Disaffiliative humor in improvised musical interactions: an experimental study

1 Introduction

Musical humor is sometimes believed to be something of a novelty, restricted to only a few selected musical compositions, genres, or practices. But it is in fact present in a wide variety of practices, from the Western classical repertoire (Arnason 2011; Lassauzet 2019) to post-war Western avant-garde music (Kippelen 2017); from ritual music (Guillebaud and Victor 2013) to film music (Mera 2002) and other video-accompanying music (Liikkanen and Salovaara 2015); from jazz, rock and popular song (Kitts & Baxter-Moore 2019) to various genres of comic opera (Loriot 2015). The ubiquity of musical humor is probably at its clearest when looking at contemporary web culture: according to Erickson et al. (2013), for every song that made an appearance to the British Top 100 Singles Chart in 2011, 24 musical videos self-declared as parodies of that song were uploaded on YouTube. All this suggests that musical humor is a crucial component of the relation we have with musical production and consumption.

It is thus not surprising that musical humor tends to be taken more and more seriously by music psychologists. In previous empirical studies, musical humor has mainly been approached by investigating the formal parameters that might elicit laughter, smiling, or amusement, such as the presence of contrasts (Rodriguez et al. 2023), stylistic incongruities (Huron 2004), imitations of laughter (Trevor and Huron 2018), or surprising endings (Nagy et al. 2022). This focus reflects a general trend in music psychology to think of music first and foremost in terms of structural or acoustic features. However, music is also very much a performative, collective and participatory affair, something that people often do together (Blacking 1973; Cook 2001) – a view that naturally invites us to consider musical material as inherently social (Clarke 2005). Thinking of music in terms of socio-interactional features opens a whole new avenue for research on musical humor, by investigating the forms of humor that might arise from the very relations in which the performers are engaged towards one another when playing music together.

Martin et al. (2003) famously distinguished between four humor styles: affiliative humor, self-enhancing humor, aggressive humor, and self-defeating humor. Interestingly, these categories have a strong social aspect, as they refer to types of humor than can be used to enhance one’s relationships with others (affiliative and self-defeating humors) or to a type of humor that directly targets the others (aggressive humor, which includes sarcasm, mockery, ridicule, etc.). Note however that these types of humor are not all “social” in the same sense. In affiliative humor and self-defeating humor, humor mainly aims at improving one’s relationship with others. So while such humor is clearly directed towards others, it does not necessarily and intrinsically rely on a interactional component: it can typically be achieved through canned joke-telling or embarrassing anecdotes from one’s past. Conversely, aggressive humor directly consists in ways of interacting with others: the interaction itself (e.g., mocking or ridiculing someone) is the material of humor here, the very source of any funniness one might ascribe to the situation. A promising path for assessing if there is such a thing as interactional forms of musical humor might thus be to focus on such aggressive humor in musical interactions – and thus look for broadly “aggressive” interactions in which musicians aim at contradicting, ignoring or even mocking one another.

That being said, the category of “aggressive humor” is not without its limits. First, it does not distinguish between various forms of negative interactional humor such as mockery, satire, sarcasm, or cynicism, which may be associated with distinct values or may not even be “aggressive” at all (Ruch and Heintz 2016). Second, it is mainly framed as a way to enhance oneself at the expense of the others, somehow downplaying the extent to which such forms of humor can also be found funny by the others (including, in some cases, the mocked, see Shoemaker 2024, for more general considerations on mockery). Third, it does not account for the fact that impolite interactions can also be funny to watch or to hear even when they are devoid of any additional humorous intent – what is referred to by Lorenzo-Dus (2009: 100) as “incivility-as-spectacle,” the sense of amusement that might result from attending to agents engaged in impolite interactions, and which can be accounted for by various mechanisms such as arousal (through the added tension that comes with aggressivity), superiority (feeling superior to someone who have been belittled), and incongruity (the violation of social/linguistic expectations). For all these reasons, we will prefer here the broader category of “disaffiliative humor” suggested by Marta Dynel (2013) to refer to the funniness (be it intentional or not) one could find in interpersonal interactions that tend to distance the interacting partners from one another, whatever the means used to create this distance.

While such disaffiliative humor has until now only been studied in the context of verbal interactions, the mechanisms identified to explain such form of humor are in no way tied to the specific resources of natural languages. It is thus only straightforward to wonder whether disaffiliation would similarly be found to be a source of humor when attending to musical interactions. This is precisely what the present paper aims to do, by experimentally assessing the extent to which disaffiliative musical interactions are found to be more amusing than affiliative musical interactions.

Instances of disaffiliative musical interactions might sometimes be encountered in the written repertoire. For example, Edward Klorman (2016) shows how Mozart’s chamber music should be understood as a deeply social practice – one that is meant to embody a social interplay between friends, capable of “deceiving, challenging, opposing, seducing, interrupting, or agreeing with one another” (p. 290), as in the Finale of Mozart’s Horn Quintet in which, to quote a recent program announcement, “the strings seem to gently mock the limitations of the solo instrument [the horn].”^[1] Closer from us, and maybe more strikingly, in Cornelius Cardew’s The Great Learning (1971) – a highly influential work in experimental music – performers are explicitly asked at some point to find a way to ridicule their co-performers’ attempt at creating “beautiful” sounds: “Players make sounds they think are beautiful, making them beautifully, spacing them and arranging them in a manner they think beautiful in the general context. Three times try and make neighboring sounds sound stupid or ugly in the general context” (Paragraph 5, “Beautiful Sound Music”).

However, we argue that collective free improvisation (henceforth abbreviated as CFI) provides a far more ideal setting to study the relation between disaffiliative interactions and musical humor. CFI is indeed a musical genre in which the very shape and content of the performance emerge from the unfolding interaction between the musicians, without relying on pre-defined plans or pre-existing musical structures (Canonne 2018b; Saint-Germier and Canonne 2022).^[2] In that sense, in CFI, interactions between the musicians are the very source of the music and not merely a by-product of the individual musicians jointly following a shared score.

Crucially for us, in CFI, musicians tend to explore the whole spectrum of socio-musical relations, from positive, supporting attitudes to more negative, agonistic ones (Golvet et al. 2024). As Eric Clarke (2005: 174) nicely puts it, CFI has thus “at least as much to do with an exploration of interpersonal dynamics as it [has] to do with a direct manipulation of musical materials.” This point of view seems to be also shared by many CFI practitioners. According to Corbett (2016: 55), saxophonist Evan Parker once said that “improvisations fall into those that express agreement, those that express disagreement, and those in which the participants agree to disagree.” And Corbett goes on writing that “improvisation is social music. The shapes it takes and the methods used to make it are collective in nature, often gregarious, interactional. But there are anti-social elements in the music, too, solitary, uncommunicative moments, times when the collective is shunned in favor of the individual. The antisocial impulse helps keep the music from growing predictable” (pp. 56–57). In other words, although free improvisers are ultimately engaged in an activity with a shared goal (i.e., creating aesthetically valuable music) which they jointly strive to achieve, such shared goal is sufficiently indeterminate – and the aesthetic background of improvisers might be distinct enough (Watson 2004) – to allow for a wide variety of interactions between the improvisers, including ones that are arguably more disaffiliative (such as contradicting, ignoring, imposing, or mocking).

That being said, in the absence of clear acoustic or musical criteria that would be non-controversially associated with such disaffiliative behaviors, it might turn out to be difficult to identify them as they happen over the course of the long and complex performances – often with a large number of musicians involved – that are typical of CFI.

For our study, we thus chose to rely on a more controlled corpus of duo improvisations recorded for the research conducted by Aucouturier and Canonne (2017), and fully available at https://archive.org/details/socialmusic. For their studies, Aucouturier and Canonne asked 10 duos of improvisers to record a series of short performances (1–2 min) in which one of the improvisers would have to endorse a given social attitude (chosen amongst five possible attitudes: being domineering, being caring, being insolent, being disdainful, and being conciliatory) towards their co-improviser, who then had to guess which attitude was being enacted. This yielded a corpus of 100 improvisations, 20 for each social attitude. Crucially, the attitudes provided to the musicians strongly differed in their degree of affiliation, thus providing us with an ideal set of experimental stimuli to investigate whether disaffiliative interactions would lead to more humorous results than affiliative interactions.

An additional benefit from using this corpus was the availability of the musicians’ individual tracks (i.e., in which each musician is heard in isolation, without any of the sounds that have been actually performed by their co-improviser), allowing us to assess whether specifically interactional cues (only present within the duo excerpts) made a difference in the perception of disaffiliative musical humor.

Finally, the chamber music-like character of these recorded duets made it equally plausible to present them to participants either as compositions or as improvisations, thus allowing us to assess whether disaffiliative behaviors were more likely to elicit amusement when they were perceived as spontaneously and directly expressed rather than as merely enacted as part of a pre-determined fictional musical dialogue orchestrated by the score’s composer.

2 Study 1

2.2 Results

In order to assess the impact of our two experimental factors (Number of musicians; Type of relation) on participants’ ratings, the data were analyzed through a 2-way ANOVA, using the EZ package in R. As shown in Figure 1, our statistical analysis revealed an absence of main effect of Number of musicians (F = 0.005, p = 0.946, η² < 0.001) and a main effect of Type of relation (F = 63.434, p < 0.001, η² = 0.201), such that participants found excerpts in which one musician expressed a disaffiliative behavior (M = 4.917; SD = 1.620) to be more amusing than the excerpts in which one musician expressed an affiliative behavior (M = 3.423; SD = 1.421). There was not interaction between our two factors (F = 1.111, p = 0.302, η² = 0.004).

Figure 1:

Results from study 1. Error bars show standard error (95 % interval), and the black asterisks shows a significant difference between excerpts in which the relation between the two musicians was disaffiliative and excerpts in which the relation was affiliative (p < 0.001).

Our results thus show that there was indeed something funnier in hearing improvisers display disaffiliative interactions, than in hearing them display affiliative interactions.

However, whether participants had access to the full interaction between the two musicians or only one individual part did not make any difference in their perception of musical humor. Two reasons can be advanced to explain this striking result.

On the one hand, it might be that participants were not really sensitive to the interactive cues contained within the excerpts (i.e., how the musicians behave towards one another) but rather relied on more global acoustic and musical properties (such as the presence of incongruous musical quotations, unusual sounds and instrumental timbres, or basic parameters, such as loudness, pitch height or brightness, that were all already present in the solo version of our excerpts). Such unsensitivity to interactional cues might be due to the lack of expertise of our participants – who had no previous exposure to free improvisation – and were thus probably not in an ideal position to assess whatever humor could have been displayed in the performances (see Huron 2006: 378: “Perhaps the closest to an acid test for musical cultural understanding is that of musical humor”). This would be in line with the results from Aucouturier and Canonne (2017), who found that non-expert listeners did in fact mostly listen to duo improvisations as if they were soli – focusing on the overall sonic result rather than on the ongoing interactional process between the two musicians.

On the other hand, it might be that the solo version of our excerpts contained additional humorous cues, that were not present in the duo versions, and that somehow compensated for the absence of humorous cues associated with the to-and-fro between the two musicians. Our solo excerpts were indeed constructed by abstracting one individual track from the overall duo improvisation. In doing that, we ended up presenting our participants with excerpts that contained abnormally long stretches of silence (i.e., there was nothing anymore to hear when only the other musician was playing). This might have reinforced the weirdness, or even “cringiness” of the music, turning our musical excerpts into absurdist or surreal performances, which might in turn have elicited a somewhat humorous reaction in our listeners.^[4] Unfortunately, our data do not allow us to decide between those two explanations, but we come back at this issue from a slightly different angle in Study 2 below.

3 Study 2

Our first study showed that humor ratings were not significantly different whether participants listened to the two musicians interacting with one another, or to only one musician from the duo. As such, it remains possible that disaffiliative relations are found to be more amusing than affiliative relations not because of how the interaction itself unfolds, but because musicians simply change the sounds they are using when they enact some kind of disaffiliative behavior (such as being insolent towards one’s co-improviser) – and that these sounds are found to be more amusing than the sounds that would be used in a more typical affiliative behavior (such as displaying a caring attitude). In short, disaffiliative relations could be found to be more amusing than affiliative relations not in virtue of the socio-musical interactions they generate but in virtue of the individual sonic behaviors they prompt.

To assess this possibility, we added another layer to the experimental design used in our first study, which contrasted the duo version of our excerpts with their solo counterparts. This time, we also did manipulate the very way the excerpts were presented to the participants: either as “improvisations, spontaneously generated by various duos of musicians,” or as “compositions for two instruments, written by various composers.” In both conditions, the music would be exactly the same; but the underlying assumption was that participants would be more likely to focus on the overall sonic result rather than on the ongoing relation between the two musicians when they imagined that the musicians were performing something that had been previously conceived by a single mind (Canonne 2018a, b). Conversely, it seemed reasonable to expect that the interactional dynamics between the two musicians would be made more salient when hearing a given duo performance as an improvisation rather than as the performance of a composition. If the kind of musical humor displayed through disaffiliative interactions has indeed something to do with how the musicians relate to one another, then duo performances should be found more amusing when listened to as improvisations, rather than as compositions.

But manipulating the mode of presentation could also affect the perception of musical humor for a different reason: as Canonne (2021) suggested, it could very well be that improvisations are found to be more amusing than compositions, all things equal otherwise, because of the wit and quick thinking that are required to generate music in such a spontaneous way, making whatever funny element we might encounter within the performance even more amusing. Having participants listen to the solo versions of our excerpts allowed to control for this possibility: if improvisation makes things funnier simply because it prompts to focus on the wit of the performer, then both solo and duo versions should be found more amusing when listened to as improvisations, rather than as compositions. Conversely, if improvisation makes things funnier because it prompts us to hear music as an ongoing interaction between musicians (and not only as the product of such interaction), then only the duo versions should be found to be more amusing when listened to as improvisations, rather than as compositions. No such difference should be found for solo performances since, obviously, there is no interaction whatsoever to be heard in that case.

3.2 Results

In order to assess the impact of our three experimental factors (Number of musicians; Mode of presentation; Type of relation) on participants’ ratings, the data were analyzed through a 3-way ANOVA, using the EZ package in R. Our statistical analysis revealed an absence of main effect of Number of musicians (F = 0.260, p = 0.612, η² = 0.004), an absence of main effect of Mode of presentation (F = 0.604, p = 0.441, η² = 0.010) and a main effect of Type of relation (F = 20.643, p < 0.001, η² = 0.081), such that, overall, Insolent improvisations (M = 5.076, SD = 2.732) were found to be funnier than Disdainful improvisations (M = 4.384, SD = 2.611). There was also a significant interaction between Mode of presentation and Type of relation (F = 4.536, p = 0.038, η² = 0.019). As shown in Figure 2, this interaction seemed to be driven by Disdainful excerpts. Post-hoc t-tests (using the Benjamini and Hochberg (1995) correction for multiple comparisons) revealed that the duo versions of Disdainful excerpts were judged to be less amusing when they were listened to as compositions (M = 4.108, SD = 2.565) than when they were listened to as improvisations (M = 4.978, SD = 2.612) (t = −2.541, df = 221.93, p = 0.047). But strikingly, there was no such difference when it came to the solo version of Disdainful excerpts (t = −0.654, df = 189.83, p = 0.685). Similarly, while there was a marginally significant difference between the solo versions (M=4.239, SD=2.629) and the duo versions (M = 4.978, SD = 2.612) of Disdainful excerpts when participants thought the excerpts were improvised (t = −2.052, df = 200, p = 0.083), no such difference appeared when participants thought the excerpts were composed (solo version: M = 3.994, SD = 2.552; duo version: M = 4.108, SD = 2.656; t = −0.320, df = 201.47, p = 0.750). As for Insolent excerpts, they were always rated in a highly similar way, whatever the condition.

Figure 2:

Results from study 2. Error bars show standard error (95 % interval). * represents significant differences, with a threshold of p < 0.05; ***: p < 0.001. + represents a marginally significant difference (p=0.083).

Our results thus suggest that believing that something is improvised does not necessarily make it more amusing. But they also show that such belief can prompt listeners to focus on aspects of the performance (i.e., the socio-musical relations between the musicians heard in the duo versions of our excerpts) that might otherwise end up being overlooked, despite playing a key role in certain forms of musical humor.

Our pattern of results indeed suggests that both types of excerpts elicited distinct kinds of humor. Whatever humor is contained within Insolent excerpts seems to be largely independent of the musicians’ interactional dynamics. However, the kind of humor afforded by Disdainful excerpts seems to be at least partly dependent on our perceiving the performance as an ongoing socio-musical interaction. While prompting our participants to pay attention to the spontaneous interactional dynamics did not make any difference in how they found Insolent improvisations to be amusing, it did make a significant difference in the case of Disdainful improvisations. Here, participants were more likely to find the performances to be amusing when they were thinking of the relation they heard between the two instruments as a live socio-musical interaction between the musicians rather than as a sonic property of a pre-composed musical structure.

Now, one remaining question is whether there is something in the acoustical structure of our two types of stimuli that might explain the pattern of results found in this study. We thus conducted additional analyses in order to further investigate the differences between our two types of stimuli, and to shed light on the types of humor they could be associated to.

4 Study 3

5 Discussion

Our studies showed that improvisations were found to be more amusing when the performers were engaged in disaffiliative behaviors towards one another than when they endorsed affiliative behaviors. This extends the relevance of disaffiliative humor to the musical case, and should invite us to consider such musical impoliteness as an interesting resource for further investigations on musical humor. We also found that believing that the music was improvised did not systematically increase the amusement of listeners, but that it could prompt them to pay more attention to the ongoing interactions between the improvisers, which can turn out to be a source of amusement in and of itself. Taken together, these results suggest an interesting connection between CFI and humor. Because of its highly social nature, free improvisation provides performers with interactional resources that might be put to use to elicit fun or amusement amongst audience members. Composers could of course strive to mimic the highly diverse socio-musical relations that are encountered in free improvisation, but this would likely end up being a very taxing process. Conversely, those socio-musical relations are readily available to performers freely improvising together: in fact, they constitute the very fabric of the music itself.

Our results are also worth further discussing in light of more general humor research.

First, we found that Insolent improvisations were found to be the most amusing of the four behaviors enacted by the improvisers, and that their unfolding was likely to be perceived as more surprising than the Disdainful performances. It might be that, in this case, improvisers were intentionally trying to be funny (e.g., by using typical forms of aggressive humor to convey insolence), explaining why, on average, the resulting music was found more amusing. Further experiments – involving e.g., improvisers that would be explicitly asked to perform humorous versus serious music – would be required however to fully disentangle humor-as-intention and humor-as-perceived in the musical case, allowing us to more systematically explore e.g., the distinction between intentional humor and unintentional humor, or cases of failed humor (Bell 2015).

Second, we found that the improvisations that were judged more amusing (i.e., Insolent improvisations) had acoustical characteristics – such as being louder and brighter – that are typically associated with higher tension arousal – although the extent to which our participants were actually aroused when listening to the excerpts remains unknown. This is line with recent humor theories that have been suggesting a positive relationship between arousal and amusement. According to Alan Roberts (2019), arousal is a key ingredient in humor: on the one hand, arousal – at least when it’s not overwhelming – can increase our feeling of amusement; on the other hand, the incongruities or cognitive dissonances that are typically elicited by funny objects (e.g., a joke) can be seen as “inherently increasing arousal” (p. 114). While establishing such a theory in the musical case would require, first, to systematically manipulate the acoustic properties of musical stimuli, and, second, to track listeners’ levels of arousal, our results can be seen as a preliminary step in assessing the connection between arousal and musical humor.

Third, and contrary to what was hypothesized by Canonne (2021), we showed that knowing that the performance was improvised was not sufficient in itself to increase the listeners’ feeling of amusement. In particular, humor ratings were pretty much the same whether the participants believed that the solo they were listening to had been improvised or composed, suggesting that the kind of experience that is entailed by improvisation (i.e., attending to the thought process of the performer) does not necessarily increase already present feelings of amusement. That being said, two alternative explanations might be brought forward to explain this result. It might be that believing that something is improvised increases amusement only in the case where we are strongly amused. Improvisers already need to achieve reasonably well their aesthetic or expressive goals in order for our emotional responses to be increased by the apprehension of the improvisers’ physical and mental virtuosity. Here, our stimuli may have been found not funny enough in the first place to elicit the kind of emotional intensification suggested by Canonne (2021). It might also be the case that (falsely) believing that our improvised performances were compositions actually increased their comic appeal, but for a totally different reason. Listening to the performance of a composition comes with its own set of normative expectations, both at the structural level (we expect some degree of internal consistency) and at the performative level (we expect the performance to exhibit a sufficient degree of instrumental mastery), which were maybe not satisfied by the corpora used in the above studies. In other words, listened to as compositions, our improvisations might have been taken as examples of “bad” music; and as Justin London (2021) argues, the absence of musical skill and taste can cause laughter. In the words of Justin London: “We are, alas, laughing at rather than laughing with the performer in these instances, and our laughter is a form of derision” (p. 331). But note that bad performances – which ostensibly fall short from all the standards we have come to expect when listening to a professional musical performance – can also be found funny when they are clearly voluntary. In that perspective, bad performances would not be funny (only) in virtue of the feeling of superiority they elicit among the listeners, but also because of the sheer incongruity between the actual performance and the social or cognitive framings it entails. Anecdotal evidence from our post-experiment debriefings partly supports this interpretation: when asking them the criteria they relied on to make their judgments, several participants indeed reported that they found it funny when performers “played their instrument badly on purpose.” Again, more systematic investigation would be required to disentangle this complex web of interrelations between normative expectations, mode of production, and musical humor.

While we did not find that believing that the performance was improvised always made it more amusing, it was nevertheless the case for the “Disdainful” performances. Here, knowing that the performance was improvised helped disclosing the music’s humorous content: the incongruence between the individual voices, which was typical of Disdainful performance, was presumably made more salient when the participants thought of the music as the emergent result of a spontaneous social interaction. In other words, at least in certain cases, properties related to the mode of production of the performance (i.e., is it improvised or pre-composed?) can make a difference in the appreciation of its humoristic dimension. These results provide additional grounds to refute aesthetic empiricism, the position according to which properties relating to an artwork’s origin or provenance are not relevant in the evaluation of its aesthetic properties (Lamarque 2010). Listening to music as an improvisation (rather than as a composition) can indeed make a difference in the evaluation of an aesthetic property as important as the music’s humorous content, at least in some cases. This is in line with the qualitative study conducted by Canonne (2018a) which showed that groups of participants attending to the very same music, but either as a duo improvisation or as a composition for two instruments, had two very different listening experiences. Our present studies complement this previous work, but through more quantitative means.

Fourth, our results point towards the existence of two distinct forms of disaffiliative musical humor: one that relies on referential properties (e.g., imitations of vocal sounds, musical quotations), and that has mainly to do with how the music actually sounds; and one that relies on relational properties (i.e., on the relations between real or imaginary agents heard within the music) and that has mainly to do with the agential perception of the music (Levinson 1996). Prototypical Insolent and Disdainful improvisations (which can be heard in full at https://archive.org/details/duo-02-07) respectively exemplify both types of disaffiliative humor.^[5] On the one hand, some Insolent improvisations (see the saxophone player in Duo 03–08) are replete with rapid, high-pitched, staccato sounds that clearly evoke laughter, which can be in itself a source of amusement (Trevor and Huron 2018); others exhibit motives played in a distorted fashion that sounds like a sarcasm or mockery (see Duo 04-02 at 12’’ for an example); others yet rely on out-of-context musical quotations that constantly disrupt the musical flux (see the piano part of Duo 02–05 for an example). On the other hand, some Disdainful improvisations show the improvisers deliberately avoiding playing at the same time (see Duo 03–05 for an example); in others, the two musicians use timbral and musical materials that are highly heterogeneous (see Duo 10-06 for an example); in others, yet, improvisers do everything they can to mimic a non-interactive situation, as if the two musicians were playing in two separate rooms (see Duo 02–07 for an example). While such musical situations might be seen as examples of “bad” compositions for duos of instruments, they can become funny once we adopt an agential perspective, and hear the music as two personae ostensibly avoiding each other.

Disaffiliative humor has been overlooked in previous taxonomies of musical humor (Eriksen 2016) – most probably because such taxonomies generally rely on individually-composed music rather than on collectively-improvised music – but we argue that it constitutes one of the important manifestations of musical humor, particularly in collective musical practices. However, we do not mean to claim that disaffiliative humor relies on wholly distinct mechanisms. In particular, disaffiliative humor clearly relies on incongruities. This is obvious in the case of the out-of-context quotations found in Insolent improvisations, but the kind of relations exhibited in Disdainful improvisations also create many kinds of sonic incongruities (e.g., between the timbral material of the two instruments), as well as a broader sense of incongruity between the apparent discoordination displayed by two musicians and our expectation that these musicians should aim at playing “together.” Disaffiliative humor also points towards the relevance of the slightly less-popular “play” theory (Apter 2013) for explaining musical humor. An important reason for why we might find Disdainful performances amusing is indeed the sense of play they entail: we hear the musicians playing a game (an “I avoid you, you avoid me” kind of game), we feel the deeply ludic nature of what is going on in the music, we imagine the musicians having fun, and that causes our feeling of amusement. While such a ludic appreciation of the music produced might have been particularly strong in the case of the Disdainful improvisations, we suspect that many other types of relations explored in collective improvisation elicit a similar sense of playfulness, which might in turn led to a humorous reaction.

Fifth, and finally, our results suggest an interesting relationship between negativity and musical humor. Many forms of humor seem to involve some kind of negativity: just think of irony (which often involves a contradiction between a verbal statement and the situation it is referring to), absurd humor (which often rely on the constant negation of social, physical, or even logical expectations) or caricature (which often focuses on the most unflattering physical attributes of a given subject). In fact, the whole superiority theory of humor (Carroll 2014) is based on the idea that we find an object or a character funny when we can feel superior to it, which often requires, in a way or another, the belittlement of said object or character. Certain genres of humor also inherently rely on negative emotions. Cringe comedy is a case in a point: the amusement it elicits is indissociable of the high level of embarrassment it aims at evoking (Hye-Knudsen 2018), and any laughter that arise might be both a sign of amusement and a way to ease the embarrassment we are meant to feel. In that perspective, there is no reason why musical humor should be fundamentally different form other forms of humor. Negative interactions between the musicians can indeed easily give rise to the various forms of humor alluded to above: they can lead in turn to ironic, absurd, or even cringe results. The “destructive” nature of many forms of musical humor has often been highlighted (Bernstein 2005). But this “destructive” element is generally conceived in relation with something that is external to the musical piece itself: broad stylistic rules that are negated, other musical genres or pieces that are parodied, etc. Our studies suggest that such negativity can in fact be embedded within the musical performance itself, in the very relations between the performers. In that sense, relational musical humor is both highly intra-musical (in the sense that it only depends on the interactions between the musicians) and highly extra-musical (in the sense that it requires to think of these interactions as not only sonic, but also fully social). And it might well be the case that the humor created through negative musical interactions stems from this ambivalence. Disaffiliative humor is undoubtedly one way for music to be funny: unpacking its many dimensions and investigating more systematically its prevalence within musical practices should pave the way for exciting new research in the domain of musical humor.