It is exciting to see the evolution of music, or rather, musicality (Honing, Reference Honing2018), neglected in the evolutionary sciences relative to the question of language, being highlighted by Savage et al. and Mehr et al. As with language, we are stymied by a sample size of one, but we can approach the evolutionary question by recognizing musicality is not one thing (Molino, Reference Molino, Wallin, Merker and Brown2000), but a collection of cognitive abilities and physiological responses, and studying its constituent elements. The origins of synchronicity, rhythmic entrainment, vocal learning, sequence learning, convergence on a series of pitches, or creative innovation can all be informed by studies on nonhumans, to explore the evolutionary forces behind human musicality. Here, we expand upon the cross-species considerations briefly mentioned by Savage et al. to argue that research into the evolutionary origins of music can be enriched by a broad comparative perspective.

Both Savage et al. and Mehr et al. emphasize social bonds in their accounts. The former emphasize social bonding and the latter the credible signalling of that bonding. The contrast is primarily about who the audience is – social partners, or potential competitors. From a signal evolution perspective, these accounts can be unified – any acoustic signal produced to support or strengthen a social bond is simultaneously a cue for any third party, containing pertinent information about that bond. We see no reason why selection for bonding and credible signalling should not work simultaneously and in the same direction – consider, for example, the apparent intermixing of these functions in duetting neotropical wrens, which are both thought to promote vocal learning (Rivera-Cáceres, Quirós-Guerrero, Araya-Salas, & Searcy, Reference Rivera-Cáceres, Quirós-Guerrero, Araya-Salas and Searcy2016; Templeton, Ríos-Chelén, Quirós-Guerrero, Mann, & Slater, Reference Templeton, Ríos-Chelén, Quirós-Guerrero, Mann and Slater2013).

Nonetheless, one surely has to have a social bond before one can credibly signal about it. Several of the social bonding functions proposed by Savage et al. can be supported by examples from animal studies. Humpback whales provide prime examples of “learned patterns of features” along with a tendency to “deviate from predicted combinations of features” (Allen, Garland, Dunlop, & Noad, Reference Allen, Garland, Dunlop and Noad2018) whereas the vocal clans of sperm whales reflect large scale social structures akin to Savage et al.'s “imagined communities” (Rendell & Whitehead, Reference Rendell and Whitehead2003). The synchronized and overlapping production of the latter is also analogous to Savage et al.'s “prediction” and overall supports a social bonding hypothesis (Whitehead & Rendell, Reference Whitehead and Rendell2015).

To illustrate how comparative approaches can both support and challenge Savage et al.'s account, consider how the absence and presence of a particular form of temporal synchronization in these two species relates to their differing social structures. Humpback song contains some music-like features: hierarchical structure, repetitive themes, and consistent phrase lengths (Martinelli, Reference Martinelli2008; Payne, Reference Payne1995). Savage et al.'s account suggests this would arise within a complex social system in which other social bonding mechanisms had proven to be insufficient for the expected group size. However, when compared with odontocetes, humpback social networks are relatively unstructured, with little evidence of long-term bonds between adults (Allen, Weinrich, Hoppitt, & Rendell, Reference Allen, Weinrich, Hoppitt and Rendell2013). Why do humpbacks have such strikingly music-like features in structurally complex song whereas sperm whales, with highly structured societies featuring enduring social bonds (Whitehead et al., Reference Whitehead, Antunes, Gero, Wong, Engelhaupt and Rendell2012) have only patterned series of clicks?

One answer relates to the kind of musical features absent in humpback song, specifically the lack of temporal synchronization between individuals, resulting in an “asynchronous chorus” (Herman, Reference Herman2017). Savage et al. present convincing evidence that synchronization can give rise to prosocial behaviour, we shouldn't be too surprised to find it absent in communities lacking long-term bonds. Sperm whales, in contrast, do coordinate their codas (Schulz, Whitehead, Gero, & Rendell, Reference Schulz, Whitehead, Gero and Rendell2008), so we have an illustration of how music-like features of vocalization systems do not come in fixed packages (Nettl, Reference Nettl, Wallin, Merker and Brown1999; Nketia, Reference Nketia1984), and how particular components of musicality may correlate with social structures. The larger point is that both communication systems and social structures exist in a plurality of different forms, and comparative study can reveal contingency in relationships that can superficially appear evolutionarily necessary when only examined in one species.

Mehr et al.'s critiques of the social bonding hypothesis, although explicitly rooted in an evolutionary psychology perspective, rely on over-simplistic evolutionary arguments and do not reflect the nuance of current evolutionary thinking. First, their “superfluous” claim assumes that the mere presence of fitness benefits negates the need for stress reduction, but this ignores at a fundamental level the real tensions between cooperation and conflict that need to be resolved with sometimes costly suppressor mechanisms in any cooperative system, even down to the genome level (Scott & West, Reference Scott and West2019). Mehr et al.'s credible signal account turns the explanation of developing evolutionarily stable social bonds in the first place into somebody else's problem. This seems to us inherently weaker than an account where the elements of musicality coevolved with social bonds, simultaneously providing useful cues that could themselves become targets of selection, as in the duetting wrens. Second, the claimed confusion between proximate and ultimate causation doesn't reflect the difficulty inherent in such distinctions engendered by gene-culture co-evolutionary dynamics (Laland, Sterelny, Odling-Smee, Hoppitt, & Uller, Reference Laland, Sterelny, Odling-Smee, Hoppitt and Uller2011). Such arguments against gene-culture co-evolutionary accounts can only endure by ignoring the accumulating evidence for gene-culture co-evolutionary dynamics in nonhumans (Whitehead, Laland, Rendell, Thorogood, & Whiten, Reference Whitehead, Laland, Rendell, Thorogood and Whiten2019), evidence that lends credence to Savage et al.'s co-evolutionary account.

Finally, the different accounts offered here map onto debates about information and influence in animal communication (see Stegmann, Reference Stegmann2013). Although Mehr et al. give an explicitly information-based account of music, musicologists more typically characterize it in terms of influence, or “affect” (e.g., Juslin, Reference Juslin2010), and this may provide both a stronger link to potential evolutionary precursors as revealed by a comparative perspective and support for an account under which music coevolved with human hypersociality.