What can you learn about a person from what music they like? Can you expect a difference in predisposition between those who prefer upbeat music versus those who like melancholic music? Do people with a taste for complex music have better cognitive skills, or is it a learned taste that results from devoting more time to the subject, or something else? Can a music streaming service make money through targeted advertising based solely on music preferences? Following from that, should we guard our musical taste closely? What might a future adversary infer about ourselves from our taste in music? Is it or will it be possible to scientifically predict which songs will be hits? If yes, can this capability be used during composition already?
I jumped the gun here by starting with the inverse (the inference) problem. The direct question is: do personality traits influence preferences for certain musical styles? This statistical framing assumes that there are hidden variables describing our personality and musical preferences are their functions, which can be observed. We will see that they do correlate.
Why do people differ so much in what music they like? More fundamentally, why do we like music? It seems most everyone likes music. Does this affection give us an evolutionary advantage, or is it just a neutral side effect of something that is advantageous? What percentage of us don’t like music? Are they a small minority, as social convention leads us to believe, or is there actually a silent majority for whom music is a nuisance? If they were a rarity, perhaps they are interesting subjects for psychological studies. How does rhythmic music suck many of us in and make us move with the beat? Shouldn’t we be better at controlling our bodies, at jamming such stimuli? Or is jamming more costly for some minds than moving with the rhythm?
It has been estimated ($) that Americans spend on average over 15% of their waking hours listening to music. What is the function of music in our lives? It is different things for different people:
- It is an external stimulus that saves us from boredom.
- It is no more than a lubricant of our social interactions, an uncontroversial convention, an ice breaker.
- At the extreme, it puts us into a different state of mind, into a trance or flow state.
- Among teens, musical taste is often a central determinant of identity and alignment. (It used to be in my time.) Why do we grow out of this? Why do we grow into it in the first place?
Some of these questions are the subject of active scientific research under the names of music psychology and music cognition. I’m going to bring you some fascinating discoveries I read about while digging into the subject.
Personality and its relationship with musical taste
First, you can get to know yourself better and help research by filling out some questionnaires. (I asked, and as of mid-November, the contributed data was still being recorded.) The study is run by a group at the University of Cambridge and PhD student David M. Greenberg in particular. The test `Musical taste’ lets you listen to 25 short musical excerpts that you score, asks about your personality and life satisfaction, and in return it scores your preference in each of five musical categories (dimensions, or, for the statistician, factors). This so-called MUSIC system is an intriguing way to think about music.
The MUSIC system transcends genres in favour of more tangible properties. Its five latent factors were suggested by findings of Rentfrow and his colleagues in 2011 ($) and 2012 ($). The categories are mellow, unpretentious, sophisticated, intense and contemporary. They can be described as follows. Mellow is like R&B and soft rock: romantic, smooth and relaxing, unaggressive, sad, slow and quiet. Unpretentious is best exemplified by country, folk and singer–songwriter genres, being sincere and rootsy, uncomplicated, acoustic. Sophisticated is complex, dynamic, inspiring, intelligent, and includes classical, operatic, avant-garde genres, world music and jazz. Intense is loud, energetic, forceful, even aggressive, often distorted, neither relaxing nor romantic, as in the genres classic rock, punk, heavy metal and power pop. Contemporary is rhythmic, percussive and electric, but not sad, such as is found in rap, electronica, Latin, funk, acid jazz and Euro pop.
This five-dimensional coordinate system can capture most of the unknown underlying structure of musical preferences. (Typical representatives are found in Table 1 of the 2011 paper ($).) Notably, alone the jazz or rock genre has enough variability to require every dimension of the MUSIC system.
The test `What’s your score on 5 musical engagement styles?’ scores how deeply music touches us on an emotional level; how analytical and cognitive our engagement is (let it be patterns, melodies, instrumentation or interplay between instruments that we focus on); the extent to which music urges us to respond by physical movement; how defining the social aspect is in our enjoyment of music through connection to both the musicians and the collective audience; and how much we are into the narrative side of music and lyrics.
Greenberg claims their work has shown that `personality and cognitive styles predict musical preferences and musical ability’, and this is exactly what we’re after. In a popular science article, he introduces Simon Baron-Cohen’s empathising–systemising (E–S) theory of thinking styles (for which the test can be found on the same site under `Brain Type/Thinking Style’). Empathisers tune into the emotions and thinking of others, systemisers have an interest in patterns, systems and governing rules. Balanced types score about equally on both aspects. Apparently, 95% of people can be classified into one of these three groups and this classification predicts a lot about their behaviour. (The remaining 5% seem to be extreme empathiser or extreme systemiser types.) Women on average have higher empathy quotient scores than men. A previous study found that women were distributed approximately 45%:29%:21% among empathiser, balanced and systemiser types, whereas men about 15%:30%:50%.
Greenberg and his co-authors found that empathisers preferred mellow music (recall: low energy, sad emotions and emotional depth—R&B, soft rock and singer–songwriter genres).1 In terms of the psychological attributes of the music they preferred, their choice featured low arousal (gentle, reflective, warm and sensual attributes), it was characterised by negative valence (it is depressing and sad) and by emotional depth (it is relaxing, poetic and thoughtful).2
On the other hand, systemisers preferred intense music (think hard rock, heavy metal or punk). The psychological attributes of their choice clearly differ from that of the empathiser type: systemisers prefer intellectual depth and complexity (as in avant-garde classical genres), high arousal (manic, strong, tense, thrilling), with aspects of positive valence (animatedness and fun). Balanced types typically had a broader range of preferred music than either of the other two types.
A significant effect of the position on the empathy scale was also found on preference for sonic attributes. Systemisers preferred acoustic features that were dense, distorted, loud, percussive, and fast in tempo. In terms of instrumental features, they preferred music that featured brass and electric guitar. Empathiser types preferred music that featured strings. No significant difference was found between the groups’ taste for heavy bass, acoustic guitar, cymbals, piano, raspy voice, woodwinds, and yelling voice.
These findings are correlational and not causational. It is not known whether the right kind of music can increase (prime) empathy. The results show that the suggestion is not entirely far-fetched. In this vein, the same research group also studies the effect of music on the listener and the promise of music therapy. Their fourth test, the Empathy and Listening Quiz compares your ability to recognise emotions on faces before and after listening to a piece of music, and reports your skill in reading the mind from the eyes. Their hypothesis is that music might increase empathy and communication skills, and they have specific recommendations for people on the autism spectrum.
In a more recent, data-driven effort, some of the same authors found ($) that music can also be naturally categorised along three dimensions. The data included 102 musical excerpts spanning 26 genres, rated on 38 perceived attributes. The organising qualities were found to be arousal (the energy or intensity level of the piece), valence (the mood or emotion, how sad or happy the music is) and depth (the complexity and sophistication, and the emotional depth of the music). Greenberg suspects that these categories are better descriptors of musical taste than genres, so liking a certain combination of them in one genre should be a good predictor of liking them in another genre, too. Streaming and recommendation platforms, music marketers, take note.
Further, self-ratings of personality traits predict musical preferences better than demographic variables. For personality classification, the Big Five personality traits paradigm was used (openness to experience, conscientiousness, extraversion, agreeableness and neuroticism/emotional stability). Nearly 9500 people were surveyed about their personalities and preferences for 50 musical excerpts. What they found was that people who ranked high on openness to experience preferred complex melodies and depth in music. Extroverted excitement-seekers preferred high arousal. While relatively neurotic people preferred negative emotions and intensity in music, self-assured (agreeable and extroverted) people preferred positive emotions.
One of the co-authors of both studies, Michal Kosinski summarised what they had learnt by saying that personality played a greater role in what kind of music you want to listen to than demographics or socioeconomics. It is important to understand this to be able to use music to make people happier and healthier. Yes, even healthier: there is evidence that music before, during and after surgery aids patient recovery rates.
Genres are not best suited for this: they developed historically due to marketing and publicity purposes and don’t necessarily reflect quantitative, definable attributes. Furthermore, one’s choices are heavily influenced by social and cultural background, such as what’s prestigious or accepted for a certain age group, social class or geographic locale. This might mean stereotypes and peer pressure win out over merit. Greenberg laments that many people don’t use the evidently vast powers of music to their full potential.
Personality and its relationship with musical ability
In another ($) study (free copy), Rentfrow, Greenberg and further co-authors found personality traits that predict general musical sophistication and musical ability. For personality classification, once again the Big Five personality traits were used, with two additional subcategories (facets) for each dimension. Musical sophistication was measured by the Goldsmiths Musical Sophistication Index (Gold-MSI) and by behavioral tests of melodic memory and rhythm perception.
The Gold-MSI is a 38-item self-report sheet which applies to both non-musicians and musicians, but which cannot distinguish extreme levels of expertise among high-level musicians. Out of these 38, 18 items together form a General Musical Sophistication assessment. Additionally, the entire 38-item set is grouped into five categories: active engagement (e.g. time and money spent on music), perceptual abilities (judging others’ musical abilities, recognising tunes or genres), musical training, singing abilities, and emotions (e.g. communicating evoked emotions or picking music to evoke emotions).
The melodic memory test required participants to compare two versions of the same short (10-to-17-notes long) melody, which were presented in different keys to filter out simple pitch memory, and decide whether they were otherwise identical or differed in melodic contour or intervallic structure.
In the beat perception test, 10–16-second rock, jazz and popular classical excerpts were played, overlaid with a metronomic beep that was either consistently on the beat of the music or off beat. Participants had to choose whether the overlaid beat was on the beat or off the beat. (The tests are available on the internet.)
The result is that the Big Five personality traits indeed predict self-reported musical sophistication and success on the two ability tests.3 The one trait to highlight is openness to aesthetics (but not the trait openness to ideas), which was the strongest trait predictor of all three aspects (always in the positive direction). Further, musical sophistication was positively associated with both facets of extraversion (assertiveness and activity) and the altruism facet of agreeableness, but less strongly than with age or with openness to aesthetics. Openness to aesthetics was also the strongest trait predictor for each five of the sophistication subcategories.
Openness to aesthetics, age and education were the strongest trait predictors for melodic memory, all with positive association.
For rhythm perception, after openness to aesthetics, age was the strongest predictor (this time negatively associated). Being male was positively associated. The order facet of conscientiousness was positively, the self-discipline facet of conscientiousness was negatively associated with rhythm perception.
The yes-or-no variable musicianship, that is, the ability to play an instrument or to sing, was significantly and positively associated with sophistication and with the two tested abilities.
Spurred on by the phrase `sex, drugs, and rock and roll’, researchers had earlier found evidence that musical preferences and substance use are linked. Qualitative studies had suggested that alcohol and recreational drug use can enhance the perceptual abilities, emotional experience and creativity of musicians, especially in the jazz and rock genres. This study found that the variables substance use (alcohol in the past 30 days or recreational drugs at any time in life) improved the models significantly for predicting both tested skills, melodic memory and rhythm perception. For rhythm perception, recreational drug use was as strong a positive predictor as being male, and just stronger than the traits order and self-discipline (which is a negative predictor). Still, the associations are weak.
Let’s call it a day here, shall we. We’ve covered plenty. The overarching motif seems to be that there are indeed connections between personality and musical taste and ability. The correlations are low but they are statistically significant.
I hope I will find the time to return with more on how and why music moves us, and on what frisson, musical anhedonia and involuntary musical imagery are. (It turns out they are more down to earth than the scientific names forebode.) There is also interesting data about how our musical horizons expand then stabilise as we grow older.
I’m admittedly no expert of this subject, meaning that I probably left out more than included. I still hope you enjoyed broadening your mind as much as I did.
At times background music was playing while I was researching and writing this essay, at other times not.
1For the statistically well versed, the correlation coefficients between empathy quotient (EQ) and musical preference for the five dimensions of the MUSIC system were reported. The observed effects display correlations with EQ of about 0.1 for mellow music, -0.1 for intense music, meaning that they are weak but statistically significant at the p<0.01 level. There were four samples with 320 to 2178 human subjects.
2In the sample used to find out preferences for psychological attributes and sonic characteristics in music, in addition to empathy quotient (EQ), Systemising Quotient–Revised (SQ-R) was also measured. The difference of these two essentially opposite measures was used as the empathy scale. Women in the sample were distributed approximately 23%:38%:40% among empathiser, balanced and systemiser types, while sampled men were about 8%:17%:76%. These are skewed towards systemisers compared to an earlier study which probably had a more representative sample of the general population. Basically all findings were confirmed when the authors controlled for gender, so they are indeed dependent on the empathising–systemising brain type and not on the gender.
3Multiple regression analyses were conducted to find which personality traits predicted musical sophistication and performance on the two behavioural tests. Not correlations but the beta coefficients of linear regression were reported, which say how much the traits influence the model prediction of musical skills.