Mehr et al. and Savage et al. have both put forward interesting and very reasonable adaptationist accounts of music – or more precisely, of certain aspects of musicality and musical behavior. I'm more sympathetic to such accounts than I was before. On balance, though, I think it's still premature to conclude that music is an adaptation, and more plausible to think that it's a byproduct. There are three main reasons for this.

First, a lot of the evidence adduced in favor of adaptationist explanations of music is equally amenable to a byproduct explanation. The cross-cultural universality of music is consistent with the claim that music is an adaptation – but it's also consistent with the claim that it's a byproduct of other adaptations that are universal but not music-specific (e.g., emotional responsiveness to the prosody in speech, which cultures might independently “learn” to trigger with melodies). The complex design evident in music could come from biological evolution – but it could also come from cumulative cultural evolution; after all, smart phones and bureaucracies exhibit complex design as well, but are clearly not adaptations. Children take to music early and easily – but they also take to iPads and TV; sometimes ease of acquisition is a result of culture evolving for our minds, rather than the other way around. Damage to certain areas of the brain impairs the ability to make or appreciate music – but none of these areas is involved exclusively in music, and it's possible that the areas in question evolved primarily for their non-musical functions (which are presumably also impaired by damage to those areas). Music-like abilities in nonhuman animals show that traits of that kind can evolve – but they don't show that they necessarily did evolve in our species, as human culture can sometimes independently discover traits that evolved in other animals: The fact that leaf-cutter ants engage in something akin to agriculture doesn't imply that human agriculture is an adaptation; similarly, the fact that various nonhuman animals produce auditory displays doesn't imply that human music is an adaptation. In short, much of the evidence is ambiguous. Given the danger of overextending the adaptationist mode of explanation, the byproduct approach seems like the safer default position in lieu of more decisive evidence.

Second, the byproduct approach has a number of advantages over its adaptationist rivals. Uncontroversial adaptations, such as arms and the basic motivations, are found in all typically developing human beings and are reasonably similar across cultures, subcultures, and historical periods. Music, in contrast, varies greatly from place to place and from time to time, and many people spend little time making or consuming it. These facts are easier to square with a byproduct explanation than an adaptationist one. Even if one argues that certain core features of music are found in every culture, it remains the case that plenty of individuals within those cultures devote little time to music, whereas almost every individual has arms and the basic motivations. And even if one argues that, in traditional cultures, almost every individual devotes substantial time to music, the fact that many individuals in modern cultures do not is still surprising on an adaptationist account – after all, even in modern cultures, every typically developing human being uses language frequently, and it would be surprising on an adaptationist account of language if this were not the case.

Third and finally, the adaptationist accounts of music proposed in this dual treatment face a number of challenges that byproduct explanations do not. If stronger social bonds are adaptive, as Savage et al. argue, why not select directly for a tendency to bond more strongly, rather than a tendency to make and enjoy rhythmically patterned pitch-sequences and to bond with others who do the same? Regarding Mehr et al.'s account, does it seem plausible that raiding parties would be less inclined to attack a group that kept perfect time than an equivalently fierce group whose rhythms were slightly off, or that such a strategy would be particularly useful? Keeping time isn't important in chimpanzee territorial displays, so the closest animal analogy doesn't support the idea. Is music-making prowess a reliable way to assess a group's potential as allies? People could make beautiful music together but be hopeless at hunting, making tools, or doing anything else that might make an alliance valuable. Why not assess the valuable abilities directly, rather than assessing people's musical chops? If rhythm evolved for territorial signaling, why aren't men notably more rhythmical than women, given that men have historically done the bulk of the territorial displaying and defense? If melody evolved for infant-directed song, why aren't women notably more melodic than men, given that women have historically done the bulk of the infant care? Although some studies suggest such differences (e.g., Miles, Miranda, & Ullman, Reference Miles, Miranda and Ullman2016), the broader literature is mixed and it's certainly not obvious that the sexes differ much in these domains. Is infant-directed song a reliable signal of commitment in any evolutionarily meaningful way? It “tells” the baby that it has the parent's undivided attention at that particular moment, while the parent is singing the song. However, the fact that it has their attention in a context where it isn't especially costly to the parent doesn't guarantee that the parent will prioritize the baby if and when difficult trade-offs need to be made – for example, if the parent has to choose to invest either in the baby or in one of the baby's siblings. A peacock can't grow a decent tail unless it's in good condition; in contrast, it's easy enough to sing a baby a song then withdraw support later on, if one's circumstances change.

I don't claim that these difficulties are necessarily insurmountable, and I concede that some of the evidence presented in favor of an evolved contribution to human musicality is at the very least suggestive. However, the difficulties do hint that it's premature to accept an adaptationist account at this stage – and if I had to make a bet today, my money would be on the byproduct approach.