Mehr et al. hypothesize that the ultimate evolutionary function of musical rhythm in its proper domain is credible social signaling of coalition quality, whereas Savage et al. propose instead that musicality developed via iterated niche construction into a gene-culture coadaptation broadly for social bonding. Interestingly, both teams of authors locate important functions and selection pressures for musicality in the multilevel social structures of behaviorally modern humanity. After highlighting several weaknesses in Mehr et al.'s argument, I will conclude by focusing on this issue.

Although Mehr et al. provide suggestive evidence of a phylogenetic association between social complexity and vocal flexibility among primates, their scenario for competitive signaling using music and dance in multilevel societies is unrealistic for several reasons. First, within such societies, intergroup allegiances and cooperative pacts are only rarely freely chosen in the sort of real-time biological marketplace that Mehr et al. envision; instead, reciprocal cooperation typically reflects some pre-standing, socially normative, or obligatory relationship, such as between nested clans or incumbents of ritualized aid roles (Hill, Wood, Baggio, Hurtado, & Boyd, Reference Hill, Wood, Baggio, Hurtado and Boyd2014; cf. Tomasello, Reference Tomasello2020). Dances typically reinvigorate or trigger these commitments rather than enabling participants to choose them; when partner choice does occur, it is likely to be between co-participants in the same dance (e.g., Rappaport, Reference Rappaport1968). Although sexual attraction between observers and dance participants can lead to new coalition memberships through marriage (e.g., Bovin, Reference Bovin2001), this sexual display function does not match what the authors have in mind, given their emphasis on the lack of sexual differentiation in music.

Second, Mehr et al.'s argument for a basic coalition-signaling function of rhythmic music partly hinges on the amount of time and shared commitment successful group dances putatively require (sect. 4.2.1). Although rehearsal is indeed crucial for many musical performances, the authors' argument on this point sits uncomfortably with the fact that many mechanistic features of rhythm perception and entrainment serve to enhance temporal predictability and perception-action coupling, and so minimize the effort required to achieve mutual synchronization (cf. Savage et al., sect. 4). For example, motor entrainment facilitates endogenous beat prediction in audiomotor brain networks, which in turn facilitates accurate motor timing – a self-reinforcing cycle (Morillon & Baillet, Reference Morillon and Baillet2017; Su & Pöppel, Reference Su and Pöppel2012). Simple forms of synchrony (e.g., clapping to a shared rhythm) can thus enable dynamic physical coordination to an isochronous or metrical beat with minimal rehearsal. It is unlikely that complex, choreographed dance performances – which require rehearsal and thus index group commitment, as the authors propose – phylogenetically preceded these simpler forms of music and dance, suggesting that choreographed group displays are a secondary function.

Finally, Mehr et al.'s claim that rhythmic music's proper domain is overtly signaling covert, prior cooperative intentions rather than catalyzing such intentions requires the rejection of wide-ranging empirical and experimental evidence that musical and rhythmic performances do, in fact, causally effect cooperation and prosociality (Mogan, Fischer, & Bulbulia, Reference Mogan, Fischer and Bulbulia2017). Indeed, research subjects may put particular effort into synchrony with outgroup members in an attempt to establish relationships – again, as co-participants in the same rhythm, not solicitations to observers (Fujiwara, Kimura, & Daibo, Reference Fujiwara, Kimura and Daibo2020). The authors offer theoretical arguments against music's social bonding role in section 3.2, but additional research poses problems for at least two of these arguments. Deep functional interconnections between audio, motor, and reward systems during rhythm perception and entrainment indicate fundamental design constraints on isolating motor entrainment from reward processing (Todd & Lee, Reference Todd and Lee2015), thus limiting the potential for free rider mutations. Meanwhile, language's ability to convey displaced, propositional information enhances the possible scope of shared action while simultaneously increasing the potential for disagreement and self-other differentiation (Fitch, Reference Fitch2006; Knoblich & Sebanz, Reference Knoblich and Sebanz2008). Music thus has a distinct advantage over language for specific kinds of social bonding (Cross, Reference Cross2009), contrary to the authors' claims.

What kinds of bonding can music better accomplish than language? Savage et al. point out that larger-scale, complex societies are more likely to feature audience–performer divides, whereas smaller-scale, politically egalitarian societies tend to exhibit participatory musical modes (often heavily featuring rhythmic dance). A parsimonious explanation both for rhythmic music's role in forging or certifying bonds between distinct groups or individuals and for its association with egalitarian political formations is that synchrony tends to reduce the salience of categorical role or group boundaries, allowing participants to bond with one another without significant regard to rank or subgroup affiliation. Synchrony thus facilitates new or temporary – often phenomenologically egalitarian – subjective in-group identities (Cross, Turgeon, & Atherton, Reference Cross, Turgeon and Atherton2019).

This suggestion points to an important question implicit but not addressed in Savage et al. (e.g., sect. 2.5) and, to a lesser extent, in Mehr et al.: What is the relationship between music and normative and symbolic social categories and structure? Multilevel human societies – and therefore the social-bonding and signaling problems humans must solve – largely rest on normative, nested categories (e.g., clans and phratries) rather than simply biological kinship. These normative roles and structural divisions are typically complementary, often hierarchical, and invariably defined by differentiated obligations. They depend on language, because language uniquely enables the creation and transmission of abstract, prescriptive concepts such as “mother-in-law” or “fellow moiety member.”

Importantly, conversational language necessarily takes a complementary, turn-taking temporal structure (Pika, Wilkinson, Kendrick, & Vernes, Reference Pika, Wilkinson, Kendrick and Vernes2018), in direct contrast to the convergence and simultaneity of mutual rhythmic entrainment. An important implication is, therefore, that language underlies the establishment of socially patterned behavior in the form of complementary, enduring roles and their (often hierarchically) differentiated obligations, whereas music and dance appear to facilitate social bonding in a different, orthogonal mode characterized by precise motor coordination with immediate, patently available physical stimuli. Rhythmic musical bonding may thus be ideally suited for addressing a need specific to multilevel human societies: Re-establishing bonds at intervals between individuals and subgroups who otherwise are at a physical but also symbolic, normative, and hierarchical remove from one another. Normatively differentiated roles, then, were likely a crucial feature of the cultural niche that drove iterated selection for musicality.