In a recent book, Aniruddh Patel argues that music behaviors are not adaptations; they are, like the control of fire, culturally propagated technologies. I critically examine both views. Patel unduly exaggerates differences he finds between music and language in arguing that only the latter is an adaptation. And there are significant differences between music behaviors and fire control, such that if the latter is a technology, it is not clear that the former is.
Music, Fire, and Evolution
In Music, Language, and the Brain (2008), Aniruddh Patel denies that music behaviors are adaptations. In other words, he holds that music behaviors are not biologically transmissible characteristics that provided selective advantage in the reproductive success of our forebears. And he denies that music behaviors are spandrels. In other words, he holds that music behaviors are not biologically transmissible side-effects of human adaptations that lack adaptive significance in themselves. He maintains instead that music is a transformational technology.
In this paper, I will briefly review Patel’s argument that music behaviors are not adaptations and, if only for the sake of moving on, I will concede his negative conclusion, because my interest lies more with the second argument. It is not my brief to defend the idea that music behaviors are a spandrel. Nevertheless, I aim to raise doubts about the argument he offers and to examine the distinction we should draw between spandrels and transformational technologies. Before I take on those tasks, I try to clarify what falls under the notion of “music behaviors.”
If we are to consider the connection with evolution, we must focus on human dispositions to act in certain ways, rather than on the artifacts or products – in this case, such things as melodies, songs, symphonies – to which those acts give rise. So if we were to speak of music as an adaptation or spandrel, by “music” we would mean music making or music responding. Music making comprises a range of different activities, such as composing, improvising, or performing. Each of these might take a variety of distinguishable forms. Performing, for example, might count as instancing a composition, creating an improvisation, rehearsing, musical doodling, singing along with someone else’s recording or performance, and so on, and could involve such actions as whistling, humming, singing, slapping one’s body, playing a musical instrument, or using some other item as if it were a musical instrument. Music responding behaviors include acts of listening to music with understanding, pleasure, and appreciation, and of dancing, of moving to music, of entrainment, and so on.
Are some of these music behaviors more central or common than others? Among our many linguistic behaviors, generating novel strings and understanding the novel strings produced by others seem central and there are musical equivalents to these in composition, improvisation, and first-time listening. But within music, as compared to language, repetition and redundancy are perhaps more common and significant. Much music making concerns the production or instancing of “works” – that is, pieces to be remembered and performed on potentially multiple occasions – with the result that most performance is re-performance. And within works and improvised music playings, repetition is almost always crucial to shaping the structure of the piece or performance (Huron 2006: 228-9). So, within music making, the most prominent behavior is probably that of repeatedly performing a pre-composed work, such as a song, or repeating one part of a work within another part of that work, though the existence of such works obviously depends on prior acts of composition. As for music responding, this also depends on music making, because it takes the product of music making as its object and focus. Here attentional listening is probably the primary activity, because coordinating one’s bodily actions with the music typically depends on it.
In evaluating possible connections between evolution and music behaviors, it is necessary to consider issues of scope and of level. The question of scope asks how many kinds of music behaviors individuals must be capable of engaging in if they are to count as musically fluent. How we answer this question will be directly relevant to assessing claims such as that music behaviors are not only pan-cultural but universal among humans. In practice, it looks as if almost all adults (without hearing defects or relevant neural deficits) are music responders, whereas, at the other end of the spectrum, composers make up a small minority. But there could be many explanations for this, and the issue is about capacities rather than the frequency of their actualization. We can reasonably expect that the kind of musical fluency that would connect music behaviors to evolution requires that most people can produce novel musical strings, not only understand them when they encounter them. And it is probably true that they are capable of this, so long as we accept that the strings in question are, for example, simple 8-measure melodies without accompaniment or are structured rhythmic patterns.
With the suggestion that musically competent adults are likely to be capable of composing at least simple melodies, we invite the question about the level at which the relevant behaviors are exercised if the connection with evolution is to be minimally plausible. There are many music behaviors and each of them could be practiced at different levels, ranging from a stumbling, incomplete grasp of the relevant behavior, through full competence, to virtuosic mastery. I take full competence in listening behaviors to involve something like the following: the listener has appropriate expectations from moment to moment about what will come next in the music, recognizes most errors in either syntax or content as such (and sometimes knows what would have been correct instead), and experiences the flux of tension, release, and closure in the music’s progress, so that she knows, for instance, if a piece was completed or, instead, interrupted. (Though I do not specify them here, I assume it is also possible similarly to spell out criteria for what counts as full competence in composition and performance.) Plainly, an individual may engage at different levels in different music behaviors – she is a good dancer and listener but a poor singer and composer – and sometimes at least, this is likely to reflect different capacities on her part for enacting the various behaviors. Moreover, other individuals will display different ranges of competence in their various music behaviors.
If we are considering possible connections between music behaviors and evolutionary forces, which music behaviors and at what level should we have in mind? Are we considering the behaviors of musical tyros, those with competence, or virtuosos and connoisseurs, and do we expect individuals in the group of interest to meet specifiable minimum levels in all music behaviors? As will emerge in later discussion, the answer to such questions is very relevant to evaluating some of the connections hypothesized as holding between evolution and music behaviors.
Are music behaviors an evolutionary adaptation? A number of writers have suggested that they are. Charles Darwin (1874), Geoffrey Miller (2000), and Daniel Levitin (2006) claim that music behaviors are a product of sexual selection, like the peacock’s tail. Ian Cross (2005, 2007) asserts that music aids in the mind’s development by promoting cognitive cross-domain flexibility and social facility. Other theorists see music as adaptive via its promoting social cohesion. For example, Robin Dunbar (2003) suggests that group singing releases endorphins, just as physical grooming in primates does, and both Sandra Trehub (2000) and Ellen Dissanayake (2000) point to its importance in establishing a bond between mother and infant. Patel (pp. 367-71) discusses and criticizes these theories in turn, but he also develops a different, powerful form of argument.
The capacity of humans to acquire oral language is an adaptation in Patel’s view. He identifies (pp. 259-366) ten human capacities for language acquisition that provide strong evidence for its adaptive significance. He then considers if music behaviors display similar marks of evolutionary selection and his conclusion is negative.
Here are the signs that there has been natural selection for the transmission of oral language: (1) Spontaneous babbling (even by deaf babies), shows that infant speech is not simply an imitation of adult speech. (2) Compared to other primates, humans have a lowered larynx, which brings the risk of choking but was necessary for speech. (3) Humans are unique among primates in their capacity for vocal learning and imitation. (4) Infants come into the world with the capacity to learn any language but soon favor their native tongue. (5) There is a critical (early) period for language acquisition. (6) Deaf signing children acquire phonology, morphology, syntax, and semantics in parallel with hearing children. That language can jump modalities testifies to the power of the drive for language. (7) The acquisition of language is robust in that, while some children get more input than others, all acquire it. (8) Successive generations of users increase grammatical complexity, for example, as in Creole. Deaf children signers in Nicaragua systematized their language with almost no access to adults who spoke a fully developed language. (9) A single gene has been identified as exercising strong influence on speech and language. Though it occurs in other animals, the human version differs from these and is almost without variation in humans. (10) It is highly likely that humans without language abilities would be at a severe disadvantage when it comes to survival and reproduction.
How do music behaviors compare (pp. 371-400)? Babbling, vocal learning, and the anatomy of the vocal tract could all reflect adaptations for an acoustic communication system that originally supported both language and vocal music, but in other respects music does not measure up well. Learning develops rather slowly; for example, at age five years sensitivity to key membership is not yet as strong as adults’ sensitivity. It is not clear there is a critical period for music acquisition; some good musicians first learned to play or compose in their teens. The development of musical skill is not nearly as robust as language. And there is no evidence that non-musical people are less successful reproducers.
Moreover, the case for music-specific innate biases is weak. Studies of neonates should be regarded as controversial both because babies are not as musically innocent as is sometimes supposed and because many of their “musical” responses can be accounted for by biases related to speech or general auditory processing. Meanwhile, there is no evidence for a “music” gene. Tone deafness runs in families but the deficit is not music-specific. Perfect pitch is heritable but is not necessary for musical ability and some non-human animals have it but lack the relative-pitch-pattern recognition that is necessary for music. Many so-called “musical” recognition abilities might be the result of separating sound sources for scene analysis, that is, they might be aspects of auditory processing evolved to help us negotiate the soundscape rather than being particular to music.
Patel’s overall conclusion is negative: based on current evidence, music does not seem to be a biological adaptation.
I find all this impressive. Unlike the usual approach to the topic of music and evolution, which is piecemeal, narrowly focused, and highly speculative, Patel offers a sustained argument that calls on a considerable spread of data that he subjects to a powerful test: if music behaviors are an evolutionary adaptation, this should be no less apparent than is the fact that oral linguistic behaviors are an adaptation, and the indicators of music’s adaptedness should parallel those for language.
Later I will assume that Patel succeeds in making the case that music behaviors are not an adaptation. (It’s what he says next that I want to focus on.) But I will raise one issue – relating to the questions of level and scope I mentioned earlier – before moving on. With language, we are all speakers as well as listeners, and we are all able to generate novel utterances. It might be thought this is not the case for music. Many music lovers cannot even hum in tune and not all performers are also composers. So, if we are checking whether music is adaptive, we might need to distinguish between various musical behaviors as I did earlier. Perhaps composing is a marker of fitness while appreciative listening is not. Patel’s approach tends to run these various musical behaviors together. This might be appropriate for language, in which competent users are equally composers of and listeners to utterances, but not for music.
Here is the reply that I believe Patel needs: at a relevantly basic level, we are all vocalists and we can all invent new musical strings. At that level, music is not so different from spoken language. When it comes to highly sophisticated forms of art music, we may not all be able to participate equally, but highly sophisticated art music should not be our focus. After all, we are not all novelists or playwrights, but that does not show that ordinary linguistic behaviors are not an adaptation.
If this is indeed the response Patel needs to make if the comparison he draws between music and linguistic behaviors is to be convincing, it might be necessary to set the standard rather low for qualifying as having acquired music behaviors. (In that case, Patel’s observation that some good musicians first learned to play or compose in their teens is probably beside the point, because he is there discussing much higher levels of competence than those that are relevant.)
Notice, however, that neither Patel’s argument nor my response to it will impress those who favor the view that sexual selection has elevated virtuosic musical skills to honest, because costly, signs of biological fitness. Their claim is not that music behaviors have become universal because they were successful as adaptations for our ancestors. It is, rather, that music behaviors serve currently as adaptations only for those who have them at the highest level, because such behaviors at such levels are positively valued by those choosing reproductive partners. By their lights, one can see music behaviors functioning as adaptations only by considering abnormally high levels of performance in those behaviors. Rather than being impressed by Patel’s approach to the issue, they might regard it as beside the point.
When Patel rejects music’s claim to be the direct product of an adaptation, one might predict that he will identify it as a spandrel, that is, as an adventitious byproduct of adaptations, such as those involved in language audition and auditory scene analysis, that is not itself of adaptive value. In fact he does not do so. He argues (pp. 400-2) instead that music is a technology and as such is not best regarded as either an adaptation or a spandrel. The idea here is that biological adaptations and spandrels should be clearly distinguished from cultural inventions that have no evolutionarily specialized biological substrates. The sign that music is a technology is that it has to be taught to each generation afresh, rather than emerging spontaneously as part of normal human development. And Patel accounts for music’s universality not by identifying it as an aspect of the evolved human nature we share, but by characterizing its high value. He compares it with fire. The ability to control fire is very likely to be a trait that does not rely on evolutionarily specialized “fire making” cognitive mechanisms in the brain. Rather, the ability to control fire is almost certainly an invention, which spread culturally and became universal because of its utility for humans.
The notion of music as a transformational technology helps us to explain why music is universal in human culture. Music is universal because what it does for humans is universally valued. Music is like the making and control of fire in this respect. The control of fire is universal in human culture because it transforms our lives in ways we value deeply, for example, allowing us to cook food, keep warm, and see in dark places. Once a culture learns fire making, there is no going back, even though we might be able to live without this ability. Similarly, music is universal because it transforms our lives in ways we value deeply, for example, in terms of emotional and aesthetic experience and identity formation. (p. 401; italics in original)
On this view, the cognitive flexibility that leads us to invent, elaborate, and develop what comes our way – to make fire or music, for instance – is an adaptation, but the results of those acts of invention, development, and elaboration are not thereby merely byproducts or spandrels. They are designed applications, not adventitious byproducts. We put the relevant capacities to work with the intention of bringing about desired outcomes and when we are successful, those outcomes are achieved, not accidentally thrown up. Moreover, such technologies are transmitted via teaching, though once again, our evolved natures might contribute some of the raw material on which this process of teaching relies. So it is that reading and writing are passed to children.
If we now apply Patel’s view to art, we might speculate that art had its genesis in our inclination to create, design, and elaborate whatever comes to hand and that this also explains why it continues to renew and transform itself. It is an expression of our biological nature, certainly, but it is not thereby either an adaptation or a spandrel. It expresses the aspect of our nature that leads us to be inventors and users of technology and to be appreciators both of what technology does for us and of the achievement that goes into its creation.
Patel has offered (in a personal communication) a further positive argument for the conclusion that music behaviors are not a spandrel. He notes that music not only uses existing brain systems but can also shape those systems through mechanisms of neural plasticity. Growing evidence from neuroscience indicates that systematic engagement with music changes neural structures and connectivities. This neural feedback loop between mental technology and brain structures is not well captured by the metaphor of a spandrel, which implies a static architectural form, he suggests.
I wonder whether spandrels are incompatible with neural plasticity, however. When London taxi drivers acquire “the knowledge” – that is, when they learn the map of London and how to navigate across the city – there is a measurable change in their brains. But the capacity to remember and navigate one’s environment is surely an adaptation, and the capacity to acquire this knowledge through reading a symbolic representation is most likely a spandrel based on that adaptation.
Patel thinks that, because music behaviors are not genetically heritable, they are neither spandrels nor adaptations. Does his position stand up? In the following I question both the claim that music behaviors are not biological (on the grounds that proto-musical behaviors develop early and, in any case, the expression of some biological behaviors is time-delayed) and also the claim that if they were not biological they could not be spandrels or adaptations (on the grounds that they could be memes or that evolution can operate at the cultural level, as described by development systems theory or multilevel selection theory). I also consider an objection to this second claim along the lines that, even if they are not themselves biological, they must be dependent on biological mechanisms, and that might be enough for them to qualify as spandrels. I go on to reject this objection for being too profligate in implying that almost all human behaviors are spandrels. Finally, I critically discuss the comparison of music with fire. I suggest that, even if there is a basis for distinguishing transformational technologies from spandrels and for counting fire behaviors among the former, disanalogies between music behaviors and fire behaviors call into doubt Patel’s claim that music is also a transformational technology.
Here is the first argument: Patel claims as evidence that music behaviors are not genetically transmitted the observation that such behaviors do not emerge spontaneously as part of normal human development. This can and should be challenged, however. Of course, the appearance of music behaviors presupposes exposure to a musical environment, but the development of language similarly requires exposure to a linguistic environment. The fact that mastery of one’s culture’s tonal or modal system is not fully formed until, say, five to eight years of age is not sufficient to show that musical behaviors do not emerge spontaneously as part of normal human development. The onset of sexual maturity is an obvious example of a part of human development that emerges spontaneously only with age. And for that matter, full linguistic competence comes years after one’s first, rudimentary meaningful utterances and comprehensions. It could be that music is similar to the extent that it relies for its appearance on the realization of appropriate facilitating conditions. Perhaps it depends on the development of cognitive sophistication, the acquisition of emotional sensitivities, or the honing of auditory processing skills, for instance.
There is also the worry that Patel here sets the bar for music ability inappropriately high, since his example concerns how long it takes the individual to acquire full competence in his culture’s tonal/modal system. As I have already suggested, if Patel is to make good on his comparison between music and language, he is probably required to set the base level of musical behaviors much lower, indeed, at a level according to which it is appropriate to count us all as performers and composers as well as listeners. Given that, it seems plausible to count pitch-structured, metrically regular vocalizing as evidence of the acquisition of music behaviors, without also requiring that what is vocalized must have a complex structure exhibiting tonal closure. Pitch-structured, metrically regular vocalizing emerges earlier than five years of age in the child’s development. Such behaviors appear to be self-motivating and are often unprompted, which suggests the environmental triggering of genetic dispositions.
It is also relevant to observe that music develops much more spontaneously than reading and writing. According to Patel (personal communication), this is because music has many cognitive links to language and because music is emotionally rewarding from a very young age. But in that case, the acquisition of music behaviors is unspontaneous only as a matter of degree in the comparison with oral language acquisition, which is rather different from suggesting that such behaviors are purely cultural. In any case, the claim of music’s relative unspontaneity compared with language is surely contestable. I doubt that complete working familiarity and facility with one’s culture’s musical system takes longer than the same level of mastery of one’s native language. And as I have already suggested, if we look instead for the first appearance of behaviors that count as musical and as linguistic, where such behaviors are initially primitive compared to what comes later through normal development, it is not obvious that the linguistic behaviors are earlier or more spontaneous than the musical ones. Additionally, many would count the intrinsic appeal of music to babies and young children as highly suggestive of its connection with behaviors of evolutionary significance. One should not explain how quickly children take to music in terms of their preference for it unless one is prepared to take on the question of how that preference arises. Perhaps what is adaptive is motherese, with its quasi-musical character, and perhaps music behaviors derive from this as a spandrel that inherits the pleasure babies take in motherese.
This first argument challenges Patel’s view that it is evident in how and when they are acquired that music behaviors are not biologically transmitted. But even if Patel were right, and here we turn to the second argument, we could challenge the assumption he recruits from standard evolutionary theory, that what is not transmitted genetically could not qualify either as an adaptation or a spandrel. Those who advocate meme theory are liable to reject the division Patel draws between biological evolution and human technology. Less controversially, at least some evolutionary theorists – advocates of what is called developmental systems theory – argue that what matters is whether the relevant resources are reliably available to each subsequent generation, not whether their transmission is biological rather than purely cultural. And the case for “multilevel selection,” a version of group selection, has been strongly made.
In fact, Patel now acknowledges (personal communication) the plausibility of multilevel selection theory. He accepts that there can be cultural group selection for traits that promote intra-group cooperation. He comments: ‘If music promoted social cohesion within ancestral human groups, and if more cooperative groups prevailed in competition with other groups for limited resources, then it is possible that music could have acted as a cultural adaptation, without having originated as a biological adaptation. Furthermore, it is possible that there was feedback between cultural group selection and biological natural selection, so that there was some degree of selection for individuals who were more biologically predisposed to be musical. This would be an example of “gene-culture co-evolution”.’ But he remains skeptical of the truth of the antecedents of this conditional.
This brings us to the third argument, which assumes that the previous discussion shows that music is heritable, either genetically or via guaranteed cultural transmission, from one generation to the next, and assumes also that such behaviors are not adaptations in their own right. Now, however low we set the bar for what counts as music, it is the case that music depends on auditory capacities that are pre-musical – for instance, on the recognition of octave equivalence – which we share with some other non-musical animals. And those pre-musical capacities are likely adaptations for fitness-enhancing forms of auditory processing. Given the dependence of musical behaviors on these pre-musical adaptive capacities, then musical behaviors must count as spandrels from the point of view of evolution. They are byproducts of adaptations and they are useless from the point of view of evolution, having nothing to do with the reproductive fitness of those who display them.
Most human behaviors (and the products they issue in) depend at some point on our evolved cognitive, perceptual, affective, and motor systems. For that reason, this third argument implies that most human behaviors should be viewed as spandrels. Some theorists would find that corollary congenial. Stephen Jay Gould (1997, 1997) suggests that, with only 10,000 years of history behind them, both writing and reading are spandrels. Indeed, he regards language, human culture, and technology generally as byproducts of the oversized human brain that evolved to address now unknown problems faced by our ancestors. But others might think this notion of a spandrel is too profligate, and I am inclined to agree. The connection between the behaviors in question and the evolved capacities on which they depend is too attenuated for spandrel-talk to have explanatory power. So, if this approach shows that music behaviors are spandrels only by showing that almost all modern cultural behaviors are the same, it wins the argument at too high a price. By contrast with this extreme view, it seems reasonable to regard some behaviors as not adaptations and as not in any direct way side-effects of selection for something else either. In other words, it seems reasonable to regard some behaviors as neither adaptations nor spandrels. In Patel’s terms, adaptations and spandrels are to be distinguished from transformational technologies and, so he maintains, transformational technologies include both fire and music.
This takes us back to the comparison of music with fire. Are there differences between fire and music such that we could concede that fire is a transformational technology that is neither an adaptation nor a spandrel, yet plausibly claim that music is different and may be a spandrel? In what follows I consider how we should determine the border between spandrels and transformational technologies and I suggest that music falls on the other side of this border than fire does.
In my earlier exposition of Patel’s view, I shifted from talk of music behaviors to talk of music. I did so because I think this follows Patel’s usage and that something like his usage is required to sustain the analogy with fire that he draws. Songs and fires certainly are neither adaptations nor spandrels – they are artifacts, not traits or features of humans – but they could be the products of behavioral dispositions that count as spandrels. So, we should be comparing music behaviors with fire behaviors with respect to their relation, or lack of it, to relevant adaptations.
That is not easy to do, however, because the idea of fire behaviors is so vague and slippery. Take fire making. There are many ways of doing it – for instance, by taking something already burning from another fire, using friction to generate sufficient heat to ignite something, focusing the rays of the sun on something flammable, striking sparks from flints, using controlled electrical discharges, and generating heat by means of certain chemical reactions (for example, by exposing phosphorous to oxygen, adding pure calcium to water, pouring sulfuric acid onto sugar, decomposing organic matter in confined places). The problem is that these various activities seemingly have in common only that they can produce sufficient heat to ignite flammable material, that is, to bring fire into existence. As actions or behaviors they do not form a coherent class. I suggest that is why our focus falls naturally on the product, fire, rather than its causes and the behaviors that activate them.
Music behaviors are importantly different from this, I think. They are unified with each other and with the product, music. We hear music as the bodying forth of sound – on hearing music, the brain’s motor as well as auditory centers are stimulated. All the actions that go into its production and reception are directly relevant to the properties the product possesses. Even if we were inclined to agree that fire should be counted as a transformational technology, we could have reservations about characterizing music in such terms.
There is a further reason for querying Patel’s parallel between music and fire. He claims that we perpetuate both because we find them valuable. That claim is certainly plausible for fire. The value of fire is not merely that it is pleasant. It promotes survival by allowing us to withstand cold and by drying clothes. With fire we can purify contaminated water. It allows access to caves and it brings the light of day to night, thereby extending the time for social intercourse, for instance. It can be used to drive way predators, to stampede game animals, to smoke out game, and to harden weapons. Above all, fire eliminates bacteria from food and makes edible foods that we would not otherwise be able to digest easily.
So profound were the effects of controlling fire on the lives of our ancestors that we should pause to consider the history of fire control. It was our hominid ancestors who first controlled fire at least 790,000 years ago, as evidenced by ancient hearths. Because of its significance in the lives our forbearers, it changed their evolution. In particular, cooking and thereby softening food significantly increased the nutrition that could be extracted and this led to reduction both in the digestive tract and in the energetic costs of digestion. Indeed, Richard Wrangham (2009) speculates that fire control goes back to our hominid ancestor Homo erectus and that it was the extra spare energy that control of fire provided by way of cooked food that permitted the development of the energy-hungry larger brain that marked the line of descent to which we Homo sapiens are heir. In any case, fire is part of what is sometimes called the extended human phenotype (Sterelny 2001) – that is, an organism-defining product of our genes that falls beyond our bodies – not merely an interesting but biologically incidental artifact. Fire making may be a technology, but it is part of gene-culture co-evolution because our long-standing dependence on fire has fundamentally altered the human genome.
The claim that music is a source of similarly profound benefits is not so clear-cut. Patel identifies the value of music in the positive effect that it has on our emotions and sense of identity. I imagine he is correct about this, and it is worth noting that both music making and music responding can produce paybacks in these areas. We do not value music merely as a means to these desirable effects, however, and we rarely bear them in mind when we make or seek out music. Rather, we are inclined to regard the primary value of music as intrinsic to it, as inherent in the activities of making and responding to it. In part, this is the common thread that unifies the various music making and music responding behaviors. Music warms our world. But note that what we value in music and the consequential benefits that come from its engaging our emotions and providing models for us to identify with are only very indirectly, if at all, connected with survival and reproduction. As well, the benefits must be balanced against the costs. Music behaviors come with significant costs in terms of the time, concentration, and effort needed to raise them to the level of full competence. There is little doubt that we value music highly, but whereas the advantages for survival that controlling fire bring in its wake are patent and are sufficient to explain why the behavior is universal, the high value we attach to music is much harder to justify in similar terms.
Here is my suggestion: with fire, the behaviors that go into its production are merely causal means to the valued end. The means are various and they do not transmit to fire the qualities for which we value it. Fire behaviors have no underlying unity of structure and no common basis of derivation from any single cluster or cohort of adaptive behaviors. With music, on the other hand, there is an intimate connection between what we value in the product and the kinds of behaviors that generate it, such that music behaviors cannot be regarded merely as dispensable means to the end product. Music behaviors share a deep unity, all being concerned with the human generation of patterned sound or with following its patterns. And it is not difficult to trace moderately close connections between music behaviors and perceptual, cognitive, motor, and affective systems that are interdependently related in ways that are adaptive, even if we agree with Patel that music behaviors are not adaptive in themselves. These differences between fire and music (or, rather, fire behaviors and music behaviors) can reasonably be thought to ground the distinction we were looking for, that is, the distinction between transformational technologies that are product-oriented, not process-oriented, and behaviors that should better be classed as spandrels or adaptations. And when we add to this the observation that, though music is highly valued, it is far from obvious how music behaviors enhance the fitness of those who display them, then if we must choose between those behaviors being spandrels or adaptations, we should favor the former.
I have agreed with Patel that we should distinguish transformational technologies from adaptations and spandrels, and that fire behaviors should count as transformational technologies. I have argued, however, that music behaviors differ from fire behaviors in being mutually integrated and unified rather than being random but for the output they happen to produce, and in being self-motivating rather than end-driven. In addition, whereas the value of fire lies in the product rather than in its production, and whereas there is an obvious connection between fire’s promoting our survival and our valuing it, with music the value is as much in the process as the product and, if it promotes our fitness at all, this seems incidental to our valuing of it. These considerations cast doubt on the claim that music behaviors are transformational technologies. Meanwhile, if the alternative is that such behaviors must be either adaptations or spandrels, the fact that they are costly and not directly connected to survival and reproduction suggests that the latter is more likely.
Blackmore, Susan. The meme machine. Oxford: Oxford University Press, 1999.
Cross, Ian. “Music and cognitive evolution,” in The Oxford handbook of evolutionary psychology, edited by R. I. M. Dunbar & L. Barrett, 649–67. Oxford: Oxford University Press, 2007.
—. “Music and meaning, ambiguity and evolution,” in Musical communication, edited by D. Miell, R. MacDonald, and D. Hargreaves, 27–42. Oxford: Oxford University Press, 2005.
Darwin, Charles. The descent of man and selection in relation to sex. London: John Murray, 1871. Second edition 1874.
Davies, Stephen. “Why art is not a spandrel,” British journal of aesthetics, forthcoming 2010.
Dawkins, Richard. The selfish gene. Oxford: Oxford University Press, 1989. Second edition.
Dennett, Daniel C. “Memes and the exploitation of imagination,” Journal of aesthetics and art criticism 48 (1990): 127–35.
Dissanayake, Ellen. “Antecedents of the temporal arts in early mother-infant interaction,” in The origins of music, edited by N. L. Wallin, B. Merjker, and S. Brown, 389–410. Cambridge, MA: MIT Press, 2000.
Dunbar, Robin I. “The origin and subsequent evolution of language,” in Language evolution, edited by M. H. Christiansen and S. Kirby, 219–34. Oxford: Oxford University Press, 2003.
Fessler, Daniel M. T. “A burning desire: Steps toward an evolutionary psychology of fire learning,” Journal of cognition and culture 6 (2006): 3–4.
Gamble, Clive. Origins and revolutions: Human identity in earliest prehistory. Cambridge: Cambridge University Press, 2007.
Gould, Stephen Jay. “Evolution: The pleasures of pluralism,” The New York review of books 44 (11) 26 June, 1997.
—. “The exaptive excellence of spandrels as a term and prototype,” Proceedings of the National Academy of Science (US) 94 (1997): 10750–5.
Griffiths, Paul, and Russell Gray. “Developmental systems and evolutionary explanation,” Journal of philosophy 91 (1994): 277–304.
—. “Discussion: Three ways to misunderstand developmental systems theory,” Biology and philosophy 20 (2005): 417–25.
Huron, David. Sweet anticipation: Music and the psychology of expectation. Cambridge, MA: MIT Press, 2006.
Janata, P., and S. T. Grafton. “Swinging in the brain: Shared neural substrates for behaviors related to sequencing and music,” Nature Neuroscience 6 (2003): 682–7.
Koelsch, S., T. Fritz, D. Yves, V. Cramon, K. Müller, and A. D. Friederici. “Investigating emotion with music: An fMRI study,” Human brain mapping 27 (2006): 239–50.
Levitin, Daniel J. This is your brain on music: The science of human obsession. New York: Dutton, 2006.
Mark, Thomas Carson. “Philosophy of piano playing: Reflections on the concept of performance,” Philosophy and phenomenological research 41 (1981): 299–324.
Miller, Geoffrey. The mating mind: How sexual choice shaped the evolution of human nature. New York: Doubleday, 2000.
Patel, Aniruddh. Music, language, and the brain. Oxford: Clarendon Press, 2008.
Pfeiffer, John E. The creative explosion: An inquiry into the origins of art and religion. New York: Harper and Row, 1982.
—. The emergence of humankind. New York: Harper and Row, 1985. Fourth edition.
Pruetz, Jill D. and LaDuke, Thomas C. “Brief communication: Reaction to fire by Savanna Chimpanzees (Pan troglodytes verus) at Fongoli, Senegal: Conceptualization of ‘fire behavior’ and the case for a chimpanzee model,” American Journal of Physical Anthropology, forthcoming 2010.
Richerson, Peter J., and Robert Boyd. Not by genes alone: How culture transformed human evolution. Chicago: University of Chicago Press, 2005.
Sterelny, Kim. “Niche construction, developmental systems, and the extended replicator,’ in Cycles of Contingency: Developmental Systems and Evolution, edited by S. Oyama, P. E. Griffiths, and R. D. Gray, 333–49. Cambridge, MA: MIT Press, 2001.
Trehub, Sandra. “Human processing predispositions and musical universals,” in The origins of music, edited by N. L. Wallin, B. Merjker, and S. Brown, 427–48. Cambridge, MA: MIT Press, 2000.
Wilson, David Sloan. “Group-level evolutionary processes,” in The Oxford handbook of evolutionary psychology, edited by R. I. M. Dunbar & L. Barrett, 49–55. Oxford: Oxford University Press, 2007.
Wilson, David Sloan, Mark Van Vugt, and Rick O’Gorman. “Multilevel selection theory and major evolutionary transitions: Implications for psychological science,” Current directions in psychological science 17 (1) (2008): 6–9.
Wrangham, Richard. Catching fire: How cooking made us human. New York: Basic Books, 2009.
 In fact, in Davies forthcoming 2010 I suggest that the human inclination to produce art could not remain a spandrel even if it started as one.
 In terms of the comparison with language, we might put this by saying that quotation is more dominant in music than novel assertion. See Mark 1981, where musical performance is modelled on quotation.
 This assumes that, as adaptations, music behaviors must be strongly heritable and universal. Patel is entitled to testing this unusually strong assumption because those who identify music as an adaptation usually subscribe to it.
 See http://news.bbc.co.uk/2/hi/science/nature/677048.stm.
 See Dawkins 1989, Dennett 1990, and Blackmore 1999.
 See Griffiths and Gray 1994, 2005.
 See Wilson 2007, Wilson, David Sloan, Mark Van Vugt, and Rick O’Gorman 2008, and Richerson and Boyd 2005.
 See Janata and Grafton 2003 and Koelsch et al. 2006.
 If we consider fire not in terms of how it is caused but with respect to the motivation for having it, the gap between fire behaviors and music behaviors might be reduced, but this is because we then move fire nearer the border with spandrels, not because music moves nearer the border with transformational technologies.
 See Pfeiffer 1982, 1985 and Fessler 2006. Gamble (2007) gives the more conservative date of 400,000 years ago for hearths. While our nearest relative, the chimpanzee does not control fire, it is interesting that in some respects it shows a more sophisticated understanding of wild fire than modern humans attain; see Pruetz and LaDuke forthcoming 2010.
 See Sterelny 2001 and Fessler 2006.
 See Wrangham 2009. Earlier I quoted Patel on fire as saying “there is no going back, even though we might be able to live without this ability” (p. 401), but this fails to acknowledge how fire has changed us in irreversible ways.
 I am grateful for comments received from Joseph Carroll, Sherri Irvin, Justine Kingsbury, Jonathan McKeown-Green, and Kim Sterelny.