Before proceeding, it is fitting to make some terminological remarks. By absolute music, I mean music that is unadorned by any explicit extramusical content. This obviously excludes song, and it also excludes programmatic music of all stripes, where a program can be anything from a descriptive title (for example, Debussy’s “Et la lune descend sur le temple qui fut”) to a lengthy description of what the music is meant to depict (Berlioz’s Symphonie Fantastique being the paradigm example).[1] These uses are not unusual and therefore should not be problematic. In addition, I will make use of a distinction suggested by Robinson between expression and expressiveness. For Robinson, expression in general is both an author-centered and an audience-centered notion, where “‘expression’ is used for the author-centered aspect,” and where “‘expressiveness’ is used for the audience-centered aspect.”[2] This distinction should lend itself to improved clarity.
As stated, I am attempting to offer a dual-level account of how absolute music expresses emotion, which is modeled on a dual-level account of mindreading, in particular that of Alvin Goldman in his Simulating Minds. Just as, for Goldman, we can distinguish between low-level mindreading and high-level mindreading, we can distinguish between low-level and high-level musical expressiveness. I shall begin by discussing low-level expressiveness, and later I shall address high-level expressiveness. Low-level mindreading tends to be “comparatively simple, primitive, automatic, and largely below the level of consciousness.”[3] The paradigm case for low-level mindreading is face-based emotion recognition (FaBER). By far, the most prominent candidate to account for low-level mindreading like FaBER, and, I contend, low-level expressiveness, is mirroring. According to Goldman, a mirroring system “obtains when there is a systematic, repeatable causal pathway, leading from one individual’s mental state to a matching (or semimatching) state in an observer.”[4] Goldman states two components of a systematic causal pathway: “(a) a subpath within the sender from his own mental state to a behavioral expression of that state and (b) a subpath within the receiver from an observation of the sender’s behavior to a mental state that matches the sender’s.”[5] How does this account for musical expressiveness? Recall that expressiveness is a relationship between an expression, in this case a musical composition, and its audience; that is, an expression is expressive to its audience. This is important in order to make sense of the systematic causal pathway required for mirroring. Music is in a problematic position with regard to (a) because music does not have mental states, but it is expressive. Expressiveness is sufficient for (b), since it is the observation that triggers the mental state in the receiver. Consider the dogs from which so-called “doggy theory” of musical expression derives its epithet: Kivy’s St. Bernard and Davies’s basset hound. These dogs have faces that are expressive of sadness, but that does not mean that the dogs are sad. Rather, it seems plausible to say that these dogs’ faces bear the features requisite for a mirror response of sadness; in other words, their faces resemble[6] human faces enough to elicit a FaBER response.[7] Thus, to have a mirror response does not require the mirroring of something that actually has the mental state in question, it only requires that it be expressive in such a way as to trigger a mirror response, and the expressiveness can be perfectly unmoving, as mirroring can still account for identification. Now, this does not seem to be quite Goldman’s definition any longer, as Goldman does not delve into the possibility of automatic mirror responses to resemblances of human expressiveness, but there is no reason to think he would deny this.
If low-level musical expressiveness is to be accounted for in terms of mirroring, what resemblances are there between human expression and musical expression? There are two parallels to be drawn that lend themselves particularly well to explaining some of the expression of absolute music in terms of mirroring. First is a parallel between vocal expression and certain elements of musical expression. Second is a parallel between bodily expression and certain elements of musical expression. This duo of parallels has been observed time and again. The classic statement likely belongs to Cone, who says that “both the verbal gestures of poetry and the bodily gestures of the dance are symbolized in the medium of pure sound.”[8] Robinson also mentions them, with reference to mirroring, observing that “doggy theory suggests one way in which [aspects of an emotion process can be mirrored by music]: music can mirror the vocal expressions and the motor activity—including expressive bodily gestures and action tendencies—that characterize particular emotions.”[9] This invocation of mirroring is quite in line with the account that I wish to give, though other uses of the term in Deeper than Reason make clear that I am using the term more narrowly.[10] In addition, Davies says, “the voice and music are more alike in dynamic structure, articulation, pitch, intensity, and periodicity of phrase lengths and shapes than in timbre or inflection as such,”[11] (though he downplays this element of musical expressiveness in favor of bodily expression) and shortly thereafter, “I think music is expressive in recalling the gait, attitude, air, carriage, posture, and comportment of the human body.”[12] Davies argues that, combined with a parallel between the temporal unfolding of music and that of human behavior, there is a resemblance between musical gestures and behavioral gestures, and this resemblance accounts for musical expressiveness. I am in agreement with Davies that there is a sense in which musical gestures resemble both vocal and bodily gestures, and it is in the sense that the resemblances are sufficient to evoke a mirror response.
The close relationship between vocal and motoric expression is something that all ordinary communicators know well. Vocal inflection can usually be expected to match bodily expression, including facial expression, posture, &c.) So the body, whose expressions are more primitive, can betray what the voice may try to obscure in an attempted deception. This relationship is even borne out by a common tendency to move the hands (among other mannerisms) when speaking. Thus it comes as no surprise that in music we find a similarly close relationship between the vocal-parallel (or resembling) expressions and the motor-parallel (or resembling) expressions.
A few comments are yet warranted on vocal mirroring in music. In addition to the litany of parallel features that Davies lists above, I am not as hesitant as he is in including the possibility that timbre, inflection, and so forth might have mirror responses, but, as I shall discuss below, low-level expressiveness is limited. That said, I think it is important to note the centrality of the human voice in the history of Western music. Unaccompanied instrumental music developed out of the styles and formalities of vocal music.[13] The idea that much instrumental music should in certain respects be heard in much the same way as vocal music is a historically grounded idea. Now, granted that instrumental lines bear important resemblances to vocal lines, it stands to reason that instrumental music can express content which, if the piece were a vocal composition, would accompany the sung text. Now, if it is also granted that sung lines and, by extension, instrumental lines are capable of resembling, likely in an exaggerated way (to be expounded below), the human voice in such ways as those Davies mentions above, it seems that even absolute music can express through mirroring much of what is expressed by ordinary vocalization apart from the words. Of course, there are ways of speaking that are expressive of certain emotions, and conversely there are certain emotions that can be roused by speaking in a particular way, which is true both for oneself and for one’s company. So certain ways of speaking are correspondent with certain emotions, and so when vocal patterns are mirrored, they give rise to those emotions. This could be minimally expressive, which would admit of little more than the identification of the intended expression, or it could be fairly deeply expressive, causing one to feel the emotion expressed.
Motoric mirroring is evidenced, in part I think, by the naturalness with which human beings incline to move with music, be it foot-tapping or ballet. The fact that there are appropriate dance responses to music, many of which are known automatically, and that such basic movements as foot-tapping are often begun unconsciously, is suggestive of mirroring activity. Of course, two of the indicators low-level mindreading, which are perfectly applicable to mirroring, are that the response is automatic and unconscious. A forthcoming paper by Gregory Currie, “Empathy for Objects,” has much to say about motoric simulation, which includes motor mirroring. Though Currie is cautious about the appropriateness of employing motor simulation to understand artwork, it is important to note that he is focused on visual art forms, for which a motoric response is not at all so natural. He says, “There is little to say in general terms about the relation between these motoric processes and the aesthetic beyond noting, unhelpfully, that these processes play an aesthetic role when they play a role in the generation of a response which is an aesthetic one,”[14] but is unsure about when this is appropriate. I suggest, on the basis of what has been said above, that music is a paradigm case for the appropriateness of motoric response, save for dance itself. If indeed we have mirror responses to musical resemblances of bodily gestures, even unconsciously, and this is as integral to understanding music as I claim, then it certainly seems that motoric simulation has its place in aesthetics. Further, the deliberate invention of dances, constrained of course by what is motorically mirrored unconsciously, is certainly an appropriate response to music, so even conscious motoric simulation can have its place. I make no attempt to comment on the appropriateness of motoric simulation in other arts, but I turn to another issue with low-level bodily expression.
An important problem remains, as Robinson notes that bodily expression in dance is often exaggerated,[15] just as vocal patterns are exaggerated in song. As in the case of song, the gestures still exaggerated resemble their ordinary forms, and they can therefore be expected to be identifiable by mirroring. One might object that such exaggeration obscures what is being expressed, but I think it is quite the contrary. Exaggeration is a means of clarifying an expressive gesture. An ordinary gesture might go unnoticed or be misinterpreted; exaggeration is a means of ensuring a very narrow range of readings.
Now, how is it that these bodily mirror responses are expressive of emotion? As in the case of vocal expression, I contend that certain bodily gestures are correlated with emotions. We speak of jumping for joy; we might slouch and slowly drag our feet when gloomy; in fear we tremble. Facial expression is a perfect example, since the facial expression of emotion is a two way street. To make a face is to dispose oneself toward a particular emotion, just as feeling that emotion is likely to cause one to make that face. FaBER does not factor into music, but it should highlight the close relationship between bodily expression and emotion. So, when music causes a mirror response, we are disposed toward feeling the emotion associated with that bodily expression, which can be highly expressive, hardly expressive, or anything in between.
Davies is somewhat conservative about what music can express, listing such possibilities as “sadness and happiness…timidity and anger…swaggering arrogance, the mechanical rigidity that goes with the repression and alienation from the physicality of existence, ethereal dreaminess, and sassy sexuality,”[16] naming candidates that he thinks are expressible by bodily gesture. Since Davies has largely limited his view to the low-level expressiveness discussed thus far, we can take a lesson from this comment; namely, low-level expression is limited, probably even more limited than Davies’s list suggests. Far more varied and complex are those emotions that can be read in a musical persona, which is the model I shall be using to account for high-level expressiveness.
Before moving on to high-level expressiveness, one more feature of low-level expressiveness is worthy of note. As Goldman points out, “mirroring doesn’t entail mindreading because the receiver may not impute anything to the sender.”[17] Mirroring can be done without reference to the composer, a musical persona, or anything of the kind. A doggy theorist would not likely find too much of what has been said up to this point particularly objectionable, except for the enormous claim that low-level expressiveness is only a part of the story. The fact is, through they do not have to be imputed to a sender, these expressions can be imputed to a musical persona to be included in his or her psycho-drama, which I will argue is the best account of high-level expressiveness. So, the claim is that whenever it is appropriate to posit a musical persona, the low-level expressions should be imputed to that persona, and they are to constrain the high-level interpretation of his or her psycho-drama.
To begin, some remarks must be made concerning what a persona’s psycho-drama is not. It is not to be understood as a drama simpliciter, for the correspondence between music and drama is only a correspondence between certain features. In his Antithetical Arts, Kivy makes much of this alleged correspondence. Though he takes himself to be arguing against the legitimacy of drawing such a correspondence at all, I think his remarks are much better suited to demonstrating the partial correspondence of music and drama, pointing instead to the features for which such a correspondence is legitimate. Kivy advances an argument from repetition. In so doing, he distinguishes between two kinds of musical repetition: external repetition, which is the repetition of lengthy, macro-structural sections of music, and internal repetition, which is the repetition of short motives throughout a piece.[18] He goes on to argue that an external repeat in music is akin to repeating a whole scene of a drama verbatim, and that internal repetition is like Hamlet repeating the line, “To be, or not to be,” over and over again.[19] By arguing thus, Kivy fails to distinguish between the persona’s psycho-drama and the overall drama, including the environment. This is illustrated quite well through the setting of poetry in song. The same music can be used to set more than one stanza of a poem, and each repeat can be just as appropriate as the last. This is because the music is not expressing exactly what is in each stanza, but what is psycho-dramatically common among the texts. So if a poetical speaker expresses the same sorrow in three different stanzas, each can properly be set to the same music. So there is a distinction between the psycho-drama that music is expressing[20] and the overall drama in which the psycho-drama is taking place. With regard to internal repetition, Kivy is guilty of the same error. A motive is not to be taken as representing the whole of a particular utterance, only certain features of it. So internal repetition is not necessarily akin to Hamlet’s repeating “To be, or not to be,” over and over again; rather, it is akin to Hamlet exhibiting the same emotion in various different utterances. Furthermore, internal repeats abound in literature, not least poetry, and it is perfectly possible for each repeat to express subtle differences. Think of Poe’s “The Raven.” This is perfectly applicable to absolute music, the difference being that the overall dramatic context of the psycho-drama is not explicit.
Furthermore, I do not wish to suggest that, when we listen to a piece like Debussy’s “La cathédrale engloutie” that the music is expressing the events that the program suggests. I wish to suggest that the music is expressing the persona’s psycho-drama in the context of the programmatic setting, as a response to those events. True, Debussy has composed portions of that piece in such a way as to imitate the sounds of church bells, but that is not musical expression; it is musical imitation. Imitative music can be highly expressive, as “La cathédrale engloutie” certainly is, but the imitation is not the expression. The music is an expression, and the imitation is a feature of the music, and the music is not just imitation. This is to further emphasize the the distinctness of the psycho-drama from the environment—it is a response to the environment, for in absolute music, the environment is a mystery, except insofar as the persona expresses his or her psycho-dramatic attitudes—both high and low-level, about it through the music.
I have been operating under the assumption that there is, properly speaking, one persona that might inhabit a piece of music, a view that I shall now defend. Davies has criticized the persona theory (which he calls “hypothetical emotionalism”) on precisely this basis, saying, “there is no constraint on the number of personas imagined as inhabiting a musical work. And there are innumerable coherent narratives about them that would coincide with and reflect the progress of the music.”[21] There are two objections here. I will answer the first by arguing that there is one persona in a given piece of music. I will answer the second by claiming that there is a sense in which “these different narratives all license the same judgment about what the music expresses,”[22] a claim that Davies acknowledges would solve the problem, but which sees no reason to accept as true.
To claim that there is but one persona in a piece of music is not arbitrary. Indeed, this concept goes back to the very beginning of the musical persona’s life as a philosophical concept. In The Composer’s Voice, Cone introduced the persona to the world, and for him, there is, properly speaking, only one persona in a piece of music. This claim includes but is not limited to absolute music; indeed, Cone uses his analysis of accompanied song to frame his discussion of the persona in instrumental music. In the case of instrumental music, Cone distinguishes the virtual persona from the agent and the idea, which correspond to the persona, protagonist, and characters of a song. The complete virtual persona in music “is the experiencing subject of the entire composition, in whose thought the play, or narrative, or reverie, takes place—whose inner life music communicates by means of symbolic gesture.”[23] Agents, “like the characters in an opera, must obey the formal demands of the music; but, again like operatic characters, they must appear to move freely—to compose their own parts, as it were.”[24] Agents come in several stripes,[25] and they could be instruments, they could be a bit of thematic or motivic material, or they could be a bit of harmonic material (perhaps Scriabin’s Promethean Chord is a good example), &c. An idea can be presented to the mind of an agent or to that of the persona: “every musical gesture conveys an idea or image in the minds of the agent making the gesture and of the musical persona.”[26] Now, I am arguing that there is properly speaking only one persona in a given piece of music, and the main reason for this is that there is only one perspective from which the psycho-drama unfolds. Someone must respond to the agents (and the ideas, with the agents), and certain features of those responses, particularly the emotional ones, are expressed in the music. If there is no response, there is no music; for the music is the psycho-dramatic response of the protagonist to whatever is happening in his or her environment. Such is my endorsement of Cone’s insistence on a single virtual persona. Where one might naturally incline to posit multiple personas, such as a piano concerto (the piano and orchestra are each personae), agents come to the rescue, and the whole of a piece of music can still be taken as an unfolding expression of a single mind, and it is the listener’s duty to both mirror and reflect on this mind’s drama, if he or she is to listen to such a psycho-dramatic work properly.
Now, my fundamental claim regarding the sense in which we distinguish between acceptable interpretations of absolute music is that our range of interpretations is constrained. The most important interpretive constraint comes from what is automatically perceived as low-level expressiveness. Whatever the persona does or feels, it must be consistent with what the low-level expressiveness dictates. This is feasible as a constraint because I take these low-level mirroring responses to be nigh-universal[27] among comprehending listeners, to borrow Levinson’s term,[28] and so they are suitable as objective constraints. After all, if I am right about low-level expressiveness the music was written by and for people who mirror sound constructs in the same way. It therefore comes as no surprise that absolute musical expression should have objective expressions like this. As we have seen, if we are to consider a high-level interpretation of musical expression, it is necessary to impute the low-level expressions to the virtual persona; this way, these constraints are immediately in place. It behooves the listener then to consider the unfolding of these expressions as part of a similarly unfolding psycho-drama, so as to account for the relations among them in suitably psycho-dramatic terms. It is also here that we can point to sections of music as the agents and ideas that the persona is experiencing, which is probably best done with the help of structural and harmonic analyses. Sometimes, it may prove helpful to bring historical considerations to the table, for example, it might be a rare case in which one might be warranted to identify the persona with the composer,[29] or perhaps we can identify someone as an agent or specify an idea. Lastly, simpler interpretations tend to be better interpretations. All this allows one to attribute complex emotions and other mental states to the musical persona; for example, if a gloomy piece comes to its close with a perfect authentic cadence, one might attribute hope or some such to the persona. Mirroring will never reveal that; only reflection, and without a persona, one is left wondering to what to attribute hope.[30]
Robinson has a few comments on consistency and legitimacy among different interpretations. Perhaps the most striking one is the limitation of her claims about the justified invocation of the musical persona almost exclusively to the music of the Romantic era. More precisely, Robinson formulates the constraint to say, “one way of specifying content is in terms of how the composer broadly intended the piece to be construed.”[31] So, listening to music as a persona’s psycho-drama is just one way of listening that is only appropriate insofar as the composer has such a thing in mind, as did nearly any composer writing under the influence of the Romantic theory of expression. I count it as an advantage of my dual-level account that it can adequately capture how both psycho-dramatic Romantic music and J. S. Bach can be expressive from this historically sensitive perspective. The Romantics take the low-level expressiveness and let it guide a grander dramatic expression, where baroque expressiveness cannot often be said to go beyond what can be derived from low-level expressiveness. The advantage here is that all sorts of music can be called expressive, to the degree that the composer conceived appropriate. No one wants to argue that, in the absence of personas and psycho-dramas, music was not expressive; that person will have a dreadful time making sense of historical ecclesiastical disputes over the primacy of text and appropriate expressiveness in sacred music. My account allows for low-level expressiveness to be highly expressive at times, albeit limited as to what it is can express.
Here, then, is the sense in which, for high-level expressiveness, “these different narratives all license the same judgment about what the music expresses.” We must distinguish between the objective constraints on interpretation, which define the range of possible interpretations. We must impute the objective, low-level expressions to the persona. We must define the agents, ideas, and the like that comprise the composite persona, for only then do we have any conception who the persona is in terms of high-level emotions that are only attributable through reflection on the music’s expressiveness, not mere mirroring. Additionally, historical considerations may be appropriate, depending on the composer and the piece. Finally, we must bear in mind that simpler interpretations tend to be better interpretations. Now, it is indeed possible to arrive at several different versions of the persona’s psycho-drama, and therefore it is doubtful that Davies could be satisfied with this response. It does not satisfy his condition as he wanted it to be satisfied, but I contend that fulfilling the condition as Davies would like would not do justice to musical expression. The reason for this is that, just as the interpretation of music in performance allows for several legitimate interpretations, as Robinson says in reference to her interpretation of the Brahms Intermezzo op. 117, no. 2, mentioning several other possible interpretations: “Possibly different pianists could perform the piece in each of these different ways and thus arrive at a different resolution to the psychological drama,”[32] so too does psycho-dramatic interpretation.
In summary, I have proposed that there are two levels of musical expressiveness, following Alvin Goldman’s high-level and low-level mindreading. I have argued that low-level expressiveness can be explained largely through mirroring, and this allows us to recognize musical expressiveness automatically and sometimes unconsciously. These are not necessarily attributed to any mind, but in cases where a musical persona is an appropriate interpretive method, according to whether or not the composer thought of his or her composition in that way, the low-level expressions are imputed to the persona, and this constitutes an instance of low-level mindreading. These imputed emotions constrain the psycho-dramatic narrative in which high-level emotions and mental states that cannot be mirrored are attributed (through high-level mindreading) to the persona (I have suggested that hope is such a state, but examples are many), as does a guiding principle of simplicity. Historical considerations, when appropriate, are also suitable constraining interpretation. This, I hope, is a framework that provides a responsible means of musical interpretation that is sensitive to a variety of considerations, be they historical, psychological, literary, or philosophical.
Bibliography
Cone, Edward. The Composer’s Voice. University of California Press, Berkeley: 1974.
Currie, Gregory. “Empathy for Objects.” (forthcoming)
Davies, Stephen. “Artistic Expression and the Hard Case of Pure Music.” Contemporary Debates in Aesthetics and the Philosophy of Art. Matthew Kieran, ed. Blackwell, Malden, Massachusetts: 2006. pp. 179-191
Goldman, Alvin. Simulating Minds: The Philosophy, Psychology, and Neuroscience of Mindreading. Oxford, New York: 2006.
Kivy, Peter. Antithetical Arts; On the Ancient Quarrel between Literature and Muisic. Clarendon, Oxford: 2007.
Levinson, Jerrold. The Pleasures of Aesthetics. Cornell, Ithaca, New York: 1996.
Robinson, Jenefer. Deeper than Reason: Emotion and its Role in Literature, Art, and Music. Oxford, New York: 2005.
---. “Expression and Expressiveness in Art.” Postgraduate Journal of Aesthetics. Vol. 4, no. 2 (Aug., 2007), pp. 19-41
[1] Edward Cone would distinguish between an explicit program of the Symphonie Fantastique and an implied program of “Et la lune,” but the definition stands: both are programmatic. See Cone, Edward. The Composer’s Voice. University of California Press, Berkeley: 1974. Pg. 83.
[2] Robinson, Jenefer. “Expression and Expressiveness in Art.” Postgraduate Journal of Aesthetics. Vol. 4, no. 2, (August 2007), pp. 19-40. Pg. 19.
[3] Goldman, Alvin. Simulating Minds: The Philosophy, Psychology, and Neuroscience of Mindreading. Oxford, New York: 2006. Pg. 113
[4] Ibid. 133
[5] Ibid. 133
[6] Resemblance is important to Davies’s theory, as we shall see below.
[7] Robinson remarks that “There is something comical about the basset-hound’s sad face, after all: it looks sad, but it doesn’t make me feel sad” (Deeper than Reason: Emotion and its Role in Literature, Art, and Music. Oxford, New York: 2005. Pg. 310). This is to emphasize that the dog’s face is not particularly expressive of sadness, and, I would add what expressiveness it might have had is easily ignored in light of the irony to be found in a human-like expression on a non-human face that is not actually expressing anything of the sort.
[8] Cone (1974) 164
[9] Robinson (2005) 311
[10] On pg. 311 and 312 of Deeper than Reason, Robinson says that music can mirror “the cognitive or evaluative aspects of emotion," such as desire. In addition, music is said to mirror memory and “evaluations of the environment.” I do not claim that any of these are beyond the capacities of musical expression, so I am not disputing what Robinson is actually saying here. I am merely noting a terminological difference.
[11] Davies, Stephen. “Artistic Expression and the Case of Pure Music.” Contemporary Debates in Aesthetics and Philosophy of Art. Matthew Kieran, ed. Blackwell, Malden, Massachusetts: 2006. Pg. 181
[12] Ibid. 182
[13] Interestingly enough, an enormous proportion of it was dance music early on. Here again we meet the duo of vocal and bodily expression.
[14] Currie, Gregory. “Empathy for Objects.” (forthcoming) pg. 16
[15] Robinson (2005) 286
[16] Davies 183
[17] Goldman 134
[18] Kivy, Peter. Antithetical Arts: On the Ancient Quarrel between Literature and Music. Clarendon, Oxford: 2007. Pg. 109
[19] Ibid. 110
[20] This is not to say that music is just expressing a persona’s psycho-drama, since music is perfectly capable to expressing more than that, including some elements of the drama external to the psycho-drama; for example, by imitating sounds in nature, such as birdcalls.
[21] Davies 190
[22] Ibid. 190
[23] Cone 94
[24] Ibid. 88
[25] Ibid. 96
[26]Ibid. 92
[27] This is not to deny that there will be variations in things like degree of affect among different listeners at different times (and even the same listener at different times).
[28] Levinson, Jerrold. “Musical Literacy.” The Pleasures of Aesthetics. Cornell, Ithaca, New York: 1996. pp. 27-41
[29] As Robinson (Deeper than Reason pp. 337-348) does with her interpretation of the Brahms Intermezzo op. 117, no. 2. Other examples arguably include Mussorgsky’s Pictures at an Exhibition and of course the Symphonie Fantastique, though the latter two are both programmatic and therefore have even greater interpretive constraints placed upon them.
[30] This is not to say there are not other accounts, but it is not apt to discuss them here.
[31] Robinson (2005) pg. 325
[32] Ibid..344
