Sunday, December 30, 2012

More Music, More Empathy

Fig. 1 (Molnar-Szakacs & Overy, 2006). Model of the possible involvement of the human mirror neuron system in representing meaning and affective responses to music. ... The shared recruitment of this neural mechanism in both the sender and the perceiver of the musical message allows for co-representation and sharing of the musical experience. Music notes from ‘The Lady Sings the Blues’ by Billie Holiday and Herbie Nichols.

The previous post examined the relationship between music and empathy, including the emotional connection that can occur between musician and audience. My thoughts on the issue were originally inspired by a quote in the book Rat Girl, a memoir by musician Kristin Hersh:
...We'd play what the audience felt and feel it at the same time and they'd feel it reflected back to them in sound and we'd all care about each other's stories and clouds of feeling and ... good luck with that I think miserably through my stage fright, trudging past the knitters, hippies, junkies, drunks, painters and psychos.

-Kristin Hersh, Rat Girl, p. 43

Molnar-Szakacs and Overy (2006) suggested that these profound human experiences are mediated by activity in the supercharged mirror neuron system (Fig. 1).2 Although the box and arrow model is lacking in explanatory value, the hypothesis raised two questions: (1) Are musicians more empathetic? and (2) Do they engage the mirror neuron system to a greater extent than those without musical training?

Are Musicians More Empathetic?

Self-absorbed rock star stereotype aside, it would be difficult to determine causality if this were the case. Do more empathetic people choose to take up music (à la the tortured artist), or does musical training make one more empathetic?

The best way to address question #1 is to look at those undergoing musical training. One such study reported that a 9 month-long program of musical group interaction (MGI) increased emotional empathy in 8-11 year old children (Rabinowitch et al., 2012). The MGI program consisted of musical "games" that seemed [to me, at least] designed to increase empathy, rather than musical prowess: entrainment games to be as rhythmically coordinated as possible, imitation games to repeat the musical phrases or gestures of the previous participant, and other games that called on the constructs of shared intentionality and intersubjectivity.

Along with a passive control group that received no training, an active control group engaged in a verbal storytelling and drama version of group interaction that didn't involve music, singing, or gesture. It's notable that 33% of all children did not play a musical instrument (which included singing), so the study didn't really ask whether musical training per se can make you more empathetic. Nonetheless, there was a p=.054 level interaction of time (pre- vs. post-training) and group (MGI vs. both controls, who did not differ) on the self-report measure of empathy [which might have resulted from a higher pre-training empathy in controls, along with less improvement].

Fig. 3 (Rabinowitch et al., 2012). Index of Empathy scores.

However, the MGI and control groups improved to an equivalent extent on an emotional face matching task, also designed to measure emotional empathy. While it's probably beneficial for children to engage in these group activities, we do not yet have a positive answer to question #1.

Do Musicians Show More Mirror Neuron Activity?

This question has a trivial element to it: of course a trained violinist will have a greater understanding of the movements and sounds involved in Beethoven's Violin Concerto in D major, so you'd expect differences in brain activity somewhere to reflect this.

There are at least two studies potentially relevant to question #2 (Chapin et al., 2010; Babiloni et al., 2012). In neither case, however, do we need to invoke the existence of the mirror neuron system.

In the first, BOLD signal changes in response to two different versions of the same musical piece (Chopin's Etude in E major) were compared in an fMRI study (Chapin et al., 2010). One version was an expressive piano performance with dynamic stimulus fluctuations, and the other was a synthesized "mechanical" version. In addition, the participants had varying levels of musical training: 7 were experienced (mean 9.2 yrs training) and 7 inexperienced (0.7 yrs training), with 7 more thrown out for various reasons. These are very small groups by modern fMRI standards.3

Participants rated their emotional arousal and emotional valence while listening to the pieces before and after scanning, but not during the fMRI experiment. The combined arousal and valence ratings were not consistently correlated across the two time points, so making inferences about what the participants were feeling during the experiment is dicey.

The fMRI results showed different activation patterns according to the main effects of performance type and musical experience (shown below). Mirror neuron-ish areas (inferior parietal lobule, inferior frontal gyrus but too anterior) showed greater activation for the expressive piece in both groups (A), but these regions didn't differ as a function of musical experience (B).

Fig. 4 (Chapin et al., 2010). fMRI ANOVA results. Brain activations (F-maps) showing a significant main effect of a) performance type (F (1,24) > 7.19, corrected p < .02), SCG = subcallosal gyrus, PHG = parahippocampal gyrus, vACC = ventral anterior cingulate, FPC = frontopolor cortex, DMPFC = dorsal medial prefrontal cortex; and b) main effect of musical experience, BG = basal ganglia, vStri = ventral striatum.

But there was an interaction between performance type and musical training, with experienced participants showing greater activation for the expressive piece in the too-anterior-for-mirror neurons IFG and the inferior parietal lobe. Importantly, activation in the mirror neuron-ish areas related to tempo changes in the expressive piece did not differ according to musical training:
An unexpected finding of this study was that, for all participants, the tempo fluctuations of the expressive performance correlated with dynamic activation changes in brain regions that are consistent with the human mirror neuron system, including bilateral BA 44/45, superior temporal sulcus, ventral PMC, and inferior parietal cortex, along with other motor-related areas and with insula.

Thus, the ability to perform a piece of music did not make a difference in the mirror neuron system BOLD response, so we do not have a positive answer to question #2. 4

Do Emotionally Empathetic Musicians Show More Mirror Neuron Activity?

Finally, the study of Babiloni et al. (2012) actually asked a third question: do trained musicians with more emotional empathy show greater mirror neuron system activity while watching their own performance, compared to musicians with a lower emotional empathy score?

The participants were the members of three internationally renowned saxophone quartets (n=12) who played a classical music piece by Domenico Scarlatti while their EEG activity was recorded. They also watched a video of their own ensemble performance while EEG was recorded.

Fig. 1A (Babiloni et al., 2012): Overview of the four musicians playing in ensemble during simultaneous EEG recordings.

The participants completed Baron-Cohen's Empathy Quotient Test (EQT) to assess emotional empathy ("emotional contagion" - the ability to mirror an emotional response observed in another person and to experience it vicariously) and cognitive empathy (perspective-taking or theory of mind). Skipping over all the technical details to the bottom line, emotional empathy scores correlated with the degree of alpha wave desynchronization over a "mirror neuron" area in the right inferior frontal cortex during observation of their own performance:
Results showed that the higher the empathy quotient test score, the higher the alpha desynchronization in right BA 44/45 during the OBSERVATION referenced to RESTING condition. Empathy trait score and alpha desynchronization were not correlated in other control areas or in EXECUTION/CONTROL conditions. These results suggest that alpha rhythms in BA 44/45 reflect “emotional” empathy in musicians observing own performance.

Finally, we have a winner! The answer to question #3 is yes. Granting all the authors' assumptions and disregarding the relatively low n (for now), a higher emotional empathy score was associated with greater cortical activity in a mirror neuron area when a musician observed his own saxophone performance (as indexed by the percentage reduction, or desynchronization, of EEG alpha power).

Did this have anything to do with their proficiency in playing with a musical ensemble? Probably not, but it might have made their music more powerful. Many punk musicians, for instance, were not very proficient, but their music was very emotionally intense and resonated with listeners.

ADDENDUM (Dec 30, 2012): An important condition that was not included in this study is the observation of OTHER sax ensembles. Although watching one's own performance includes self and close colleagues, a true test of empathy would be to watch the performance of OTHERS. Watching a video of yourself involves a distinct memory of playing the piece, as well as emotions related to self-consciousness (e.g., pride, modesty, embarrassment at a minor slip, etc.). 5

ADDENDUM #2 (Dec 31, 2012): An anonymous commenter said that the EEG paper is totally bogus due to major unavoidable artifacts and to the uncertainty of EEG source localization. I address these technical problems and limitations in a subsequent comment.


1 I was also struck by this passage, which sounds a lot like synesthesia for pain:
...people should be able to touch one another and feel each other's pain. Physically, like you should be able to touch someone's cheek and feel their toothache; and emotionally, if you move someone, touch them deeply, you have to take responsibility for that depth of feeling and care about them.

-Kristin Hersh, Rat Girl, p. 42
Some have suggested that synesthesia for pain is mediated by (wait for it...) mirror neurons! But Hersh actually does have color-music synesthesia. E major is red, for instance. And this: "New song is done. It's burgundy and ochre with a sort of Day-Glo turquoise bridge—another tattoo on this pathetic little body." (ibid, p. 90).

2 The human mirror neuron system extends well beyond the traditional ventral premotor and inferior parietal areas of monkey neurophysiology fame.

3 The scanner is a very noisy environment, and in case you're wondering the authors used a sparse temporal sampling technique to increase the signal and to avoid interaction of the scanner noise with the music.

4 Five of the seven experienced participants were trained pianists. They were unfamiliar with this specific piece, but I'm assuming they have the ability to perform it themselves.

5 This was pointed out to me by Sandra Kiume of Channel N.


Babiloni, C., Buffo, P., Vecchio, F., Marzano, N., Del Percio, C., Spada, D., Rossi, S., Bruni, I., Rossini, P., & Perani, D. (2012). Brains “in concert”: Frontal oscillatory alpha rhythms and empathy in professional musicians NeuroImage, 60 (1), 105-116 DOI: 10.1016/j.neuroimage.2011.12.008

Chapin, H., Jantzen, K., Scott Kelso, J., Steinberg, F., & Large, E. (2010). Dynamic Emotional and Neural Responses to Music Depend on Performance Expression and Listener Experience PLoS ONE, 5 (12) DOI: 10.1371/journal.pone.0013812

Molnar-Szakacs I, Overy K. (2006). Music and mirror neurons: from motion to 'e'motion. Social Cognitive and Affective Neuroscience 1(3): 235-241.

Rabinowitch, T., Cross, I., & Burnard, P. (2012). Long-term musical group interaction has a positive influence on empathy in children. Psychology of Music DOI: 10.1177/0305735612440609

Rizzolatti G, Sinigaglia C. (2010). The functional role of the parieto-frontal mirror circuit: interpretations and misinterpretations. Nat Rev Neurosci. 11:264-74.

I have a gun in my head
I'm invisible
I can't find the ice

-Throwing Muses, "Hate My Way"

Subscribe to Post Comments [Atom]


At December 31, 2012 5:29 AM, Anonymous Anonymous said...

I did not bother reading the entire paper, but any EEG/ERP expert can tell you right away that this is totally bogus. Methodologically, the major and unavoidable artifacts present in these datasets, plus the uncertainty of EEG localization (even with clean data and massive Ns), make this study extremely weak. Most likely, nobody will be able to replicate these results. It must have been reviewed by some of these "cognitive neuroscientists" who only know the techniques because they read about them in some textbook!

At December 31, 2012 7:34 AM, Blogger The Neurocritic said...

Anonymous - To be honest, I glossed over the methods and did not read them in detail. I'd say that yes, EEG recorded while performing a saxophone piece would be riddled with artifacts, but that objection wouldn't apply to the critical OBSERVE condition. I'll quickly list some potential issues with the EEG methods.

(1) Artifacts - these would be substantial during EXECUTE but not during OBSERVE, CONTROL (video of quartet turning pages of the score), and REST.

(2) Ocular, muscle, and other artifacts were identified and corrected by an unfamiliar-to-me autoregressive method. I've heard of artifact correction for eye blinks but not EMG and other artifacts. EOG artifact correction algorithms can be pretty good but I'm not so sure about correcting for other types of artifacts.

(3) Cortical source localization is not precise, because there is no unique solution to the inverse problem, i.e. numerous configurations of cortical sources can generate the pattern of EEG activity recorded from scalp. It's an ill-posed question. So yes, we have to take the sLORETA solutions that identified bilateral BA44/45 sources with a grain of salt. This is one reason to throw out the mirror neuron notion here, because source localization isn't that precise.

(4) Nonetheless, the authors could have shown data from the F3/F7 and F4/F8 electrodes as a rough proxy for lateralized frontal activity. However, the location of F7/8 in particular would be yield many artifacts if you're blowing into a saxophone.

(5) Again, this objection wouldn't apply to the OBSERVE condition. As I said in the first addendum, the important 'observation of OTHERS' condition is missing here. Direct comparisons of the two, and between OBSERVE vs. CONTROL and OTHER vs. OTHER CONTROL might be informative.

(6) I think the correlations were undercorrected for multiple comparisons. They had 4 ROIs, 2 alpha frequency bands (low/high), and 3 recording conditions but only used a Bonferroni correction of p < 0.016. This is too lenient. One of the R44/45 correlations (p=.01) would no longer be significant, and the other might be marginal (p=.004) with adequate correction.

(7) I mentioned the low number of subjects too, but really, how many professional saxophone quartets can you recruit? So that aspect of the study is impressive. But for the correlations with empathy in particular, 12 is a low n.

At December 31, 2012 11:08 AM, Anonymous Roger Bigod said...

This is analytically less than pristine. For starters, it isn't clear what's being mirrored. At a simple physical level, there's identification with the performer as a mechanic. For the emotional response, there's a issue of whether the identification is with the point of view of the performer or a response to it. Then there's the effect on later state of mind.

The mechanical stuff might be better studied in sports. It would be interesting to see responses while watching a golf shot, comparing people who play the sport with those who don't. Putts would probably show a rising level of tension as the ball approaches the hole, followed by some sort of discharge depending on the outcome. Pass plays in US football would probably show the same changes. There's probably not so much identification with the passer, but rising tension while the ball is in the air.

Classical music and jazz have passages that show off virtuosity. But they often coincide with an emotional climax, so it would be hard to isolate effects. The performers, especially pianists and violinists, appear to have the highest levels of "temperament", possibly the result of pushing technique to the limits.

Interpreting the emotional aspects is even messier. General emotional valence (happy, sad) sounds like a mirror neuron effect. But if the performer plays a role like vulnerable child or hopeful lover, different listeners may choose to identify or play a counter-role.

Another set of questions concerns music and pathology. Do psychopaths have mirror neurons? Any emotional responses at all?

Yet another object of study would be the recurrent "hook" that repeats involuntarily so that "you can't get it out of your head". I'm hoping not to have a relapse by mentioning "MacArthur Park". It would be highly unethical to expose a newbie to that, however noble the goal of the experiment.

At January 01, 2013 11:33 AM, Blogger The Neurocritic said...

Roger Bigod - Thanks for the comment. You made a number of good points, enough to fuel several research programs.

I'm not really an expert, but in the realm of sports we have Modulation of motor area activity by the outcome for a player during observation of a baseball game
Neural correlates related to action observation in expert archers.

The first involves greater emotion, because of the requirement to identify with either pitcher or batter. The second seems more mechanical.

At January 02, 2013 10:49 AM, Anonymous Anonymous said...

For additional detailed, replicated, peer-reviewed studies on this topic, many of which support these findings, reference the International Society of Music Perception and Cognition whose members are musicians, researchers, neurologists, neuroscientists, psychiatrists, psychologists, etc. While there may be room for improvement in some of the study's methodology, this is a relevant topic and the questions and conclusions are areas under much discussion.

At January 03, 2013 11:49 AM, Anonymous Roger Bigod said...

I did get carried away, but empathy is such a big, amorphous subject and there are many ways of relating to music. It's difficult to design experiments to approach your original question.

Studies of responses to sports isolate physical performance from the emotional and cultural valence of the music. Ball sports introduce an element of numerical win/loss for each play, but aesthetic sports like gymnastics or diving don't have this.

One pro technique for learning classical piano pieces is a stereotyped process. Too bad it isn't compatible with fMRI. To do it right, the student spends a long time, over days to weeks, playing small stretches mechanically for one hand at a greatly reduced tempo. it's a dispiriting process, like making a dog do nothing but submissive commands for an hour. Then after the music starts to "get into the fingers", the tempo is raised gradually with longer passages as a reward for good behavior. Finally, one day there's an event I think of as he Eliza Doolittle moment when it suddenly comes out as music. Raininspain tralala. After that, issues of tempo, dynamics and phrasing. I have no taste, so it's a matter of deciding whether to copy Glenn or Sviatoslav, whoever. But the ability to discern fine points of interpretation has interesting neurological correlates.

If someone could capture an Eliza Doolittle moment on fMRI, preferably with a sound tract, it would be a sensation at TED.


Post a Comment

<< Home

eXTReMe Tracker