The target articles present themselves as diametrically opposed. Mehr and coauthors argue that human musical skill and appreciation (“musicality”) has its origins in the use of rhythm and sound to communicate trustworthy information, ensuring honesty in contexts where partners are tempted to lie. Savage and coauthors, by contrast, argue that musicality evolved to establish and cement “social bonds,” and to coordinate behaviours when partners have aligned interests. We believe that the hypotheses overlap much more than either side admits, given the breadth of “credible signals” in the natural world.

Savage et al.'s argument that musicality provides adaptive “social-bonding” benefits is split into three components (see their Fig. 3): (1) strengthening “bonds” with groupmates; (2) providing cues of group membership; and (3) enhancing within-group coordination. Yet, the first and second of these must ultimately be based on credible signalling, which is Mehr et al.'s proposed mechanism.

Our first point is that explaining the existence of exclusive and enduring “bonds” between individuals remains an open question in social evolution. Social bonds are “stable, equitable, and strong” dyadic relationships (Ostner & Schülke, Reference Ostner and Schülke2014). A “social-bonding mechanism” is an adaptive behaviour that increases the strength of a specific bond. It must make it more profitable for the two individuals to invest in the dyadic relationship than to leave it, overcoming temptations to defect. Social-bonding mechanisms only make sense where conflicts of interest can arise (which do not feature explicitly in Savage et al.): if there is automatic alignment of interest, there is no need for any “bonding” to establish and stabilize a relationship.

Nonhuman animals can increase the reliability of a relationship in two main ways: (1) by basing relationships on a series of small reciprocal trades (e.g., removal of parasites by allogrooming in primates, ungulates, and birds; Akinyi et al., Reference Akinyi, Tung, Jeneby, Patel, Altmann and Alberts2013; Mooring, Blumstein, & Stoner, Reference Mooring, Blumstein and Stoner2004; Radford & Du Plessis, Reference Radford and Du Plessis2006), reducing the stakes of each step (Dixit & Nalebuff, Reference Dixit and Nalebuff1991), and rewarding good behaviour; or (2) by using credible signals of ability and/or commitment to advertise a future resource (e.g., males honestly advertising parenting quality; Buchanan & Catchpole, Reference Buchanan and Catchpole2000; Pettitt, Bourne, & Bee, Reference Pettitt, Bourne and Bee2020). It is difficult to see how music is a tradable resource – there is no immediate fitness benefit – and neither paper provides an account of music trading. This leaves credible signalling of trustworthiness or ability as the implicit adaptive device for strengthening “social bonds.” This may occur in myriad ways; for instance, paying the cost of investing in lengthy bouts of music-making may only be in the interest of partners pursuing long-term alliance benefits (e.g., support during within-group contests). In this view, both hypotheses rely on the same logic.

Drawing on Dunbar's “grooming-at-a-distance” hypothesis (Dunbar, Reference Dunbar and Bannan2012), Savage et al. appear to argue that the benefit of music was release of dopaminergic rewards, and consequent “bonding.” We agree with Mehr et al.: this argument is circular. Selection builds proximate rewards to drive animals to pursue behaviours that are ultimately adaptive. Invoking dopaminergic rewards begs the question: the aim is to explain why brains are initially attracted to music (i.e., why they are rewarded with dopamine or other drivers). Savage et al. might escape the charge of circular reasoning by more explicitly distinguishing traits representing “senders” (ability to produce music) and “receivers” (attraction to hearing music). In principle at least, the second might be a non-adaptive by-product of some other feature of the brain (a “sensory bias”) to which the first initially evolved as an adaptation (cf. fish swordtails; Basolo, Reference Basolo1995).

Confusion of proximate and ultimate explanations for affiliation extends beyond music to other aspects of animal behaviour. Affiliative acts between groupmates abound: for example, body contacts in ants (Birch, Cant, & Thompson, Reference Birch, Cant and Thompson2019), soft bumps in cichlid fish (Bruintjes, Lynton-Jenkins, Jones, & Radford, Reference Bruintjes, Lynton-Jenkins, Jones and Radford2015), and preening or grooming in birds and mammals (Borgeaud & Bshary, Reference Borgeaud and Bshary2015; Kern & Radford, Reference Kern and Radford2018; Radford, Reference Radford2011). Simply invoking the terms “social bonding,” “social cohesion,” and “affiliation” (or proximately “relieving stress”) for these stops short of real explanations, which are explicit about ultimate costs and benefits, the origin of constraints, and – in cases where signalling plays a role – the information communicated and the guarantees of credibility.

The second component of Savage et al.'s social-bonding hypothesis is that musicality provides cues of group membership. This argument is explicitly about credible signals, and indeed also features in Mehr et al. Here, music would be analogous to the group-specific cuticular hydrocarbons of insects (van Zweden & D'Ettorre, Reference van Zweden, D'Ettorre, Blomquist and Bagnères2010), call signatures of birds (Hopp, Jablonski, & Brown, Reference Hopp, Jablonski and Brown2001; Radford, Reference Radford2005), and olfactory cues of mammals (Christensen, Kern, Bennitt, & Radford, Reference Christensen, Kern, Bennitt and Radford2016; Henkel & Setchell, Reference Henkel and Setchell2018): difficult-to-fake indices of group identity.

Only the third component of Savage et al.'s hypothesis does not imply that music is a device allowing credible signalling. Here, musicality promotes within-group coordination by synchronizing emotions or behaviours. It is unclear how or why this coordination should occur; we read it as describing individuals that have no reason not to trust one another, sharing a common interest in coordination and looking to exchange information. This is signalling, with the credibility of the signals assumed. Here, music would be analogous to trustworthy acoustic signals that coordinate movement or recruitment within animal groups (Braune, Schmidt, & Zimmermann, Reference Braune, Schmidt and Zimmermann2005; Radford, Reference Radford2004; Radford & Ridley, Reference Radford and Ridley2006).

The hypotheses proposed in each paper have much overlap: both must see music as an honest signal facilitating cooperation. The authors differ in the specific contexts of cooperation that they favour as important at the origins of musicality: Mehr et al. focus on inter-group coalitions and mother-offspring bonding, whereas Savage et al. focus on within-group coalitions. There exists sparse empirical evidence either way for these interesting, if speculative, scenarios. More generally, costs maintain the credibility of diverse signals across the natural world (Biernaskie, Perry, & Grafen, Reference Biernaskie, Perry and Grafen2018). If music is a signal, perhaps the most progress will therefore be made by taking a leaf out of the animal behaviour book and asking empirically where the costs lie.