Methodology in aesthetics: the case of musical expressivity

Philos Stud DOI 10.1007/s11098-010-9571-7 Methodology in aesthetics: the case of musical expressivity Erkki Huovinen Tobias Pontara Ó Springer Science+Business Media B.V. 2010 Abstract A central method within analytic philosophy has been to construct thought experiments in order to subject philosophical theories to intuitive evaluation. According to a widely held view, philosophical intuitions provide an evidential basis for arguments against such theories, thus rendering the discussion rational. This method has been the predominant way to approach theories formulated as conditional or biconditional statements. In this paper, we examine selected theories of musical expressivity presented in such logical forms, analyzing the possibilities for constructing thought experiments against them. We will argue that philosophical intuitions are not available for the evaluation of the types of counterarguments that would need to be constructed. Instead, the evaluation of these theories, to the extent that it can succeed at all, will centrally rely on inferential, non-immediate access to our subjective musical experiences. Furthermore, attempted thought experiments lose their methodological function because no proper distinction can be drawn between the persons figuring in the thought-experimental scenario and the evaluator of the scenario. Consequently, some of the central contributions to what is generally understood to be analytic philosophy of art are shown to represent a form of aesthetic criticism, offering much less basis for rational argumentation than is often thought. Keywords Intuition Evidence Thought experiments Aesthetics Philosophy of music Expressivity E. Huovinen (&) University of Minnesota/School of Music, 80 Orlin Ave. SE, Minneapolis, MN 55414, USA e-mail: huovinen@umn.edu T. Pontara Department of Musicology, Åbo Akademi University, Piispankatu 17, 20500 Turku, Finland

E. Huovinen, T. Pontara 1 Introduction At least since several notable efforts during the 1950s and 1960s, the framework of analytic philosophy has customarily been taken to include aesthetics, or, more generally, the philosophy of art (see, e.g., Kennick 1964; Goodman 1968; Wollheim 1968). Apart from some philosophers such as Nelson Goodman, who pursued aesthetics as a part of a more wide-ranging philosophical program, work within this field has arguably suffered from limited integration to general philosophical discussion. It is not that many philosophers of art would not be well-versed in general philosophical issues, but rather that the debates of aestheticians have not always reached the attention of general analytic philosophers in the way that the latter are usually aware of advances in other sub-fields of analytic philosophy. There may be rather mundane reasons for this. For one, central discussions of analytic aesthetics have appeared in specialized journals that do not cover other areas of philosophy. For another, some philosophers may view the realm of art as somewhat mysterious, if not for any other reason than that art forms such as music are accompanied by their own special notations and formal practices which may seem repulsive to the non-initiated. In any case, it may be argued that the analytic philosophy of art has been left to develop in relative seclusion from general analytic philosophy often taking impulses from it but less often taking initiative in general philosophical developments or being itself critically examined by the larger philosophical community. Commenting on the early analytic philosophy of art, Scruton (1983, p. 5) implied that an emphasis upon logical competence and detailed analysis of critical argument had often eclipsed willingness to explore the philosophical foundations of the subject. Even if much progress in this sense may have been made since Scruton s comments, the problem still exists. Moreover, we would like to suggest that pace Scruton, the lack of exploring the philosophical foundations pertains to some extent exactly to the formal, methodological issues relevant to philosophical argumentation. Writing on practicing the philosophy of music, Kivy (2002, p. 13) summarizes this non-reflective attitude succinctly: it should seem obvious [ ] that you needn t have a philosophy of philosophy to do, to know how to practice philosophy. While acknowledging that even Kivy himself would not probably take such a stance to its extremes by relying merely on instinctive methodological choices, we will take the opposite view, demonstrating how some of the discussions within the philosophy of art may be severely crippled by the lack of concern to philosophical methodology. Even though our essay will focus on one salient issue in recent aesthetics the question of expressivity our wider purpose is to awaken interest in related methodological issues within the field of aesthetics at large, prompting larger-scale examinations of the disciplinary status of aesthetics as a part of analytic philosophy. This said, we may turn to introduce the issues at hand. The question of expressivity in art is certainly one of the characteristic questions that has occupied aestheticians almost across the board. What does it mean to say that an artwork such as a painting, a poem, or a musical composition expresses or in the often-used phrase is expressive of anxiety, melancholy, or happiness? It seems that

Methodology in aesthetics experiencing such expressivity is among the things that keep drawing people to art, and that for this reason alone it merits philosophical scrutiny. In the following, we will focus on the specific case of musical expressivity, but the assumption is that similar issues would arise even in analogous theories concerning the nature of expressivity within other artistic media. Thus, we will avoid technical treatments concerning musical structure, keeping our discussion on a level that allows the reader to draw potential analogies to similar questions within other art forms. Following the general approach among analytic philosophers of art, we will also not deal with empirical theories concerning the causes or bases of musical expressiveness, but rather on the question of what musical expressiveness is, what it consists in (cf. Levinson 1996, p. 90). There is a multitude of philosophical theories addressing this question on the market, some of them elaborately argued for. We may now put forward our basic question: what evidence is there for or against such theories and how do we go about to reach our verdicts when evaluating them? Let us begin by considering the version of the so-called arousal theory put forward by Derek Matravers: A work of art x expresses the emotion e if, for a qualified observer p experiencing x in normal conditions, x arouses in p a feeling which would be an aspect of the appropriate reaction to the expression of e by a person, or to a representation the content of which was the expression of e by a person (Matravers 1998, p. 146). As Kivy (2001, p. 140) has noted, this is a modification (and according to Kivy, a very radical modification) of earlier arousal theories. Given that a sad person typically arouses, say, pity in his fellow-men, Matravers basic idea would be that a musical work is expressive of sadness if a qualified listener would experience a feeling of pity while listening to it. This implies two main alterations to the arousal theory (ibid., 120) according to which sad music is sad in virtue of arousing the emotion of sadness in its listeners. First, what is felt by the listener can but need not necessarily be the same emotion that the music is consequently taken to be expressive of. The joyfulness of the music might conceivably consist in listeners experiencing joy while listening to it, but it is also possible that the reaction emotion the one that it is appropriate to feel in reaction to expressed joy is something else than joy. Second, what is aroused need not be the full-blown emotion itself but merely a feeling component that is an aspect of the emotion. That is, if you feel joy while listening to music, your emotion need not have an object (as when you feel joy for your son), and you need not have beliefs about such an object (as when you believe that your son has won the gold medal). How are we supposed to argue against such a theory of musical expressivity? Matravers (2003, p. 362) himself recommends that his critics would have to demonstrate clear cases in which all of the elements and relations were present, but there was no experience of expression, adding, [i]t is not clear to me that this can be done. Given that Matravers (1998, p. 183) takes his theory to be an account both of the experience of the listener and of what is meant by expressive judgments, we can readily understand this as a call to logically refute his theory as

E. Huovinen, T. Pontara formulated above in the form of an if-sentence. Obviously, the procedure should be similar to what is supposed to happen in many philosophical thought experiments. To recall a classic example, Gettier (1963) showed that the traditional tripartite definition of knowledge was unsatisfactory in that the definiens failed to provide an adequate account of the definiendum. This he did by showing that we can construct hypothetical situations where the conditions specified by the traditional account of knowledge (i.e. justified, true belief) are fulfilled, but in which we would nevertheless not ascribe knowledge to the imagined persons involved in those situations. In this article, our task will be to examine whether analogous argumentative procedures are applicable with reference to theories of expressivity in the arts. Before proceeding to examine our main target theories of musical expressivity we will first set the stage by reviewing the rationale for relying on intuitive evidence in philosophy. 2 Philosophical intuitions and standard philosophical analysis Notwithstanding recent movements of experimental and empirical philosophy (see, e.g., Prinz 2008), analytic philosophy has traditionally neglected empirical observation as its central evidential base. This does not mean, however, that considerations concerning evidential support are absent from it (see, e.g., Williamson 2007, pp. 208 246). According to a common and currently muchdiscussed view (see, e.g., the articles in Cohnitz and Häggqvist 2009), evidence is here provided by what are called philosophical, or rational, intuitions, and this is what largely sets philosophy apart from empirical science as a special sort of discipline. It has been argued by Hales (2006) that beliefs about philosophical propositions might also be acquired by other methods such as Christian revelation or the use of hallucinogens, and that this may ultimately lead to a certain relativism concerning philosophical propositions. However, for at least philosophers in the analytic tradition, the basic belief-acquiring method has been reliance on what can broadly be characterized as philosophical or rational intuitions. This evidential role of intuitions can be made clearer by comparing the functioning of philosophical thought experiments to the experiments of empirical science. As Devitt (forthcoming, p. 4) puts it, scientific experiments confront the expert with phenomena and ask her whether they are F s, whereas philosophical thought experiments confront her with descriptions of phenomena and ask her whether she would say that they were F s. Obviously, in both cases the aim of the procedure is typically to provide evidence for or against a theory of (or related to) F- ness, and it counts as a virtue of a theory that it admits of critical examination by way of such experiments. In the case of philosophical thought experiments, then, using rational intuition amounts to nothing more extraordinary than deciding what we would say about the F-ness of the described phenomena. Understood in this general sense, it is clear that intuitions may come into conflict with each other, and they thus provide no infallible, Cartesian way to indubitable knowledge. Obviously, relying on intuitions will be less risky when the thought-experimental scenarios stay close to our knowledge concerning the actual world, whereas their reliability may

Methodology in aesthetics begin to dwindle in, say, a twin-earth scenario, requiring us to imagine twin-people who do not contain H 2 O-molecules (cf. Häggqvist 1996, p. 159). Indeed, it seems reasonable to treat intuitions as a form of prima facie evidence the ultimate acceptance of which will depend on our ability to fit them into a more comprehensive framework of reflective equilibrium (cf. Hales 2006, pp. 36 43). According to one influential account, a philosophical intuition ideally appears as an intellectual seeming (Bealer 1998, 2000; Pust 2000) that analogously to the sensory seemings of empirical observation may function as non-empirical evidence for or, most typically and usefully, against a given theory. Hales (2006, p. 20) simple example of how such intuitions works makes this clear: one can just see that it is impossible for if p then p to be false without further proof. Being intellectual seemings, philosophical intuitions are not the result of inferential procedures, and although they are generally considered fallible and the beliefs they yield revisable, they are often taken to involve an apparent inevitability, or apparent necessity (Pust 2000, pp. 44 46; cf. Hales 2006, pp. 42 43). This, of course, becomes most evident when one considers propositions like necessarily, if p then p, which we could hardly think of as being otherwise, whereas such certitude may be contestable in many philosophically interesting thought experiments. Indeed, even if intuitions may sometimes seem to involve an apparent necessity of some kind, this does not mean that having the intuition would be necessary for any given person. If recent experimental approaches to philosophical questions have taught us something, then at least they have taught us the fact that thought experiments can and do elicit mixed reactions when they are presented to culturally or socioeconomically heterogeneous groups of evaluators (see, e.g., Weinberg et al. 2001), or even when the presentation of the scenario is slightly modified (see, e.g., Gendler 2007). Still, it is hard to escape the view that philosophical analysis crucially relies on these intellectual seemings, and that discarding their role as a source of evidence would question the status of philosophy as a discipline. If we want to stay clear from belief-acquiring methods such as religious revelation or the use of hallucinogens, and if we do not wish to reduce philosophy naturalistically to empirical science, this is in fact the best we can do. It is against such a background that we should approach talk of Gettier cases as canonical examples of thought experiments. As Williamson (2007, pp. 179 180) notes, the discussion concerning Gettier often proceeds on the background working hypothesis [ ] that his thought experiments are paradigmatic, in the sense that if any thought experiments can succeed in philosophy, his do: thus to determine whether Gettier s thought experiments succeed is in effect to determine whether there can be successful thought experiments in philosophy. It is true that evaluating philosophical thought experiments by intuitions provides no automatic access to indubitable truths. It is also true that no clear consensus exists about what the possible evidential role of philosophical intuitions ultimately is based on, or indeed what they are. In the context of the present article, however, we cannot pretend to resolve these fundamental questions (for a recent attempt to address some these issues, see Fedyk 2009). Instead, we will provisionally place ourselves in the camp of those philosophers who want to continue relying on their intuitions as a basic method of acquiring beliefs about philosophical propositions.

E. Huovinen, T. Pontara From this perspective, there seems no route around the fact that the rationality of philosophical discussion relies on the possibility of finding evidence in philosophical intuitions in a similar manner to how presenting empirical theories only makes sense against the backdrop of possibilities for gathering empirical evidence for or against them. Hence, we are sympathetic to the position taken by Pust (2000, pp. 2 11) when he argues for the centrality of rational intuitions in what he calls the method of standard philosophical analysis. One of the desiderata that Pust (2000, p. 45) posits for philosophical intuitions is that such intuitions are distinct from perceptual states: while perception shows us only what is actually the case, rational intutions are supposed to demonstrate what is or is not necessarily the case. It should be noted, though, that all philosophers do not draw such a sharp distinction between philosophical intuitions and perceptual states. Devitt (forthcoming, p. 4), for example, mentions a paleontologist identifying what seems to be a white stone as a pig s jawbone, and describes this as an intuitive judgment. What the layman sees as a stone, the paleontologist is directly able to see as a jawbone. Here it will nevertheless be important to keep philosophical intuitions distinct from such perceptual states in order to avoid confusions with other uses of the term intuition within aesthetics, where the term is sometimes used in referring to a certain sort of direct perceptual relation between the mind and a work of art. According to such usage, an intuitive access to the aesthetic character of the work would be one in which the perceiver directly grasps this character without conscious inference (see, e.g., Perkins 1977; Reid 1981). In such a perceptual sense, then, it might also be said that one intuits the expressivity of an artwork, meaning that the relationship between the artwork and one s understanding of its expressivity would be direct and noninferential. Notice that neither the aesthetician in such a situation nor Devitt s paleontologist is apparently testing a specific theory, but rather making a spontaneous identification on the basis of her acquired expert competence. We will here call both the paleontologists and the aestheticians intuitions (as described above) empirical intuitions (cf. Sosa 2008), distinguishing them from the nonperceptual rational (or philosophical) intuitions that are used as evidence in standard philosophical analysis. Thought experiments and intuitive evidence have a firm place not only in such areas as epistemology, philosophy of mind, or ethics, but also in the philosophy of art. To illustrate, let us briefly consider an example from philosophical aesthetics that is analogous to Gettier s argumentation. In the Gettier case, we evaluate a philosophical theory by way of (i) setting up a hypothetical case scenario that satisfies the sufficient conditions given in the analysis, and (ii) intuitively testing whether the concept analyzed applies to the scenario. For an analogous example from the philosophy of art, one might consider Jerrold Levinson s well-known rejection of the account of the nature of musical works in terms of their sound structures. As a part of his argument, Levinson (1990, pp. 68 73) gives several hypothetical scenarios in which a single sound structure is independently created in two music-historical contexts, the reader s task being to intuitively evaluate the theory in question in the light of these cases. In each case, the reader will come to see that placing identical sound structures in different music-historical contexts will result in two distinct musical works. This is a fairly typical example of

Methodology in aesthetics argumentation intended to elicit an immediate conviction to the effect that the theory under investigation is inadequate. The success of Levinson s thought experiments may, of course, be contested just as Gettier s has been. Nevertheless, the rationale is the same in both cases: when confronted with a well constructed thought experiment, the reader should be able to directly see the falsity of the theory at hand; at least ideally, she should also come to believe that others in her position could not help but make the same intuitive judgment when confronted with the same hypothetical scenario. Thus, it seems that the philosophy of art, at least sometimes, neatly proceeds according to the standard philosophical analysis. 3 Philosophical theories of musical expressivity Philosophical thought experiments are used to assess philosophical propositions that is, propositions such as knowledge is justified true belief, all of the parts of a whole are necessary to it, psychological facts supervene on microphysical facts, or the like. A curious fact about these propositions is that even though some of them could be false, it is nevertheless the case that if they are true, they have to be necessarily true. This may sound like a grand claim, but it is, in fact, something that simply follows from how philosophers reason about philosophical propositions. As Hales (2006, p. 22) explains, [i]f those propositions were not taken to be necessary truths, then refutation by demonstration of what s possible would not work. If we oppose to, say, a utilitarian theory of morally right action for example, the actutilitarian theory that action x is morally right if and only if it produces more pleasure, overall, than any other alternative action by devising a (merely) possible scenario that works as a counterexample, the power of our argument lies only in the fact that as a putative necessary truth, the theory is vulnerable (if it is) to any (merely) possible counterexample and the counter-utilitarian intuitions evoked by that example. Thus, the logic of thought-experimental counterexamples only applies to philosophical propositions in the sense explained above. Philosophical propositions are often readily recognized by the use of terms such as necessary or essential, but sometimes the modal force of the propositions may be hidden under vaguer linguistic expressions. However, nowhere does the modal intent of the proposition stand out better than in theories that try to account for the nature of something in terms of necessary and sufficient conditions (cf. Pust 2000). From, say, debates concerning musical ontology, we know that such theories are not uncommon in the philosophy of music either. In the discussion concerning our present topic, musical expressivity, biconditional formulas may not have been quite as common, although there are indications that many of the theorists may, in fact, have necessary and sufficient conditions in mind. One should be cautious, however. For example, Malcolm Budd states his make-believe-based theory of musical expressivity not by an iff-sentence but rather using the formula M is anguished is to be understood as [ ] It is make-believedly true that anguish is being experienced, and this make-believe truth is generated by M (Budd 1989, p. 135; emphasis added; the details of the theory are inconsequential to our present concerns). Although Budd perhaps somewhat incautiously refers to what follows the

E. Huovinen, T. Pontara phrase is to be understood as as a definiens (ibid., p. 134), his concession later in the same article that there is, apart from the make-believe view, also another way in which music can be heard as expressive of emotion (ibid., p. 137) should make any rendition of his primary theory by an iff-sentence suspect (see, e.g., Levinson 1996, p. 96). Substituting if and only if for Budd s is to be understood as would make his primary theory fall already on the exceptions that he himself admits to it in the form of an alternative theory. On the other hand, a more charitable reading without the biconditional has the unfavourable consequence of leaving Budd s theory out of reach from critical examination by means of philosophical thought experiments, appreciably diminishing its philosophical appeal. Despite the methodological importance of whether or not a theory is formulated in terms of necessary and sufficient conditions, it seems that this point has not generally received as much consideration in discussions of expressivity in the arts as it should have. Derek Matravers account of musical expressivity in his 1998 book seems particularly ambiguous in this regard. As we have seen, Matravers formulates his theory as a simple if-sentence, that is, without giving his criteria the status of necessary conditions. Analogously to Budd s case, this is consistent with Matravers (1998, p. 183) assertion that [a]lthough the arousal theory captures the primary sense in which music is expressive, there are other ways in which music can be expressive which are not directly tied to the arousal of feelings in a listener. Despite this, Matravers still claims his theory to have captured the central notion of musical expressiveness, upon which these other ways in which music is expressive ultimately rely (ibid., p. 185). Indeed, right after stating his if-sentence, he claims that he intends it as the analysis of expressive judgments; it is what we mean when we (for example) call music sad (ibid., p. 146; emphasis added). Later he even claims his theory to be deliberately modelled on the familiar basic equation for colour, which he formulates as an iff-sentence, that is, in terms of sufficient and necessary conditions (ibid., p. 188). Concerning the specific case of sadness, he then restates his theory in the following form: the belief that a work of art is sad will be true if and only if a suitably qualified observer in normal conditions would experience it as sad (ibid., 190; emphasis added). Considering such statements it does not come as a surprise that he sets his theory the task of defining the emotional qualities of works of art in terms of the reactions of qualified observers in normal conditions (ibid., p. 192). All this strongly suggests that Matravers is indeed after an account of the nature of musical expressivity. 1 These examples should be enough for suggesting that the philosophical discussions concerning musical expressivity may have partly suffered from insufficient concern with the methodological consequences of logical sentence forms. In order to pursue our discussion further, we need to focus on a theory that presents no such formal ambiguities, and candidly opens the way to examination through standard philosophical methodology by way of thought experiments constructed around possible scenarios. Jerrold Levinson s widely known theory of musical 1 Curiously, also Kivy (2001, p. 141) appears to treat Matravers theory as a definition, noting that [i]t is stated in the traditional philosopher s formula of necessary and sufficient conditions, even though the logical form of Matravers theory is clearly just a conditional statement, formulated in terms of sufficient conditions.

Methodology in aesthetics expressiveness serves this purpose well, its biconditional form signaling a philosophical proposition that should be necessarily true if true at all: P expresses (or is expressive of) a iff P is most readily and aptly heard by the appropriate reference class of listeners as (or as if it were) a sui generis personal expression of a by some (imaginatively indeterminate) individual. (Levinson 1990, p. 338.) 2 For short, Levinson also refers to the indefinitely imagined [ ] subject of the state being expressed as the persona of the music (Levinson 1990, p. 338), 3 and suggests that, in hearing a passage as expressive, we may, first, be imagining a persona for it, and second, be imagining this persona to have the power to achieve in unmediated fashion an outpouring of sound that normally requires [ ] the intermediary of musical instruments and the possession of musical training (ibid., p. 339). As formally stated, Levinson s theory specifies the exact conditions under which a piece of music is expressive of a. Briefly put, expressivity is present in music in all and only those situations in which the appropriate reference class of listeners will tend to hear the music as an expression by an imagined persona. Now, our main question will be whether it is possible to argue against such a theory along the lines of standard thought-experimental methodology. In principle, disproving a theory of musical expressivity in biconditional form could happen in two ways. First, one might set up an imaginary situation in which the right side of the biconditional is satisfied, but in which it would be intuitively clear that it is not musical expressivity that the situation is about. This would correspond to what happens in a Gettier scenario. The second possibility is that we focus on the only if -part of the biconditional and set up a scenario in which the conditions specified are not, in fact, fulfilled in what intuitively appears as a clear case of musical expressivity (cf. Pust 2000, p. 3). In the following section, we will consider these two possibilities in order. All the way through, it should be remembered that it would only be to the expressivity theorist s credit if the required kinds of scenarios can, at least in principle, be set up and evaluated: only by opening the door to refutation by possible cases can the discussion methodologically stand in line with the way in which counterarguments are standardly set up in (the analytic tradition of) contemporary epistemology, philosophy of mind, or moral and political philosophy. Indeed, supposing for a moment that such counterarguments could be fashioned, their exact details, the strength of our intuitions in evaluating them and, consequently, their persuasiveness would still remain a different issue. Ingenious thought experiments such as Gettier s do not grow on trees, and it is not possible to prove that there don t exist some yet-to-be-explored variants of the required scenarios that would do the trick against, say, Levinson s theory. But again, this is only on the hypothesis that 2 Cf. Levinson 1996, 2006a, 2006b. In the last statements of his theory, Levinson has dropped the potentially problematic reference to sui generis expression. As our argument will not rest on the exact details on the right side of the biconditional, however, we will use the original version since it is the one that is most clearly formalized. 3 In a later formulation of the theory, the phrase (imaginatively indeterminate) individual is, indeed, changed to an indefinite agent, the music s persona ; see Levinson (1996, p. 107).

E. Huovinen, T. Pontara the requisite kinds of thought experiments can be, in principle, made to work. It seems to us that in the discussion concerning the nature of expressivity in the arts, designing thought-experimental scenarios has never really proceeded far enough in order for discussions to arise concerning their exact details, their persuasiveness, or the possible individual variability of our intuitive reactions to them. Our claim will be that there is a good reason for this methodological eccentricity: the required thought-experimental scenarios just cannot be constructed in a manner that allows them to be properly evaluated. 4 Counterarguments against theories of musical expressivity To begin with, one could try to invalidate Levinson s theory by pointing out that the conditions he gives are not sufficient for a correct account of musical expressiveness. As explained above, the standard way of doing this would be to construct a hypothetical scenario in which it would be intuitively clear to us that the satisfaction of the biconditional s right side does not amount to an adequate account of musical expressiveness. Constructing the scenario would involve imagining that a qualified listener hears the musical passage P as a sui generis personal expression of a by some (imaginatively indeterminate) individual. In order to count as a methodologically sound counterargument, this scenario would have to elicit the intuition that what is at issue does not involve musical expressivity. First of all, we would need to construct the relevant sort of scenario. Frankly, we do not know how the exact details of the scenario should be layed out, but what we do know is what sorts of ingredients it should minimally involve in order to do the work that is required. Let us, for instance, imagine a situation in which Tom, who is thoroughly acquainted with the tradition of Western art music, is listening to a piece of classical music, which he hears as an expression of sadness by an imaginatively indeterminate persona. We might furnish this description with further details by supposing, for example, that no feelings are aroused in Tom. Now, given a more elaborately refined scenario based on these ingredients, our task would be to refute Levinson s theory by evaluating whether music is expressive in the scenario. Could we do this? If we are, say, arousal theorists that are already convinced of a necessary connection between musical expressivity and the listener s feeling of emotions, we might be inclined to think that the kind of scenario envisaged provides some sort of counterexample: as described, the situation does not involve the arousal of feelings. Methodologically, however, this would not be a sound move. As we recall from the Gettier case, no alternative theories need to be in place in order for an intuitive judgment of the hypothetical scenario to work. Gettier s project is a totally negative one: no alternative theory of knowledge is assumed in order for the argument to go through. In the above example, on the other hand, we would be evaluating the scenario from the perspective of its compatibility with our antecedently accepted arousal theory. Contrary to standard philosophical methodology, we would be assuming the correctness of a competing theory of musical expressivity in testing the theory under consideration. This is not the way thought experiments are supposed to work.

Methodology in aesthetics We now face the question: is it possible, in principle, to refute a theory of musical expressivity without relying on such an alternative theory? Can we obtain the relevant insight immediately, without considering anything else beyond the hypothetical scenario, as described? It might seem that a more direct and less theory-dependent evaluation of the scenario would indeed be possible, by carefully reflecting over our earlier and very personal experiences with music. Considering the case in which Tom hears an expression of sadness by a persona in the music that he is listening to, we would thus have to consider his situation from the point of view of our own musical experiences. What would we say about the expressivity of the music in Tom s situation? We might, then, consider the description of the situation that an expression of sadness by a persona is heard in the music, but no feelings are aroused and imagine what it would be like to experience music under those conditions, comparing this to personal cases of listening in the past in which we recall attributing expressivity to the music. It may be suggested that something like this in fact should happen if one manages to evaluate Tom s situation and give a negative answer: No, the music is not expressive in the situation, and thus the theory will be disproved. We will argue, however, that an evaluative procedure along the lines of standard philosophical methodology cannot, in fact, be carried out if reference to one s own personal musical experiences is the only kind of evidence available. Before turning to this major methodological qualm, however, we will first give two interrelated arguments intended to show that in turning to our personal musical experiences we are not dealing with philosophical intuitions. First, when we reflect on our past (or present) musical experiences in order to evaluate whether the music is expressive in Tom s situation, our judgment is not immediate: music happens in time, and so does our reflection of our past musical experiences. Even if reflecting upon the experience of a given piece of music hardly requires that we imaginatively listen to the whole piece in real time, it seems reasonable to suppose that we minimally need to imagine some temporally organized fragments or samples of the music in order to access the phenomenal character of our past experience. In the standard philosophical cases, on the other hand, no such re-enlivening of a concrete, particular experience needs to occur. On the contrary, the Gettier case (supposing it works for us) allows us to directly see that knowledge is not present in the situation, without any imaginative simulation of a phenomenological character of some previous experience. There is a possible objection, however. Maybe our evaluation of Tom s situation will not require the re-enlivening of particular, temporally extended experiences, but instead relies on implicit knowledge that we have collected concerning these experiences. This knowledge might be represented in such a manner that its application would not require running through any single sample of a temporally extended experience. In other words, we might possess an implicit theory concerning our experiences that would be built into our system a theory, say, with the content that in all of our past musical experiences, ascriptions of expressivity have coincided with the arousal of emotions in us. According to such a counterargument, then, no recourse to a phenomenological re-enlivening of particular experiences would necessarily be needed in order to assess Tom s case.

E. Huovinen, T. Pontara A quick and confident reply No, the music cannot be expressive in the scenario would be possible just on the basis of an implicit theory. We would like to suggest, however, that an evaluation of Tom s case by an implicit theory would not be well-grounded. Indeed, the situation would be reminiscent to that of encountering a friend s description of the sky as red and quickly judging it as false on the basis of an implicit theory concerning the blueness of the sky. Even if the implicit theory itself (that the sky is always blue) is based on previous experiences, it is highly generalized and does not account for the possibility for which one might still find evidence by re-examining particular experiences that the sky may, indeed, have sometimes been experienced as red. Similarly, in the musical case, the experiences themselves are there to be tapped for all of those who may have made implicit generalizations concerning them. We may also go back to these experiences themselves by imaginatively re-enlivening them and contemplating their phenomenological character. When faced with the task of testing a philosophical theory by way of evaluating a hypothetical scenario, a snap reply based on an implicit theory would fail to take heed of the ultimate evidence that is available in our past musical experiences themselves. For a well-grounded evaluation of Tom s situation, we would arguably have to go back to these experiences. And to repeat the earlier argument, in re-enlivening them, we would be doing something else than relying on philosophical intuitions. Whether or not the above reasoning succeeds in showing that our immediate judgments concerning Tom s situation are not based on philosophical intuitions, there is still a stronger argument in store. Recall that in the standard philosophical situation, our intuitions are directed toward the hypothetical scenario only. In the musical case, as described above, we would proceed in a different manner. First, we would presumably have to imagine for ourselves what it would be like to fulfill the criteria specified in Levinson s theory that is, we would imagine ourselves to be hearing expression by a persona in the music. Secondly, however, we would also have to have access to something external to the imagined scenario. Namely, we would inspect our earlier particular experiences of music, thus employing some perception-like introspective faculty. Or, to use the term introduced earlier, we would apply some kind of empirical intuitions concerning our previous musical experiences. It would be only through a comparison between these empirical intuitions and the imagined scenario that we could test the theory at hand. This means that unlike what happens when we proceed according to standard philosophical methodology, our evaluation of the scenario would be inferential in the sense that it would rely on some particular states of affairs outside of the scenario itself namely our empirical intuitions concerning our earlier musical experiences. The above argument could also be formulated from another perspective. Instead of saying that we inferentially reach beyond what was originally described in the scenario by referring to our own empirical intuitions, we can say that we evaluate the scenario by way of placing ourselves in it, which comes to the same thing. Let us summon up the situation once more: Tom is hearing an expression of sadness by a persona in the music but no feelings are aroused in him, and our task is to evaluate whether the music, in itself, is expressive of sadness. Now, we submit that we have

Methodology in aesthetics to place ourselves in Tom s shoes in order to pass the judgment concerning the presence of musical expressivity in his situation. If so, it is clear that we cannot view the situation from afar, as outsiders, as in the Gettier case. Whether we describe the matter in terms of inferentially reaching beyond the original scenario or in terms of placing ourselves in the scenario, what happens is that we replace Tom and make him and the scenario in which he appears completely irrelevant. This leads us to discuss the most serious methodological problem of Levinson s theory: the sheer impossibilty of forming Gettier-like counterarguments with reference to our own personal musical experiences. In philosophy, what is typically required of a thought experiment designed to provide evidential support for or against a theory is a certain degree of objectivity that crucially relies on the distinction between the evaluator and the agents figuring in the imagined scenario. Here we have seen, however, that the scenario ultimately turns out to be about our very personal experiences, and this means that the required distinction between the person or persons figuring in the scenario, on the one hand, and the evaluator of the scenario, on the other, breaks down. This, in its turn, may lead to two different consequences depending on whether or not we are able to find, among our previous musical experiences, ones that would correspond to Tom s situation. First, assuming that we can recall experiences of as-if-expession of sadness by a persona in music, no room seems to be left for a judgement that these situations would not be about musical expressivity. The step from the concept of expression (in the definiens of the theory to be tested) to that of expressivity (in the definiendum) is too small to leave any room for such judgement. The case is closed: when considering a situation in which we have heard expression of sadness by a persona in music, it just does not seem conceivable that we could even in principle say that these situations haven t, for us, been about musical expressivity. As we turn from evaluating the original scenario to evaluating our own experiences that correspond to Levinson s criteria, not only does the scenario lose its methodological function, but it seems that no in-principle possibility will remain for disproving the theory that we are dealing with. The situation gets still worse if, when reaching back to our personal experiences to test the sufficiency of Levinson s criteria for musical expressivity, we don t find experiences that would correspond to these criteria. Given that you cannot recall any experience where you would have heard music as if it were a personal expression of an emotion by a persona, you won t even be able to start evaluating whether these conditions could fail to be sufficient for musical expressivity to occur. To be sure, you might still recall other personal experiences in which you actually have experienced music as expressive, but they will not be of any use in evaluating the sufficiency of Levinsonian as-if-personal-expression for musical expressiveness. For a formally valid refutation of the proposed sufficient conditions (along the lines of the Gettier case), you would have to start from a case in which these conditions themselves are fulfilled as they were in the original thought-experimental scenario and proceed to see whether musical expressivity might fail to occur under these conditions. This game cannot even get started if we try to kick it off by referring to other kinds of personal experiences than those corresponding to Levinson s criteria.

E. Huovinen, T. Pontara Now, it is instructive to see what would happen if we tried to re-establish the methodologically obligatory distinction between the agent in the thoughtexperimental scenario and the evaluator of the scenario by ascribing propositional attitudes to the former. In a Gettier scenario, even if the person in the scenario should be described as being convinced of having knowledge that p, we the evaluators of the scenario would still be intuitively convinced that the person does not know that p (again, supposing that the thought experiment works for us). As Sorensen (1992, p. 273) puts it, [t]here is a difference between what the characters within a thought experiment can know and what the audience can know. In the musical case, however, if Tom is described as sincerely reporting that the music is expressive of sadness, no further evidence can be given to the contrary. At least as a prima facie stance, we simply have to trust him. Again, the scenario loses its methodological function this time because our own evaluation of it plays no role whatsoever. In sum, depending on whether propositional attitudes concerning the expressivity of the music are ascribed to Tom or not, the evidence lies either in Tom s experiences or in our own, but neither case will be about evaluating the original thought-experimental scenario as an outside observer with freedom to disagree with Tom on the correct analysis of the situation. It may thus be concluded that the sufficient conditions of Levinson s theory cannot be evaluated in accordance with standard philosophical methodology. The same goes for the sufficient conditions of musical expressivity in other theories. Earlier, we saw how Matravers urged his critics to find situations in which his sufficient conditions for musical expressivity would be present, but nevertheless there would be no experience of expressivity. Matravers it is not clear to me that this can be done may have been intended as a statement that the critic would be unable to find the situations asked for, and that the arousal theory would thus survive the attack. We are now in a position to see that Matravers is indeed right in his this can t be done, but that far from being an indication of the superiority of his theory, it actually points out a major defect. Matravers arousal theory, stated in terms of an if-sentence, is in principle impossible to formally refute, because the requisite philosophical intuitions will remain absent, no matter how the critic will try to construct his counterargument. The critic will begin to construct his case by seeing to it that the conditions indicated in Matravers theory as sufficient are fulfilled. Analogously as in Levinson s case, however, the absence of musical expressivity may then be ascertained (if at all) only by means which do not count as applications of standard philosophical methodology. On the one hand, the critic may presuppose some alternative (necessary) conditions for musical expressivity, and construct his case scenario so that these conditions are absent from it, but this theory-dependent solution hardly counts as a fair strategy. On the other hand, he may attempt to judge the presence or absence of musical expressivity by relating it to his own previous musical experiences, but this is not to use philosophical intuitions either. We leave it for the reader to examine Matravers theory in this light to see that, again, the evaluator of the argument will have to substitute herself for the subject figuring in the scenario.

Methodology in aesthetics We may finally turn to examine the other possible form of counterargument, namely the one that would be directed towards the only if -part of a theory of musical expressivity, that is, the necessary conditions that such a theory specifies. Above, we saw how Budd and Matravers allowed other ways in which music can be expressive alongside their primary theories, thereby implicitly acknowledging that the criteria stated in their primary theories do not work as necessary conditions. For an explicit biconditional such as Levinson s, however, any other ways would be fatal. In Levinson s case, a working counterargument along these lines would require a situation in which the music is experienced as expressive, but without any expression by an (imaginatively indeterminate) individual heard in it. Such a situation would, in the simplest possible case, be one in which (i) we have an aesthetic intuition that the music is expressive, but (ii) we do not observe anything that would correspond to Levinsonian individuals or personae. The right kind of situation might of course be found among our actual, particular experiences of concrete musical situations, without any recourse to a counterfactual scenario. You may simply be hearing music now and judging it as expressive without noticing any indeterminate individuals. However, this would hardly satisfy Levinson (2006a, p. 104) who insists that by his theory, listeners are not required to always explicitly [be] aware that personae are involved in their hearing music as expressive. What about using a hypothetical scenario, then? Analogously to the subjective, actual case just mentioned, we might begin by imagining a hypothetical listener Tom who, in some hypothetical scenario, would hear a piece of music as expressive of sadness. Now, what would have to be shown by Levinson s critic (through the carefully constructed details of the scenario) is that this could be possible without Tom hearing the music as if it were an expression of sadness by a persona. After being told the details of the scenario, we as readers of the counterargument would have to evaluate whether it really is the case that Tom, under the specified conditions, would not hear the music s expressiveness as issuing from a persona in the music. Again, however, no intellectual seemings would appear to be obtainable: we would not directly see whether Tom as an individual distinct from ourselves hears expression by a persona or not. Again, the only way we could carry out the evaluation is by setting ourselves in Tom s situation, this time reflecting on our own previous (or present) experiences of expressivity in music. Unlike when testing the sufficiency of Levinson s criteria, here we might indeed reach a subjective refutation of the necessary conditions proposed, given that we recall any single personal experience of musical expressivity which was not an experience of expression by a persona. As before, however, Tom himself would remain entirely irrelevant as we would not be able to draw a distinction between him and ourselves as evaluators of the scenario. Once more, we reach the conclusion that Levinson s theory cannot be evaluated by way of standard philosophical methodology. In fact, the only way of evaluating the hypothetical scenario would be through the same sort of introspective judgment that Levinson already renounced in the case of actual musical situations.