Abstract
The relationship between music training and executive functions has remained inconsistent in previous studies, possibly due to methodological limitations. This study aims to investigate cognitive inhibitory control in children (9–12 years old) with and without musical training, while carefully considering confounding variables. To assess executive functions, the Simon task was used, measuring reaction times (RTs) and error rates on congruent and incongruent trials. Information on important variables such as bilingualism, socio-economic status (SES), music pedagogy and amount of musical training was collected through a parental questionnaire. Furthermore, verbal and non-verbal intelligence were assessed with validated tests to consider their effects as well. The results showed that the samples did not significantly differ in background variables. The analysis of the RT data on the Simon task revealed a significant group × congruency interaction, such that musically trained children showed a reduced magnitude of the congruency effect (RTs on incongruent trials – RTs on congruent trials) compared to non-musicians. To conclude, music training seems to be associated with enhanced cognitive inhibitory control in well-matched samples.
Previous studies have shown that music training results in more than instrumental skill acquisition. Musical training has been positively associated with various cognitive aspects (e.g., Moreno et al., 2011; Zuk, Benjamin, Kenyon, & Gaab, 2014). Additionally, music training can take different forms, such as individual instrumental classes, group instrumental instruction and classroom music lessons.
Music training and cognitive abilities
Music training has been associated with general IQ outcomes and academic achievement (Schellenberg, 2004, 2006, 2011). In a study by Schellenberg (2004), 6-year-old children were randomly assigned to music classes, drama classes or no extra classes. After one year of training, only the music group showed higher increases in general IQ and academic achievement compared to the drama and the control group. Music training has also been shown to be associated with linguistic skills (Forgeard, Winner, Norton, & Schlaug, 2008; Moreno, Bialystok, et al., 2011). For instance, in Moreno, Bialystok, et al. (2011), children were quasi-randomly assigned to a computer-based music or visual arts training. Although there were no differences between groups before the training, the musicians outperformed the other group after training on a measure assessing vocabulary. Moreover, musical training has also been positively associated with enhanced early reading skills (Moreno, Friesen, & Bialystok, 2011), visuo-spatial skills (Hetland, 2000), mathematical skills (Vaughn, 2000) and prosody perception (Thompson, Schellenberg, & Husain, 2003). For a more extensive overview of studies assessing the associations between music training and cognitive abilities, see Schellenberg and Weiss (2013).
Music training and executive functions
Less research has been conducted on the relationship between music training and executive functions (EF). EF is a general concept that refers to goal-directed behaviour, planning and problem-solving and includes several components such as working memory (updating), task shifting and inhibitory control (Miyake et al., 2000). According to Miyake et al. (2000), working memory (updating) refers to an “active manipulation of working memory content”, and shifting is defined as “shifting back and forth between multiple tasks, operations, or mental sets”. Finally, inhibitory control is considered the “ability to deliberately inhibit dominant, automatic, or prepotent responses when necessary”. Although EFs seem to have a fixed developmental trajectory (Diamond, 2002), various experiences may influence them as well (Diamond & Lee, 2011). Hence, some studies have shown that bilinguals seem to have more developed EFs (Bialystok & DePape, 2009; Bialystok, Martin, & Viswanathan, 2005; Costa, Hernández, Costa-Faidella, & Sebastián-Gallés, 2009; Costa, Hernández, & Sebastián-Gallés, 2008). Other types of training that have been shown to improve EF are, for example, video games (Green & Bavelier, 2003; Oei & Patterson, 2013) and physical exercise (Davis et al., 2011; Voss et al., 2011). However, music training could also enhance EF (Bialystok & DePape, 2009; Degé, Kubicek, & Schwarzer, 2011; Zuk et al., 2014). The intensity and the need for selective attention in all these activities might explain why they have been associated with enhanced EF. However, what could make music unique is the requirement to respond to others’ rhythms, which helps develop one’s sensorimotor coordination and integration and selective attentional processing, which are all crucial for EF development (Miendlarzewska & Trost, 2014). Moreover, playing a musical instrument may rely on several elements of EF such as working memory when memorizing musical excerpts, inhibitory control when playing with others in an orchestra or in a chamber music setting, and shifting when dealing with different time signatures, rhythmic patterns, tempi and dynamics in a musical piece.
Previous studies examining the possible relationships between music training and EF have shown more developed working memory skills in musically trained children and adults (Lee, Lu, & Ko, 2007; Pallesen et al., 2010). However, with respect to shifting, the findings are less consistent. On the one hand, Zuk et al. (2014) found that musically trained children and adults outperformed non-musicians on several tasks assessing shifting, such as the Trail Making Task. This positive association between music training and shifting was confirmed by Bugos, Perlstein, McCrae, Brophy, and Bedenbaugh (2007), who examined the effects of 6 months of individualized piano training in older people. After this training, the participants showed enhanced shifting skills as well as other improvements. On the other hand, these advantages in shifting have been contradicted by other studies that have found no advantages in musically trained children compared to controls (Schellenberg, 2011). Considering cognitive inhibitory control, similar inconsistencies exist. Musicians showed an inhibitory control advantage on a go/no-go inhibition task after a 20-day computerized music training compared to children receiving a computerized visual arts training (Moreno, Bialystok, et al., 2011). Moreover, Degé et al. (2011) found that musically trained 9- to 12-year-old children showed enhanced performance on a subtest of the Developmental NEuroPSYchological Assessment (NEPSY II test; see Korkman, Kirk, & Kemp, 2007), which measures inhibitory control. Conversely, no advantages on inhibitory control were found in 9- to 12-year-old musically trained children in a recent study by Zuk et al. (2014) that used the Colour-Word Interference Test, a measure related to the Stroop test. Similarly, Schellenberg (2011) found no inhibitory control advantages in musically trained children using the Sun-Moon Stroop task.
Some authors have hypothesized that EF might have a mediating role between music training and general intelligence (Hannon & Trainor, 2007; Schellenberg & Peretz, 2008). Indeed, Degé et al. (2011) found positive associations between music instruction, fluid intelligence and EF in 9- to 12-year-old children assessing five different aspects of EF (selective attention, inhibitory control, set shifting, planning and fluency), with inhibitory control and selective attention as the strongest mediators between music training and intelligence. In contrast, Schellenberg (2011) assessed five different aspects of EF in 9- to 12-year-old musicians and non-musicians and did not find evidence for this mediation theory, with musically trained children showing no advantages on any of the EF measures.
The inconsistent findings between studies might be explained by several factors. One reason could be that some important background variables were not always taken into account in previous studies, possibly differing between music and non-music groups. For instance, differences in or lack of information about the socio-economic status (SES) or linguistic background of the participants in both music and non-music groups appear in some studies (Bialystok & DePape, 2009; Degé et al., 2011; Schellenberg, 2011), and these variables have been shown to influence EF (Bialystok, 2001; Mezzacappa, 2004). In addition, more precision concerning the definition of “musical training” is needed. As Degé et al. (2011) and Zuk et al. (2014) suggested, the samples should not only be categorized as “musicians” and “non-musicians” but should incorporate the different durations of musical training received, the weekly amount of music practice and the number of instruments played. In addition, we believe that the variations in music pedagogy (e.g., Kodály method, Suzuki method, solfège-based methods) might also be responsible for differences within music groups. This latter element has often been neglected in previous studies. For instance, in the Suzuki method, children learn to play music before being able to read notes. They thus need to memorize musical pieces rather than reading from notes. In contrast, in solfège-based methods, children first learn how to read notes and thereafter use these notes when playing musical pieces. This may result in working memory differences between both music pedagogies. Similarly, in both the Kodály and Suzuki methods, children mostly begin training at a very young age (e.g., 4 years old), which is not the case in solfège-based methods (e.g., 9 years old in Flanders, Belgium). In addition, the Suzuki method is an intensive music learning method, requiring daily practice with one parent (who also has to learn to play the musical instrument) and frequent (weekly) group classes in addition to individual classes. While intensive practice might also occur in other pedagogies, it is not always required. For example, in solfège-based methods, parents do not need to practice on a daily basis with their child, and group classes are not mandatory. A recent study assessing the variables modulating the relationship between music training and cognitive development reported that the degree of the effects of music training on cognitive development (e.g., EF) depends, among other variables, on the age of music training commencement (Miendlarzewska & Trost, 2014). Hence, the authors refer to a “sensitive period”, a limited period in development in which the effects of training are unusually strong (Knudsen, 2004). It has indeed been shown that the effects of music training seem to be stronger when training starts in this sensitive period (before age 7) (Habib & Besson, 2009). In addition, Miendlarzewska and Trost (2014) suggest that training duration and intensity modulate the relationship between music training and EF as well. Because music pedagogies differ in these elements, it is possible that they influence EF differently, and these variables may also need to be considered. Finally, different musical styles (e.g., jazz) and different formats of learning music (e.g., communal learning) may also differentially affect EF (for a more in-depth discussion of the definition of music training, see Chin & Rickard, 2012 and Müllensiefen, Gingras, Musil, & Stewart, 2014).
Purpose of the present study
The present study aims to shed light on the association between music training and cognitive inhibitory control, an element of EF, in 9- to 12-year-old children using a cross-sectional design and carefully considering multiple background confounders. This age group was chosen because it reflects a period in which individual performance on interference tasks measuring inhibitory control develops considerably (Diamond, 2002). Furthermore, certain neurobiological developments linked to cognitive performance occur in this age group as well, such as changes in the grey matter volume of the prefrontal cortex (Giedd et al., 1999; Sowell, Delis, Stiles, & Jernigan, 2001). In addition, because it has been shown that the duration of music training is associated with EF outcomes (Degé et al., 2011), children in that age range might have received sufficient musical training for EF advantages to appear.
To measure inhibitory control, the Simon task (Simon & Rudell, 1967) has been used and is a standard measure of EF (Lu & Proctor, 1995). Inhibitory control was measured because it appears to be inconsistently associated with music training in studies testing 9- to 12-year-old children (Degé et al., 2011; Zuk et al., 2014). The interference effect (the difference between the reaction times on the incongruent trials and the reaction times on the congruent trials) has been calculated as an index of cognitive inhibitory control. We hypothesized that the musically trained group would have a reduced interference effect compared to the control group. In addition, we rigorously selected our experimental and control groups in relation to a few possible confounders that were reviewed above. Hence, for the music group, we only included monolingual children receiving music classes following the Suzuki method, a special musical education method that enables children to play a musical instrument from a young age. In this method, children learn to play music before being able to read notes, by listening to and imitating their teacher. We also accounted for SES, verbal and non-verbal intelligence, number of years in musical training, extent of weekly practice and number of instruments played. For the control group, monolingual children who were not involved in extracurricular musical classes beyond the music classes offered at school were included. SES and verbal and non-verbal intelligence were assessed in both groups.
Method
Participants
The participants included 63 children (Mage = 10.44 years, SD = 0.53, range = 9.50–11.50 years; n = 35 boys). All participants had normal vision and hearing and had no developmental disorders. To determine the sample size, we conducted a power analysis on the interference effect (RT difference = RT incongruent – RT congruent) using G*Power software (Faul, Erdfelder, Lang, & Buchner, 2007). To this end, we assumed a moderate-to-large estimate of effect size, d = .65. Assuming equally sized groups, a directional hypothesis (one-sided test; i.e., a reduced interference effect in the music group), and an alpha level of .05, we needed at least 60 participants to achieve 80% power (30 participants in each group).
The music group consisted of 32 Dutch-speaking children (Mage = 10.44 years, SD = 0.51, range = 9.75–11.50 years; n = 21 boys). All these children had been taking music classes since the age of 5 or younger. They were recruited from three music schools in Flanders (Belgium) that offered Suzuki-method classes with qualified teachers. The group included 22 violin players, 8 viola players, 1 flute player and 1 cellist. Two of the violin players played a second instrument (piano and viola). Because bilingualism might be associated with enhanced EF (Bialystok, 2001), two bilingual children were excluded from the analysis.
The control group consisted of 31 monolingual Dutch-speaking children (Mage = 10.44 years, SD = 0.56, range = 9.50–11.42 years; n = 14 boys). The children in this group did not have music training outside of the classes offered at their school. These children were selected from two regular primary schools in Flanders (Belgium).
Cognitive inhibitory control measure
Inhibitory control was measured with the Simon task (Simon & Rudell, 1967). Stimuli were shown on a desktop with a 15-inch CRT screen. The responses had to be provided on a RB-844 response pad (Cedrus). The Simon test was programmed with E-Prime 2.0 software (Schneider, Eschman, & Zuccolotto, 2002). The distance between the child and the screen was approximately 50 cm. A trial started with the presentation of a fixation cross for a duration of 800 ms at the centre of the screen. Next, a green or red circle was shown at the extreme right or left of the screen, above the response keys. The circle was shown until a response was given (with a maximum duration of 3,000 ms). The test included two blocks of 100 trials in which the four different types of trials (2 colours × 2 locations) were presented in a counterbalanced order. Hence, half of the participants were asked to press the left button (covered with a red colour) as fast and as accurately as possible when a red circle was shown on the screen and the right button (covered with a green colour) when a green circle was presented (and vice versa for the other half of the participants). Of the trials, 50% were congruent, meaning that the position of the circle on the screen matched the position of the required response button. The remaining 50% of the trials were incongruent, with the position of the circle on the screen not matching the position of the required response key. Prior to testing, 10 practice trials were offered to familiarize the participants with the task. Personalized feedback was given during these practice trials by the examiner. In between blocks, a short break was provided. Reaction times (RTs) and error rates on congruent and incongruent trials were measured for each participant as the dependent variables.
Background measures
Socio-economic status (SES)
Information on SES was obtained via a parental questionnaire (see Appendix) and concerned parental education. Four levels of education could be indicated: level one for primary education, level two for education until 16 years of age, level three for education up to 18 years of age, and level four for higher education (Nicolay & Poncelet, 2015). In our analyses, mother’s and father’s education were considered.
Musical training
Information concerning the number of years playing a musical instrument and the amount of weekly instrument practice was collected via a parental questionnaire (see Appendix). Regarding the years playing a musical instrument, parents were asked about the number of years their child had been involved in music classes. To gather information on the amount of weekly instrument practice, parents were asked to indicate the average amount their child practised at home in minutes for each day of a regular week. In addition, parents had to indicate the average weekly duration of their child’s individual music classes and group classes in minutes. These outcomes were summed and considered the amount of weekly instrument practice. If a child played two musical instruments, the practice time and time spent in music classes for the two instruments were summed. The average amount of time spent on weekly training in minutes was calculated for each child and used for the analyses.
Raven’s Coloured Progressive Matrices
Raven’s Coloured Progressive Matrices (Raven, Court, & Raven, 1998) were used to measure non-verbal intelligence. On each trial, the participants had to select one of six pieces to complete an image that was missing one piece. Only one correct answer was possible for each trial. For every correct trial, the participants were given one point. The maximum score was 36. For the analysis, raw scores were used.
Peabody Picture Vocabulary Test-III-NL (PPVT-III-NL)
The PPVT-III-NL (Dunn & Dunn, 2005), a Dutch adaptation of the Peabody Picture Vocabulary Test (Dunn & Dunn, 1981), was used to measure verbal intelligence. The children were asked to indicate the correct image that corresponded to a word read out loud by the examiner. Only one image was correct in every trial. Raw scores were used for the analysis. The maximum score was 204.
Design and procedure
Children were assessed in a single session lasting approximately 30 minutes by the same examiner. Musicians and non-musicians were tested in a quiet room in their respective schools. The order of the tasks was the Simon task (Simon & Rudell, 1967), followed by Raven’s Coloured Progressive matrices (Raven et al., 1998) and the Peabody Picture Vocabulary Test III (PPVT III-NL–Dutch version (Dunn & Dunn, 2005). All sessions were administered in Dutch, which was the mother tongue of the children and the examiner. Children received a sweet after every completed task.
Statistical analyses
Mean RTs and error rates for the congruent and incongruent trials in the Simon task were analysed for each participant as the dependent variables. For the RT analysis, only trials with correct responses were considered. RTs below 150 ms or above 2,000 ms were excluded from the analyses (.05% of the data) as outliers. The main analyses were performed by a mixed-design 2 × 2 analysis of variance (ANOVA), in which congruency was the within-subject variable (congruent vs. non-congruent trials) and group was the between-subject variable (music vs. control group). To analyse possible group differences in background information, independent t-tests were used for continuous data and chi-square tests were performed for categorical data.
Results
Demographic and control variables
Table 1 shows the demographic variables of each group. There were no significant differences between both groups in age, mother’s education, father’s education and verbal intelligence (all p values > .23). In fact, in both groups, all the mothers had the highest level of education. However, although small and not statistically significant, the music group tended to have a higher mean score for non-verbal intelligence (Raven score: 31.5) than controls (30.1), t(59) = 1.77, p = .08. This difference in non-verbal intelligence will be considered in subsequent analyses. The composition of both groups in terms of gender did not differ significantly, χ2(1) = 0.89, p > .43. As expected, the average amount of years receiving music training (5.67 years) and the weekly practice (240 minutes) were significantly greater in the music group compared to the control group (p < .001 for both).
Background variables by group.
Note. Y = years; Mother and father education consisted of 4 levels; min. = minutes; PPVT III: Peabody Picture Vocabulary Test-III-NL.; max. = maximum.
Cognitive inhibitory control
Figure 1 shows the mean RTs for both groups as a function of congruency. Table 2 describes the mean RTs and error rates for both groups on both congruent and incongruent trials.
Mean RTs in congruent and incongruent trials for the music and the control group. Error bars indicate the SE of the mean.
Means and standard deviations (SD) of the Reaction Times (ms) and Error Rates (%) in congruent and incongruent conditions in both groups.
The analysis of the RTs revealed a robust congruency effect, F(1, 59) = 78.94, p < .001. Moreover, a significant congruency × group interaction, F(1, 59) = 6.74, p < .02, was observed (this main result remained significant when statistically controlling for non-verbal intelligence). Simple effects analyses revealed a significant congruency effect in both groups. However, the congruency effect was considerably smaller in the music group (18.5 ms), F(1, 29) = 15.1, p < .001, η2 p = .34, compared to the control group (33.8 ms), F(1, 30) = 92.6, p < .001, η2 p = .75. Indeed, an independent t-test showed that the congruency effect in the musicians was significantly smaller than the congruency effect in the controls, t(59) = 2.6, p = .01. Finally, the effect sizes for congruency, congruency by group and group were: η2 p = .57, η2 p = .10, η2 p = .00, respectively.
The analysis of the error rates again revealed a significant main effect of congruency, F(1, 59) = 26.57, p < .001, with more errors in incongruent compared to congruent trials (5.2% vs. 3.0%). The main effect of group, F(1, 59) = .54, p = .46, and the group × congruency interaction effect, F(1, 59) = .89, p = .35, were not significant.
Discussion
The present study aimed to examine previous inconsistencies concerning cognitive inhibitory control in musically trained children, with a strong emphasis on methodological issues such as participant inclusion criteria and confounders. Among all aspects of EF, we focused on cognitive inhibitory control because it appeared to show inconsistent findings between some studies using age groups similar to those in the present study – 9–12 years old (e.g., Degé et al., 2011; Zuk et al., 2014). To measure inhibitory control, the Simon task was used. A robust congruency effect appeared in RTs and error rates in both groups, validating the test. Importantly, a significant congruency by group interaction emerged for RTs, which revealed that the magnitude of the congruency effect was significantly larger in the control group compared to the music group. This finding supports the presence of better cognitive inhibitory control in children with music education compared to controls. These enhanced inhibitory control skills in musically trained children might be explained by several elements related to music training. For instance, playing a musical instrument requires high levels of selective attention. Hence, children need to focus their attention on playing music while ignoring other distracting elements. Furthermore, in the Suzuki method, children often play in groups. Thus, they might need to “ignore” the sometimes different and possibly distracting parts played by the other musicians (e.g., the pianist accompanying the violinists) to focus on their own part.
Our results confirm previous findings suggesting that musical training seems to be associated with enhanced cognitive inhibitory control (Bialystok & DePape, 2009; Degé et al., 2011; Moreno, Bialystok, et al., 2011). For instance, Bialystok and DePape (2009) showed that adult musicians outperformed non-musicians on the Simon arrow task. In addition, Degé et al. (2011) found similar cognitive inhibitory control advantages in musically trained children using the inhibition subtest of the NEPSY II (Korkman et al., 2007). However, other studies reported no enhanced inhibitory control in musicians compared to non-musicians (Schellenberg, 2011; Zuk et al., 2014). This inconsistency might be due to several factors. First of all, the rather small sample size (n = 15 musically trained children) in Zuk et al. (2014) might explain the lack of findings in their study. Another possible explanation could be the appropriateness of the chosen EF task for the age of the participants. For example, Bialystok (2011) noted that the Sun-Moon Stroop task, used in Schellenberg (2011), is not suitable for children between 9 and 12 years old and seems to be designed for younger children. However, the inconsistency in the results might also be due to differences between the tasks measuring inhibitory control (e.g., visual stimuli in the Simon task vs. verbal stimuli in the Stroop task). Furthermore, the inclusion criteria regarding the amount and intensity of music training and the age of music commencement might also differ between studies and could account for this inconsistency. In Schellenberg (2011), there was an average training of about two years in children between 9 and 10 years old and an average training of about three years in children aged between 11 and 12 years old. In our study, however, children started at a younger age, with an average of five years of training via an early childhood music education method. As Miendlarzewska and Trost (2014) mention, both age of commencement and training duration seem to influence the relationship between music training and cognitive development. Hence, it could be that the effects of music training on EF were not visible (yet) in Schellenberg’s study due to the insufficient duration of music training or due to the later age of commencement.
The limitations of our study are its cross-sectional design, which makes it impossible to draw conclusions about the causality of the advantage in EF in musically trained children. Although we tried to account for important confounders, such as SES and IQ, it is equally possible that children who had better pre-existing EF are more inclined to take and pursue music training. In addition, other variables, such as the personality of the child and of the parents, may have influenced the likelihood of enrolling and staying in music training (Corrigal & Schellenberg, 2015; Corrigall, Schellenberg, & Misura, 2013). Furthermore, the observed effects in our sample could be related to the method of the early childhood education itself and not to the musical training per se. For instance, as mentioned above, pursuing music classes at an early age, especially via the Suzuki method, requires considerable parental involvement to succeed: parents are asked to practise music with the child at home, preferably on a daily basis. Hence, it could also be possible that the results are linked to the type of child–parent interaction, which might differ when children are following these early childhood methods. One previous study has indeed shown that the quality of parenting also influences EF (Bernier, Carlson, & Whipple, 2010). Especially when assessing the effects of early childhood music education on EF, in which parental involvement is important and even required, future studies could account for the quality of parenting. In addition, most of the studies assessing the association between music training and EF have examined EF in musicians playing Western classical music. It could be interesting to examine the association between music training and EF in musicians of other cultures and in musicians playing other musical styles as well. Finally, part of the process of learning music involves a shift from conscious effort to a more automatic and implicit way of playing music. Implicit learning has been related to cognitive control (Deroost, Vandenbossche, Zeischka, Coomans, & Soetens, 2012). Hence, future studies could assess the role of implicit learning in explaining any differences in EF between musicians and non-musicians. In summary, our results indicate that musical training is associated with enhanced EF and, in particular, with enhanced inhibitory control, at least when measuring EF with the Simon task. This advantage in EF linked to music training is significant given its importance in academic achievement (van der Sluis, de Jong, & van der Leij, 2007) and other key life domains (e.g., job success; Prince et al., 2007).
Footnotes
Appendix: Parental questionnaire
Acknowledgements
The authors would like to thank Professor Emeritus Eric Soetens for his help and advice. In addition, the authors are grateful to the children, schools and parents who participated in this study.
Funding
This study was funded by a grant from the Research Foundation Flanders (FWO), which was awarded to Marie-Eve Joret (No.11M0115N).