Abstract
Although individual differences in engagement with and response to music are well documented, little is known about variations in musical empathizing and systemizing (E-S) traits and their relation to musical sophistication, including musical training. The current study examines the relationship between musical and general (non-musical) E-S traits and how musical sophistication and specific aspects of musical training are related to musical E-S traits. A total of 81 respondents reported on their level of musical sophistication and training (e.g., musical abilities, formal training, and engagement in musical activities) and endorsement of musical and general E-S traits. Participants were asked to complete the Goldsmiths Musical Sophistication Index (musical sophistication and training), the Empathizing and Systemizing quotients (general, non-musical E-S traits), and the Musical Empathizing and Systemizing inventory (musical E-S traits). Results suggest that general E-S traits are related to musical E-S traits and that musical sophistication, including but not limited to formal training, is positively associated with musical E-S traits. Furthermore, greater music training, as measured by the number of instruments played and years of formal instrumental and theory training, is related to greater endorsement of E-S traits. This study provides grounds for assessing the link between musical sophistication and training and musical E-S traits within clinical populations that have atypicalities in empathizing (e.g., autism spectrum disorder).
A primary interest for the study of music psychology is the way individuals experience and appreciate music in everyday life (Gabrielsson & Lindstrom Wik, 2003). This complex phenomenon is commonly explored by examining the influence of individual differences on music listening, such as gender (e.g., Kreutz, Schubert, & Mitchell, 2008; Nater, Abbruzzese, Krebs, & Ehlert, 2006), culture (Demorest et al., 2009), mood (Vuoskoski & Eerola, 2011), personality (Vuoskoski, Thompson, McIlwain, & Eerola, 2012), musical preferences (Kreutz, Ott, Teichmann, Osawa, & Vaitl, 2008), and musical expertise (Baraff & Coley, 2003; Gromko, 1993). Kreutz, Schubert, and Mitchell (2008) identified specific musical traits in the general population that affect music listening and that align with the perspective of Baron-Cohen’s (2003) empathizing-systemizing (E-S) theory. The current study assesses the relationship between musical and general (non-musical) E-S traits, as well as the relationship between individual differences in musical E-S traits and musical sophistication and training.
The E-S theory aims to account for the behaviours of individuals with autism spectrum disorder and behavioural differences between men and women by distinguishing two fundamental cognitive and behavioural traits: empathizing and systemizing (Baron-Cohen, 2003, 2010), herein referred to as general (non-musical) empathizing and systemizing. Empathizing is the drive to identify and respond accordingly to the thoughts and emotions of others, while systemizing is the drive to analyze and respond to systemic patterns in objects and events (Baron-Cohen, 2003, 2010). For the purpose of this research, we will focus solely on traits, which reflects a consistent and stable preponderance towards behaving or responding to situations in a certain way, which is different from state, where responding may vary across contexts and time (Cupach & Spitzberg, 1983). Generally, empathizing and systemizing traits are considered independent of each other (Wakabayashi et al., 2006; Baron-Cohen, 2010), however there is evidence to suggest that these traits may be inversely related among males and unrelated in females (Valla et al., 2010).
Individual differences in tendency to empathize and systemize are measured by the Empathizing and Systemizing Quotients, respectively (EQ: Baron-Cohen & Wheelwright, 2004; SQ: Baron-Cohen, Richler, Bisarya, Gurunathan, & Wheelwright, 2003). Together, scores on the EQ and SQ quantify different cognitive styles (Baron-Cohen, 2010) that are associated with psychological sex differences. People who score higher on the EQ compared to the SQ (“Type E”) tend to perform better on emotion recognition tasks such as the “Reading the Mind in the Eyes” test (Chapman et al., 2006). Females generally have a stronger drive to empathize as evidenced by their high performance on tasks that require the recognition of subtle facial expressions of emotion (Hoffmann, Kessler, Eppel, Rukavina, & Traue, 2010), verbal abilities (Hyde & Linn, 1988), and social sensitivity (Christov-Moore et al., 2014). People who score higher on the SQ compared to EQ (“Type S”) tend to demonstrate talent and interest in the fields of mathematics, technology, science, and engineering (Billington, Baron-Cohen, & Wheelwright, 2007; Wei, Yu, Shattuck, McCracken, & Blackorby, 2013). “Type S” cognitive style is associated with males, who are more proficient at tasks where systemizing skills are favoured, such as mental rotation (Shepard & Metzler, 1971) and map reading (Kimura, 2000). An extreme version of this cognitive style (Extreme Type S) is characteristic of people with autism spectrum disorder (Baron-Cohen, 2002), who demonstrate lower levels of empathizing skills, such as recognizing the mental states of others and identifying emotions from facial expressions (Baron-Cohen, Leslie, & Frith, 1985; Harms, Martin, & Wallace, 2010), combined with intact or even superior interest in mechanical or numerical systems (Baron-Cohen & Wheelwright, 1999). However, people with autism spectrum disorder do not present with deficits in perceiving music-evoked emotions (Heaton, Hermelin, & Pring, 1999; Heaton, Allen, Williams, Cummins, & Happé, 2008; Quintin, Bhatara, Poissant, Fombonne, & Levitin, 2011).
Well-grounded in empirical evidence, the E-S theory provides a solid framework in which to study musical phenomena. People likely empathize and systemize when listening to or performing music. As such, music is a unique domain in which empathizing and systemizing traits can be simultaneously assessed. Specifically, people may empathize when perceiving musical information and responding affectively and physiologically to it (Egermann & McAdams, 2013; Rabinowitch, Cross, & Burnard, 2013) and may systemize when analyzing musical elements and structure (e.g., tempo, melody, harmony) to understand how they relate to the whole musical piece (Greenberg, Rentfrow, & Baron-Cohen, 2015; Vuoskoski, 2015). Kreutz, Schubert, and Mitchell (2008) developed a questionnaire called the Musical Empathizing-Systemizing (ME-MS) Inventory to investigate empathizing and systemizing as it specifically relates to music, rather than using the EQ-SQ which assesses general (non-musical) E-S traits in separate domains. Their findings revealed that general E-S traits have musical equivalents, musical empathizing and musical systemizing, which are related to variations in music appreciation. Individuals high on music empathizing (ME) were found to have a drive for responding to the emotional content of a musical piece and for attending to the emotions of the composer or performer, while individuals high on music systemizing (MS) were found to have an interest in understanding the mechanics of musical instruments and the structure and organization of a musical piece. Poli, Canazza, Roda, and Schubert (2015) reported that ME and MS profiles aligned with judgments of musical expression. Specifically, adults endorsing ME traits were more likely to judge musical excerpts based on emotional aspects while those endorsing MS traits based their judgments according to structural components of the musical excerpts. In addition, Garrido, Schubert, Kreutz, and Halpern (2011) and Greenberg, Baron-Cohen, Stillwell, Kosinski, and Rentfrow (2015) more recently found that musical E-S traits differentially influenced musical preferences, such that individuals reporting higher ME traits preferred mellow genres like R&B while those with greater MS traits preferred more intense genres like rock and techno. Thus, it appears that ME and MS traits are associated with other traits.
In addition to the limited research that examines the relationship between musical E-S traits and music listening, there is even less, if any, research on factors that are correlated with musical E-S traits within the general population. Kreutz, Schubert, and Mitchell (2008) did report that musical expertise has a positive association on musical systemizing traits, but not on empathizing; this is surprising given that both systemizing and empathizing are likely involved in the processing of music. Kreutz, Schubert, and Mitchell (2008) did not use a multi-faceted measure of musical expertise and simply asked participants to categorize their level of musicianship (e.g., professional vs amateur). Thus, it is possible that this methodological procedure was not sensitive enough to catch the association between ME and musical expertise.
There is mounting evidence for the influence of musical sophistication, operationalized as the skills and behaviours that include but are not limited to musical training (Levitin, 2012; Müllensiefen, Gingras, Musil, & Stewart, 2014a), on the development of music ability and for the transference of such skills to other cognitive domains. This nuanced definition allows for the consideration of people who are immersed within the musical domain, yet did not receive formal musical training. Nonetheless, musical training has been associated with improvements in the memory for musical rhythms, pitches, and temporal intervals (Aagten-Murphy, Cappagli, & Burr, 2014; Gaab & Schlaug, 2003; Schaal, Banissy, & Lange, 2015), pitch discrimination (Kishon-Rabin, Amir, Vexler, & Zaltz, 2001), melodic contour perception (Fujioka, Trainor, Ross, Kakigi, & Pantev, 2004), reading musical notation, improvisation, and fine motor synchronization (Schlaug, Norton, Overy, & Winner, 2005). People with greater musical training demonstrate increased proficiency in perceiving music-evoked emotions in pitches and in musical performances (Besson, Schön, Moreno, Santos, & Magne, 2007; Bhatara, Tirovolas, Duan, Levy, & Levitin, 2011; Castro & Lima, 2014), and show stronger affective responses to music (Blood & Zatorre, 2001; Kantor-Martynuska & Horabik, 2015), particularly for negative emotions like sad and fearful (Park et al., 2014), and appear to experience greater musically induced chills (Panksepp, 1995).
Moreover, advantages of musical training on music-related ability are shown to cascade onto cognitive and emotional functions outside of the music domain, such as intellectual functioning (Jaschke, Eggermont, Honing, & Scherder, 2013; Schellenberg, 2004, 2006), attention (Roden et al., 2014; Wang, Ossher, & Reuter-Lorenz, 2015), motor and tactile abilities (Costa-Giomi, 2005), verbal and working memory (George & Coch, 2011; Taylor & Dewhurst, 2017; Zuk, Benjamin, Kenyon, & Gaab, 2014), auditory and visual perception (Anaya, Pisoni, & Kronenberger, 2016; Weijkamp & Sadakata, 2017), metacognition (Bathgate, Sims-Knight, & Schunn, 2012), nonverbal reasoning (Forgeard, Winner, Norton, & Schlaug, 2008) and speech and language processing (Kraus et al., 2014; Moreno, 2009; Moreno & Besson, 2006). Benefits of musical training also include increases in emotional empathy (Rabinowitch et al., 2013) and greater proficiency in recognizing emotions in speech prosody (Lima & Castro, 2011).
Given the evidence for the association between musical training and musical and non-musical abilities that likely involve systemizing or empathizing, we aim to investigate the relationship between music sophistication and training and musical E-S traits, using the Goldsmiths Musical Sophistication Index (Gold-MSI; Müllensiefen, Gingras, Musil, & Stewart, 2014b) and ME-MS inventories, respectively. Our specific objectives are to i) examine the relationship between musical and general (non-musical) E-S traits; and assess the relationship between musical E-S traits and ii) musical sophistication, iii) overall level of musical training, and iv) specific aspects of musical training (i.e., number of instruments played, years of formal instrumental training, and years of formal theory training). We hypothesized that 1) musical (ME-MS inventory) and general E-S (EQ and SQ-R inventories) traits would be positively correlated as per Kreutz, Schubert, and Mitchell (2008), 2) greater musical sophistication (Gold-MSI inventory) would be related to greater endorsement of musical E-S traits, 3) individuals with more musical training (Musical Training subscale of the Gold-MSI inventory) would have higher musical E-S traits, and 4) more instruments played and more years of formal instrumental and theoretical training (3 items from the Musical Training subscale of the Gold-MSI inventory) would be related to greater levels of musical E-S traits.
Method
Participants
A total of 81 respondents (52 females; 29 males; Mage = 21.33 years, SD = 3.34 years, range = 18 to 33 years) were included in this survey, which was part of a larger study. Participants were recruited from McGill University and the general Montreal area through official channels, social media, or word of mouth. Exclusionary criteria included frequent use of recreational drugs and/or use of prescription medication (e.g., antidepressants). Aside from 4 participants, all respondents were current students with 78% in an undergraduate program (n = 63) and 17% in a graduate program (n = 14). Of the 81 participants, only 11% indicated music as an area of study (other fields of study included physics, biology, pharmacology, kinesiology, dietetics, education, linguistics, history, psychology, anthropology, philosophy, sociology, business, engineering, architecture, computer science, political science, and law). In reporting their level of musicianship, 3.7% indicated a “professional” status, 8.6% “semi-professional”, 38.3% “amateur”, 14.8% “occasional”, 32.1% “hardly ever play/sing”, and 2.5% “other”. One participant was a multivariate outlier and was excluded from analyses (original sample: N = 82).
Measures
General E-S traits
The Empathizing Quotient (EQ; Baron-Cohen & Wheelwright, 2004) and the Systemizing Quotient-revised (SQ-R; Wheelwright et al., 2006) were used as valid and reliable measures of general empathizing and systemizing traits, respectively. The EQ consists of 40 items (e.g. “I am good at predicting how someone will feel”), and the SQ-R consists of 75 items (e.g. “I am fascinated by how machines work”). Participants rated individual items on both quotients according to a 4-point Likert scale with the following possible choices: “strongly agree”, “slightly agree”, “slightly disagree”, “strongly disagree”. Items that were rated strongly were assigned 2 points and those that were rated slightly were given 1 point. As such, the total score for the EQ was out of 80 and the score for SQ-R was out of 150.
Musical E-S traits
The Musical Empathizing and Systemizing inventory (ME-MS; Kreutz, Schubert, & Mitchell, 2008), adapted from the short forms of the EQ and SQ (Wakabayashi et al., 2006), was used to examine “musical empathizing (ME)” and “musical systemizing (MS)”. The inventory is comprised of two scales (ME and MS) with 9 items each for a total of 18 items. Participants rate individual items according to a 4-point Likert scale from 3 (strongly agree) to −3 (strongly disagree), with middle responses (agree/disagree) scored as 1 or −1. The items from the ME and MS scales were presented in alternating order.
The negatively stated items for ME (e.g., ME#3: “I never guess the emotions of the performer(s).”) and MS (e.g., MS#12: “I do not find it interesting how music is created from different parts.”) scores were recoded by multiplying a simplified unit weighting score of −1. For example, a score of 3 (“strongly agree”) for item ME#3 was recoded to a score of −3 (“strongly disagree”). As described by Kreutz, Schubert, and Mitchell (2008), scores (including those recoded) were then summed and subtracted by 13 for the ME scale and 10 for the MS scale (see scoring description in Kreutz, Schubert, & Mitchell, 2008, p. 61). ME scores have a possible range from −40 to 14 and ME scores range from −37 to 17, with higher scores indicating higher self-reported ME and MS and lower scores indicating lower ME and MS.
Musical sophistication
Consistent with current research on musical abilities (Vromans & Postma-Nilsenová, 2016), the Goldsmiths Musical Sophistication Index (Gold-MSI; Müllensiefen et al., 2014b) was used to assess self-reported levels of musical abilities, expertise, and sophistication of musical behaviours. The Gold-MSI is comprised of five subscales: Active Engagement (involvement in musical activities, e.g., “I spend a lot of my free time doing music-related activities”), Perceptual Abilities (musical listening skills, e.g., “I can tell when people sing or play out of tune.”), Musical Training (history of formal training, e.g., “I engaged in regular, daily practice of musical instrument (including voice) for __ years”), Singing Abilities (skills related to singing performance, e.g., “I am able to hit the right notes when I sing along with a recording.”), and Emotions (emotional responses to music, e.g., “I sometimes choose music that can trigger shivers down my spine.”).
The Active Engagement and Perceptual Abilities subscales consist of 9 items and each have a maximum total score of 63. The Musical Training and Singing Abilities subscales consist of 7 items for a maximum total score of 49, and the Singing Abilities subscale includes 6 items with a maximum total score of 42. A general measure of Musical Sophistication is also derived by summing scores from some items from each scale.
All items for each scale were scored according to a 7-point Likert scale. Responses ranged from “completely agree” to “completely disagree”, except for items that asked participants to select one of seven options from a drop-down menu that best describes their level of musicianship. Responses to these items were re-scored onto a 7-point Likert scale in order to generate the Musical Training total score. For the present study, 3 items from the Musical Training subscale were used to characterize musicianship and explore hypothesis 4 (response options are presented in brackets): 1) I can play [0; 1; 2; 3; 4; 5; 6 or more] musical instruments; 2) I have had [0; 0.5; 1, 2, 3–5, 6–9; 10 or more] years of formal training on a musical instrument; 3) I have had [0; 0.5;1; 2; 3; 4–6; or 7+] years of formal training in music theory.
Procedure
Respondents were invited to participate by email containing a link that directed them to an online survey platform (RedCap; Harris et al., 2009). They filled out a consent form and completed a demographic questionnaire that assessed their gender, age, nationality, language fluency, level of education, income, handedness, and if they took medication. Participants were then asked to complete the questionnaires described above (see Measures), and additional measures that were part of a larger study. Participants received $20 as compensation upon completion of the larger study. Ethical approval to conduct this study was granted from McGill University’s Research Ethics Board.
Statistical analysis
Planned analyses
Hypotheses #1 and #2 were explored through Pearson’s correlation coefficients. For hypothesis #1, we examined the relationship between musical E-S and general (non-musical) E-S traits using EQ and SQ-R scores (E-S Quotients) and ME and MS scores (ME-MS inventory). A Bonferroni adjusted alpha level of .013 (α = .05/4) was used to account for multiple comparisons. For hypothesis #2, we examined the relationship between musical E-S traits and musical sophistication using ME and MS scores and the Gold-MSI general and subscale scores (Gold-MSI inventory). A Bonferroni adjusted alpha level of .004 (α = .05/12) was used. An a priori power analysis using the software G*Power (Faul, Erdfelder, Buchner, & Lang, 2013) indicated that a total sample of 40 people would be needed to detect large effects (r = .5) with 80% power using Pearson coefficient correlations with alpha at .013 and a sample of 49 with alpha at .004.
Hypotheses #3 and #4 were explored using four one-way multivariate analyses of variance (MANOVAs). The participants were divided into groups in the analyses that follow, thus the results are descriptive as there was no true experimental manipulation. For hypothesis #3, we examined the relationship between musical E-S traits and level of musical training. The participants were divided into low and high musical training groups using the median split of the Gold-MSI Musical Training subscale scores. For hypothesis #4, we examined the relationship between specific aspects of musical training (i.e., number of instruments played, years of formal instrumental training, and years of formal music theory training) and musical E-S traits. The participants were divided into two or three groups using scores from 3 items of the Gold-MSI Musical Training subscale. ME and MS scores were the dependent variables across the analyses. Bonferroni adjusted alpha levels of .025 (α = .05/2) were used to account for multiple ANOVAs being run. An a priori power analysis for a MANOVA with two or three levels, and two dependent variables using an alpha level of .025, a power of 1-β = .80, and a large effect size (f = .40) revealed a total sample size of N = 62 and 78, respectively.
Test assumptions
Assumptions for parametric correlations were met. Bivariate scatterplots showed linearity between key pairs of variables. Variables were normally distributed except for slight skewness in the distribution of ME and MS scores: ME: skewness = −.854, SE = .267, kurtosis = .847, SE = .529; MS: skewness = −.694, SE = .267, kurtosis = .638, SE = .529 as well as slight skewness and kurtosis in the distribution of Gold-MSI Emotions and Musical Training subscale scores: Emotions: skewness = .685, SE = .267, kurtosis = 1.156, SE = .529; Musical Training: skewness = .053, SE = .267, kurtosis = −1.118, SE = .529. However, visual examination of pp-plots and boxplots demonstrated that the asymmetry in the distribution was minor and not problematic. Pearson correlations between the two dependent variables (ME and MS) within each independent group included in the MANOVAs were moderately correlated with each other (r = .40 to .62, p <.05) and other assumptions for the MANOVAs were also met (Box’s M test: p > .05, except for hypothesis #3 due to unequal variance across groups for MS, which was accounted for with an independent t-test with Levene’s correction to supplement the MANOVA).
Results
Musical E-S and general (non-musical) E-S traits
Hypothesis #1 was assessed by calculating Pearson’s correlation coefficients between EQ and SQ-R scores (EQ and SQ-R inventories) and ME and MS scores (ME-MS inventory). A significant moderate positive association between ME and EQ scores, r = .28, p = .011, and a trending significance between MS and SQ-R scores, r = .27, p = .014 was found (with Bonferroni adjusted α = .013) with small to moderate effect sizes, suggesting that people report similar levels of musical and general empathizing, which may also be the case for musical and general systemizing. Correlations between MS and EQ scores, r = .06, p = .614, and ME and SQ-R scores, r = −.03, p = .771, were not significant (see Table 1). Although not related to our hypothesis, we calculated Pearson’s correlation coefficients between ME and MS which were highly correlated and EQ and SQ-R which were not correlated (see Table 1).
Correlation matrix showing Pearson correlation coefficients between musical and general (non-musical) E-S traits.
Note. BCa bootstrap 95% CIs for r reported in brackets.
significant and
Musical E-S and musical sophistication
Hypothesis #2 was assessed by calculating Pearson’s correlation coefficients between ME and MS and the Gold-MSI general and subscale scores. ME and MS were positively associated to the General Musical Sophistication scale, and the Active Engagement, Emotions, Musical Training, Perceptual Abilities, and Singing Abilities subscales (see Table 2), with moderate to large effect sizes. Although not related to our hypothesis, we calculated the correlations between EQ and SQ-R and the Gold-MSI subscales and found that EQ was positively associated with several Gold-MSI subscales, but SQ-R was not (see Table 2).
Pearson’s correlation coefficients between musical sophistication and musical and general (non-musical) E-S traits.
Note. BCa bootstrap 95% CIs for r reported in brackets.
p < .05**p < .01, ***p < .001.
Musical E-S traits and overall musical training
The Musical Training subscale score of the Gold-MSI was used to divide the participants into a low or high overall musical training groups to investigate hypothesis #3. Given that the median score for the Musical Training subscale was 26 (out of a maximum possible of 49), the participants with scores equal to or less than 26 were classified in the low musical training group (M =17.11, SD = 6.52), and those with scores greater than 26 were classified in the high musical training group (M = 35.92, SD = 5.60; see Table 3).
Demographic information for overall musical training, number of instruments played, years of instrumental training, and years of music theory training groups.
A MANOVA revealed a significant effect of overall musical training on Musical E-S traits, Pillai’s trace = .123, F(2, 78) = 5.49, p = .006, ηp2 = .12, observed power = .84. A significant difference was found between participants in the low and high musical training groups in ME scores, F(1, 79) = 5.35, p = .023, ηp2 = .06, observed power = .63, with a small to moderate effect size, and in MS scores, F(1, 79) = 10.65, p = .002, ηp2 = .12, observed power = .90, with a moderate to large effect size. The participants in the high musical training group reported having higher ME traits (M = .65, SD = 9.11) and MS traits (M = 1.65, SD = 6.99) than those in the low musical training group (ME: M = −3.95, SD = 8.79; MS: M = −5.44, SD = 11.88; see Figure 1). The finding with MS was replicated with an independent t-test with Levene’s correction for unequal variance across groups (MS: t(65) = −3.28, p = .002).
Average scores for ME and MS participants in the low and high musical training groups. The high musical training group reported having higher musical E-S traits than the low musical training group. Error bars represent standard errors.
Musical E-S traits and specific aspects of musical training
The participants were divided into groups based on their answers to the 3 items of the Gold-MSI to investigate hypothesis #4.
Musical E-S traits and number of instruments played
Of the participants, 25.9% did not play any musical instruments (including voice), 24.7% played 1 instrument, 21% played 2 instruments, 11.1% played 3 instruments, 0% played 4 instruments, 8.6% played 5 instruments, and 8.6% played 6 or more instruments. Since participants were unevenly distributed across the categories, participants were divided into two groups: those who played 0 or 1 instruments and those who played 2 or more (see Table 3).
A MANOVA revealed a significant effect of number of musical instruments played on musical E-S traits, Pillai’s trace = .158, F(2, 78) = 7.33, p = .001, ηp2 = .16, observed power = .93. A significant difference was found between participants who played 0 or 1 instrument and those who played 2 or more instruments in ME scores, F(2, 78) = 8.80, p = .004, ηp2 = .10, observed power = .83, and MS scores, F(2, 78) = 13.19, p < .001, ηp2 = .14, observed power = .95, with moderate to large effect sizes. Participants who played 2 or more instruments reported having higher ME traits (M = 1.25, SD = 7.49) and MS traits (M = 2.00, SD = 7.90) than those who played 0 or 1 instrument (ME: M = −4.54, SD = 9.87; MS: M = −5.78, SD = 11.08; see Figure 2).
Average scores for ME and MS participants in the 0 to 1 instrument and 2 or more instruments groups. Participants who played 2 or more instruments reported having higher musical E-S traits than participants who played 0 to 1 instrument. Error bars represent standard errors.
Musical E-S traits and years of formal instrumental training
In our sample, 24.7% had no years, 2.5% had 0.5 years, 3.7% had 1 year, 6.2% had 2 years, 30.9% had 3 to 5 years, 19.8% had 6 to 9 years, and 12.3% had 10 or more years of instrumental training. To have more evenly distributed groups, the following categories were created: 0 to 1 year, 2 to 5 years, and 6 or more years of formal instrumental training (including voice; see Table 3).
A MANOVA revealed a significant effect of years of formal instrumental training on musical E-S traits, Pillai’s trace = .151, F(4, 156) = 3.18, p = .015, ηp2 = .08, observed power = .82. Years of formal instrumental training significantly influenced ME scores, F(2, 78) = 4.27, p = .017, ηp2 = .10, observed power = .73, and MS scores, F(2, 78) = 4.85, p = .01, ηp2 = .11, observed power = .79, with moderate to large effect sizes. Gabriel’s post hoc tests were used since the sample sizes between the groups were slightly uneven. Participants with 6 or more years of instrumental training reported having higher ME and MS traits (ME: M = 1.31, SD = 7.59, p = .016; MS: M = 2.77, SD = 6.96; p = .01) than those with 0 to 1 year (ME: M = −5.76, SD = 8.86; MS: M = −5.64, SD = 11.28). Participants with 2 to 5 years did not significantly differ in ME and MS scores (ME: M = −.87, SD = 9.78, MS: M = −2.93, SD = 10.76) to those with 0 to 1 year and those with 6 or more years of training (p > .05; see Figure 3).
Average scores for ME and MS participants in the 0 to 1 year, 2 to 5 years, and 6 or more years groups. Participants with 6 or more years of instrumental training reported having higher musical E-S traits than participants with 0 to 1 year. Error bars represent standard errors.
Musical E-S traits and years of formal music theory training
Of the participants, 39.5% did not receive any formal music theory training, 8.6% had 0.5 years, 7.4% had 1 year of training, 9.9% had 2 years, 8.6% had 3 years, 19.8% had 4 to 6 years, and 6.2% had 7 or more years of training. To address the uneven distribution of groups, participants were categorized into the following groups: 0 years, 0.5 to 3 years, and 4 or more years of experience (see Table 3).
A MANOVA revealed a significant effect of years of formal music theory training on musical E-S traits, Pillai’s trace = .163, F(4, 156) = 3.46, p = .01, ηp2 = .08, observed power = .85. Years of formal music theory training significantly influenced MS scores, F(2, 78) = 5.40, p = .006, ηp2 = .12, observed power = .83, with a moderate to large effect size, but not ME scores, F(2, 78) = 3.79, p = .027 (Bonferroni adjusted alpha level of .025). Gabriel’s post hoc tests were used. Participants with 4 or more years of formal music theory training reported having higher MS traits (M = 2.14, SD = 8.01, p = .009) than individuals with 0 years of formal training (M = −6.25, SD = 12.42). MS scores did not significantly differ between individuals with 0 years and 0.5 to 3 years of formal training, although this difference trended towards significance (p = .051), nor between individuals with 0.5 to 3 years and those with 4 or more years (p > .05; see Figure 4).
Average scores for ME and MS participants in the 0 years, 0.5 to 3 years, and 4 or more years groups. Participants with 4 or more years of theory training reported having higher MS traits than participants with 0 years. Differences in ME scores across training groups were non-significant after correcting for multiple comparisons (p < .025). Error bars represent standard errors.
Discussion
In an attempt to explain traits associated with the perception of and responsivity to music, Kreutz, Schubert, and Mitchell (2008) extended the Empathizing-Systemizing (E-S) theory to the musical domain to form its musical equivalents: musical empathizing (ME) and musical systemizing (MS). In the present study, we corroborated that ME traits are related to general (non-musical) empathizing, as measured with the Musical Empathizing and Systemizing inventory (ME-MS; Kreutz, Schubert, & Mitchell, 2008) and Empathizing Quotient (EQ; Baron-Cohen & Wheelwright, 2004), and a trending relationship between MS traits and general (non-musical) systemizing, as measured with the ME-MS and Systemizing Quotient-revised (SQ-R; Wheelwright et al., 2006). ME traits were not related to general systemizing and MS traits were not related to general empathizing. However, ME and MS traits were positively associated, a relationship that was not previously reported by Kreutz, Schubert, and Mitchell (2008) and that differs from the unrelated general E-S traits (Wakabayashi et al., 2006; Baron-Cohen, 2010). This finding that musical empathizing and systemizing traits are related, further supports the assertion that music is a unique domain to directly examine the E-S theory (Greenberg, Rentfrow, & Baron-Cohen, 2015).
The results from this study also revealed a relationship between various aspects of musical sophistication, as measured with the Gold-MSI, and musical E-S traits. Specifically, we found that engagement in music-related activities, knowledge for and abilities in the perceptual aspects of music, singing abilities, and emotional responsiveness to music were positively associated with musical E-S traits. This finding suggests that aspects of musical expertise other than musical training play a role in influencing musical E-S traits. Further, it lends credence to the argument that musical expertise encompasses more than just technical training or abilities in playing instruments (Levitin, 2012; Müllensiefen et al., 2014a). Adopting this nuanced perspective would pave the way for future research to examine music perception and responsivity among people who are enculturated within this domain but may otherwise have no musical training in the conventional sense (e.g. avid listeners).
Nonetheless, we found that musical training was related to higher levels of endorsement of musical E-S traits, which partially supports that of Kreutz, Schubert, and Mitchell (2008) who found that musical training is related to MS, but not ME traits. Such a discrepant finding in the relation between musical training and ME traits can be attributed to the way in which Kreutz, Schubert, and Mitchell (2008) assessed level of musicianship, which consisted of one question where the participants classified themselves along a continuum that ranged from professional to musically naïve. In the current study, the musical training factor comprised of several questions that quantified specific aspects of training including, but not limited to, years of formal instrumental and music theory training, number of instruments played and amount of daily practice. Thus, responses to such questions allowed for a more comprehensive measure of musical training, which may have led to the finding that musical training is related to ME traits in our sample but not in that of Kreutz, Schubert, and Mitchell (2008).
Aside from finding that musical training is related to musical E-S traits, we also found that specific aspects of musical training, particularly the number of instruments played, years of formal instrumental and music theory training, are associated with differing levels of endorsement of ME and MS traits. Regarding the number of instruments played, we found that individuals who played two or more instruments endorsed higher levels of musical E-S traits compared to those who play no or one instrument. Likewise, individuals with six or more years of formal instrumental training also endorsed higher levels of musical E-S traits compared to those who received less than one year of training. However, years of formal music theory training revealed a slightly different story, in that individuals with four or more years of training reported higher levels of MS but not ME traits compared to those with no years of formal theory training.
Overall, these findings suggest that individuals with the most musical experience endorsed highest levels of musical E-S traits. Such a finding is reminiscent of the literature that similarly found that musical expertise is related to an increase in processing of musical structures such as pitch (Gaab & Schlaug, 2003; Kishon-Rabin et al., 2001), melodic contour (Fujioka et al., 2004), and rhythm (Schaal et al., 2015) as well as emotion perception from music (Besson et al., 2007; Castro & Lima, 2014). However, the fact that individuals who received the highest levels of music theory training did not differ in ME traits compared to those with no formal music theory training runs counter to expectations, especially given that individuals within this group were also more likely to have more years of instrumental training and played more instruments. The relationship between ME traits and music theory training thus seems to follow an inverted U-shape with highest levels of ME traits for the subgroup who received a moderate (compared to low and high) number of years of music theory training. This would suggest an optimal or critical window at which music theory training increases ME traits. Alternatively, low power in comparing groups with divergent sample sizes could account for this intriguing finding.
This study extends our understanding of the impact of musical sophistication and training on individual differences in musical E-S traits. However, it is not without its limitations. The first is the generalizability of these findings to the general population, given that the majority of our sample were undergraduate students between the ages of 18 to 33. There is evidence to suggest that age is an important individual factor in perceiving music-evoked emotions (Castro & Lima, 2014) and music listening preferences (Bonneville-Roussy & Eerola, 2017; LeBlanc, Sims, Siivola, & Obert, 1996). Thus, it is possible that the relationship between musical sophistication and musical E-S traits may be mediated by age. A larger scaled study is needed to confirm our findings in a sample with varying age groups and demographic backgrounds (i.e., nonstudents and cultural backgrounds). The second limitation is related to the uneven ratio of males to females, thereby possibly confounding our current findings as well as limiting our exploration of the impact of sex differences on the relationship between musical sophistication and musical E-S traits, especially for our exploration of the impact of formal musical instrumental and theory training where there were fewer males in the subgroups with many years of training than those with less years of training. Previous research has demonstrated that males and females process structural (Koelsch, Maess, Grossman, & Friederici, 2003) and emotional (Nater et al., 2006) aspects of music differently at a neurological and physiological level. Kreutz, Schubert, & Mitchell (2008) equally found that sex differences differentially contributed to the relationship between level of musicianship (i.e., professional versus amateur) and endorsement of musical E-S traits. Given previous findings indicating sex differences in E-S traits, further study is needed to explore the role of sex differences in the relationship between musical sophistication, different aspects of musical training, and endorsement of musical E-S traits.
Our findings based on a correlational design do not imply causality, but provide grounds for future longitudinal studies to investigate the impact of musical sophistication and training on musical E-S traits. Furthermore, it also has implications for clinical populations that have atypicalities in empathizing, such as autism spectrum disorder, Williams syndrome, and schizophrenia. Given the evidence that the benefits of musical training cascade onto other areas, specifically cognitive (Schellenberg, 2004, 2006; Taylor & Dewhurst, 2017; Wang et al., 2015) and socio-emotional functioning (Rabinowitch et al., 2013), assessing the link between musical training and musical E-S traits within these clinical populations is warranted. Further, engaging in music-related activities and interventions may hold the promise of increasing both musical and general (non-musical) empathizing abilities of people with these disorders (Greenberg, Rentfrow, & Baron-Cohen, 2015).
In conclusion, people vary in their appreciation of specific aspects of music. In the current study, we have demonstrated that such variation is associated with differences in a preponderance towards musical empathizing or systemizing. General musical expertise, that extends beyond training, such as engagement in music-related activities, is positively related to cognitive traits involved in music processing. In particular, receiving instrumental or theoretical training in music may lead to stronger musical E-S traits.
