We recruited 97 participants from Hong Kong, aged 18–33, with a first language of Cantonese and an educational qualification of college or above. They had normal or corrected-to-normal vision, despite three reporting mild visual impairments (e.g., color deficiency) which did not affect their music reading abilities in the study. No participants reported any auditory impairments. The participants were classified into three groups according to their music training background, including musicians (n = 38), music learners (n = 26), and non-musicians (n = 33).

A priori power analysis with F-tests ANOVA: Repeated measures, within-between interaction, was conducted using G^*Power version 3.1.9.6 for Mac OS X 10.7 to 14 (Faul et al., 2007) to examine the required minimum sample size for this study. The result revealed that, at least 60 participants (i.e., 20 participants per group) are needed to achieve 80% power for detecting a small effect at a significance criterion of α = 0.05. Thus, the recruited sample size of the musicians, music learners, and non-musicians fulfilled the suggested sample size indicated by G^*Power.

The musicians (n = 38; 19 males, 19 females; mean age = 21, range = 18–32) were well-trained pianists/western instrumentalists who read music scores regularly. They had attained at least a Grade 8 or above in the Associated Board of the Royal Schools of Music (ABRSM) graded music examinations or equivalent in their primary instruments, including piano (n = 24), violin (n = 8), flute (n = 4), saxophone (n = 1), and clarinet (n = 1). They had started music learning between age 3 and 16 (mean = 5.92, SD = 2.59), and had 7–26 years of experience in music playing (mean = 14.26, SD = 3.74). Some musicians engaged in orchestral playing (n = 13), wind-band playing (n = 1), and choral conducting (n = 2).

The music learners (n = 26; 10 males, 16 females; mean age = 20.35, range = 18–33) were piano/western instrumental learners who read music scores regularly. They had attained Grades 2–5 in the Associated Board of the Royal Schools of Music (ABRSM) graded music examinations or equivalent in their primary instruments, including piano (n = 14), violin (n = 4), flute (n = 2), saxophone (n = 1), clarinet (n = 1), trumpet (n = 2), horn (n = 1), and guitar (n = 1). They had started music learning between age 3 and 16 (mean = 7.92, SD = 3.31), and had 1–18 years of experience in music playing (mean = 8.27, SD = 4.53). Some music learners engaged in orchestral playing (n = 5), wind-band playing (n = 1), and choir singing (n = 1).

The non-musicians (n = 33; 17 males, 16 females; mean age = 20.70, range = 18–27) were counterparts who had not received any formal music training in western instruments and did not read music scores, but matched in terms of demographics, handedness and working memory with the musicians and music learners.

Despite music training backgrounds, differential musical abilities among musicians, music learners, and non-musicians had been reflected in other measurements in this study. A set of music expertise tasks, comprising a visual memory and a visual-auditory music task, was included in this study. The visual memory task examined participants' ability to judge whether a bar has been shown in the previously shown four-bar, diatonic musical phrases (i.e., phrases made up of notes of a prevailing key) or non-diatonic musical phrases (i.e., phrases made up of chromatic notes that are out of a prevailing key), while the visual-auditory music task examined participants' ability to judge whether a visually shown musical phrase matched with an auditorily played musical phrase. The analysis of the music expertise task was based on the average accuracy (ACC) and response time (RT) of the visual memory task and visual-auditory music task. Musicians, music learners, and non-musicians were significantly different in ACC, with a large effect size [F_{(2, 94)} = 12.44, p < 0.001, ηp² = 0.21], while no significant differences were observed in the RT [F_{(2, 92)} = 1.10, p = 0.33]. In ACC, post-hoc comparisons with Tukey HSD correction showed that musicians (mean = 67.68; SD = 10.43) significantly outperformed music learners (mean = 58.05; SD = 9.06) and non-musicians (mean = 58.43; SD = 7.22), while no significant differences were observed between music learners and non-musicians. Table 1 shows the participants' performance of the visual memory task and visual-auditory music tasks respectively.

Similar differential musicality ratings among musicians, music learners, and non-musicians were also observed in the Goldsmiths Musical Sophistication Index (Gold-MSI; Müllensiefen et al., 2014). Gold-MSI is a self-reported questionnaire measuring musicality (i.e., multifaceted and broad musical skills and behaviors) among the general population with diverse musical background and expertise. Gold-MSI consisted of 38 items in five subscales, including (i) active engagement, representing active musical engagement behaviors, and the use of time and money on musical activities; (ii) perceptual abilities, representing cognitive musical abilities, and music listening skills; (iii) musical training, representing musical training and practice, and the degree of musicianship; (iv) singing abilities, representing singing skills and activities; and (v) emotions, representing emotional responses to music. The five factors contributed to a general musical sophistication factor, demonstrating the musicality among the general population (Müllensiefen et al., 2014). The participants indicated whether each of the statement described their music-related behaviors in a 7-point Likert scale, ranging from 1 (completely disagree) to 7 (completely agree). Using Gold-MSI in the current study allowed a more precise measurement of individual differences in musical skills and behaviors among musicians, music learners, and non-musicians.

As shown in the Gold-MSI (Müllensiefen et al., 2014), musicians, music learners, and non-musicians were significantly different in the sum scores of general musical sophistication, with a large effect size [F_{(2, 94)} = 65.57, p < 0.001, ηp² = 0.58]. Post-hoc comparisons with Tukey HSD correction showed that musicians (mean = 87.21; SD = 9.41) significantly outperformed music learners (mean = 79.15; SD = 14.07) and non-musicians (mean = 55.67; SD = 12.45) in general musical sophistication, while music learners also significantly outperformed non-musicians. In addition, the three groups also significantly differed in the sum scores of other five subscales, with medium to large effect sizes, including active engagement [F_{(2, 94)} = 9.96, p < 0.001, ηp² = 0.18], perceptual abilities [F_{(2, 94)} = 24.32, p < 0.001, ηp² = 0.34], musical training [F_{(2, 94)} = 364.01, p < 0.001, ηp² = 0.89], singing abilities [F_{(2, 94)} = 16.31, p < 0.001, ηp² = 0.26], and emotions [F_{(2, 94)} = 4.93, p < 0.001, ηp² = 0.10]. Table 2 shows the post-hoc comparisons with Tukey HSD correction, in which suggesting musicians outperformed non-musicians in all subscales, while music learners outperformed non-musicians in active engagement, perceptual abilities, musical training, and singing abilities. Musicians also outperformed music learners in perceptual abilities and musical training.

A similar finding was observed among the three groups in a self-report familiarity rating of a musical note (i.e., a D5 crotchet). The participants indicated their familiarity toward the note based on a 10-point Likert scale (i.e., 1 represents having no knowledge in that musical note at all, and 10 represents being extremely familiar to that musical note). A significant difference was found in the familiarity rating of the musical note among musicians, music learners, and non-musicians, with a large effect size [F_{(2, 94)} = 163.65, p < 0.001, ηp² = 0.78]. Post-hoc comparisons with Tukey HSD correction showed that musicians (mean = 9.66; SD = 0.75) had a significantly higher familiarity rating of the musical note than music learners (mean = 7.42; SD = 2.52) and non-musicians (mean = 2.24; SD = 1.84), while music learners had a significantly higher familiarity rating of the musical note than non-musicians.

To further explore the differential music expertise between musicians and music learners, the two groups completed self-report questions based on a 10-point Likert scale (i.e., 1 represents do not read music at all, and 10 represents being extremely confident in music reading and relevant knowledge). Musicians significantly outperformed music learners in sight-reading, with a large effect size [musicians: mean = 7.26, SD = 1.67, music learners: mean = 5.58, SD = 1.79; t₍₆₀₎ = 3.75, p < 0.001, d = 0.97] and knowledge of music theory, with a medium effect size [musicians: mean = 6.16, SD = 1.67, music learners: mean = 5.08, SD = 1.74; t₍₆₀₎ = 2.43, p = 0.018, d = 0.63]. In addition, musicians (mean = 5.58, SD = 7.72) spent more time in music reading per week than music learners, with a medium effect size [mean = 2.89, SD = 1.95; t_(44.15) = 2.07, p = 0.044, d = 0.49].

The participants of all three groups were matched in handedness and working memory. The Edinburgh Handedness Inventory (Oldfield, 1971; Cohen, 2008) was used to assess participants' handedness based on the 10-item (Oldfield, 1971) and 15-item versions (Cohen, 2008). No significant differences were observed among the three groups in the 10-item laterality index [musicians: mean = 74.87, SD = 35.46, music learners: mean = 66.73, SD = 46.78, non-musicians: mean = 66.06, SD = 44.93; F_{(2, 94)} = 0.48, p = 0.62] and 15-item augmented laterality index [musicians: mean = 73.60, SD = 34.20, music learners: mean = 65.51, SD = 44.79, non-musicians: mean = 64.95, SD = 42.98; F_{(2, 94)} = 0.50, p = 0.61]. Most participants were right-handed (musicians: n = 36, music learners: n = 22, non-musicians: n = 26), while a few of them were left-handed (musicians: n = 2, music learners: n = 2, non-musicians: n = 3) or with mixed-handedness (musicians: n = 0, music learners: n = 2, non-musicians: n = 4). Using participants with matched handedness was essential to rule out the possibility that the potential differences observed in hemispheric lateralization among musicians, music learners, and non-musicians were due to the actual difference in hemispheric lateralization per se, instead of an effect observed based on their music training background.

In addition, a set of n-back tasks comprised verbal and spatial two-back tasks was used to assess participants' verbal and spatial working memory. In the verbal two-back task, participants judged whether an English letter was the same as the one previously presented in a two-letter interval. While in the spatial two-back task, participants judged whether a symbol's location was the same as the location that was previously presented in a two-stimulus interval. Similarly, no significant differences were observed in the ACC [F_{(2, 94)} = 2.55, p = 0.08] and correct RT [F_{(2, 93)} =1.07, p = 0.35] in the set of n-back tasks, comprising a verbal and a spatial two-back task, among the three groups. Table 3 shows the participants' performance of the verbal and spatial two-back tasks respectively. Using participants with matched working memory could minimize the possibility that the potential differences observed in task performance among musicians, music learners, and non-musicians were due to their differences in working memory, instead of their music training background.

The experimental design of this study comprise a set of divided visual field sequential matching tasks with a 4 × 2 × 3 mixed design. It included two within-subject variables [musical elements: pitch, harmony, rhythm, and articulation; visual fields (VFs): LVF vs. RVF] and one between-subject variable (groups: musicians vs. music learners vs. non-musicians).

The divided visual field sequential matching task examines hemispheric lateralization as a behavioral technique, with its theoretical basis based on Bourne (2006), in which suggested the contralateral relationships between the visual fields and hemispheres. For example, the stimulus shown in the right visual field projects more directly to the left hemisphere (RVF/LH), while the stimulus shown in the left visual field projects more directly to the right hemisphere (LVF/RH).

Additionally, the divided visual field sequential matching task paradigm was designed based on the recommendations in Hunter and Brysbaert (2008) for naming tasks. For instance, the current study consists of sufficient trials, as shown in the blocks with pitch (224 trials), harmony (224 trials), rhythm (216 trials), and articulation (208 trials). Each stimulus was shown with an equal number of occurrences in the LVF and RVF. No stimulus degradation was applied. Participants' central fixation was controlled and monitored using an eye-tracker. A bilateral presentation of stimuli for 200 ms was used to assess hemispheric lateralization. The design was based on the recommendation stated in Hunter and Brysbaert (2008) to limit voluntary saccade that has 340 ms average onset latency (Walker and McSorley, 2006, as cited in Hunter and Brysbaert, 2008). Presenting the stimuli for 200 ms could ensure participants' central fixation during bilateral stimuli presentation, and thus result in an accurate measurement of hemispheric lateralization using the divided visual field sequential matching task. Last but not least, masks were also applied after stimulus offset.

The computerized divided visual field sequential matching tasks were conducted using Eprime 3.0 (Psychology Software Tools, Pittsburgh, PA, USA) for stimulus presentation with a 24″ screen with a resolution of 1,920 × 1,080. A Tobii Pro Spectrum eye-tracker (Tobii, Danderyd, Sweden) is a high-frequency screen-based eye-tracker that used to monitor participants' central fixation with binocular eye-tracking with a sampling frequency at 600 Hz. A green rectangle was shown if the participant has fixated at the central fixation for 250 ms as detected by the eye-tracker, prior to the 200 ms bilateral presentation of musical elements in the LVF and RVF. The 250 ms fixation detection threshold was set with respect to the average fixation duration of normal reading (Rayner, 1993). A chin and forehead rest (i.e., SR Research Head Support; SR-Research Ltd., Mississauga, Ontario, Canada) was used to minimize participants' head movement during the tasks at a viewing distance of 60 cm. A nine-point binocular calibration was performed before the start of each block. Recalibration was conducted if the initial calibration was unsatisfactory or unsuccessful. All blocks were counterbalanced with randomized trials. An equal number of same and different trials was arranged. For different trials, the only difference between the first and second target stimuli was either in pitch, harmony, rhythm, and or articulation in each separate block respectively, while other musical elements were all controlled.

Participants indicated their responses on a Chronos response box (Psychology Software Tools, Pittsburgh, PA, USA) for all tasks using both index and middle fingers. Fingerings were counterbalanced among participants. This design was to avoid the possible lateralization effects induced by single-hand responses (Mohr et al., 1994).

This study sought the Ethical Review for Research involving Human Participants as Research Subjects from the Research Ethics Committee (REC), Hong Kong Metropolitan University, and was conducted in accordance to the ethical guidelines stated in the World Medical Association's Declaration of Helsinki.

Each experimental session was conducted in the Experimental room of the Psychology and Behavioral Sciences Laboratory at the Hong Kong Metropolitan University, Hong Kong, China. Participants were recruited through mass email and poster promotions within the University and referrals. A few preliminary screening questions regarding music training background were applied when participants signed up for the study.

On the day of experiment, the experimenter checked once again with participants regarding their music training backgrounds upon arrival. Participants who fulfilled the inclusion criteria of musicians, music learners, and non-musicians would then proceed to the study. A short briefing regarding the experimental procedures was given to each participant. A written consent was obtained before the start of each experimental session. Each experimental session lasted for about 1.5 h.

Each experimental session comprised a set of computerized divided visual field sequential matching tasks. Each block, namely pitch, harmony, rhythm, and articulation, consisted of four sections respectively. In addition, three questionnaires [i.e., a demographic questionnaire, Goldsmiths Musical Sophistication Index (Gold-MSI; Müllensiefen et al., 2014), and Edinburgh Handedness Inventory (Oldfield, 1971; Cohen, 2008)] and two computerized preliminary tasks (i.e., n-back tasks and music expertise tasks) were included.

The divided visual field sequential matching tasks were conducted on Eprime 3.0 (Psychology Software Tools, Pittsburgh, PA, USA). For each trial, participants first fixated at the center of the screen (+). A Tobii Pro Spectrum eye-tracker (Tobii, Danderyd, Sweden) was used to monitor participants' central fixation. For each trial, participants were instructed to fixate at the center of the screen. When their eye gazes stayed at the central fixation for 250 ms, a green rectangle surrounding the central fixation emerged. Then, two stimuli (pitch/harmony/rhythm/articulation in each respective block) were presented briefly and simultaneously for 200 ms on the left visual field (LVF) and right visual field (RVF). A central arrow either pointing to the left (←) or right ( → ) was also simultaneously presented for 200 ms, indicating the first target stimulus. Masks were then presented for 200 ms to avoid afterimages being formed. Then, one stimulus (pitch/harmony/rhythm/articulation in each respective block) was presented at the center of the screen as the second target stimulus and waited for participants' responses (Figure 4). Participants judged whether the first target stimulus was the same as the second target stimulus by pressing the Chronos response box (Psychology Software Tools; Pittsburgh, PA, USA) as quickly and accurately as possible. ACC and RT were recorded.

Besides, the demographic questionnaire and Goldsmiths Musical Sophistication Index (Gold-MSI; Müllensiefen et al., 2014) were conducted on Qualtrics (Qualtrics, Provo, UT, USA). The Edinburgh Handedness Inventory (Oldfield, 1971; Cohen, 2008) was conducted on Cohen (2008) webpage, https://www.brainmapping.org/shared/Edinburgh.php#. The sets of n-back tasks and music expertise tasks were carried out on Eprime 3.0 (Psychology Software Tools, Pittsburgh, PA, USA). A few practice trials were provided for all computerized tasks before its respective experimental tasks. Short breaks were provided between sections within each block.

A debriefing was delivered to the participants upon the completion of all tasks. A small honorarium was given to the participants to thank them for their support.

In the ACC data, a significant two-way interaction between musical elements and groups was observed, F_{(6, 186)} = 4.57, p < 0.001, ηp² = 0.13, suggesting that musicians, music learners, and non-musicians performed differently in the four musical elements, with a medium effect size. To further understand this two-way interaction, we examined the data on different musical elements (Figure 5A). In terms of pitch, musicians (mean = 81.42; SD = 12.14) significantly outperformed music learners (mean = 71.88; SD = 13.22) and non-musicians (mean = 62.01; SD = 9.16). Music learners also significantly outperformed non-musicians. Regarding harmony, musicians (mean = 82.52; SD = 12.84) significantly outperformed non-musicians (mean = 73.27; SD = 13.09), but not music learners (mean = 78.38; SD = 13.81). For rhythm, no significant differences were observed among musicians, music learners, and non-musicians. In terms of articulation, musicians (mean = 87.40; SD = 13.18) significantly outperformed non-musicians (mean = 77.06; SD = 15.84), but not music learners (mean = 83.48; SD = 17.92). In addition, we also examined the data based on musicians, music learners, and non-musicians. For musicians, a significant main effect of musical elements was observed, F_{(3, 35)} = 43.38, p < 0.001, ηp² = 0.79, suggesting that pitch (mean = 81.43; SD = 12.13) and harmony (mean = 82.52; SD = 12.84) had similar ACC, as supported by a large effect size. However, rhythm (mean = 72.32; SD = 13.43) had a lower ACC than pitch, harmony, and articulation (mean = 87.38; SD = 13.19; Figure 5B). In contrast, articulation had a higher ACC than pitch, harmony and articulation. For music learners, a significant main effect of musical elements, with a large effect size, was observed, F_{(3, 23)} = 9.69, p < 0.001, ηp² = 0.56, suggesting that pitch (mean = 71.88; SD = 13.22) had a lower ACC than harmony (mean = 78.38; SD = 13.81), and articulation (mean = 83.49; SD = 17.92), but not rhythm (mean = 68.53; SD = 14.90). Harmony had a higher ACC than rhythm, while rhythm had a lower ACC than articulation (Figure 5C). For non-musicians, a significant main effect of musical elements with a large effect size was observed, F_{(3, 30)} = 25.52, p < 0.001, ηp² = 0.72, suggesting that pitch (mean = 62.01; SD = 9.16) had a lower ACC than harmony (mean = 73.27; SD = 13.09) and articulation (mean = 77.06; SD = 15.84), but not rhythm (mean = 65.13; SD = 13.09). Harmony had a higher ACC than rhythm, while rhythm had a lower ACC than articulation (Figure 5D).

In addition, a significant two-way interaction between musical elements and VFs was observed, F_{(3, 92)} = 4.54, p = 0.005, ηp² = 0.13, suggesting that different musical elements had different VF processing advantages, with a medium effect size. To further understand this two-way interaction, we examined the data on different musical elements (Figure 6). No significant main effects of VF were observed for pitch [F_{(1, 94)} = 0.48, p = 0.49], harmony [F_{(1, 94)} = 2.65, p = 0.11], rhythm [F_{(1, 94)} = 1.24, p = 0.27], and articulation [F_{(1, 94)} = 1.57, p = 0.21], suggesting no particular processing advantages were found in either the LVF or RVF for each musical element. In addition, we examined the data from separate VFs for the four musical elements. A significant main effect of musical elements was observed in the LVF, F_{(3, 92)} = 39.09, p < 0.001, ηp² = 0.56, suggesting that the four musical elements showed different ACC values in the LVF, with a large effect size. The post-hoc comparisons showed that pitch (mean = 71.34; SE = 1.24) had a significantly lower ACC than harmony (mean = 79.33; SE = 1.35) and articulation (mean = 81.68; SE = 1.51), but not rhythm (mean = 69.33; SE = 1.61). Harmony had a higher ACC than rhythm, and rhythm had a lower ACC than articulation. In addition, a significant main effect of musical elements was also observed in the RVF, F_{(3, 92)} = 41.69, p < 0.001, ηp² = 0.58, suggesting that the four musical elements showed different ACC values in the RVF, with a large effect size. The post-hoc comparisons showed that rhythm (mean = 67.77; SE = 1.64) had a significantly lower ACC than pitch (mean = 72.20; SE = 1.43), harmony (mean = 76.78; SE = 1.75), and articulation (mean = 83.61; SE = 2.00). In addition, articulatory processing had a significantly higher ACC than pitch, harmony, and rhythm. Harmony also had a higher ACC than pitch.

In the ACC data, a significant main effect of musical elements was observed [F_{(3, 92)} = 60.11, p < 0.001, ηp² = 0.66], suggesting the four musical elements had different ACC values, with a large effect size. The post-hoc comparisons showed that rhythm (mean = 68.66; SE = 1.41) had a significantly lower ACC than pitch (mean = 71.77; SE = 1.19), harmony (mean = 78.06; SE = 1.36), and articulation (mean = 82.65; SE = 1.59; Figure 7). In addition, articulation has also showed a significantly higher ACC than pitch, harmony, and rhythm. Harmony also had a higher ACC than pitch. No other significant comparisons were observed.

A significant main effect of groups was also observed [F_{(2, 94)} = 8.69, p < 0.001, ηp² = 0.16], suggesting that musicians, music learners, and non-musicians had different ACC values, with a large effect size. The post-hoc comparisons showed that musicians (mean = 80.91; SD = 11.77) and music learners (mean = 75.57; SD = 12.19) had a significantly higher ACC than non-musicians (mean = 69.37; SD = 10.99). No significant difference was observed between musicians and music learners.

However, no significant three-way interactions between musical elements, VFs, and groups were observed [F_{(6, 186)} = 0.28, p = 0.95], suggesting that musicians, music learners, and non-musicians did not show any different VFs effects among the four musical elements (pitch, harmony, rhythm, and articulation), in the ACC data.

In the correct RT data, a significant two-way interaction between VFs and groups was observed, F_{(2, 94)} = 5.94, p = 0.004, ηp² = 0.11, suggesting that there were different VF effects for musicians, music learners, and non-musicians during music reading, with a medium effect size. To further understand this two-way interaction, we first examined the data in separate groups. For musicians, no main effect of VF was observed [F_{(1, 37)} = 0.01, p = 0.94]. However, for music learners, a significant main effect of VF was observed, F_{(1, 25)} = 10.72, p = 0.003, ηp² = 0.30, showing that music learners had a faster RT when musical elements were presented in the RVF (824.79 ms) than in the LVF (860.86 ms), with a large effect size. In contrast, for non-musicians, a significant main effect of VF was found, F_{(1, 32)} = 8.49, p = 0.006, ηp² = 0.21, suggesting that non-musicians had a faster RT when musical elements were presented in the LVF (696.89 ms) than in the RVF (726.85 ms), with a large effect size (Figure 8). In addition, we examined the data from separate VFs among the three groups. In the LVF, a significant main effect of group was observed, with a large effect size, F_{(2, 94)} = 4.77, p = 0.011, ηp² = 0.92. The post-hoc comparisons with Tukey HSD correction showed that non-musicians (mean = 696.89; SD = 203.59) had a faster correct RT in the LVF than music learners (mean = 860.86; SD = 252.93). In the RVF, no significant main effect of group was observed, F_{(2, 94)} = 1.62, p = 0.20, suggesting that the three groups did not differ in music reading in the RVF.

In addition, a significant two-way interaction between musical elements and VFs was observed, F_{(3, 92)} = 5.27, p = 0.001, ηp² = 0.15, suggesting that various musical elements had different VF processing advantages, with a large effect size. To further understand this two-way interaction, we examined the data on different musical elements (Figure 9). No significant main effects of VF were observed for pitch [F_{(1, 94)} = 0.53, p = 0.47], harmony [F_{(1, 94)} = 2.38, p = 0.13], and rhythm [F_{(1, 94)} = 0.24, p = 0.63]. However, articulation showed a significant main effect of VFs, with a medium effect size [F_{(1, 94)} = 6.14, p = 0.015, ηp² = 0.06]. These findings suggested that no particular processing advantages were found in either the LVF or RVF for pitch, harmony, and rhythm, while articulation showed a RVF (mean = 737.17; SD = 239.50) advantage over the LVF (mean = 754.20; SD = 236.81) in processing. In addition, we examined the data from separate VFs for the four musical elements. A significant main effect of musical elements was observed in the LVF, F_{(3, 92)} = 4.18, p = 0.008, ηp² = 0.12, suggesting that the four musical elements resulted in different correct RTs in the LVF, with a medium effect size. The post-hoc comparisons showed that rhythm (mean = 836.51; SE = 28.94) had a significantly longer RT than pitch (mean = 777.15; SE = 25.50), harmony (mean = 751.43; SE = 21.36), and articulation (mean = 757.48; SE = 23.99). No other significant results were observed in the post-hoc comparisons. In addition, a significant main effect of musical elements was also observed in the RVF, F_{(3, 92)} = 4.57, p = 0.005, ηp² = 0.13, suggesting that the four musical elements had different RTs in the RVF, with a medium effect size. The post-hoc comparisons showed that rhythm (mean = 829.80; SE = 30.26) had a significantly longer RT than harmony (mean = 769.59; SE = 23.28), and articulation (mean = 736.92; SE = 24.76), but not pitch (mean = 784.22; SE = 25.64). In addition, articulation also showed a significantly shorter RT than pitch and rhythm, but not harmony.

In the correct RT data, a significant main effect of musical elements was observed, F_{(3, 92)} = 4.50, p = 0.005, ηp² = 0.13, suggesting that the four musical elements had different RTs, with a medium effect size (Figure 10). The post-hoc comparisons showed that rhythm (mean = 833.16; SE = 28.81) had a significantly longer RT than pitch (mean = 780.69; SE = 25.11), harmony (mean = 758.27; SE = 21.89), and articulation (mean = 747.20; SE = 24.03).

A marginally significant main effect of groups was also observed [F_{(2, 94)} = 3.05, p = 0.052, ηp² = 0.06], suggesting that musicians, music learners, and non-musicians had different RTs, with a medium effect size. The post-hoc comparisons showed that music learners (mean = 842.82; SE = 40.10) had a significantly longer RT than non-musicians (mean = 711.87; SE = 35.60). No significant difference was observed between musicians and music learners.

However, no significant three-way interactions between musical elements, VFs, and groups were observed [F_{(6, 186)} = 0.97, p = 0.45], suggesting that musicians, music learners, and non-musicians did not show any different VF effects among the four musical elements (pitch, harmony, rhythm, and articulation) in the correct RT data.

The hemispheric lateralization effects among musicians, music learners, and non-musicians in music reading were further examined using the laterality index (LI). The laterality index ranges from −1 (indicating a right-lateralized processing advantage) to +1 (indicating a left-lateralized processing advantage; Seghier, 2008; Seghier et al., 2011). A repeated measures ANOVA was conducted with musical elements and groups as within-subject variables, while groups was the between-subject variables.

A laterality index analyses were conducted on the ACC data. In order to prevent possible misinterpretation of the laterality index based on the ACC data, a negative sign was inserted in the following formular to allow the same interpretation of the laterality index as in the RT data (i.e., −1 (indicating a right-lateralized processing advantage) to +1 (indicating a left-lateralized processing advantage; Seghier, 2008; Seghier et al., 2011).

In the laterality index based on the ACC data, no significant interaction and main effect of groups were found. A significant main effect of musical elements was found, F_{(3, 92)} = 2.96, p = 0.04, ηp² = 0.09, suggesting that the four musical elements had different hemispheric lateralization with a medium effect size. The post-hoc comparisons showed that pitch (mean = 0.003; SE = 0.01) was more left lateralized than harmony (mean = −0.02; SE = 0.02), while articulation (mean = 0.002; SE = 0.01) was more left lateralized than harmony. No other significant comparisons were observed.

In the laterality index based on the correct RT data, no significant interaction was found. A significant main effect of groups was found, F_{(2, 92)} = 4.45, p = 0.01, ηp² = 0.09, suggesting that musicians, music learners, and non-musicians differed in hemispheric lateralization of music reading, with a medium effect size. The post-hoc comparisons showed that music learners (mean = 0.019; SD = 0.03) had a more left-lateralized processing advantage than non-musicians (mean = −0.017; SD = 0.04), while non-musicians were more right lateralized than music learners in music reading. No significant comparisons were observed between musicians (mean = 0.003; SD = 0.57) and music learners or non-musicians in the laterality index.

In addition, a significant main effect of musical elements was found, F_{(3, 92)} = 4.87, p = 0.003, ηp² = 0.14, suggesting that the four musical elements had different hemispheric lateralization, with a large effect size. The post-hoc comparisons showed that articulation (mean = 0.01; SE = 0.006) was significantly more left-lateralized than pitch (mean = −0.005; SE = 0.006), harmony (mean = −0.007; SE = 0.006), and rhythm (mean = 0.007; SD = 0.009). No other significant comparisons were observed.

To further examine whether our findings of hemispheric lateralization observed in the correct RT data among the three groups fit into the predefined thresholds (LI_TH) indicating left/bilateral/right dominance, a follow-up analysis was conducted. As shown in Seghier (2008), left hemispheric dominance is indicated by LI > LI_TH, while right hemispheric dominance is indicated by LI < –LI_TH. Bilateral dominance is indicated by |LI| ≤ LI_TH. The threshold was set as 0.15 (see Baciu et al., 2005).

A one-sample t-test was used to examine whether the laterality index in the correct RT data among musicians, music learners, and non-musicians was significantly different from 0.15.

In the correct RT data, musicians' laterality index was significantly smaller than the predefined thresholds (0.15) (mean = 0.003, SD = 0.06), t₍₃₇₎ = −16.00, p < 0.001, suggesting musicians' bilateral dominance in music reading. Nonetheless, music learners' laterality index was significantly smaller than the predefined thresholds (0.15) (mean = 0.019, SD = 0.03), t₍₂₅₎ = −22.05, p < 0.001, suggesting music learners' bilateral dominance in music reading. Similarly, non-musicians' laterality index was significantly smaller than the predefined thresholds (0.15) (mean = |−0.017|, SD = 0.42), t₍₃₂₎ = −22.94, p < 0.001, suggesting non-musicians' bilateral dominance in music reading when the laterality indices were compared with the predefined thresholds.

To further explore how other measures correlated with the laterality index measures in the correct RT data, correlational analyses were conducted as follow.

A significant positive correlation was found between the ACC of music expertise tasks and the laterality index measured in the correct RT data [r₍₉₇₎ = 0.21, p = 0.04; Figure 11A left]. The higher the ACC in the music expertise tasks, the higher the observed laterality index value. This suggests that participants with a higher ACC in music expertise tasks tended to be more left lateralized in music reading. No other significant correlations were observed among ACC, or correct RT of music expertise tasks, and the laterality index in the correct RT and ACC data.

In addition, a significant positive correlation was found between the total scores of music training as measured in Gold-MSI (Müllensiefen et al., 2014) and the laterality index measured in the correct RT data [r₍₉₇₎ = 0.27; p = 0.009; Figure 11B right]. No similar correlations were found between other subscales, including active engagement, perceptual abilities, singing abilities, emotions, and general musical sophistication.

To further explore how musicians' and music learners' music training background related to the ACC and the correct RT data, correlational analyses were conducted as follow.

Musicians' and music learners' onset of music learning was significantly correlated with overall ACC [r₍₆₂₎ = −0.27, p = 0.037], overall correct RT [r₍₆₂₎ = 0.35, p = 0.005], ACC in the LVF [r₍₆₂₎ = −0.32, p = 0.011], correct RT in the LVF [r₍₆₂₎ = 0.32, p = 0.011], and correct RT in the RVF [r₍₃₈₎ = 0.36, p = 0.004]. These findings suggest that the earlier the onset of music learning, the higher the overall ACC, and the ACC in the LVF as shown in the divided visual field sequential matching task. Also, the earlier the onset of music learning, the faster the correct RT, and the correct RT in the LVF and RVF. No significant correlations were found between musicians' and music learners' onset of music learning and the ACC in the RVF and the laterality index in the ACC and RT data.

Similar correlational analyses between musicians' and music learners' years of experience in music playing and the ACC and correct RT data were conducted. Musicians' and music learners' years of experience in music playing was significantly correlated with overall ACC [r₍₆₄₎ = 0.30, p = 0.015] and ACC in the RVF [r₍₆₄₎ = 0.29, p = 0.021]. These findings suggested that the longer the years of experience in music playing, the higher the overall ACC and ACC in the RVF in the divided visual field sequential matching task. No significant correlations were found between musicians' and music learners' years of experience in music playing and overall correct RT, ACC in the LVF, correct RT in the LVF and RVF, and laterality index values in the ACC and RT data.

In this study, musicians, music learners, and non-musicians had shown different VF advantages in music reading. From the correct RT data of the divided visual field matching task, musicians had shown a bilateral representation in music reading based on the similar performance in the LVF and the RVF in the task. In addition, music learners tended to be more left-lateralized in music reading based their RVF/LH processing advantage over the LVF/RH in the task. In contrast, non-musicians tended to be more right-lateralized in music reading based on their LVF/RH processing advantage over RVF/LH in the task.

Some other findings were shown in the analysis of the laterality index (LI) comparing with the predefined threshold (LI_TH; Baciu et al., 2005; Seghier, 2008). This classic measurement of hemispheric dominance showed that musicians had a bilateral dominance in music reading (indicated by |LI| ≤ LI_TH). In addition, although music learners and non-musicians had also shown a bilateral dominance in music reading in music reading. When the laterality indices were compared with the predefined thresholds, some differential tendencies might still be able to observe based on the raw laterality indices across the three groups.

Additionally, music expertise was shown to be positively correlated with the laterality index (LI) in music reading. The finding suggested that the higher the music expertise, the greater the tendency to be more left-lateralized in music reading.

Our finding of music expertise differentially modulates the hemispheric lateralization of music reading was reflected in previous studies. Musicians' bilateral processing advantage in music reading has been shown in neuroimaging studies. For example, an early study suggested a bilateral activation of the extrastriate visual areas during music reading (Sergent et al., 1992). Additionally, in an fMRI study, Mongelli et al. (2017) has shown that musicians had bilateral left-predominant cortical activations in the lateral and mesial prefrontal and rolandic cortex, IPS, temporal and occipital lobe, and right-predominant cerebellum during music reading. It is important to note that the music expertise-related network was found to be predominated in the left hemisphere (Mongelli et al., 2017). The left-lateralized perisylvian regions, left intraparietal cortex, and bilateral ventral temporal cortex were regions specified for music expertise. These findings provided insights to explain musicians' bilateral processing advantage in music reading. In addition, musicians had active representations of musical elements that are readily engaged in music reading (Sloboda, 1974). This idea is further supported by Patston et al. (2007), in which both groups performed similarly when the dots were presented to the left of the line, musicians outperformed non-musicians when the dots were presented to the right of the line. This finding has further suggested that musicians had a more bilateral attention capacity as compared with non-musicians in visual processing. Thus, musicians could process music more flexibly in both hemispheres.

Another speculation to explain musicians' bilateral processing advantage in music reading could be related to notation-evoked sound imagery (i.e., notational audiation, in which professional music players are able to image the sound of music from musical notations; Wolf et al., 2018). The sound imagery may potentially involve various cognitive processes with different hemispheric lateralization effects, such as music auditory processing, that tends to be more right-lateralized (Tervaniemi and Hugdahl, 2003), and auditory imagery, that tends to be more left-lateralized (Prete et al., 2016). On the other hand, individual differences among musicians may further complicate the investigation. Musicians who were absolute pitch possessors have shown stronger activations in the right-sided perisylvian network when compared to musicians who were relative pitch possessors (Burkhard et al., 2019). It remains unclear how these cognitive processes interact with one another during music reading among musicians. More studies are needed to figure out the underlying cognitive processes and neural correlates of notational audiation, and whether and how it relates to musicians' hemispheric lateralization in music reading.

Concerning the findings of music learners' left-lateralized processing advantage in music reading, this finding was supported by an fMRI study conducted by Muayqil et al. (2015). Participants with some basic music literacy has shown different activations in the left inferior frontal gyrus between musical notation and its non-symbolic equivalent. In addition, among the few studies examining music learners' hemispheric lateralization in music reading, Stewart et al. (2003) has shown that participants who learned piano and score reading for 15 weeks showed a bilateral activation in the superior parietal cortex in a sight-reading task. Some specific training effects were also observed in the left supramarginal gyrus, left inferior frontal sulcus, and right frontal pole. Both studies further highlighted the possibility that music learners had a left-lateralized processing tendency in music reading.

Additionally, music learners' left-lateralized processing advantage in music reading may be explained further by the use of analytical processing strategy in the divided visual field sequential matching task. Music learners may focus more on details on compare between the two target stimuli given their basic music literacy. The analytical processing was shown to be left-lateralized (Bever and Chiarello, 1974), which might further support our findings.

Our findings of non-musicians' right-lateralized processing advantage in music reading are also consistent with previous findings in neuroimaging studies (e.g., Proverbio et al., 2013; Pantaleo et al., 2024). As novices, musical notations did not carry specific meanings to non-musicians, and thus they might employ different processing strategies to finish the divided visual field sequential matching task. One possibility is that non-musicians may employ different visuo-spatial processing strategies (e.g., comparing the location of notes in pitch/harmony matching; differentiating visual features when matching articulations), which tended to be right-lateralized (Bever and Chiarello, 1974; Patston et al., 2006).

To further explore how different musical elements were processed among different music expertise, this study examined whether musicians, music learners, and non-musicians differ in hemispheric lateralization in pitch elements (e.g., pitch, harmony), temporal elements (e.g., rhythm), and expressive elements (e.g., articulation), respectively. No significant interactions were observed among visual fields, musical elements, and groups, indicating that musicians, music learners, and non-musicians did not show different visual field effects in any individual musical elements respectively.

This finding is also consistent with other previous studies showing that no hemispheric lateralization effects were observed in the musical features examined in the auditory domain, including pitch, chord, rhythm, and timbre, among musicians and non-musicians (Ono et al., 2011). Our study may also be among the first few to show that no significant hemispheric lateralization differences were shown in individual musical elements among participants with different music expertise. A few speculations may explain this finding. First, it may suggest that the processing of pitch, harmony, rhythm, and articulation involve resources from the two hemispheres, while musicians, music learners, and non-musicians may not differ much in the cognitive processes when only one particular musical element is examined. Or alternatively, it may suggest that the processing of pitch, harmony, rhythm, and articulation might involve different cognitive processes among musicians, music learners, and non-musicians when only one particular musical element was examined, but then these cognitive processes might share similar lateralization effects as observed.

The qualitative difference between music reading in general and music reading focusing on a musical element remains uncertain. To speculate, it may be that further suggest hemispheric lateralization is sensitive to music reading in general, but not to music reading focusing on a particular musical element. More research is needed to investigate this matter.

Our findings showed that pitch, harmony, and rhythm tended to be bilateral in processing, while articulation remained either bilateral or left-lateralized in processing as shown in different analyses.

The bilateral processing advantage of pitch tends to be similar to previous studies showing no hemispheric lateralization effect in behavioral studies (e.g., Li and Hsiao, 2018) or bilateral processing in neuroimaging studies (Lu et al., 2019; Proverbio et al., 2024). This may potentially reflect the right-lateralized global processing and left-lateralized local processing (Fink et al., 1997) involved in pitch reading. Music readers needed to attend to the five-line staff as global information, as well as the specific note locations across the five lines and four spaces as local information.

A similar bilateral processing advantage was also observed in harmony as shown in behavioral studies (e.g., Li and Hsiao, 2018). This again may reflect the right-lateralized global processing of staff and left-lateralized local processing (Fink et al., 1997) of specific note locations involved in harmony reading. Some previous studies had shown that chord reading was found to be left-lateralized in musicians or amateur musicians (Segalowitz et al., 1979; Salis, 1980). However, this may possibly due to the task nature of the studies, in which participants needed to play the harmony on a keyboard after reading chords on scores. Motor planning and control was found to be left-lateralized (Mutha et al., 2012). Since our divided visual field sequential matching task did not involve music playing, and thus no left-lateralized processing advantage was observed in chords.

Furthermore, bilateral processing advantages were also observed in rhythm and articulation. These findings were among one of the few studies examining the hemispheric lateralization effects of reading rhythmic and articulatory patterns. These could be explained by the right-lateralized global processing (Fink et al., 1997) of the overall rhythmic and articulatory patterns, and left-lateralized local processing (Fink et al., 1997) of specific rhythmic and expressive markings. The findings from Salis (1980) examining the hemispheric lateralization of dot enumeration (i.e., perceiving spatial relations and groupings among dots) among musicians may also provide some insight to explain the role of the right hemisphere perceiving spatial relations and groupings among items, as in rhythmic and articulatory processing.

Among the few studies examining hemispheric lateralization of different musical elements, this study showed that the processing of pitch, harmony, rhythm, and articulation involved resources from the two hemispheres. The similar results of our study and previous studies (e.g., Ono et al., 2011) indicate that it remains uncertain whether the visual and auditory processing of musical elements may have some overlaps in the cognitive processing and neurocognitive bases.

Music training was the only subscale in Gold-MSI (Müllensiefen et al., 2014) that showed a positive correlation with the laterality index (LI) in music reading. This finding highlights the uniqueness of music training on hemispheric lateralization among other measurements of musicality in Gold-MSI, including active engagement, perceptual abilities, singing abilities, emotions, and general musical sophistication. Nonetheless, we remain conscious about the interpretation regarding correlational nature of this finding.