Conceived and designed the experiments: MF EW AN GS. Performed the experiments: MF EW AN GS. Analyzed the data: MF EW AN GS. Contributed reagents/materials/analysis tools: MF EW AN GS. Wrote the paper: MF EW AN GS.
The authors have declared that no competing interests exist.
In this study we investigated the association between instrumental music training in childhood and outcomes closely related to music training as well as those more distantly related.
Children who received at least three years (
While these results are correlational only, the strong predictive effect of training duration suggests that instrumental music training may enhance auditory discrimination, fine motor skills, vocabulary, and nonverbal reasoning. Alternative explanations for these results are discussed.
There is a widespread view that learning to play a musical instrument in childhood stimulates cognitive development and leads to enhanced skills in a wide variety of areas
The effect that training (or skill acquisition) in one domain might have on skills and cognitive performances in other domains is commonly referred to as transfer. The study of transfer has a long and contentious history
Although only experimental/longitudinal studies can demonstrate transfer, the results of many correlational studies have been used to suggest that transfer may occur from music training to other domains. We differentiate here between experimental, longitudinal studies and correlational studies testing for transfer.
Past research has clearly demonstrated that near transfer occurs from music training to music perception skills. In a longitudinal study, Flohr
There is also evidence for near transfer from instrumental music training to motor skills. A longitudinal investigation by Costa-Giomi
While near transfer effects are relatively common, it is notoriously difficult to demonstrate far transfer
A number of studies have investigated the relationship between music training and spatial abilities. Some have argued that spatial reasoning could be enhanced by music training because music notation itself is spatial, since specific pitches are indicated by their particular position on a series of lines and spaces
Parallels between music and language have been used to support the hypothesis that music training may strengthen verbal skills. Both music and written language involve formal notation read from left to right; music notation consists of symbols that represent information about sound (pitch, harmony, melody) and time (rhythm, meter), and listening to both music and speech requires attention to the temporal order of rapidly changing acoustic events
A number of correlational studies have reported an association between musical and language skills. Anvari, Trainor, Woodside and Levy
The existence of a transfer effect between music training and language skills has also been supported by experimental studies, although the evidence is not unequivocal. In a study by Overy
Although many explanations could be given for potential transfer between music training and mathematical performance (e.g., musical rhythm is based on mathematical relations), little experimental evidence has shown that such transfer occurs. A meta-analysis of six experimental studies testing whether music training leads to improved mathematics performance yielded a small, but significant overall effect size (
Schellenberg
Most of the studies investigating the effect of instrumental music training on cognitive abilities have not tested for near transfer. We argue here that any study of far transfer must include measures of near transfer in order to ascertain that learning in the parent domain has actually occurred. In addition, as shown above, the findings on far transfer are not always consistent, and hence further research is called for.
Here we report the results of a correlational study testing the hypotheses that children receiving instrumental music training perform better than those without such training in four areas of cognition that are distantly related to music: spatial, verbal, nonverbal, and mathematical. We also investigated whether children with music training perform better on two areas closely associated with music: fine motor and auditory skill. We also examined whether duration of training predicted performance on either the distant or closely associated tests. Finally, we discuss possible explanations for our findings and suggest which variables should be further investigated in future research.
Fifty-nine children were recruited from public schools and community music schools in the Boston area, as well as by word-of-mouth, to participate in this study which was approved by the Institutional Review Board (IRB) of Beth Israel Deaconess Medical Center (BIDMC). All children as well as their parents gave written informed consent to participate in this study. The mean age at the median testing session for each child was 9.96 years (
Children participated in 3–4 testing sessions (about six hours), over the course of 3–4 weeks.
Parents reported their highest level of education on a questionnaire and responses were scored on a 6-point scale: (1) some high school; (2) high school diploma or GED; (3) some college, vocational degree, or associate's degree; (4) 4-year college degree (BA, BS); (5) master's degree (MA, MS, MBA); (6) doctoral degree (PhD, MD, JD, EdD, ThD). Final SES scores represent either the single parent's score in a one-parent family, or the average of both parents' scores in a two-parent family. While education alone is not a complete measure of SES, it is considered to be an acceptable indicator
Duration of training for children in the Instrumental group (in weeks) was calculated from the child's first music lesson to the median time-point of the child's testing cycle.
In addition to reporting the date of commencement of music training, parents also indicated how much their children played their instrument(s) (in minutes per week), including home practice time as well as instruction and ensemble time. We specifically asked parents to estimate practice intensity a
Handedness was assessed using four measures adapted from Annett
Children received Gordon's Intermediate Measures of Music Audiation (IMMA)
A Melodic/Rhythmic Discrimination task designed in our lab
Beginning with their non-dominant hand, children performed three, 30-second trials of a 4-finger sequence task on the number keys of an alpha-numeric computer keyboard
Children received the Block Design subtest of the WISC-III
Children received the Object Assembly subtest of the WISC-III
Three levels of the Raven's Progressive Matrices (RPM) were administered in the following order: Colored Progressive Matrices, Standard Progressive Matrices and Advanced Progressive Matrices-Set I
Children were given the Vocabulary subtest of the WISC-III
Children received the Auditory Analysis Test
Children received
A first set of preliminary analyses examined whether or not the type of instrumental instruction received (Suzuki vs. traditional) affected the outcomes measured. One-way ANOVAs (with the type of instruction as the between-subjects factor) showed that the two Instrumental subgroups did not differ in age, gender, SES or duration of training (all
Preliminary ANOVAs comparing the Instrumental and Control groups showed no difference between groups in SES (
A MANCOVA (covarying age) was conducted to test for between-groups differences on all outcomes (except math since our sample was smaller for this test, as explained earlier). Missing values were replaced by the series' means (for 5.87% of all values). There was an overall significant effect of group, Wilks' Lambda = .54,
Outcomes | Control | Instrumental | Results | |||
Left Hand Motor Learning | 7.05 | .93 | 10.27 | .60 | 8.17 | <.01 |
Right Hand Motor Learning | 7.48 | .10 | 11.93 | .65 | 13.57 | <.01 |
IMMA Tonal | 33.76 | .66 | 37.14 | .43 | 17.732 | <.01 |
IMMA Rhythm | 33.24 | .71 | 33.78 | .46 | .391 | .53 |
Melodic Discrimination | 71.29 | 2.82 | 82.16 | 1.83 | 10.04 | <.01 |
Rhythmic Discrimination | 64.42 | 3.60 | 71.69 | 2.34 | 2.75 | .10 |
Vocabulary | 13.47 | .61 | 15.50 | .40 | 7.39 | <.01 |
Auditory Analysis | 31.01 | 1.10 | 32.99 | .71 | 2.21 | .14 |
Block Design | 13.84 | .76 | 14.22 | .49 | .168 | .68 |
Object Assembly | 10.87 | .69 | 11.74 | .45 | 1.07 | .31 |
Raven's Colored PM | 33.06 | .53 | 34.13 | .35 | 2.72 | .11 |
Raven's Standard PM | 22.64 | 1.04 | 25.15 | .68 | 3.97 | .05 |
Raven's Advanced PM | 8.44 | .39 | 9.45 | .26 | 4.50 | .04 |
KeyMath-Basic Concepts | 123.16 | 2.47 | 127.62 | 1.94 | 1.89 | .18 |
KeyMath-Operations | 115.73 | 3.76 | 123.69 | 2.96 | 2.59 | .12 |
KeyMath-Applications | 119.17 | 2.22 | 122.45 | 1.75 | 1.26 | .27 |
A second MANCOVA (covarying age) was carried out in order to compare groups on mathematical outcomes (the Basic Concepts, Operations and Applications areas of the Keymath-R test). A separate analysis had to be conducted because only 41 out of the 59 children received this test (16 Controls and 25 Instrumentals). The MANCOVA did not reveal an overall significant effect, Wilks' Lambda = .92,
In the above analyses, we did not attempt to equate the Instrumental and Control groups for either verbal or non-verbal IQ since these outcomes may well be effects of music training. However, to determine whether findings could be explained by pre-existing differences in either verbal or non-verbal IQ, we repeated the above analyses, once adding Vocabulary as a covariate, and once adding Raven's Progressive Matrices as a covariate (and covarying age both times, as in the previous analyses). When Vocabulary was added as a covariate, the Instrumental group was no longer superior on any of the Raven's Progressive Matrices (but remained superior on the motor and the auditory discrimination tests). However, when Raven's Colored, Standard, or Advanced Matrices scores were added as covariates (one at a time), the Instrumental group remained superior on the Vocabulary subtest of the WISC (
We investigated the relationship, within the Instrumental group, between weekly practice intensity at the time of testing and other independent variables. Practice intensity (in minutes per week) was significantly correlated with training duration (
A series of multiple regressions were performed in order to determine the effects of training duration on outcomes, controlling for age. Those in the Control group were entered as having zero weeks of training.
Controlling for age, training duration predicted four near outcomes: Motor learning on both left (partial
Duration of training did not predict any other outcomes (all
We did not include practice intensity as an independent variable in the above regression analyses for two reasons: (1) practice intensity was only reported at the time of testing and therefore did not accurately reflect the amount of practice achieved over the years; (2) training duration and practice intensity were significantly correlated with each other and thus entering both factors at the same time would be redundant and reduce power.
The results of the present study showed that children who had received instrumental music training for three years or more outperformed their control counterparts in areas closely related to music: fine motor skills (both hands) and discrimination between melodies (both on the Gordon's IMMA and the Melodic Discrimination Task). Strengthening these results is the finding that duration of music training predicted these results as well. These results are consistent with previous reports in the literature that music training is associated with enhanced fine motor skills in children
The results also showed that instrumental children outperformed their control counterparts in verbal ability (Vocabulary) and in non-verbal reasoning (both Raven's Standard and Advanced PM). Strengthening these results is the finding that duration of music training predicted performance on the Vocabulary test and on Raven's Advanced Progressive Matrices.
Contrary to some previous research, the Instrumental group did not outperform the Control group on phonemic awareness (as measured by the Auditory Analysis Test) or spatial skills (as measured by the WISC-III Block Design and Object Assembly subtests), nor were duration of training effects found for these outcomes. While Hetland
Three different types of explanations, which are detailed below, could account for our findings. The superiority of the Instrumental group may be due to: (1) domain-specific transfer effects (instrumental music training may causally enhance selected cognitive abilities) (2) a domain-general transfer effect (instrumental music training may enhance general IQ and lead to improvements in all cognitive domains); (3) non-causal associations mediated by third variables which were not accounted for in this study.
The first two explanations, which a correlational study can only suggest but not demonstrate, propose that transfer effects may have occurred between music training and distantly related domains. This explanation is supported by the fact that training duration predicted performance in these outcomes. The association between music training and vocabulary is consistent with past research suggesting that instrumental music training enhances verbal memory
The first explanation suggested by our results is that transfer effects may have occurred between music training and a selection of related domains. These transfers can be explained by the fact that some aspects of music are shared with other activities. Learning to decode written music notation may, for instance, increase reading ability. Learning to categorize sounds may enhance phonological awareness in nonmusical settings. The honing of visual pattern recognition and pattern matching skills resulting from instrumental practice and notation reading may explain our surprising results on the Raven's Progressive Matrices, since many of the items on this test can be solved using a visual pattern recognition/matching strategy. In the present study, the domain-specific hypothesis was supported by the fact that verbal ability was superior in the Instrumental group even when nonverbal reasoning skills were controlled.
A second hypothesis can explain our findings and clarify how transfer effects take place. As suggested by Schellenberg
The correlational design of this study does not allow us to rule out a variety of non-causal explanations for the associations found. First, family dynamics may account for the findings. Children whose parents enroll them in music lessons may be more engaged with their children's education, and provide more enriched environments than do parents who do not enroll their children in music lessons. Children who practice their instruments more than others may do so because of parental expectations and insistence. These same parents may insist that their children work hard in school, do their homework, and read. Miller and Orsmond
Second, superior motivational skills in instrumental children may account for our findings. Our study only included children who had persisted with music lessons for at least three years. We did not include those who had begun lessons but dropped them early on. Thus the children in the Instrumental group may have been more persistent and motivated than average. Children with superior motivational skills and ability to persist on difficult tasks may not only stick with music lessons but may also practice more than is typical of children taking lessons. In addition, such children may work harder at school and read more, thereby learning more and resulting in heightened performance on the kinds of cognitive tests administered in this study.
The results of our study confirm previous research showing that children who take instrumental music lessons are ahead on a number of cognitive abilities. However, the correlational design of this study does not allow us to determine definitively whether music causally enhanced verbal and nonverbal reasoning skills, or whether other variables were responsible for the effects found. The causal explanation was supported by the fact that duration of training linearly predicted cognitive outcomes.
The results of this study, and the issues they raised, will guide our expectations for an ongoing quasi-experimental longitudinal study currently underway in our laboratory. At the final timepoint of this longitudinal study, the children participating will have received about the same amount of music training as the children included in the present study. At baseline, children in the Instrumental group did not differ from those in the Control group on any outcomes
We thank Katie Overy, Karl Cronin, Lucy Forbes, Lucy Blake, Charlotte Alexander, Kathleen Brumm, Lin Zhu, Udita Iyengar, Camilla Rosam and the many other part-time research assistants and summer interns who helped us with data collection, as well as the participating children and their families.