modified by musical training and dance training surprisingly had no effect.

The results of the study are surprising since the expected effects of musical training are limited, and the dance training were not observed.

The main limitation is that the information about the training is very limited: we don’t know when the training was and what its intensity and level was. This an important limitation in this study that is discussed in the discussion

There is also an important lack of information on Parkinson’s disease. Hoehn and Yahr stage is very insufficient to evaluate PD severity. Disease duration Ldopa Equlivalent daily dose are needed to understand the kind of patients that were explored. These two information should be added.

Also the recent reference on rhtythm disturabances in PD and prePD should be added and discussed

Rhythm disturbances as a potential early marker of Parkinson's disease in idiopathic REM sleep behavior disorder.

Cochen De Cock V, de Verbizier D, Picot MC, Damm L, Abril B, Galtier F, Driss V, Lebrun C, Pageot N, Giordano A, Gonzalvez C, Homeyer P, Carlander B, Castelnovo G, Geny C, Bardy B, Dalla Bella S.Ann Clin Transl Neurol. 2020 Mar;7(3):280-287. doi: 10.1002/acn3.50982. Epub 2020 Feb 14.PMID: 32059086

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Valentin Bégel

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

6 Jan 2022

Dear Dr. Sonja Kotz,

Thank you for handling our manuscript, “The effects of Parkinson’s disease, music training, and dance training on beat perception and production abilities,” for possible publication in Plos One. We also thank the reviewers for their careful feedback and helpful comments. We now submit a revised version, with additions about beat production tapping variability. Point-by-point responses to the reviewers’ comments are attached below. Note that the line numbers refer to the Revised Manuscript with Tracked Changes file. We hope that you might find this version to be suitable for publication for the wide readership of Plos One.

Sincerely,

Prisca Hsu, Emily Ready, Jessica Grahn

Reviewer #1

Introduction

- “simple timing tasks [8] and more complex beat-based rhythm discrimination tasks [2].” (l65). What are the simple tasks that the authors refer to? Interval-based? It would be helpful to describe briefly the tasks typically used in these experiments here. In the same paragraph, two other types of tasks are introduced (“rhythm discrimination tasks”, l67, and “temporal discrimination task”, l69). Please clarify by giving a short description of the tasks, as very few readers will be familiar with that.

Indeed, there is room to clarify what is meant by simple and complex here. We have added task clarification in lines 64-66:

Patients are impaired on tapping tasks involving finger tapping to a metronome followed by paced tapping without a metronome [8] and more complex rhythm discrimination tasks that required participants to decipher whether two beat-based rhythms were identical [2].

- Note that in some studies, the BAT is only the perceptual task ([11], [13], [16] in the article reference list), whereas in others it includes a variety of other tasks ([14] in the article reference list). It would be worth mentioning it to avoid possible confusion.

Yes, good point. Reference 11,13 and 16 in the original submission, only used the perception component of the BAT and used another, different, synchronization task to measure beat production. We have now noted that these studies only utilized the beat perception component of the BAT.

L137-142: Though beat perception and production are thought to be related, there is evidence that the abilities dissociate [24,25], thus many assessments include both perceptual and production tasks [16,26,27]. The Beat Alignment Test (BAT) is one such assessment [26]. The beat perception task of the BAT has been extensively used in various studies [15,24] and has successfully identified people with impaired beat perception but intact beat production, and vice versa [24,25].

- The second paragraph of the “Rhythm disturbance in Parkinson’s disease” section does not seem to fit in well with the section. The authors can consider moving it to another subsection.

We agree that the second paragraph seems more about the relationship between production/perception and the BAT, not on rhythm disturbances in PD and not about the neural underpinnings. We have moved this paragraph to the Study Rationale section.

- Because the article does not include any brain imaging measures, I suggest not to start by describing the brain areas involved in PD, music training, and dance training in the different sections. Focusing on the behavior is more relevant. Neuronal underpinnings can be briefly mentioned at the end of each paragraph, probably in one sentence, especially because they are not discussed in the rest of the article.

We agree that the descriptions of anatomical differences between musicians/non-musicians and dancers/non-dancers aren’t relevant to the current study, we have now removed that information.

- “As the basal ganglia have not been implicated in the neural changes associated with music and dance training” (l126). I would tone down this statement. It may be true that no changes have been directly observed in the basal ganglia, and I trust the authors who have certainly carefully reviewed the literature on this topic. Nevertheless, this absence of observed change in the basal ganglia may have different causes, and the basal ganglia are part of networks that are associated with music and dance training.

Fair point. We have tempered the sentence accordingly:

L155-159: We hypothesized that music and dance training would correlate with better beat perception and production abilities, while Parkinson’s disease would reduce these abilities relative to controls. The behavioral benefits of music and dance training may still be preserved in Parkinson’s disease. Therefore, Parkinson’s disease patients with previous training may have better abilities than patients without training.

Methods

- Is there a significant difference in age between the older adults and the PD groups?

The mean age for older adults is 64.5 and Parkinson’s disease patients is 68.2. A Welch two-sample t-test does show a significant difference between mean ages for the two groups (p = .01), but linear models fitting age and beat perception, asynchrony, coefficient of variation and coefficient of deviation indicate no significant effects of age on the dependent variables (all p’s > .05). We believe that the age difference between older adults and Parkinson’s disease patients is therefore not driving any group differences in beat perception or production.

Each graph is showing age of the participant along the x-axis and the relevant dependent variable along the y-axis. These graphs include data points from both healthy older and Parkinson’s groups.

We have added this information to our results section, clarifying that there is a significant difference between the ages of healthy older adults and Parkinson’s patients, but this age difference does not reliably affect in beat perception and production abilities.

L298-317: The mean age for older adults (64.5) significantly differed from that of Parkinson’s disease patients (68.2), as shown by a Welch’s unequal variance two sample t-test (t = -2.58, p = .012). However, linear models fitting age and beat perception, asynchrony, coefficient of variation and coefficient of deviation indicated age did not predict performance on any of the dependent variables in these groups. Therefore, age differences in the range found in the older adults and Parkinson’s disease group do not appear to reliably affect beat perception or production (all p’s > .05).

- “Six participants who did not complete both beat perception and production tasks of the BAT and participants who did not indicate the years of previous music and dance training experience were excluded from the analyses” (l140-142). Does the final N of participants presented in table 1 include those participants? If so, I suggest to rather present only the participants that were included in the analyses.

The final N=458 does not include participants who did not complete both beat perception and production tasks or did not indicate years of music/dance training. We have clarified this in the Participants section.

L169-173: Six participants who did not complete both beat perception and production tasks of the BAT and eight participants who did not indicate the years of previous music and dance training experience were excluded from the analyses. The final N=458 only includes participants who completed the BAT and indicated the years of previous music and dance training experience.

- The initial battery developed by Müllensiefen, Gingras, Musil, & Stewart (2014) that is used in this study contains 18 stimuli. Why did the authors use only 17? I would find it useful to give some precision about the stimuli (e.g., N of trials with phase and period changes, percentage of change etc.) so that the reader does not have to read the article by Müllensiefen et al. to understand the methods. This is particularly important if the number of stimuli differs.

Indeed, this is correct. The musical stimuli for the BAT were taken from the Beat Alignment Test of the Goldsmiths Musical Sophistication Index (Gold-MSI) v1.0 downloaded from https://www.gold.ac.uk/music-mind-brain/gold-msi/download/. This version was recommended to us by the authors of the 2014 study when we contacted them for stimuli, as Version 1.0 of the Gold-MSI is optimized relative to the version reported in Müllensiefen et al., 2014’s study, with 17 items, selected as described in the documentation available at https://www.gold.ac.uk/music-mind-brain/gold-msi/download/). We added a table outlining the BAT stimuli as supplementary material, and added a reference to the website and a specific version number of the BAT to clarify exactly which version was used. This also adds relevant information about the number of trials with phase and period shifts, as well as the amounts of those shifts on each trial.

L195-199: Musical stimuli were taken from the Beat Alignment Test of the Goldsmiths Musical Sophistication Index (Gold-MSI) v1.0 [26] downloaded from https://www.gold.ac.uk/music-mind-brain/gold-msi/download/. Version 1.0 of the Gold-MSI is optimized relative to the version reported in Müllensiefen et al., (2014), with 17 items, selected as described in the documentation available at https://www.gold.ac.uk/music-mind-brain/gold-msi/download/ (S1 Table).

L213-216: Phase shifts of the superimposed beeps were adjusted 10% or 17.5% ahead relative to the musical beat. Period shifts of the superimposed beeps were adjusted 2% slower or faster relative to the musical tempo (S1 Table).

- “low asynchrony scores reflected less variable and more consistent tap times” (l180). I do not think the asynchrony score reflect variability and consistency. It tells how far from the beat the participants tapped; a high score can reflect a very consistent performance in antiphase, for example. None of the two indexes reflect variability; it could be worth using a coefficient of variation of the asynchronies [SD(asynch)/Mean(asynch)], where asynch = Mean(Response - Beat). This index might reveal additional effects that the tempo matching index failed to capture. Notably, I expect that participants with dance and music training are less variable in their tapping performance. For example, a recent study on children with cerebellar anomalies showed that a dance training protocol reduced their variability in a synchronization task (Bégel et al., 2021).

This is a great suggestion. We have adjusted the interpretation for high/low asynchrony scores and analyzed coefficient of variability scores. The CoV data is now included in all results.

Results

- “When assessing the effects of dance training on beat perception, the null model had a BF10 of 1 (Table 2b), suggesting that the data could not discriminate between the alternative hypothesis from the null hypothesis” (l288-290). This is also the case for the music & group model. Please explain a bit more how you come to this conclusion and what is the difference between the two Null models. This remark applies to the phase matching analyses as well.

We added a brief explanation of the 3 logical states (null hypothesis supported, null hypothesis rejected, insufficient information to discriminate between the two hypotheses) in relation to BF10 values in the “Bayesian analyses” paragraph.

L391-401: The Bayesian ANOVA compares the predictive performance of each model with and without each independent variable and their interactions. The P(M) column is the prior model probability which assumes that all rival models are equally likely to represent the data. The P(M|data) column indicates the probability of each model given the actual data. The BFm column indicates the relative likelihood of each model compared to the average of all other models, and the BF10 column indicates the relative likelihood of each model compared to the null model. Generally, a BF10 < 1/10 provides strong support for the null hypothesis, BF10 > 10 provides strong support for the alternative model, and 1/3 < BF10 < 3 indicates that the data is insufficient to support either hypothesis [29,30]. The strength of the evidence can be quantified based on the Bayes Factor (e.g., BF10 =8 is twice as strong as BF10 = 4 in supporting the alternative hypothesis).

- Figure 2. Because the PD group was not included in the analyses with dance training, I think their results should not appear in the figure.

We removed the PD group from the dance figures.

- Figure 3. It is not clear what the three lines (user, wide, ultrawide) are. Most readers are not familiar with Bayesian statistics. It is important to give more information.

Thanks for noting this, we realized the sequential aspect of the analysis isn’t relevant to the analysis and have decided to remove the figure entirely.

Discussion

- The absence of difference between the PD and healthy adults groups in beat perception is inconsistent with the results of Benoit et al. (2014). There are interindividual difference in Parkinson’s patients that may explain the contradictory results (Cochen de Cock et al., 2018). I think this should be mentioned in the discussion.

Thanks for the suggestion. We have referenced and discussed the Benoit et al. (2014) and Cochen et al., (2018) papers in our discussion section.

L499-504: In contrast, our results differed from Benoit et al.’s findings that Parkinson’s patients showed worse timing perception than healthy adults [31]. However, in their study, Parkinson’s patients were tasked to detect misaligned beats in a two-measure music excerpt, as opposed to several seconds of tones overlaid on music in the BAT. This task is less taxing on memory and attention, and thus may not be able to differentiate between healthy adult groups and Parkinson’s patients.

L528-530: Cochen de Cock et al. corroborated these findings by suggesting that cognitive abilities such as attention, executive function and cognitive flexibility could influence beat perception abilities [32].

- The “Implications” and “Limitations” sections are quite long. Condensing them into one section seems appropriate.

We have condensed these paragraphs into one section.

Reviewer #2

The main limitation is that the information about the training is very limited: we don’t know when the training was and what its intensity and level was. This an important limitation in this study that is discussed in the discussion.

We agree this is a limitation in our study and have noted that in our discussion section.

L543-544: Years of training is an imprecise quantification of true training effects, as the rigor of different training programs and hours of deliberate practice varies across individuals.

There is also an important lack of information on Parkinson’s disease. Hoehn and Yahr stage is very insufficient to evaluate PD severity. Disease duration Ldopa Equlivalent daily dose are needed to understand the kind of patients that were explored. These two information should be added.

Thanks for noting this limitation, we agree that the Hoehn and Yahr scale provides limited information on the PD severity. We unfortunately do not have other information on the PD participants regarding to their disease and have noted this limitation in our discussion section.

L611-615: Hoehn and Yahr stage is also the only indicator of disease severity collected in the study, and future work addressing the impact of these factors across disease stages would benefit from including more specific disease severity information such as disease duration or the levodopa equivalent daily dose and by including participants with moderate to severe stages of Parkinson’s.

Also the recent reference on rhtythm disturabances in PD and prePD should be added and discussed

The reference was added to our introduction section.

L73-76: Furthermore, beat perception and production deficits are correlated with idiopathic REM sleep disorder which commonly occurs prior to Parkinson’s disease onset and is often considered a prodromal-Parkinson’s disease symptom [11].

Submitted filename: Response_To_Reviewers.docx

Click here for additional data file.

19 Jan 2022

20 Jan 2022

Dear Dr. Sonja Kotz,

Thank you for handling our manuscript, “The effects of Parkinson’s disease, music training, and dance training on beat perception and production abilities,” for possible publication in Plos One. We also thank the reviewers for their careful feedback and helpful comments. We now submit a revised version, with further clarifications about our Bayesian analyses. Point-by-point responses to the reviewers’ comments are attached below. Note that the line numbers refer to the Revised Manuscript with Tracked Changes file. We hope that you might find this version to be suitable for publication for the wide readership of Plos One.

Sincerely,

Prisca Hsu, Emily Ready, Jessica Grahn

Reviewer #1

I believe there is still a problem with the sentence “When assessing the effects of dance training on beat perception, the null model had a BF10 of 1 (Table 2b), suggesting that the data could not discriminate between the alternative hypothesis from the null hypothesis”. Description of table 4 is similar. Again, all null models have a BF10 of 1, so how can that be supporting the null hypothesis in itself? I believe the reason is that all other models have a BF10 comprised between 1/3 and 3.

Thanks for pointing this out. We realize that the description of the BF10 may be confusing to our readers. As noted in L311-317, a BF10 value of 1 falls between 0.33 and 3 and thus indicates inconclusive results that do not clearly support H0 or H1 models. We have rephrased those sentences.

L331-333: When assessing the effects of dance training on beat perception, the null model had a BF10 of 1 (Table 2b), suggesting that the data are inconclusive and do not support either null or alternative models.

L367-369: For tempo matching, the Bayesian ANOVAs for both music and dance revealed a BF10 = 1 (Tables 4a & 4b). A BF10 = 1 value suggests that the data are inconclusive and do not support either null or alternative models.

I suggest presenting the interpretation for the full range of possible BF10 values: “BF10 < 1/10 provides strong support for the null hypothesis, 1/10 < BF10 < 1/3 provides moderate support for the null hypothesis”, and again 3 < BF10 < 10. It would also be useful to explicitly say that the BF10 value is a ratio that is expressed in decimal in the tables. I had difficulties to understand that.

This is great suggestion. We have adjusted the Bayesian statistics description and changed the BF10 values to decimal notation in the following lines.

L311-317: As BF10 deviates from 1, support for the null or alternative hypothesis increases. Generally, 0.33 < BF10 < 3 indicates that the data is insufficient to support either null or alternative hypothesis. 0.1 < BF10 < 0.33 provides moderate support for the null hypothesis and 3 < BF10 < 10 provides moderate support for the alternative hypothesis. Finally, a BF10 < 0.1 provides strong support for the null hypothesis and BF10 > 10 provides strong support for the alternative hypothesis. [29,30]. Strength of the evidence can be quantified based on the Bayes Factor (e.g., BF10 =8 is twice as strong as BF10 = 4 in supporting the alternative hypothesis).

I am not convinced that scientific notation should be used in table 2. Numbers are not that big and scientific notation can be confusing. Three numbers presentations are used for BF10 when presenting the interpretation of the index and the values (ratios, decimals and scientific), which make it very hard to follow.

Thanks for noting this. We have changed the notation of the numbers in table 2. We have also changed the BF10 values to decimal notation in the Bayesian Analyses section.

Submitted filename: Response_To_Reviewers_Rev2.docx

Click here for additional data file.

25 Jan 2022

28 Jan 2022

Dear Dr. Sonja Kotz,

Thank you for handling our manuscript, “The effects of Parkinson’s disease, music training, and dance training on beat perception and production abilities,” for possible publication in Plos One. We thank reviewer 1 for their careful feedback and helpful comments. We now submit a revised version, with further clarifications about our Bayesian analyses. We hope that you might find this version to be suitable for publication for the wide readership of Plos One.

Sincerely,

Prisca Hsu, Emily Ready, Jessica Grahn

Reviewer #1:

Thanks a lot for addressing my comments; I think the presentation of the Bayesian analyses is much improved, which is very important as most readers are not familiar with it.

Nevertheless, I still think there is a major mistake with the sentence ‘When assessing the effects of dance training on beat perception, the null model had a BF10 of 1 (Table 2b), suggesting that the data are inconclusive’. I already pointed that in my previous reviews: all null models have a BF10 of 1, included the one presented in table 2a, where the authors say that data are conclusive. I believe that the null model’s BF is always 1. Unless I am missing something, this sentence is wrong and makes the Bayesian results really confusing. The sentence describing the model in the ‘Beat production: tempo matching’ section is also incorrect. Please modify that.

Thank you for your careful review. You are correct. All BF10 values are relative to the BF10 value of the null model. As such, the BF10 value for the null model will always be 1. We realize that our indication of interpreting the BF10=1 is confusing and have modified those lines to be clearer about how the BF10 values should be interpreted.

L325-327 When assessing the effects of dance training on beat perception, the dance model had a BF10 value of 0.33 (Table 2b), indicating insufficient support for either the null hypothesis or the alternative hypothesis [29,30].

L362-364 For tempo matching, the music model had a BF10 of 0.50 (Table 4a), indicating insufficient support for either the null hypothesis or alternative hypothesis. For dance, BF10 = 0.22 (Table 4b), indicating moderate support for the null hypothesis.

Submitted filename: Response_To_Reviewers_Rev3.docx

Click here for additional data file.

14 Feb 2022

The effects of Parkinson’s disease, music training, and dance training on beat perception and production abilities

PONE-D-21-33421R3

Dear Dr. Hsu,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Sonja Kotz

Academic Editor

PLOS ONE

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thank you for addressing my final comments. It was important to clarify the presentation of the Bayesian analyses. In my view, the article is now suitable for publication and makes an important contribution to the field

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Valentin Bégel

23 Feb 2022

PONE-D-21-33421R3

The effects of Parkinson’s disease, music training, and dance training on beat perception and production abilities

Dear Dr. Hsu:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Sonja Kotz

Academic Editor

PLOS ONE