The authors have declared that no competing interests exist.
Conceived and designed the experiments: CB NG. Performed the experiments: JZ CB AK NG. Analyzed the data: JZ CB AK NG. Contributed reagents/materials/analysis tools: CB NG. Wrote the paper: JZ CB AK NG.
Executive functions (EF) are cognitive capacities that allow for planned, controlled behavior and strongly correlate with academic abilities. Several extracurricular activities have been shown to improve EF, however, the relationship between musical training and EF remains unclear due to methodological limitations in previous studies. To explore this further, two experiments were performed; one with 30 adults with and without musical training and one with 27 musically trained and untrained children (matched for general cognitive abilities and socioeconomic variables) with a standardized EF battery. Furthermore, the neural correlates of EF skills in musically trained and untrained children were investigated using fMRI. Adult musicians compared to non-musicians showed enhanced performance on measures of cognitive flexibility, working memory, and verbal fluency. Musically trained children showed enhanced performance on measures of verbal fluency and processing speed, and significantly greater activation in pre-SMA/SMA and right VLPFC during rule representation and task-switching compared to musically untrained children. Overall, musicians show enhanced performance on several constructs of EF, and musically trained children further show heightened brain activation in traditional EF regions during task-switching. These results support the working hypothesis that musical training may promote the development and maintenance of certain EF skills, which could mediate the previously reported links between musical training and enhanced cognitive skills and academic achievement.
Executive functions (EF) encompass a number of cognitive processes that allow for independent and self-regulated behavior
The development of executive function occurs rapidly during early childhood
Overall, EF abilities have been shown to be more predictive of academic readiness for schooling than intelligence
Various extra and intra-curricular activities have been shown to improve EF skills in children. For instance, Tools of the Mind, a curriculum especially designed to enhance EF skills and social/emotional development in preschool children
One extracurricular activity of recent interest to researchers is music, and its link to EF skills has been debated
To date, only a few studies have investigated the relationship between musical training and EF constructs in children and adults. Superior performance has been demonstrated in children and adults with musical training over non-musician controls on measures of auditory and visual working memory
For studies that have examined EF performance in trained musicians, the mixed findings reported are likely due to various methodological limitations regarding the validity of the assessments employed and subject inclusion criteria. Enhanced processing in adult musicians has been reported for components of EF, demonstrated through a nonverbal spatial task and both auditory and visual Stroop tasks
In order to address the causal nature of this hypothesized connection between musical expertise and EF abilities, the influence of musical training on EF development has also been examined longitudinally. Six months of individualized piano instruction demonstrated improved EF abilities, specifically cognitive flexibility and working memory, in elderly subjects with minimal musical experience
It is evident that musical training relates to cognitive abilities, but it remains somewhat unclear which constructs of EF, if any, may mediate this connection. In the present study, we seek to (a) evaluate the relationship between intensive instrumental musical training and EF skills through a cross-sectional design that addresses the limiting factors of previous studies that resulted in mixed findings, and (b) compare the neural correlates of EF skills in musically trained as compared to untrained children. We assessed adults with extensive musical training and school-age musically trained children, documenting the intensity and longevity of training, and included only adult non-musicians and musically untrained children that were carefully screened to have no prior musical training beyond general curricular requirements. Several indicators of socioeconomic status were reported and matched between musicians and non-musicians, and our groups were matched for IQ to avoid any confounds of higher intelligence in the group comparisons. Further, we implemented a standardized battery of EF measures that assessed cognitive flexibility, inhibition, verbal fluency, working memory, and processing speed. In addition, this is the first study to examine the neural correlates of executive functioning, specifically task-switching (rule representation and task-set reconfiguration, adapted after Crone and colleagues
30 healthy, right-handed, monolingual, English-speaking adults (15 musicians (9 male, 6 female) and 15 non-musicians (9 male, 6 female), age range: 18-35 yrs, mean: 24.80 yrs; STD: 3.48 yrs) took part in the present study. Adult musicians were either seeking or had obtained a music performance degree and were working professionals. Adult musicians had commenced musical study by or before the age of 9 (mean start: 5.73 yrs, STD: 1.62 yrs), had received private lessons, were presently playing at least 8 hours per week (mean: 21.87 hrs/wk, STD: 11.49 hrs) and had studied music continuously since the onset of training. All musicians actively pursued multiple instruments while maintaining one principal instrument (type of principal instrument described in
Mean ± SD | |
Age at onset of musical training (years) | 5.73±1.62 |
Intensity of practice time/week (hours) | 21.87±11.49 |
Duration of musical training (years) | 5.2±1.33 |
Piano | 6 |
Strings | 5 |
Woodwinds | 1 |
Brass | 2 |
Harp | 1 |
Age at onset of musical training (years) | 5.86±1.41 |
Intensity of practice time/week (hours) | 3.74±2.63 |
Duration of musical training (years) | 5.2±1.33 |
Piano | 5 |
Strings | 5 |
Woodwinds | 2 |
Guitar | 1 |
Percussion | 2 |
27 children (15 musically trained (7 male, 8 female) and 12 untrained (4 male 8 female), age range: 9-12 yrs; mean 10.9 yrs, STD: 1.2 yrs) took part in this study. Musically trained children had played an instrument for a minimum of two years in regular private music lessons, started training on average at age 5 (mean: 5.86 yrs, STD: 1.41 yrs) and had been studying their instrument on average 5.2 years (STD: 1.33 yrs). More information on the details of musical training can be found in
No significant group differences in age, gender, or IQ were observed for adults or children (
Musicians | Non-musicians | P-Values | |||
Mean ± SD | Mean ± SD | Mus vs. Non | |||
IQ | |||||
WASI | Verbal Ability | 63.73±5.79 | 61.80±7.63 | 0.441 | |
Nonverbal Ability | 60.80±6.01 | 57.20±5.80 | 0.106 | ||
(Mean Rank) | (Mean Rank) | ||||
Adult Education | 15.27 | 14.71 | 0.852 | ||
Current job responsibility | 14.13 | 15.93 | 0.524 | ||
Parent Education | 14.83 | 16.17 | 0.668 | ||
Total Combined Family Income |
14.62 | 13.43 | 0.693 | ||
DKEFS | Trail Making | 9.00±2.39 | 8.80±3.19 | 0.423 | |
Verbal Fluency | 11.80±3.90 | 8.87±3.38 | 0.018 | * + | |
Color-Word Interference | 11.27±1.10 | 10.73±1.79 | 0.160 | ||
Design Fluency | 15.07±2.37 | 12.33±2.72 | 0.003 | ** + | |
WAIS | Digit Span Backwards | 14.47±3.25 | 10.40±3.42 | 0.001 | ** + |
Coding | 13.40±2.90 | 11.93±3.15 | 0.098 | ||
IQ | |||||
KBIT | Non-Verbal Ability |
119.60±9.34 | 117.70±11.24 | 0.665 | |
(Mean Rank) | (Mean Rank) | ||||
Parent Education | 11.75 | 12.39 | 0.817 | ||
Current job responsibility of parent | 13.54 | 9.61 | 0.156 | ||
Total Combined Family Income |
13.79 | 9.22 | 0.080 | ||
DKEFS | Trail Making |
9.33±1.76 | 7.33±2.24 | 0.026 | * + |
Verbal Fluency | 10.80±2.51 | 8.17±3.56 | 0.016 | * + | |
Color-Word Interference | 10.20±1.21 | 9.92±2.19 | 0.336 | ||
WISC | Digit Span Backwards | 9.80±2.36 | 10.81±2.52 | 0.151 | |
Coding | 11.13±1.99 | 9.17±2.41 | 0.013 | * | |
% | % | ||||
Univalent Rule Accuracy | 95.97±0.09 | 95.41±0.08 | 0.865 | ||
Bivalent Rule Accuracy | 90.56±0.12 | 85.03±0.16 | 0.307 | ||
Switching Accuracy | 92.09±0.12 | 87.76±0.13 | 0.367 | ||
Rule Representation | 92.72±0.10 | 89.18±0.12 | 0.424 |
+ significant with FDR Correction.
one child did not finish all testing.
Scale where 1 = $25 000–34 999, 2 = $35 000–49 999, 3 = $50 000–74 999, 4 = $75 000–99 999, 5 = $100 000+.
This study was approved by the Boston Children's Hospital's Committee on Clinical Investigation (CCI). Written assent and informed consent were obtained from each child participant and guardian, respectively. All adult participants provided written informed consent.
Adult participants completed the Delis-Kaplan Executive Function System as part of a larger study (DKEFS;
The Trail Making subtest assesses visual scanning, numeric and alphabetic sequencing, motor speed, and cognitive flexibility. Participants are timed on their ability to trace objects within a specified order when scrambled across a large sheet of paper, and corrected for errors throughout. The task includes five trials, (1) line tracing, (2) number tracing, (3) letter tracing, (4) number-letter switching, and (5) motor speed. The task of interest is a number-letter switching test in which the participant is required to draw straight lines to connect numbered and lettered circles in numerical and chronological order while switching between numbers and letters as quickly as possible. The output variable contrasted time to completion of the switching task to time required for the combined outcome of the two separate trials measuring number tracing and letter switching.
The Verbal Fluency subtest contains three conditions that measure letter fluency, category fluency, and category switching fluency. Our output of interest compared achievement on letter fluency with category fluency. In letter fluency, participants were prompted with a single letter and asked to state as many words starting with that letter as possible within 60 seconds, excluding names of people, places, or numbers. In category fluency, participants were prompted with a category (e.g. boy's names, animals) and asked to name as many objects within the category of interest as possible within 60 seconds. Category switching fluency required participants to switch naming between two categories simultaneously. Responses were standardized based on the number of correct words named.
The Color-Word Inference Test, based on the Stroop test
Adults additionally completed the Design Fluency subtest of the DKEFS. Design Fluency involves three subtests that require the participant to connect a set series of dots to make as many different designs possible within 60 seconds. Performance on the third task, creating designs while switching between empty and filled dots, was compared in musicians and non-musicians.
Working memory and processing speed were evaluated through the Digit Span Backwards and Coding subtests (respectively) of the Wechsler Abbreviated Intelligence Scale, 4th Edition (WAIS-IV;
In order to match general cognitive ability across groups, nonverbal IQ was tested in our children with the Kaufman Brief Intelligence Test (KBIT;
A multi-modal version of a traditional set-shifting task was developed (
In each trial a cue [arrow; circle; or triangle] representing a rule was followed by a sound. Children responded with a left or right button press (arrow: horse = right; dog = left; circle: frog = right; bird = left; triangle: bird = right; frog = left). Critically, in one instance the rule consistently maps to single auditory stimuli (univalent rule) while in the latter two the auditory stimulus-response relationship changes with the visual cue (bivalent rules).
Trial type and switch type were both randomized within each run using optseq (
The following regressors were modeled: bivalent rule repetitions and switches, univalent rule repetitions and switches, and bivalent rule reconfigurations. Correct and incorrect trials were modeled separately, as were misses. Trials commenced with the visual cue and terminated at the end of the auditory stimulus (3.5 s). Each child's session level models were combined into fixed effects models; children were then combined in random effects analyses. Statistical inference was completed using Z (Gaussianised t) images, cluster thresholded (Z>2.3;
Independent t-tests (one-tailed, FDR corrected
Musically trained children performed better than untrained children (independent t-tests, one-tailed,
In-scanner performance revealed that both groups achieved high performance accuracy in rule representation and task-switching. Accordingly, no significant differences in behavioral scanner performance were observed for musically trained versus untrained children on univalent rule accuracy, bivalent rule accuracy, switching accuracy, or rule representation (see
Whole brain analyses of rule representation (contrast: all bivalent > all univalent rule trials) demonstrated significant activation for both groups in several brain regions including the SMA/paracingulate cortex and VLPFC bilaterally for the musically trained group only (
Note: activation is displayed with the FSL radiological convention.
Coordinates | |||||||
Voxels | Maximum (Z) | x | y | z | Cerebrum | BA | Region |
1619 | 3.82 | 8 | 22 | 40 | R | 8 | Middle/Superior frontal gyrus (pre-SMA/SMA) |
1355 | 3.72 | −34 | −60 | 44 | L | 7 | Lateral Occipital Cortex/Superior Parietal Cortex |
1253 | 3.92 | 42 | −42 | 48 | R | 40 | Supramarginal Gyrus (Inferior Parietal Lobule) |
933 | 3.55 | −46 | 12 | 30 | L | 9 | Middle Frontal Gyrus (DLPFC) |
854 | 4.03 | −6 | −58 | 44 | L | 7 | Precuneous |
645 | 3.71 | −34 | 2 | 56 | L | 6 | Middle Frontal Gyrus (pre-SMA/SMA) |
590 | 3.35 | 8 | −80 | −26 | R | Cerebellum | |
456 | 3.53 | −30 | −60 | −34 | L | Cerebellum | |
409 | 3.39 | 32 | 2 | 46 | R | 6 | Middle Frontal Gyrus (pre-SMA/SMA) |
396 | 3.22 | 40 | 30 | 24 | R | 46 | Middle Frontal Gyrus (VLPFC) |
308 | 3.77 | −30 | 24 | −6 | L | 47 | Inferior Frontal Gyrus (VLPFC) |
Coordinates | |||||||
Voxels | Maximum (Z) | x | y | z | Cerebrum | BA | Region |
1531 | 4.12 | −30 | −70 | 40 | L | 19 | Lateral Occipital/Superior Parietal Cortex (Precuneous) |
936 | 3.53 | 34 | −60 | 66 | L | Superior Lateral Occipital Cortex | |
514 | 3.77 | 2 | 18 | 50 | R | 8 | Middle/Superior Frontal Gyrus (pre-SMA/SMA) |
Coordinates | |||||||
Voxels | Maximum (Z) | x | y | z | Cerebrum | BA | Region |
24 | 2.98 | 24 | 78 | 63 | L | 47 | Inferior Frontal Gyrus (VLPFC) |
14 | 2.9 | 24 | 78 | 39 | L | 44 | Inferior Frontal Gyrus/Frontal Operculum |
11 | 3 | 21 | 30 | 36 | L | 41 | Heschl's Gyrus/Planum Temporale |
Coordinates in MNI space, gray matter activations significant at p<0.05 with a cluster threshold >50 voxels for musically trained and untrained groups separately; p<0.005 uncorrected threshold for the two-sample t-test.
Whole brain analysis of task-switching (contrast: bivalent switches and reconfigurations > univalent switches) revealed a similar pattern of activation in bilateral SMA, VLPFC, and superior parietal regions in both groups of children (
Note: activation is displayed with the FSL radiological convention.
Coordinates | |||||||
Voxels | Maximum (Z) | x | y | z | Cerebrum | BA | Region |
1171 | 3.64 | 49 | 96 | 99 | R | 8 | Middle/Superior Frontal Gyrus (pre-SMA/SMA) |
988 | 3.63 | 69 | −24 | 99 | R | 7 | Lateral Occipital/Superior Parietal Cortex (Precuneous, Angular Gyrus) |
532 | 4.29 | 42 | −24 | 102 | L | 19 | Lateral Occipital/Superior Parietal Cortex (Precuneous) |
485 | 3 | 61 | 66 | 102 | R | 8 | Middle/Superior Frontal Gyrus (pre-SMA/SMA) |
384 | 3.58 | 29 | −27 | 102 | L | 7 | Lateral Occipital/Superior Parietal Cortex (Precuneous, Angular Gyrus) |
Coordinates | |||||||
Voxels | Maximum (Z) | x | y | z | Cerebrum | BA | Region |
1367 | 4.32 | 28 | −36 | 96 | L | 7 | Lateral Occipital/Superior Parietal Cortex (Precuneous, Angular Gyrus) |
1273 | 4.31 | 62 | −14 | 114 | R | 7 | Lateral Occipital/Superior Parietal Cortex (Angular Gyrus) |
662 | 4.26 | 47 | −39 | 114 | R | 7 | Precuneous Cortex |
533 | 3.67 | 29 | 72 | 114 | L | 8 | Middle/Superior Frontal Gyrus (pre-SMA/SMA) |
283 | 3.62 | 59 | 72 | 108 | R | 8 | Middle/Superior Frontal Gyrus (pre-SMA/SMA) |
274 | 3.57 | 42 | 90 | 108 | L | 8 | Superior Frontal Gyrus |
Coordinates | |||||||
Voxels | Maximum (Z) | x | y | z | Cerebrum | BA | Region |
52 | 3.56 | 27 | 75 | 45 | L | 44 | Inferior Frontal Gyrus/Frontal Operculum (VLPFC) |
25 | 2.91 | 64 | 72 | 51 | R | 44 | Inferior Frontal Gyrus/Frontal Operculum (VLPFC) |
21 | 2.95 | 75 | 27 | 75 | R | 40 | Supramarginal Gyrus (Inferior Parietal Lobule) |
Coordinates in MNI space, gray matter activations significant at p<0.05 with a cluster threshold >50 voxels for musically trained and untrained groups separately; p<0.005 uncorrected threshold for the two-sample t-test.
Our study employed strict participant inclusion criteria and utilized standardized psychometric measures to clarify the mixed findings to date on the relationship between musical training and EF abilities. We further explored the associated neural correlates of task-switching in musically trained over untrained children through fMRI. Overall, adult musicians and musically trained children showed heightened performance on several but not all constructs of EF, and children further demonstrated enhanced brain activation in traditional EF regions during a task-switching paradigm. Behavioral differences between adult musicians and non-musicians were observed for measures of cognitive flexibility (such as Verbal Fluency, Design Fluency, Trail Making) and working memory. Musically trained and untrained children also significantly differed on measures of cognitive flexibility (Verbal Fluency and Trail Making) and processing speed as well. The investigation of the neural correlates of rule representation and task-switching revealed greater activation in the SMA and right VLFPC for musically trained as compared to untrained children. These results support the working hypothesis that musical training may promote the development and maintenance of EF, which could mediate the previous reported links between musical training and heightened academic achievement, though our cross-sectional study presently cannot address whether prior EF abilities may have promoted the development of musical training.
The connection between musical skill and specific components of EF is conceivable, given the demands of sustained attention, goal-directed behavior, and cognitive flexibility that are involved in musical training. Our results are in line with some prior evidence of a connection between musical training and certain EF constructs, particularly the previously reported association between several EF constructs in children with varying intensities of musical training
Interestingly, significant differences in processing speed were only observed between musically trained and untrained children. The developmental trajectory of processing speed has been described to begin in childhood and continue until adolescence
Only one previous study did not observe any differences between intensive musical training and executive function
As for our neuroimaging evidence, we evaluated rule representation and task-switching in our sample of children using a multi-modal traditional EF task. Rule representation (contrast: bivalent > univalent rule trials) and task-switching (contrast: bivalent switches and reconfigurations > univalent switches) was associated with an activation increase within the VLPFC, SMA, and superior parietal cortex in all children regardless of musical training. This is consistent with previous results utilizing a similar task in adults and children
As for activation differences based on musical training during task-switching, musically trained children demonstrated enhanced activation in the bilateral VLPFC in the whole brain two-sample comparison over untrained children. Although no significant differences in the superior parietal ROIs were found between groups, untrained children demonstrated more activation in the left superior parietal regions over musically trained children at the whole-brain level. This finding suggests that children with versus without musical training differentially recruit specific brain regions during task-switching. In particular, untrained children appear to recruit parietal regions within a network of activation that is typical for task-switching
Interestingly, musically trained and untrained children both showed high performance accuracy on the neuroimaging task and no significant behavioral differences were observed. It may be possible that we did not find more robust activation differences in other regions, such as the parietal areas, due to this strong behavioral performance in both musicians and non-musicians. An alternative explanation to no significant activation differences in parietal regions could be that the effects of musical training do not enhance the aspects of executive functioning that are represented by parietal areas, but instead specifically engage prefrontal regions of the brain. No findings in other brain regions supports the evidence that distinct neural components are involved in task-switching
All children exhibited activation in the VLPFC for rule representation (contrast: bivariate > univariate rule trials). We further observed greater activation for musically trained as compared to untrained children in the right VLPFC during rule representation in the whole brain findings and for more complex rather than simple rule representations (bivariate rule reconfiguration > univariate switches) in the ROI analysis. This is in line with the study reported by Pallesen and colleagues
Undoubtedly, the connection between musical training and cognitive ability is highly complex, as previously argued by Schellenberg
Overall, we conclude that children and adults with extensive musical training show enhanced performance on a number of EF constructs compared to non-musicians, especially for cognitive flexibility, working memory, and processing speed. Investigation of the neural correlates of rule representation and task-switching further revealed heightened activation in bilateral SMA and left VLPFC for musically trained as compared to untrained children through direct whole brain comparison and ROI analysis. Thus, our results support the working hypothesis that executive functioning may be one of the mechanisms mediating the often reported link between musical training and heightened academic skills, as EF skills and academic skills are highly correlated. However, more longitudinal studies and interventions are needed in order to examine a possible causal relationship between musical training, EF skills and academic achievement. Furthermore, behavioral and neural developmental trajectories for various EF skills need to be examined for musicians as compared to non-musicians. Nevertheless, future studies examining cognitive and academic skills between musicians and non-musicians should control for various components of EF. Likewise, it is important to consider that replacing music programs with reading or math instruction in our nation's school curricula in order to boost standardized test scores may actually lead to deficient skills in other cognitive areas.
We thank Michelle Chang for her early contributions, and all the families and adults who participated in this study.