Debating the Work of Stephen Davies

Ryan Paul Doran

Universities of Sheffield & Antwerp

 Shelby Moser

University of Kent

This first issue of Debates in Aesthetics is concerned with the work of Stephen Davies; and, perhaps unsurprisingly given Davies’ prodigious interest in music, all three of the commentaries contained within this issue focus on his contributions to debates in the philosophy of music.


Anton Killin brings Davies’ most recent views on the evolution of art, and particularly music, to bear on his contour account of musical expression.[1] Killin provides additional motivation for the arousal account of musical expression, and specifically the phylogenetic dependence of emotional expression via contour on emotional expression via arousal. Indeed, drawing together a series of Davies’ most recent claims about the evolution of music, Killin observes that there is some precedent for such a view in Davies’ recent work: Killin notes that Davies suggests that the presence of musical practices in our first musical ancestor—which Davies believes is Homo heidelbergensis—may be phylogenetically dependent on the tendency (and “know how”) to “vent her feelings in a musical fashion”.[2]

Killin suggests that Hiedelbergensian crooners may at least sometimes have registered and communicated the emotions they felt in music, and presumably aroused these emotions in others, and that this might have proved adaptive. In support of this, Killin seems to have in mind the duel songs of the Inuit (following anthropologist Jean Briggs). As Inuit communities transformed from small nomadic societies to larger settlement societies, music in the form of duel songs allowed the safe relief and management (through laughter) of negative emotions that would presumably otherwise have been socially destructive.

In response, Davies suggests that Killin’s point should be understood as one about the importance of arousal and self-expression in certain music-making and -appreciating practices.[3] Davies denies, however, that this has any bearing on the question of the truth-makers of statements such as ‘this music is expressive of sadness’. Davies may be partly motivated by the thought that the empirical question of how we, as a matter of fact, come to find music expressive is logically independent from the normative question of what makes statements about the emotional expressivity of music true or false.

Those sympathetic with stronger forms of philosophical naturalism—as Davies himself surely is—may worry about the extent to which this response hits its target. Killin may be understood as making the point that the phylogenetic history of music might come to bear on the truth-makers of statements about expression. For Killin, the truth-maker of such statements about expression may need to be relativized to the species of hominin concerned, and furthermore, the stage of phylogenetic development of that hominin species, together with the type of music concerned.

Indeed, once we see that the ability to hear music as emotionally expressive is likely to have emerged slowly, and that there is surely a true account of its phylogenetic development (which may, as Killin believes, plausibly involve emotional arousal), it is difficult to see how these empirical facts wouldn’t have a bearing on the question of what makes it true that a given piece of music is emotionally expressive.

If that thought is right, then one might think that the force of Killin’s suggestion is at least linked to the plausibility of his empirical assertions about the phylogenetic history of music. And here, readers may harbour some doubt. One aspect that speaks in favour of Killin’s proposal of ‘arousal first, contour second’ is that the adaptive advantages of expressing an occurrently felt emotion or of emotionally stirring someone by musical means are clear and plausible. Someone who expresses their anger in music may be able to manage conflict in a more controlled fashion. But, on further inspection, the examples Killin suggests may not clearly support the link with the arousal theory of musical expression by themselves. For where the emotion aroused by the Inuit duel songs is mirth, the emotion expressed by the music is anger. A better example may be the use of music to emotionally entrain group members. As a result, while it may be clear that the arousal of certain emotions may be advantageous, it is not clear that this is linked to emotional expression.


Matteo Ravasio takes aim at Davies’ contour theory of musical expression.[4] Whilst Ravasio does not deny that a great deal of music may be literally expressive of emotions in virtue of presenting the behavioural correlates of emotion, Ravasio argues that this cannot account for all emotional expression in music. Ravasio presents three examples that put pressure on Davies’ account. The first is the ascending glissando at the beginning of Gershwin’s Rhapsody in Blue. Ravasio claims that the glissando is underdetermined as to the behavioural correlate it presents. As a consequence, Ravasio suggests that if the ascending glissando were expressive in virtue of the behavioural correlate of emotion it presents, it should be ambiguous in the emotion it expresses. Since Ravasio believes that it is not, he concludes that it cannot be expressive in virtue of presenting the behavioural correlates of emotion. Similarly, Ravasio suggests that his second example—the menacing character of a saturated, dark timbre—cannot be expressive in virtue of a link between the timbre and some behavioural correlate, since these timbres can occur in the absence of menace.

Davies’ counters these examples by denying that the glissando in Gershwin’s Rhapsody, and the dark saturated timbre, are determinatively expressive; and claiming that, as a result, they do not constitute counterexamples to the contour account.[5] Indeed, if they are expressively ambiguous, and if the emotional appearance that is represented is ambiguous (as Ravasio claims), then this might be taken as further support for the contour account.

Readers might also worry that Ravasio’s examples seem to set the bar for emotional expression by contour too high. Davies’ account need not involve the claims that something is emotionally expressive only if it can be seen to present the behavioural correlate of one emotion; nor that a given behavioural correlate is emotionally expressive in music only if it always co-occurs with the corresponding emotional episode in non-musical contexts. Rather, the contour account can more plausibly be taken to involve the claim that we learn to recognise music as emotionally expressive by picking out statistical regularities between behavioural correlates and emotional episodes. As a result, Davies’ contour account may both allow for dark saturated timbres to be expressive of menace and Geschwind’s glissando in his Rhapsody to be determinatively expressive (in some manner).

This latter potential response, however, doesn’t apply to Ravasio’s third example. Ravasio points out that it can’t be the case that the tenseness of a tritone is the result of experiencing some behavioural correlate. Ravasio observes that it cannot be the case that it resembles the sound of a tense voice, since we ordinarily cannot produce three pitches at once; nor can it resemble a tense movement, since the tritone does not consist of a succession of musical sounds (which Ravasio notes is necessary for hearing movement in music). Davies responds by denying that tritones (and perhaps similar musical entities such as an isolated minor triad) are straightforwardly emotionally expressive.

Readers might wonder whether this response does enough to assuage Ravasio’s objection. For whilst tension is not an emotion per se, it certainly seems to be a state that one can feel and express (much like other affective states that are not fully-fledged emotions, such as moods). As such, there doesn’t seem to be any obvious and principled reason why the contour account shouldn’t be expected to accommodate such cases. [6] Given this, readers might think that Ravasio and Davies’ exchange leaves Ravasio’s preferred account—namely emotional expression through secondary polysemy—a live possibility in future debate surrounding emotional expression.


Nemesio Puy takes up the issue of the relative ontological weight of transcriptions, versions, and musical works. Puy pursues three main lines of argument against Davies’ view that transcriptions are new works distinct from the original.[7]

Puy argues against Davies’ claim that the features relevant for the individuation of musical works are determined by the musical practices and conventions in place at the time at which the piece was composed. Since instrumentation has, according to Puy’s Davies, at least at times been conventionally considered to be an essential feature of the work, a change in instrumentation may be sufficient for the individuation of a new work. Puy claims that one can capture the fact that instrumentation was seen as increasingly relevant to the correct appreciation of works in a certain musical epoch without subscribing to the view that those conventions and practices determine that a change in the medium of a work is sufficient for a change in identity. Rather, Puy claims that one should understand this fact as mandating a preferred way for the work to be performed rather than an essential feature of the work itself.

In support of this, Puy notes that even in the nineteenth century—where, according to Puy’s Davies, the relevant musical practices and conventions dictate that instrumentation is essential to a work of art—the practices for work individuation, as indicated by naming practices, show that a change of instrumentation was not sufficient for a change in the work’s identity. Puy also offers two thought experiments which he thinks show that Davies’ view runs against our intuitions: Puy notes that a transcription of a work for the practicalities of a rehearsal would not be identified as a distinct work belonging to the transcriber; and that an audience member who expects the presentation of a new work would feel cheated by a transcription of another work even if it were a medium-specific work whose transcription involved creative skill.

In response, Davies makes two points.[8] The first is that transcriptions of works that are stripped down for the purposes of rehearsal should be considered reductions rather than transcriptions—and so, even if one holds the intuition that such works are distinct from the original, this case is not a counterexample to his view. The second is to argue that Puy’s challenge here doesn’t take account of the fact that there is an established, and apt, way of referring to such works, which acknowledges both the primacy of the original composer and the contribution of the transcriber. One might think, like Davies, that the latter response at least blunts the force of Puy’s objection; but one might also think that it doesn’t seem to unequivocally support the view he defends. For if the original composer is primary, and the title of the work remains the same, then arguably it doesn’t seem obvious that they are two different works.

Puy’s argument doesn’t rest on mere naming conventions, or on plumbing intuitions. And perhaps this is for the best, since readers might think that one of the conclusions that emerges from Puy and Davies’ exchange is that our naming practices are at least indeterminate with regard to their respective positions. Puy challenges what he takes to be two of Davies’ other theoretical reasons for considering transcriptions of medium-specific works (at least), but not versions, to be new works.

Puy interprets Davies as holding the view that a change of medium can result in a different work on the grounds that two works that differ in their aesthetic properties are different works, and that some aesthetic properties such as virtuosity are medium-dependent. Puy rejects these cases by arguing that all such aesthetic and medium-specific properties are more appropriately attributable to the performance, and only derivatively to the work itself. Taking the example of virtuosity, Puy argues that works can vary in the extent to which they are virtuosic—depending on the skill of the musician, technological change, or otherwise—without thereby changing their identities. Taking himself to have shown that the property of virtuosity, at least, is not essential to the identity of the work, Puy claims that there are no such aesthetic properties of a work that are medium-dependent and predicated of the work rather than the performance.

Davies makes two responses to Puy’s claim. The first is that virtuosity may indeed be a property of works rather than merely of performances. That performances of works can vary to the extent that they are virtuosic without changing their identity doesn’t show that virtuosity isn’t ever an essential property of the work. As Davies notes, there may be circumstances in which the work itself will be the source of the virtuosity. Even a work that is easily played may be seen as virtuostic under certain conditions. Davies’ second response is that, virtuosity aside, there are other aesthetic properties that are medium-dependent and apply to the work. Davies proposes that the properties of being rushed, and perhaps noisy, might qualify.

But perhaps there is less of a disagreement between Puy and Davies than might appear at first sight. After all, Puy also recognises the role of normal conditions, at least in identifying some properties that apply to performances. He claims, for example, that a work wouldn’t be considered virtuosic if an appropriate rendition were not virtuosic; whereas a performance cannot be atonal if the work is not atonal. Since one might think that the performance of a work may be atonal even if the work is not—imagine someone playing a work on an instrument that is somewhat out of tune—Puy may yet also wish to turn to normal conditions in identifying the properties that attach to works themselves.

On the other hand, Puy might agree with Davies that The Rite of Spring is a noisy work, but deny that this counts as an aesthetic property that is medium-specific. A performance utilising brass instruments alone will surely be noisier than one performed entirely with woodwind instruments, but the work itself might still be noisy to the extent that this is the upshot of the sound structure itself—irrespective of the particular instrumentation chosen.

Finally, Puy argues that Davies presses his view by drawing an infelicitous analogy between colour and form on the one hand, and timbre and musical structure on the other. Puy argues that the supervenience bases of colour and timbre are not sufficiently alike for Davies to be able to infer that timbre influences musical form from the fact that colour influences visual form.

Davies replies by countering that Puy misunderstands the purpose of the analogy. The relation between timbre and musical form is importantly similar to, but not the same as, the relation between colour and form. Moreover, Davies clarifies that the analogy is supposed to pick out a similarity between timbre’s contribution to the experience of musical form and colour’s contribution to the experience of visual form, and not the structural relation that holds within each pair.

A reader might worry that it is not easy to disentangle experience from structure here. The experience of colour might plausibly be thought to influence the experience of form in virtue of some fact about the physics of colour. As such, to the extent that the structures are different, one might not expect the experiences to be similar.

In light of this, one might think that if the point can be made without using an analogy with colour then so much the better, and Davies clearly thinks that it can. Davies claims that if we were to imagine Ravel’s Bolero rendered in pure sine waves we would “hear how the work fades away…”.[9] Clearly there would be a difference between this and a standard performance, but how is this difference to be glossed philosophically? One might worry that it is not clear that the timbre contributes to the form of the work. Indeed, it may be precisely because it is not clear that (let alone how) timbre influences form that the analogy with colour is so often invoked. But even if it is not true that timbre contributes to sound structure in any precise sense, Davies’ example may yet have the resources to deal with Puy’s objection. For are any of the differences between the original and the sine-wave version aesthetic, and medium-dependent, differences?

In closing, we would like to thank Anton Killin, Matteo Ravasio, Nemesio Garcia Puy, and Stephen Davies for their contributions to the issue, and forbearance whilst we organised the re-launch of the journal. We hope that readers will find sustenance for further debate about Stephen Davies’ work in these pages, as well as the issues he addresses.


Davies, Stephen. (2015). “How Ancient is Art?” Evental Aesthetics 4 (2), pp. 22–45.

—. (2017). “Music Matters: Responding to Killin, Ravasio, and Puy”. Debates in Aesthetics 13 (1), pp. 91-106.

Killin, Anton. (2017). “Hominin Musicality and Musical Expressivity: Revisiting Davies’ Contour Account”. Debates in Aesthetics 13 (1), pp. 18–27.

Puy, Nemesio. (2017). “Musical Versions and Transcriptions: Disputing Stephen Davies”. Debates in Aesthetics 13 (1), pp. 53-74.

Ravasio, M. (2017). “Stephen Davies on the Issue of Literalism”. Debates in Aesthetics 13 (1), pp. 28-52.

[1] Killin 2017.

[2] Ibid., p. 14, citing Davies 2015, pp. 28-29.

[3] Davies 2017.

[4] Ravasio 2017.

[5] Davies 2017.

[6] We are grateful to Matteo Ravasio for clarifying some of Davies’ existing views in relation to this point.

[7] Puy 2017.

[8] Davies, op. cit.

[9] Ibid., p. 64.