Cognitive neuroscience, phenomenology and dance: questions of approach (introduction)

Shared interest in dance

A revival of interest in phenomenological analysis of dance (e.g. Legrand and Ravn 2009; Parviainen 2002; Rothfield 2005) has coincided, in the last decade, with the increased presence of dance in neuroscientific research. Most neuroscientific studies used dancers as experts (Jola and Mast 2005a) and/or dances as stimuli (e.g. Brown et al. 2006; Calvo-Merino et al. 2005, 2006; Calvo-Merino et al. 2006; Cross et al. 2006; Jola and Mast 2005b; MacFarlane et al. 2004) primarily to investigate processes involved in movement execution and observation. Generally, studies with a more phenomenological approach emphasise the first-person ‘lived’ experience of phenomena, such as the experience of what it is like to be a dance spectator, while those with a neuroscientific approach have a more third-person objective perspective such as trying to find out what processes implicitly occur in the brain when a person watches dance. In the literature to date related to investigating dance spectatorship, phenomenology and neuroscience have mainly remained distinct rather than complementary.

In the present paper, we discuss research processes and methods used to try to bridge the perceived gap between these two approaches in an interdisciplinary collaborative project. As neuroscientists and audience researchers informed by phenomenological approaches, we are aiming to understand the relationship between experiences of audience members as described in the first person and the corresponding neurophysiological data. Due to the complexity of the stimuli and the diversity of the audiences’ experiences, we cannot simply match the knowledge of elementary neural processes in visual perception to the experience of an audience member watching a dance performance in the theatre. How could we, for example, explain the very individual responses to one performance? A recently published essay by the father of psychophysics, Gustav Fechner (2009), using introspection provides some insight into the intrinsic difficulty of matching experience to psychological processes.

Our research specifically explores audiences’ responses to live theatre danceFootnote 1 using methods drawn from cognitive neuroscience and qualitative audience research informed by phenomenology. We argue that while these fields carve out different and sometimes incommensurable areas of knowledge, it is possible to open up channels of communication between them, which can enrich our overall understanding of spectators’ responses to this significant art form, which is currently enjoying a high profile on the UK cultural scene. Styles of dance we refer to in this paper are bharatanatyam, which is a classical form of Indian dance, contemporary dance, and ballet.

The neuroscientific focus on dance has been strongly motivated by a surge of research outputs related to the identification of so-called mirror neurons (Gallese et al. 1996). These neurons in the frontal and parietal cortices of monkey brain fire both when a monkey grasps for food as well as when the monkey simply observes the experimenter executing the same grasping action. In the search for understanding the functional properties of equivalent neurons in the human brain, dance has proven to be a useful tool. Dance enables neuroscience to study basic functional properties of movement observation which include non-verbal communication and aesthetic appreciation different to many goal and object-related ‘everyday actions’, such as grasping a cup, and sporting activities. In some ways, spectators’ responses to dance, in particular motor simulation, can be assumed to be similar to those when watching other physical activities like gymnastics (Munzert et al. 2008) or basketball (Aglioti et al. 2008). However, theatre dance combines physical (athletic) components with artistic (aesthetic) values. Unlike competitive sports, gymnastics or athletics, or even competitive dance (such as ballroom competitions), the quality of theatre dance cannot be measured in terms of a ‘score’ by which one beats an opponent (see also Friesen 1975). A theatre dance performance is a multifunctional socio-cultural event constituted of many diverse strands, including the dancers, the music that accompanies the movements, the costumes, the lighting, the set, the other audience members, and so on, that work together to impact on the spectator in the moment of watching.

Certain attributes of theatre dance, such as the interaction of several modalities like music and movement and also the long duration of dance performances (in contrast to the 3-s presentation duration of stimuli commonly used in neuroscientific experiments), make its use in scientific experiments problematic. Qualitative methods of audience research, informed by phenomenology, are well placed to address some issues, although this approach can in turn be challenging for a neuroscientific approach. For instance, qualitative audience research emphasises describing experience in an exploratory, interactive process which takes place after, rather than during the event and itself contributes to uncovering and shaping spectators’ experience. Moreover, unlike neuroscience, there is no clear-cut distinction between collecting and analysing data, as the methods (e.g. the questions asked in interviews) are already in themselves part of the analysis. This means that each process is unrepeatable and aims to be valid on its own terms rather than to provide a predictive model. However, audience research does not have access to the brain functions studied by neuroscience. We feel that neither qualitative research nor neuroscience can present a complete picture of spectator response and we are interested in combining both first-person experiences and third-person neurophysiological data to explore spectators’ kinesthetic responses to watching dance.

The collaborative research project

The work presented here comes out of a collaborative research project called “Watching Dance: Kinesthetic Empathy”, funded in the UK by the Arts and Humanities Research Council. The project uses audience research and neuroscience to explore how dance spectators respond to and identify with dance based upon the assumption of the concept of kinesthetic empathy. The following essay addresses some questions emerging from descriptive data gathered in qualitative interviews with audience members and from data produced by questionnaires, by functional magnetic resonance imaging (fMRI) and by transcranial magnetic stimulation (TMS, see “Dance turns science round and round” section for further methodological details), which were analysed quantitatively. The crucial question for us was to what extent (if at all) do spectators of dance feel that they are emotionally and cognitively absorbed by and engaged in the movement of the dancer(s)? And in terms of the methods employed to answer this question, how might spectators’ experiences as reported through qualitative audience research correspond to neurophysiological data as derived through fMRI and TMS? And in particular, what can we learn from combining these approaches that we could not discover by either method used separately?

In the present paper, our main aim is to argue for the importance of interdisciplinary collaboration, rather than to engage with the concept of kinesthetic empathy itself and the debates surrounding it. However, a brief overview of this topic will be useful at this juncture. In dance research, the concept of kinesthetic empathy has emerged to describe the response of some spectators when watching dance. Spectators frequently report that even while sitting still, they feel they are participating in the dance they observe, experiencing movement sensations and related feelings and ideas. This type of response has been described in terms of kinesthesia (Daly 1992) and kinesthetic empathy (Hagendoorn 2004; Lipps 1906; Martin 1939), and was a major focus of our collaborative research. The concept of kinesthetic empathy as it has been discussed in dance studies is controversial, both in terms of how it is defined and how it has been regarded by scholars and critics.Footnote 2 For some, it is integral to the experience of the ideal dance spectator, while others see it as falsely universalist (see discussion in Reason and Reynolds 2010). However, it is currently under the spotlight, partly as a result of the impetus provided by neuroscience research on ‘mirror neurons’, as mentioned above.

Some neuroscience researchers claim that there are brain areas in action observation linked with empathic responses (Keysers and Gazzola 2009). Empathy has been described as “the drive or ability to attribute mental states to another person/animal, and entails an appropriate affective response in the observer to the other person’s mental state” (Baron-Cohen and Wheelwright 2004). Some researchers focus on the affective rather than the cognitive aspects of empathy and acknowledge any form of affective response as empathic, whether matching the experience of another or showing compassion for another, or simply any type of emotional response (e.g. pleasure at somebody else’s pain). Definitions of kinesthesia and proprioception have a complex history (see Foster 2008), but at least since Sherrington’s work on the ‘muscular sense’ (Sherrington 1907), proprioception (or kinaesthesia) is widely accepted in modern cognitive psychology as a distinct multicomponent sensory modality, often referred to as the sixth sense (Wade 2003).Footnote 3 Many studies of proprioception and kinaesthesia involve angle or position matching, or balance tests. Examining the role of motor simulation in affective empathy is, however, novel.

Neurophysiological approaches have the advantage of measurement at the moment of the activity of watching dance. However, the popular neuroimaging techniques which enable study of the functional role of cortical areas have spatial and temporal restrictions which make it impossible to expose participants to real life dance events. Therefore, even if one agrees with Zeki (2002) in theory, that there is a neurobiological network responsible for the spectator’s response, in practise, we may not be able to explain it at this point in time by the use of a single methodological approach. As mentioned above, dance is more than complex abstract movement sequences of 3-s in front of a blue background, as it is frequently presented in neuroscience studies. In our research, we wanted to try to maintain the variables of performance, such as costume, music, and longer duration.

The diversity of dance theatre aesthetics and the wide range of subjective preferences prompted us to study individual differences in spectators’ experience when watching dance and to explore links with their neuronal activity. For instance, we measured how the visual experience of watching dance affected different dance spectators’ verbal and neurophysiological responses to watching their preferred, familiar style as compared to a novel, unfamiliar style of dance (Jola et al. 2011, Enhanced motor cortex excitability induced by watching dance in emphatic and visually experienced dance spectators, under review; see also Grosbras et al. 2010; Jola et al. 2011). In brief, we found evidence that motor simulation, which we related to kinesthetic experience, was modified by both visual expertise and by the spectators’ empathic abilities. However, the link between motor simulation and empathic responses remains hypothetical. The simultaneous occurrence in some individuals of kinesthetic sensation and motor simulation was the only evidence of their relatedness. Similarly, we cannot yet conclude whether enhanced empathic abilities led to enhanced empathic responses.

“Watching Dance: Kinesthetic Empathy” thus faces at least two methodological challenges. First, investigating dance as a live art form represents a difficulty for conventional neuroscience research in particular due to time constraints (dance lasts much longer than conventional experimental stimuli) and less rigorously controlled stimuli (dance is multilayered). Second, linking spectators’ experiences with their neuronal underpinnings challenges qualitative audience research in terms of the need to generalise across individual experiences in order to make the findings comparable with neuroscientific results.

Dance turns science round and round

Why research on watching dance differs from existing neuroscience research on visual motion perception

In visual perception research, how neuronal processes give rise to a unified experiential perception of objects or events has been investigated for over half a century (Spillmann 2009). In particular, scientists have found neurophysiological mechanisms that are indispensable in the explanation of such visual percepts. Quite a large number of these are related to the ways in which we perceive motion visually, such as grouping by coherent motion,Footnote 4 apparent motion,Footnote 5 or our extraordinary ability in biological motion perception.Footnote 6 For instance, grouping of coherent motion has been described as the most powerful example of a correlation between neurophysiological findings and perceptual experience (Spillmann 1999, p. 1479). An example of such grouping is when a circle of regularly spaced dots within a field of random dots can be perceived in the instant they start moving coherently. This so-called pop-out effect has been explained by selective neuronal activity for coherent motion in the visual areas (V1 and MT) of the monkey’s brain (for details see Spillmann 2009). Also, some scientists are working bi-directionally, which means that they are not just studying neural correlates of perceptual illusions, but they also create perceptual illusions based on their understanding of perceptual mechanisms. This is a rare approach; but if successful, it allows an experiential way of evaluating scientific models of perception.Footnote 7 In sum, in visual perception research, the descriptions of the experiences of perceptual illusions and the neurophysiological data gathered during the perceptual processes have been successfully complementing each other.Footnote 8 However, several factors distinguish this type of research from studying audiences’ experience of watching dance.

In visual perception research, the approach is confined to the study of mostly low-level visual perceptual mechanisms (which means early in the process of visual perception) whereas watching dance involves processes which culminate at higher levels of perception. Also, in contrast to the diverse complex strands we have referred to in the context of watching dance, the engagement with the stimuli in research into low-level perception is often limited to visual aspects of percepts in the form of a perceptual illusion which is perceived consistently across many participants. Finally, the stimuli used to study visual perception are predominantly controlled with precise millisecond timing, and retraceable to an experimentally controlled event. A good example is the immediate response of a neuron in the visual cortex to a light being switched on-and-off as shown by Jung et al. (see Spillmann 1999, p. 1462). Watching dance, however, is not based on a single clear indicative event such as switching on-and-off of sensory stimulation. Dance is multimodal and ephemeral, and since no two performances of the same work are identical, reproducibility is limited and conducting a scientific experiment is a challenge.

Due to the complexity of the stimuli and the diversity of the audiences’ experiences, we cannot simply match the knowledge of elementary neural processes in visual perception to the experience of an audience member watching a dance performance in the theatre. While visual and notably auditory information are the main sensory stimulations the spectator is engaged with, what is interesting about dance (and other art forms) is that they bring together experiences on different levels, e.g. perceptual, emotional, and cognitive. Research in visual perception rarely connects the visual phenomena with diverse levels of individual experience and generally uses reduced stimuli with short presentation times that do not allow individual processes based on extended time exposure to kick in (Hasson et al. 2008). Hence, any researcher seriously investigating audiences’ engagement in watching dance has to deal with manifold phenomenological descriptions or neuronal/behavioural responses.

The evolution of ‘naturalistic’ stimuli in cognitive neuroscience

The experimental setup classically used in cognitive neuroscience consists of manipulating a couple of features of a particular predominantly visual or auditory stimulus, in order to identify the effect of one feature upon another with a certain statistical significance and power. This approach has provided much scientific insight into many areas, beyond human perception. However, the inferences made from such highly controlled approaches are not necessarily transferrable to complex situations. In other words, the ‘artificial’ experiments have little in common with the ‘real world’. We have outlined some of the issues below.

There has been previous criticism in scientific fields of studies whose models were based on reductionist approaches (e.g. Dehaene 1989). An example of change of stimulus complexity in experimental research is illustrated in Fig. 1. Graphical displays with a number of simplified objects or letters as illustrated in the upper row of Fig. 1 were frequently used by experimental psychologists to test their models of perception and attention. Treisman and Gelade (1980), for example, verified their feature-integration theory (i.e. complex visual scenes are perceived via single features which are then combined in a second step) against the Gestalt account (i.e. a complex visual scene is perceived as a whole or ‘gestalt’ which can then be dissected into its components) by means of similar stimuli. Also, the study of biological motion perception mostly involves short movement sequences. These are often presented in the form of point light displays, as shown in the middle row of Fig. 1, in order to remove any visual information that may interfere with the research question or just add noise to the data (Pollick et al. 2010). However, the visual scene a dance spectator is exposed to is rarely that simple. An audience member may—from a good seat—see several dancers, involved in different actions, as well as moving in synchrony as shown at the bottom of Fig. 1, all of which are part of the experience of watching dance. And importantly, the inferences made from highly controlled approaches as exemplified above are not necessarily transferrable to complex situations. In particular, recent studies showed that the complexity of experimental stimuli does affect visual and auditory perception (Cook and Van Valkenburg 2009).

Fig. 1
figure 1

Examples of experimental stimuli used in scientific research from pop-out feature detection to motion perception, and a scene from a piece that an audience may see (from top left clockwise). Figures by Austin Mcwiggin (feature integration, wikipedia project), Corinne Jola (artificially created point-light display), Scottish Ballet (Gregory Dean as the French Prince, Jarkko Lehmus as The King, Eve Mutso as The Queen, Tama Barry as the Russian Prince and Christopher Harrison as the Austrian Prince in Page’s The Sleeping Beauty. Photograph by Andrew Ross). Please see online publication for colour effects

Full size image

As one might expect, in the vast majority of published studies that used ‘dance’ to investigate neurophysiological processes, the movement sequences have not yet reached the full potential required to study the multiple levels involved when watching theatre dance. The stimuli used in empirical neuroscientific studies differ from stimuli experienced outside the lab environment on several levels. For one, simple, constrained stimuli have an artificially reduced number of variables. A reasonable number of studies investigating human motion do present a movie of the full human body in motion but control for factors such as face recognition by blurring the face of the performer (e.g. Calvo-Merino et al. 2006). Also, in most neuroscientific studies, the combinations of features are limited. For example, in most experimental studies, movements are performed in neutral clothing, presented in front of a monotonous two-dimensional background without music. However, in ‘real-life’, dance frequently plays with sound, costume, and light effects. Finally, the time is constrained, ranging from presentations in the millisecond range in vision perception research to a few seconds in neuroscientific studies on motor control.

We recognise that these previous studies were important to the development of dance and neuroscience collaborations and using simplified stimuli is still an accepted way to study human cognitive and emotional processes. After half a century of increasingly proficient reductionist approaches, however, a number of cognitive neuroscientists are eager to develop methods that allow experimental investigation by means of so-called ‘natural’ stimuli (Hasson et al. 2004; Wade and Bruce 2001).

Complex data in qualitative research

For our purposes, the research has to engage with dance as a Western theatre art form, unconstrained by laboratory conditions and presented as an art form to theatre-going spectators, thereby raising questions of aesthetics (e.g. judgment/taste, pleasure). If studying complex, unpredictable data presents difficulties for neuroscience, this is precisely the forte of qualitative research methodologies. Qualitative research is concerned with the various ways in which the social world is ‘interpreted, understood, experienced, produced or constituted’ (Mason 2002a, p. 3). It emerges from different traditions and disciplines in the human and social sciences and encompasses a variety of methods related to divergent disciplinary approaches and philosophies. The main methods associated with qualitative research are ethnography/participant observation; qualitative interviewing; focus groups; language-based approaches (such as discourse analysis); collection and analysis of documents and texts, and visual methods. Within this methodological variety there are common strands which distinguish qualitative research from perspectives commonly associated with cognitive science. Most obviously, a large amount of research in the cognitive sciences is ‘quantitative’, with outcomes that can be expressed numerically. Qualitative researchers, by contrast, maximise the depth, complexity and nuance provided by the data. Although it may involve some quantification, this is not regarded as central. Importantly, audience research is concerned that meaning would be lost if data were reduced to numbers. In qualitative research, there is a two-way relationship between researcher and data where the researcher aims to learn from the data.

“All qualitative methods seek to discover understanding or to achieve explanation from the data instead of from (or in addition to) prior knowledge or theory. Thus the goals always include learning from, and doing justice to, complex data, and in order to achieve such understanding, the researcher needs ways of exploring complexity.” (Richards and Morse 2007, p. 2).

This means that the qualitative researcher needs to be flexible and to adopt a critical, reflexive approach to the research process.Footnote 9 The methods of generating data can be adapted to changing ‘real life’ contexts, and to changes in the researcher’s approach generated by the data. The integration of collection and analysis is seen in itself as a major strength of qualitative inquiry (Richards and Morse 2007, p. 1). This unremitting adaptation of question, data and analysis means that each case is quite specific and is not reproducible. However, reproducibility is not necessary, as by the same token, qualitative research does not set out to produce predictive statements. The aim is to provide descriptions of and insights into specific cases, which may be informed by, and in turn inform, conceptualisations or theories. Descriptions produced in this way are sometimes described as ‘thick’, a term originally used by Ryle (1949), meaning that they incorporate considerable detail and depth. Although the outcomes of qualitative research may not be immediately transferable or predictive, if the descriptions are sufficiently ‘thick’, others will be able to evaluate the extent to which the conclusions are transferable to different contexts.

Complex data in cognitive neuroscience

The cognitive sciences have always aimed at being able to make predictions about perceptual, emotional or cognitive processes that are valid in the real world. This proves particularly challenging in the case of dance, with its high degree of complexity, as well as the effects of liveness (no two performances are the same or seen in the same context). Indeed, one may doubt whether it is possible at all to make consistent statements about the perceptual, emotional and cognitive processes associated with dance as seen from a phenomenological perspective.

In general, with advances in brain imaging technology it has been increasingly viable to relate brain activity with experimentally driven studies of participants’ individual, subjective evaluations and responses. Such an approach can add substantial information to the brain data and can be either predictive (e.g. effect of independent varied stimuli features on the dependent measures) or reversed (enhanced brain activity is traced back to the event in the stimuli). Calvo-Merino et al. (2008) related the brain activity of naïve observers when watching dance movements to their consensus preference ratings. However, little attention has been paid to linking brain activity with individual experiential awareness in the field of movement observation. This may be partly due to technological limitations of fMRI. For instance, with EEG (electroencephalogram) and TMS, functional processing can be identified in the human brain in the range of milliseconds whereas fMRI has a higher spatial (in the range of few millimetres) but lower temporal resolution (in the range of seconds). By means of TMS, researchers can trigger the motor cortex of a participant at a particular time-point and they will know exactly what the participant was shown in relation to the data measured. In fMRI, brain activity analysis is restricted to the time it takes to scan the brain areas of interest, which is usually around 1–3-s. Therefore, if using more complex stimuli, several events are combined into one signal. In some studies, the participants’ feedback is nevertheless required to identify the underlying processes of the brain activity measured. This is for example the case in terms of research on consciousness where the participants’ state of mind is the subject of investigation. One of the main topics in these studies is ‘change blindness’ (e.g. Simons and Chabris 1999; see also Simons and Rensink 2005), where the participant is not aware of what is in front of him. This raises interesting questions for phenomenological approaches, as the participants’ descriptions may not correspond to the actual stimulation. In our research, we are not questioning the spectators’ reports, or that they ‘miss out on one thing or another’. What the spectator experiences is what we want to know. This approach informed our neurophysiological design in the way we ignored—to a certain degree—issues of stimulus complexity but instead focused on differences between audience members. This allowed us the freedom to evaluate our measures of cortical excitability without defining a precise physical event.

The qualitative lens

We have suggested that qualitative research can complement some of the limitations of neuroscience, and have indicated that our approach is informed by phenomenology. This means that we focus on dance spectatorship as lived, embodied experience, on which research participants are invited to actively reflect. Rather than being abstracted and decontextualised, the experience of watching dance is connected with the spectator’s ‘lifeworld’: “Phenomenology attempts to explicate the meanings as we live them in our everyday existence, our lifeworld.” (Van Manen 1990, p.11). To date, academic empirical research on Western theatre dance audiences is extremely rare, and existing research leans towards approaches informed by cognitive psychology rather than qualitative methods (Hanna 1983; Glass 2005; Glass and Stevens 2005). While a small but growing body of qualitative research on theatre audiences exists (Reason 2010a,b, 2006a,b; Sauter 2000; Schoenmakers 1990), our approach is pioneering in the context of dance audiences. Our audience research activities take the form of interactions (usually but not always verbal), which take place after the performance, and therefore effectively deal with memories and reflections rather than with the experience of watching as it takes place in the moment. Reason writes, for instance, about his research with young audiences: “in bringing participatory enquiry and phenomenology together one major objective was to invite the young audience members to reflect upon their experience and the construction of that experience through memory” (Reason 2010a,b, p. 133).

To date, hermeneutical phenomenology has been the most influential strand in phenomenologically informed audience research. Hermeneutical phenomenology emphasises the role of conscious reflection and also the use of language as part of lived experience (e.g. Van Manen 1990). Indeed, no experience can be accessed or communicated without some form of mediation. Van Manen states that because “a person cannot reflect on lived experience while living through the experience”, phenomenological reflection is therefore “not introspective but retrospective”. Reflection on lived experience is always recollective; it is reflection on experience that is already passed or lived through (Van Manen 1990, p. 10). From this point of view, a spectator’s ‘experience’ of the performance is not located solely in what takes place while they are watching, but encompasses re-construction and interpretation through subsequent reflection in social contexts, which can include activities such as interviews and focus groups. Rather than purely personal and private, experience is treated as socially mediated. Audiences are considered as active agents in constituting the meaning of the performance through articulating their experiences. In describing and discussing their responses to dance performances, the spectators we dialogue with have the opportunity to tease out what they themselves consider to be important about what they have seen. It is interesting to note that Glass (2005) found in her research on audiences of contemporary dance that experiences of dance, and also pre-performance information sessions, had relatively little impact on response, whereas audiences derived pleasure from post-performance discussions and enjoyed being asked their opinion and giving their interpretations of the work. This led her to hypothesise that “Perhaps it is not pre-performance information that is important, but the observers’ opportunity to reflect on their interpretation, understanding and connection with the piece” (p. 120).

There is a strong tradition linking the study of dance and phenomenology (Fraleigh 1987; Sheets-Johnstone 1980). These earlier writings have been critiqued for their universalist approach to lived, embodied dance experience (e.g. Legrand and Ravn 2009), and Rothfield (2005) has argued for the need to extend the phenomenological approach to include corporeal specificity and to differentiate dance experience “according to its social, historical and kinaesthetic milieux” (p. 44). She advocates a ‘pluralization of the lived body’ and argues that phenomenology should’recast‘the lived body as lived bodies. With its empirical base and its strong links with the social sciences, qualitative audience research is ideally placed to counter universalism. This can mean approaching interview data in a number of ways, diversifying the traditional phenomenological approach, by endeavouring to take account of factors such as race, gender, and cultural capital.

Our duet

An approach to studying spectators’ kinesthetic empathy using TMS and qualitative audience interviews

Thus far, we have outlined the interdisciplinary qualitative and quantitative approaches that are being employed in our research and indicated some of the broader methodological and epistemological issues of our collaboration. In this section, we illustrate ways in which we have attempted to link our disciplines by means of a study which combined methodologies of cognitive neuroscience and qualitative interviews informed by phenomenology.

For the study in question, volunteers watched three live solo performances: one short dance piece in the style of ballet, one in the style of bharatanatyam, and a non-dance acting performance (without voice).Footnote 10 The aim of this study was to explore neuronal correlates of visual expertise in watching dance. Previous studies showed that a putative mirror neuron network is more strongly activated for actions that were shared between the observer and the performer, such as when expert dancers watched dance movements they were physically familiar with (e.g. Calvo-Merino et al. 2005). This seminal study, however, did not differentiate between visual and motor experience or measure the spectators’ response by any means other than the neuronal activity in the form of the BOLD (blood oxygenation level dependent) signal. Following on subsequent studies (e.g. Calvo-Merino et al. 2009), which explored visual and motor experience, our aim was to ascertain what, if any, was the impact of different degrees of viewing-only experience in each style on the viewers’ neurophysiological response. This was evaluated by examining cortex excitability induced with TMS and by exploring data produced through qualitative audience research (see below). To separate visual from motor expertise, we only measured participants who had no physical dance training. Roughly, a third of the participants were experienced spectators of ballet but not bharatanatyam, and a third were experienced spectators of bharatanatyam but not ballet, while the remainder had no experience in watching either. As argued above, it was important that the experiment corresponded as closely as possible to the ‘normal’ setting of the experience. Hence, the performers were professionals in each style, dressed in full costume, and danced a solo of 4.5 min to style-specific recorded music. This is an unconventional setting for both neuroscientific experiments as well as qualitative audience research (see also “Conclusions: situating experience across disciplines” section).

The TMS experiment

For the neuroscience research, we assessed cortical excitability as a sign of how much the audience was engaged with the performance. In general terms, the threshold for moto-corticospinal excitability is reduced when the subject is engaged with observing or executing a movement, e.g. during motor imagery, action observation, and motor preparation.Footnote 11 In this experiment, participants’ responses were measured while they were watching a live performance. The point of measure was clear and reproducible, but nonetheless movement selective. In other words, the participants’ cortical excitability was measured every 8-s with a random jitter of between zero and 2-s a total of 30 times. Hence, at first, we ignored which particular type of movement the performer executed at each time point of the TMS trigger. This experiment was not designed to control systematically for particular movements within each style for each trial. Nevertheless, each style has a systematic movement style. Since we do know that our individuals were different in their visual experience and they would experience individual levels of enjoyment, we focused on correlating their subjective responses with differences in neurophysiological data (for detailed methods please see Jola et al. 2011, Enhanced motor cortex excitability induced by watching dance in emphatic and visually experienced dance spectators, under review; see also Grosbras et al. 2010).

The experiment showed that the excitability of the cortex of spectators who have previous visual experience in watching ballet is higher when they watch ballet than any of the other performance forms, with which they were less familiar. This was not the case for either the control group or the experienced bharatanatyam spectators (whose level of experience was less consistent).Footnote 12 We also found statistically significant correlations between the participants’ levels of empathic abilities and their response to watching dance: participants who scored higher on empathyFootnote 13 showed higher cortical excitability than those who scored low (Jola et al. 2011, Enhanced motor cortex excitability induced by watching dance in emphatic and visually experienced dance spectators, under review). However, these results were based on comparing the responses to the different styles by different groups of participants (those who scored high or low on empathy, those who were visually experienced or inexperienced spectators of ballet or bharatanatyam, or novices in both forms), rather than comparing data between individual subjects. So-called case studies could be pursued (i.e. Kosslyn et al. 2002) but we have yet to define valid scientific selection criteria to single out individuals as case studies from the 32 participants.

The qualitative interviews

For the audience research, each subject participated (after watching the performances) in a semi-structured interview, i.e. an interview which uses certain pre-set questions, but where the order can be varied and the interviewer can also respond to directions suggested by the interlocutor’s response (see Mason 2002b). We used this interview method because we did not want to be confined by our own questions, which would inevitably influence interviewees’ responses. The emphasis was on encouraging participants to describe freely what they had seen or felt and their responses to it. However, the interviews were also themed around certain questions, such as identifying parts of the performance participants particularly liked. The number of themes and responses were rich and complex, as one would hope and expect in qualitative research. Some participants said that they felt they were responding in their face to the facial expression of the dancer, such as focusing on the bharatanatyam dancer’s smile, and also that they valued the physical closeness to the dancer which resulted from the small performance space, as opposed to the less intimate environment of a theatre. In sum, from asking about embodied response we also learned about what made the performances enjoyable for each individual spectator and made them feel involved (or not).

To facilitate qualitative interview analysis categories were devised using QSR NVivo 8,Footnote 14 starting off with themes which were anticipated through our research questions such as ‘familiarity’ and ‘unfamiliarity’, ‘connection to dancer’, ‘desire to move’, ‘emotional response’, and ‘embodied response’. We noted that the majority of experienced ballet spectators expressed high levels of enjoyment from the ballet and their responses frequently came into the categories of desiring to move, feeling a connection to dancer, and having an emotional response to the performance. By contrast, people who enjoyed the performance less made very few references to these categories. Moreover, detailed analysis of the interview material, known as ‘thick description’ (see above), also prompted us to add new categories, such as ‘admiration of virtuosity’, ‘evaluation of quality’ and even ‘can’t remember’. The latter was because lack of familiarity or understanding sometimes made it difficult for participants to take in or remember much of what they had seen. Another category that we added was ‘motivation’ as it became evident that what the participants were looking for in a dance performance had a huge effect on how they responded to what they saw. This impacted on our understanding of the original research questions.

Spectators’ enhanced excitability

Whereas we had started out to explore the importance of familiarity in terms of viewing experience, people’s responses underlined for us the importance of wider factors relating to cultural competence, as described by Bourdieu (1984): “A work of art has meaning and interest only for someone who possesses the cultural competence, that is, the code, into which it is encoded. [...] A beholder who lacks the specific code feels lost in a chaos of sounds and rhythms, colours and lines, without rhyme or reason.” (p. 3). By the same token, an individual’s general disposition to consume and seek exposure to certain kinds of things is not random or even necessarily wholly self-aware but instead deeply ingrained into embodied cultural practises and their associated social and economic conditions (see Reason and Reynolds 2010). While the audience research provided examples which corroborated the TMS findings about the effects of visual expertise on movement simulation, it also emphasised the aspect of enjoyment and brought into play factors of cultural experience and motivations, as it was not only what the participants had seen of a given dance style (their level of viewing experience), but also their preconceived/culturally received ideas about it and their personal preferences or motivations which affected how they experienced it.Footnote 15

Matching different ways of representing kinesthetic empathy

Our question in the next part of the research was: what is the correspondence between the data from the neuroscience research for each individual’s average responses to the three performances and their experiences as reported verbally in the interviews? In particular, we were interested to find out whether stronger responses as measured through TMS corresponded to stronger kinesthetic responses and/or levels of enjoyment as articulated through interviews. If we were to find participants’ described experiences reflected in their neurophysiological responses, would the explanatory gap between cognitive neuroscience and qualitative audience research be narrowed? However, while attempting to match individual verbal reports with individual cortical excitability, we reminded ourselves that this is not much different from the psychological phenomenon of pareidolia, i.e. seeing a random stimulus as being significant, as it often is described by the example of seeing figures in the clouds. How can we see the connection between the data gathered without over-interpreting? The human perceptual system acts in a creative manner (see “Data turns science round and round” section above), so if we tried to match a subject’s verbal report with the neurophysiological data, we might identify some connections, but at the risk of a bias. Thus, the best way to reduce the chances of this risk, from a neuroscientific perspective, would be a ‘blind review’ in form of a researcher who was kept uninformed about the studies’ aims.

Hence, similar to an inter-rater reliability test, a former student who had no experience with this study was paid to match both types of data, i.e. cortical excitability with verbal reports. First, she was asked to find the neurophysiological responses to each performance that belong to the same individual.Footnote 16 Clearly, these data were from the same individual, and should therefore be related, however, if the movements between hand and arm were different at any time point of the randomly applied triggers between the three types of performances observed (which we hypothesised) the data could only be matched according to overall individual characteristics (such as the subjects’ average level, when high empathic abilities give higher cortex excitability independent of the movement observed). Second, the student was asked to select a few salient audience research transcripts and find the corresponding data. However, as expected, her task was a very difficult one, if not impossible: of all participants, the cortical excitability of the hand and arm were matched correctly for one participant only. This means that she was not able to identify one person’s data, i.e. matching graphs of the hand (as measured during the bharatanatyam performance) to the arm (as measured during the ballet performance). This is not surprising, as the location of the measure, the embodiment, varied between different types of performances (in ballet, expressive gestures emphasise arm movements, while in bharatanatyam they emphasise finger movements). Interestingly, of the incorrectly matched data, the student matched six out of 11 ballet spectators with other ballet spectators, three out of ten novices, and none out of eight experienced bharatanatyam spectators. It seems thus that the student’s naïve matching corresponds with the scientific observation, namely, that the responses of the bharatanatyam spectators were the most inconsistent ones. As noted above, the ballet spectators were more experienced than the bharatanatyam spectators. Hence, the objective raters’ outcome supported the information we gained for the qualitative audience research.

Conclusions: situating experience across disciplines

“All my knowledge of the world, even my scientific knowledge, is gained from my own particular point of view, or from some experience of the world without which the symbols of science would be meaningless” (Merleau-Ponty 1945, p. ix).

In this paper, we have mapped parts of our experiences to date in combining different epistemological and methodological approaches in our exploration of spectators’ responses to watching dance. This process requires openness to each researcher’s position at the same time as being grounded in our own disciplines, combining the readiness and skill to perform ‘grand jetés’ across disciplinary boundaries with core strength and balance.

Generally, neuroscience works within a paradigm which values simplified stimuli, precision, and statistically significant results that should be valid for a wider population. Phenomenologically based audience research values more open-ended exploration and rich data situated culturally, historically, and/or socially. This contrast could make neuroscience seem conclusive and fixed, though we have found that this is not always the case, and we value Merleau–Ponty’s statement, cited above, which acknowledges the grounding of all knowledge in lived experience.

By letting neither neuroscience nor audience research methodology take primacy over the other, but by working bi-directionally, we have deepened understanding of how audience members respond to watching dance. The neuroscientific data indicated that empathic abilities enhance cortical excitability for mimed everyday hand actions, while visual experience enhances cortical excitability for formal—stylised—movements, such as performed in ballet. The outcomes of the qualitative research, where responses from spectators who enjoyed the performance more also indicated stronger levels of kinesthetic and empathetic engagement, lend weight to the argument that kinesthetic empathy is connected with spectators’ pleasure in watching dance (see Reason and Reynolds 2010). Also, although the neuroscience doesn’t tell us about conscious experience and qualitative research doesn’t tell us anything about the brain, the fact that the experienced ballet spectators showed distinctive characteristics in both the neuroscience and qualitative research may indicate connections or at least commensurability between conscious experience and its ‘neural substrates’ which merit further exploration.

The interactive approach further provided insights into the other’s methodologies and fostered exploration of ways in which the other discipline and their results could be applied in our own discipline. For instance, the questionnaire used to correlate empathy scores with cortex excitability in cognitive neuroscience highlighted that the qualitative research does not use such a tool, but the results have informed qualitative discussions of how kinesthetic response relates to empathy. The qualitative researchers have also introduced a quantitative aspect to their research by developing a questionnaire in conversation with the neuroscientists aimed to further understanding of audiences’ motivations, which will be posted online and distributed at theatres, and will be analysed quantitatively. In the other direction, it emerged through the interviews that there were significant differences in the extent to which audience members focused on certain parts of the body, e.g. legs and feet/arms and upper body. Exploring the reasons for this is likely to prove a fruitful area of investigation for cognitive neuroscience.

We are learning to value contrasting aspects of each other’s fields, and to challenge fixed conceptions. Rather than eliminating disciplinary differences, we aim to advance knowledge and understanding of the other position (e.g. through reading each other’s work, reading material across disciplines, and even training in aspects of each other’s research methodologies). Flexibility on both sides is required to achieve these aims, including challenging our methodological approaches by considering new ways of thinking via collaboration (Moran 2010, p. 181). For instance, the dance performances for the TMS experiments and qualitative interviews stretched the parameters of the TMS procedures in the direction of empirical neuroscience stimuli, but also required acceptance of deviation from ‘normal’ theatre conditions for the audience researchers. There may be the risk here of compromising at both ends and thus being open to critique from within our chosen disciplines about the rigour of our methods. However, without this flexibility we would not be able to collaborate and communicate across different disciplinary approaches.