Musical pluralism and the science of music Anton Killin & Adrian Currie Penultimate version, forthcoming in the European Journal for Philosophy of Science Abstract: The scientific investigation of music requires contributions from a diverse array of disciplines (e.g. anthropology, musicology, neuroscience, psychology, music theory, music therapy, sociology, computer science, evolutionary biology, archaeology, acoustics and philosophy). Given the diverse methodologies, interests and research targets of the disciplines involved, we argue that there is a plurality of legitimate research questions about music, necessitating a focus on integration. In light of this we recommend a pluralistic conception of music—that there is no unitary definition divorced from some discipline, research question or context. This has important implications for how the scientific study of music ought to proceed: we show that some definitions are complementary, that is, they reflect different research interests and ought to be retained and, where possible, integrated, while others are antagonistic, they represent real empirical disagreement about music’s nature and how to account for it. We illustrate this in discussion of two related issues: questions about the evolutionary function (if any) of music, and questions of the innateness (or otherwise) of music. These debates have been, in light of pluralism, misconceived. We suggest that, in both cases, scientists ought to proceed by constructing integrated models which take into account the dynamic interaction between different aspects of music. Keywords: Definitions of music; musicality; pluralism; musical cognition; evolution of music. 1. Introduction Music is a complex and fascinating target of scientific inquiry. Nevertheless, theorists disagree on what music is—on how best to characterise or define it (see e.g. Davies, 2012a; McKeown-Green, 2014). We argue that although some of this disagreement should be retained, some should be resolved. That is, contra several 1 attempts to provide a unitary definition of music (e.g. Kania, 2011; Levinson, 1990; Godt, 2005), we advocate a form of conceptual pluralism—that is, there is no objective, definitive definition of music independent of some explanatory context; rather, there are multiple, non-equivalent, legitimate concepts of music. Our argument is based on an examination of musical science itself: it is plausible, given the range of questions and evidence which scientific investigation of music requires, that there is no unitary definition of music simpliciter. Moreover, we argue that musical pluralism has positive consequences: it can lead and drive integration in the study of music. To this end, we distinguish complementary and antagonistic relationships between definitions. Complementary definitions target different aspects of music, or music in different contexts, while antagonistic definitions give rise to empirical disagreements. Embracing a plurality of concepts does not entail accepting every one. Indeed, definitions and empirical work are not independent. Given a particular research question, some definitions can be more or less appropriate, and determining the best one for the job can itself be an empirical task. In other words, good definitions are not necessarily stipulated assumptions: they can be legitimate research findings. Indeed, to render some aspect of a real-world phenomenon amenable to scientific investigation, scientists often need to approach definitions somewhat instrumentally: operational definitions are called for.1 The relationship between 1 An anonymous referee points out that such operational, instrumental definitions often require a stipulated, general account of ‘music’ to get off the ground, otherwise the relationship between some of the concepts we discuss, for instance the distinction between ‘music’ and ‘musicality’, is decidedly ambiguous. This might be right, the relationship between the various concepts required to make music scientifically tractable can be very complex. We take this to be grist for our mill: on our view, different research agendas require a number of different, but importantly related concepts. Considering how these concepts interact is an important part of integrating such perspectives. 2 evaluative and instrumental definitions is complex, and we think the framework we present in §2 gives some guidance for navigating this. We proceed as follows. In the next section we provide a rough and partial sketch of the complexity of musical investigation. In §3 we draw on this to argue for pluralism. In the remainder of the paper, we illustrate our point and tie discussion to scientific practice. In §4 we discuss the relationship between music and evolution and in §5 we turn to music’s innateness (or otherwise). 2. The Science of Music: A Sketch This section sketches the science of music along two lines. We discuss the research agendas which guide inquiry (§2.1) and the lines of evidence (§2.2) which are drawn upon. This picture forms the basis of our argument for pluralism in §3. It is important to note that the view we present here is both schematic and heuristic. Particularly in terms of research agendas, our account is inadequate as an attempt to capture the actual landscape of music research. However, in regards to our argument about musical pluralism, this is a feature rather than a bug. Our aim is to show how complex and multi-faceted musical research is, and this is used to motivate pluralism in §3. That the sketch fails to capture all of this complexity (in ways we shall point out) strengthens our argument. 2.1 Research Agendas Different aspects of music must be approached from a variety of stances: there is no one question about music, but a plurality. Following Love and Brigandt (see below) we call these stances ‘research agendas’. Of course, a plurality of agendas does not entail 3 a plurality of music concepts, but as we shall see, the complexity of musical research at least makes pluralism attractive. We take a ‘research agenda’ to be a perspective from which one might approach the study of music. These perspectives will include a set of investigative concerns. For instance, the scientist interested in psychological models of music will be largely interested in how adult humans process musical information, and the psychological features which underlie musical behaviour. In contrast, the scientist interested in psychological musical development will focus on how these features develop over ontogeny; the changes taking place as children develop. The agendas come apart in terms of the kinds of questions they ask, and the kinds of answers that they want. There is an issue regarding the commensurability or otherwise of various research agendas. How, indeed, might evolutionary and aesthetic approaches be meshed? For the purposes of this paper, we will put this to one side. Below we give an account of how integration between agendas may occur, but take it that whether some are ultimately incommensurable is a question which can only be answered by doing the science, that is, whether and how specific approaches might be integrated (or not) will be decided by the outcome of research, not a priori. We characterize the complexity of musical research in a rough and ready way, along two dimensions: level of organization and temporal scale. Fig. 1 loosely sketches this landscape: 4 Fig. 1: Research areas along dimensions of abstraction and temporal scale Fig. 1 is heuristic: it is not an authoritative analysis of the layout of music scholarship. Indeed, there will be many other useful ways of characterising the research space too; we illustrate our conception’s utility below. Even so, some caveats are important before we explain the dimensions. Firstly, some domains are probably misplaced: if, as we argue, the biological and cultural aspects of music evolved dynamically, then it might be a mistake to separate the evolution of musical psychology and music’s cultural evolution. We return to this thought in §4. Secondly, for reasons of parsimony, the figure does not clearly distinguish the object of some investigation (for instance, music’s evolution) from the method by which some investigation proceeds (for instance, neuroscientific versus psychological). To be sure, we do not endorse conflating objects and methods, as will be clear from our discussion. A more thorough attempt to characterise the voluminous landscape of musical science, beyond the scope of this paper, will by necessity incorporate more dimensions. Temporal Scale 5 Research questions about music can be organised along a temporal scale: some questions are long-range, concerning glacially paced processes in the deep past, while others cover shorter, quicker time-frames.2 Ernst Mayr (1961) famously distinguished between ultimate and proximate causes of biological phenomena. Although we do not commit to the distinction (indeed, it is highly problematic, see Ariew, 2003; Beatty, 1994; Calcott, 2013; Laland et al., 2013), it is a useful way of illustrating our point. An ‘ultimate cause’ explains traits in terms of evolutionary history. We might ask, for instance, whether music is an adaptation, whether musical traits were positively selected for in our evolutionary history. Proximate causes, in contrast, shape an organism over its lifetime—we might wonder from what age a child is able to recognise and remember melodies, for instance. A proximate explanation of birdsong might be that increased light in spring precipitates the release of a hormone in a male songbird’s brain (e.g., Moore et al., 2006). The ultimate, evolutionary explanation might be that birdsong indexes the quality of the male bird against its conspecifics, in an effort to attract females with which to breed. In other words, our temporal scale resembles the phylogeny/ontogeny continuum (Gould 1977). However, the temporal scale is not restricted to questions of evolution and development. As we move from left to right on the temporal axis we shift from research agendas covering the origin of music, its evolution and subsequent historical development, to questions about how musical cognition develops over a lifetime, to theoretical questions about musical structure. At its extreme right end, questions become near-atemporal. Western-style analytic music theory (analysis of music in the Western terms of meter, scales, and so on), for instance, is hardly 2 Note that our use of the term ‘scale’ is not meant to imply literal measurement. 6 temporally bound.3 Much of modern cognitive science and psychology is concerned with characterising contemporary human minds, not explaining them in terms of historical forces (for example, see Eerola, 2012 for a computational model of musical emotion perception; Lerud et al., 2014 for a neurodynamic model of auditory response to musical intervals; Neuhaus, 2013 for approaches to musical form processing; Tesoriero and Rickard, 2012 for frameworks for elucidating music-facilitated recall). How a researcher approaches the study of music, then—that is, the nature of their research agenda—depends in part on temporal scale. Some might target relatively narrow scales, such as the emergence of musical capacities during ontogeny, or anthropological studies of musical expression across human cultures. Others target wider scales like the historian tracing the development of musical styles across centuries, or wider yet, the question of when and why musical capacities appeared in our evolutionary history. Levels of Organization Our other dimension organises research agendas by the ‘level of organization’ they target. This is a common notion which is nonetheless difficult to specify, philosophically speaking. We can understand a ‘level of organization’ in terms of either the ontological relationships between targets of investigation, or in terms of how 3 Musical features associated with Western music-theoretic concepts, however, are arranged temporally—meter, tonality and so on have specific meanings relative to different eras in music history. A chronology of Western music-theoretic concepts might fall under the broad study of History of music on Fig. 1. Of course, Western music theory is not the only music theory! One of the great weaknesses of much musical research is that it assumes that a Western musical experience is universal (or very nearly) and thus fails to take into account the nature of music/musicality and its social correlates in a variety of non-Western contexts. Some music researchers are acutely aware of this, and keen to point out the problem, and we are deeply sympathetic. Thanks to an anonymous referee for pressing us on this issue. 7 scientists represent them, their ‘level of description’.4 Atoms combine to form molecules which themselves form biological entities like proteins, and so on. A psychologist might be interested in positing mental models to explain human musical behaviour and cognition, while a cognitive neuroscientist would posit mechanisms in neural anatomy to ground both those models and (by extension) those same behaviours. The research agendas of cognitive science have typically focused on fairly low levels of organization, such as neural and psychological targets. Proponents of a growing unrest about the limits of this focus urge researchers to take socio-cultural matters into consideration (Bender et al., 2010; Cross, 2012a; Stevens, 2012; Trehub et al., 2015), thus widening focus to include higher levels of organization. Moreover, theoretic musical properties, such as key, pitch, mode and meter (and non-Western analogues), can play an important role in the construction of theories and studies in less abstract sciences. This picture immediately recalls reductionist models of science, most vividly the ‘layer-cake’ model (Oppenheim and Putnam, 1958). However, in presenting this picture we do not commit to a reductionist view. First, dividing research questions by level of organization (or description) doesn’t commit to there being reductive relationships between those levels (see, for instance Potochnik, 2010; Wimsatt, 1994; Fodor, 1974; Darden and Maull, 1977). Second, as we have stated, our purpose here is to illustrate the wide variety of perspectives from which music can be studied, and this is a helpful way of doing so. Helpful, but deeply flawed. Most strikingly problematic is our placement of highly abstract investigations such as analytic music theory at the highest level of organization—this hardly makes sense! Human social groups do not together form the mathematical structures of musical theory! This would be problematic if our aim was to 4 See Craver (2005, 2007) for discussions of both different conceptions of ‘levels’ and integration in cognitive neuroscience. 8 provide a thorough characterization of the landscape of musical research. However, given that our purpose is to provide a preliminary sketch which motivates pluralism, we think this is adequate for our purposes. As we have said, that our two dimensions fail to capture some aspects of musical research is a feature rather than a bug: it shows that to capture musical research adequately more complexity is required. Combining the two dimensions provides a set of possible questions, or perspectives, from which one might embark on the project of understanding music; these are different research agendas. Embracing the cluster of research agendas represented by the combination of these two dimensions will drive our pluralism about music concepts. 2.2 Lines of Evidence and Integration Musical science requires a broad range of evidential sources. Here we present some of these, before discussing integration: the relationship between such evidence and research agendas. Lines of evidence are generated by some technology or technique. A novel line of evidence gains relevance and garners support via background theory which (1) underwrites how the technology in question operates, and (2) links it to some hypothesis. Lines of evidence utilised by music researchers include—but are by no means limited to—brain scans, behavioural responses (from various stages of development), archaeological finds, cross-cultural ethnography, cross-species comparison, data on happiness/wellbeing, psychiatry, physiology and genetics. Often these lines of evidence are best harnessed by representatives of distinct disciplines—psychology, neuroscience, 9 archaeology, anthropology, biology, to name a few. Musical science is a truly interdisciplinary game. In short, there is a wide range of research agendas guiding us, and a plurality of available empirical resources. Understanding the relationships between these different questions and evidence is important for the field’s progress (cf. Fitch, 2015). Brigandt (2010) and Love (2006) have discussed the relationship between different research agendas and evidence in reference to evolutionary developmental theory and functional morphology, respectively (see also Potochnik, 2010). By their picture, different sources of evidence play different roles depending on which research agenda is salient. For some contexts, a particular agenda, and thus some evidence, carries the explanatory force, while other evidence plays a supporting role. We can determine explanatory force (at least as a rough ‘first pass’) by considering the pragmatics of explanation. We can ask which agenda provides the relevant answer to a question, and which agendas play supporting roles—that is, provide reasons to believe that answer. Let us explain. Consider Grahn’s (2012) discussion of interval and entrainment models. She asks, which psychological features allow humans to recognise rhythm? The two kinds of models posit different mechanisms to explain human rhythm recognition. An interval model contrasts the beat of a fixed internal clock (or clocks) to an external rhythm, while entrainment models fit an internal oscillator to external stimuli (see Grondin, 2010; McAuley, 2010). As Grahn points out, it is not obvious that models of time intervals are fine-grained enough to handle rhythms (which are sequences of time 10