An earned presence: studying the effect of multi-task improvisation systems on cognitive and learning capacity

ABSTRACT

In this article, we articulate preliminary insights from two pilot studies. These studies contribute to an ongoing process of developing empirical, cross-disciplinary measures to understand the cognitive and learning effects of complex artistic practices – effects that we situate between theory of embodied concepts and conceptually calibrated physical attention and action. The stage of this process that we report on here was led by the cognitive performance studies scholar and dramaturge, Pil Hansen, and undertaken in collaboration with the experimental psychologist, Vina Goghari, and the behavioural economist, Robert Oxoby, assisted by four research assistants from Drama, Music, and Psychology at the University of Calgary. Our team set out to test the following hypothesis: Active participation in performance generating systems has a positive effect on advanced student performers’ working memory capacity, executive functions, and learning. Our results have implications, in particular, for understandings of embodied learning in the educational sector, however a perhaps more significant contribution is a better understanding of the measures and constructs needed to arrive at a more complex, yet operational concept of embodied learning and forward the experimental study of relationships between performing arts practices, cognition, and learning.

KEYWORDS:

• Performance generating systems
• improvisation
• dance
• learning
• embodiment
• cognition
• empirical research methods

Abductive inferences behind the hypothesis

Our experimental intervention, Performance Generating Systems, has been defined elsewhere by Hansen as a systematic and semi-closed form of improvisation in dance, theatre, and music: a dramaturgy of multiple, pre-identified tasks, rules of engagement, and source materials within the boundaries of which performers interact on stage (Hansen, 2015, pp. 124–125). Within a Performance Generating System, as opposed to more open forms of improvisation, performers must work attentively and simultaneously on multiple tasks while remaining aware of limiting rules. The source materials (such as movement sequences, memories of movement, or score text) are predefined and recycled by the performers while working on these tasks. These systems generate a performance on stage and in front of audiences where self-organising patterns of interaction arise over time, but are neither set nor necessarily repeatable.

Here, we report on and discuss two pilot studies that were specifically designed to empirically test interesting, but previously anecdotal observations made in other contexts. Within a past research project, Acts of Memory, Hansen conducted open, qualitative interviews with 14 dance and theatre performers from the Canadian companies (Theatre Replacement and Public Recordings) regarding their investment of autobiographical memory and trained memorisation skills in performance generating systems. The performers also discussed the specific aspects of the performance tasks they had experienced as easy or challenging. Without being asked directly, six performers independently commented on how these systems began to affect what they attended to, how they perceived both their actions and those of others, and when their general learning (within the system) made challenging tasks easier. In other words, they reported a cognitive effect, which could be caused by the specific praxis. Hansen has co-taught the systems to students whose ability to work analytically with complex concepts advanced beyond the level those same students were able to achieve in her other courses. In the open-ended comments section of course evaluations, students ascribed their learning achievements to experiences they had during practical, physical workshops. In addition to testing these anecdotal observations in our pilot studies, we also sought to explore whether there could be a connection between the cognitive effect reported by experienced performers and the learning effect reported by students.

As is typical for pilot studies, we turned towards well-established tools within our respective disciplines when selecting tests and data analysis methods to measure the cognitive and learning effects of our intervention: performance generating systems. That said, these measures reflect what we thought an effect might be related to and what we more specifically thought was affected. In short, we chose to utilise quantitative tests of working memory and fluid intelligence as measures, hypothesising that the cognitive needs of participating in a performance generating system may manifest in such measures. To study learning effects, we applied a combination of deductive and inductive qualitative coding categories. These were designed to reveal relationships between conceptual and physical learning engagement with reference to the notion of embodied learning while also identifying the primary factors that impacted these relationships over time. A more detailed introduction of these tools will be offered later. First we take a closer look at the demands our intervention may place on cognitive systems and how it can be understood through an embodied cognition framework.

In a performance generating system, the performers’ work on complex tasks within restrictive rules that demand sustained activation of often competing foci of attention (see Hansen & House, 2015). In the example of the choreographer Ame Henderson and Public Recordings’ “futuring memory” system, the dancers’ meta-task was to work towards unison together and without surrendering the individual. Unison is a form of simultaneous dancing that normally depends on the perfect execution and timing of a choreographer’s set and rehearsed choreography, effectively demanding that the dancer surrenders his or her individuality to become a seamless part of the group. In the futuring memory system, rules were established to ensure that dancers did not replace their conventional role in a unison formation with that of a leader or follower (e.g. dancers were not allowed to copy each other). A task called “futuring” was developed as a solution to the problem of how to work on unison without copying; dancers were tasked to form conscious hypotheses about where their co-dancers would go next, and then meet them there. Honouring the embodied memory and repertoire of choreography danced in the past, which dancers implicitly draw upon for improvised work, a task of “futuring memory” was added to the system. Dancers were invited to bring pre-identified sources (i.e. specific memories of past dances) into the work by recalling them when prompted by a physical association. The rest of the group would continue to future the recalling dancer and each other, while the recalling dancer had to recall transparently – that is, adapting to the physicalised hypotheses of the group rather than developing their own hypotheses of other’s movement. When this performance generating system is engaged, the dancers have to sustain simultaneous attention to five competing foci:

1. Perceiving the difference between the trajectory of one’s own movement and that of everyone else’s movement.

2. Forming hypotheses about how the group will move in the moment after right now.

3. Realising these hypotheses in movement, trying to meet the group where it is going next.

4. Registering moments when a physical posture or movement aligns associatively with a set of past choreography, which can then be performed.

5. Minding rules: leading, following, and copying are not allowed.

The challenge proposed by the system is largely impossible to meet; dancers try, sustain the attention for a while, start shifting, catch themselves copying, adjust and refocus, again and again. The effort involved is exhausting and the learning curve that sustained attention to these foci depends upon is steep and relentless. When dancers’ skills improve and they get better at futuring each other’s memories, new dancers, memory sources, or rules are introduced in order to keep the dancers’ efforts alive and revitalise the system’s ability to generate unpredictable performance (Hansen, 2015, pp. 129–130).

Our theoretical foundation and choice of measures are based on concepts from experimental, cognitive psychology that informed our understanding of these foci. Consider a “futuring dancer” who is attending to task execution while abiding by the rules of a given performance generating system. This dancer’s perceptual orientation depends on using current information to form conscious hypotheses about where performers are going to meet in the next moment of the performance (i.e. the moment after right now). This orientation is kinaesthetic (i.e. cross-modal), including the performer’s proprioception and motor system, peripheral visual perception, and auditory perception. At the same time, the direction of attention needed to train and presently sustain these modes of perception demands executive control within working memory. Although there are many theories of working memory, core to most models is the proposal that working memory involves the temporary storage and processing of information that is either perceived in the present or drawn from long-term memory. Such working memory processes include attending to and mentally manipulating information while suppressing distractions (Lawlor-Savage & Goghari, 2016; Miyake & Shah, 1999), both of which are involved in the task of futuring memory. The problem-solving involved in forming futuring hypotheses also depends on higher executive functions, applied with fluency and in flow. As performers learn their co-dancers’ patterns of response, apply multiple foci of attention simultaneously, and shift between the futuring and recall tasks, they may train cognitive flexibility and generation (Lövdén, Bäckman, Lindenberger, Schaefer, & Schmiedek, 2010). The rules (e.g. do not follow, lead, or copy) counter the learned skills and habits that are coded in the performers’ implicit procedural and episodic long-term memory. Thus, the executive function of inhibition is added to our cognitive matrix. In other words, instead of primarily perceiving and responding through memory without conscious awareness (as is the norm), performance generating systems aim to inhibit long-term memory, expanding the performer’s attention to sensory stimuli in the present and enhancing the performer’s ability to process stimuli using executive function skills (e.g. working memory capacity, cognitive flexibility, and generation).

Within a framework more familiar to dance and performance studies, what likely is earned through the effort of performing this impossible multi-task is an enhanced performance presence that does not rely on “getting out of your head” but trains a mindful body and embodied mind. As Henderson expresses it, when performing in these systems “thinking is doing and doing is thinking” (Hansen, Kaeja, & Henderson, 2014, p. 30). As such, the practices of performance generating systems simultaneously require heightened executive control, motor awareness, sensory attention to co-performers, and analytical reflection on relations among performers, thereby pushing past the Cartesian body–mind dualism found in some dance, theatre, and performance studies frameworks. These frameworks include extremes such as forms of semiotic structuralism that regard physical action as a semiotic sign available for meaningful interpretation, constructivist approaches that dissolve the sensory capacity of the biological body into social constructions, and counter-responses that default to using a metaphysics of the body as a framework of study.

Embodiment and embodied learning capacity

Body–mind dualism is so engrained in social practices within and beyond dance and performance that direct attempts to apply theories of embodied cognition can come up against boundaries. There are ample and well-established theories of embodied cognition, action-based language, and concept-motor interactions with varying degrees of grounding in 4E philosophy and the cognitive sciences, including experimental psychology (e.g. Edelman & Tononi, 2000; Glenberg & Gallese, 2012; Noë, 2004; Rodriguez, McCabe, Nocera, Reilly, & Norbury, 2012; Thompson, 2007). These understandings have inspired artists, dance scientists, and cognitive performance scholars, giving rise to a growing research field (e.g. Hansen, 2016; Kemp, 2017; Lutterbie, 2011; Sofia, 2016). This work has found its way into higher education as theory of acting and performance reception, but, to our knowledge, the field has not yet advanced the same approach to enhance learning, nor tested the effects on learning of creative practices that challenge body–mind dualisms. Yet, a comparatively far more widespread application of embodied cognition is in the area of education and learning, where the frameworks closely match Piaget’s influential theories from the 1950s. As demonstrated by Carly Kontra et al. in a current overview of embodied learning approaches, non-artistic pedagogical applications tend to place the body at the service of semantic and conceptual learning (Kontra, Goldin-Meadow, & Beilock, 2012, pp. 736–737). For example, physical movement is used to create simple bodily representations of mathematical terms and concepts to help students grasp math. While the enhanced learning effect of such embodied learning instruments is well documented (e.g. Broaders, Cook, Mitchell, & Goldin-Meadow, 2007; Goldin-Meadow, Cook, & Mitchell, 2009), the use of predominantly semantic and conceptual learning objectives makes these studies less relevant in the context of creative and emergent learning processes. An example of an inversed hierarchy from the field of dance, which presents comparable strengths and limitations, can be found in Martin Puttke’s application of cognitive imaging and memorisation exercises when teaching advanced classical ballet. Using cognition as a tool to arrive at precise physical execution, Puttke places cognitive tasks at the service of codified physical learning objectives (2010, pp. 110–111).

The embodied learning potential Hansen finds relevant in the context of contemporary performing arts education lies in feedback effects between (i) embodied engagement with concepts and (ii) conceptual engagement with embodied experience. Here, the focus is on identifying the kinds of learning that produce conscious, enactive insight (self-reflexive learning capacity), instead of definitions or perfect execution (absorption of predefined knowledge). This form of self-reflexive learning is what we hypothesise that participation in performance generating systems will facilitate. It is possible that these systems reconcile body–mind dualism through tasks and rules of performance that simultaneously demand conscious intellectual control, heightened sensory-motor awareness, and attention to relational dynamics while “unlearning” trained habits and learning new cognitive practices.

When taking this step we would like to specify how we position our research in current discussions of embodied cognition. Several empirical theories exist regarding the close relationships between the motor and linguistic systems, ranging in arguments from all concepts are embodied (i.e. understood through or mapped onto physical experience; see Gallese & Lakoff, 2005; Glenberg, 2015; Glenberg & Gallese, 2012), to the perspective that either embodied language comprehension or the impact of language on movement depend on the extent to which concepts are motor action salient (e.g. Dove, 2011; Rodriguez et al., 2012; Sidhu, Heard, & Pexman, 2016). It is widely accepted that the motor system plays a role in perceiving action verbs and nouns that are closely associated with physical actions (see Pexman, in press) and there is significant evidence that the production of such verbs can affect physical posture and movement (Rodriguez et al., 2012). However, to our knowledge, there is no empirical evidence for the embodiment of abstract concepts (see Dove, 2016; Pecher, Boot, & Van Dantzig, 2011; Zdrazilova & Pexman, 2013 for recent discussions of this problem). As Pexman indicates in her review paper on embodied conceptual development, empirical research finds greater support for emotional salience than motor salience in the development and comprehension of abstract concepts (Pexman, in press, pp. 13–17). To us, these discussions indicate that the field of embodied cognition is maturing: the early phase of perhaps overgeneralised theories has now transitioned to a phase of asking critical questions, revising past research results, and experimentally examining ranges of embodiment.

It is not coincidental that we focus on two braches of otherwise distinct research into the relationship between motor and linguistic systems: cognitive linguistics and kinesiology. In the performing arts, the discovery that action-salient linguistic concepts affect body posture and movement is as relevant as the insight that motor memory (i.e. physical experience) affects the perception of action-salient concepts. This feedback effect and reciprocal reliance does reflect the practice of assigning concepts to physical actions within dance and theatre, both for purposes of memorisation and instant recall (Kemp, 2017) and in order to bring out certain qualities of expression. To a lesser extent, this feedback effect provides a framework for understanding the productivity of creative strategies and tasks that are simultaneously demanding attention to intellectual conceptualisation and heightened sensory-motor awareness (Hansen, 2016), as in performance generating systems. However, in order to arrive at self-reflexivity in creative learning processes, we need to consider more than action-salient concepts and their related motor actions, it becomes necessary to work with abstract concepts that allow a higher degree of generalisation of experience as well as complex, specialised movement approaches that are not reflected in action-salient concepts. For example, how can “presence” be experienced as conscious inhibition of trained movement skill and heightened sensory attention to the surroundings? In the performing arts, such an understanding of an abstract concept can neither be grasped through physical praxis nor conceptual analysis alone.

As it will be evident from our later discussion of learning capacity pilot results, participants reached an advanced understanding of the concept of presence through both the sensory-motor experience of, for example, falling into copying while “futuring” and the cognitive attempt to inhibit this trained response and instead redirect perceptual attention to the movement of others, possibly sustaining attention past normative working memory capacity. In turn, these physical, cognitive, and sensory experiences were reflected upon, understood, and advanced with reference to the concept of presence. The ability to generalise from sensory-motor experience to an abstract concept through more complex and less directly action-salient movement is exactly the embodied learning capacity that Hansen seeks to better understand and enhance in the the performing arts. In the context of performance generating systems, the cognitive demands of the performance tasks (operation of dual tasks, inhibition, cognitive flexibility, and working memory capacity), could be central to the engagement with and development of embodied learning capacity. It is possibly by meeting such demands that participants in our pilot studies became aware and able to relate abstract concepts to the cognitive challenges experienced while producing complex movement.

Two pilots

Our two pilots were designed around an intensive five-day course on performance generating systems with 20 advanced dance, theatre, and music students. Hansen taught this course with Valentina Bertolani (PhD Candidate in Music) and Christopher House (Artistic Director of Toronto Dance Theatre) in January 2016. Each of the first four days was dedicated to a collective investigation of how a specific system works (how it affects the performers and generates performance) with the fifth day dedicated to students creating their own systems.

The following systems were introduced as interventions (see original production images in Figure 1):

• Day 1: Ame Henderson and Public Recording’s “futuring memory” system from the dance work relay (2010);

• Day 2: Paul Bettis’s theatrical rule play The Freud Project (1996);

• Day 3: Cornelius Cardew’s Paragraph 7 from his music composition The Great Learning (1960s);

• Day 4: Christopher House’s adaptation of Deborah Hay’s solo dance score I’ll Crane for You (2015).

Figure 1. l-r: Relay 2010, photo by Ömer Yükseker; I’ll Crane for You 2015, photo by Alejandro Santiago; The Freud Project 1996, photo by Paul Bettis.

For each system, we typically discussed the contextual practices of improvisation, notation, and artistic inquiry and identified specific system tasks, rules, and sources based on analysis of provided readings. Students then watched archival recordings of the systems while extracting observations for later exchange about the performers’ perceptual orientation and attention, rules in action, and arising patterns. The second half of the day was dedicated to practical, physical workshops and performance of a system, with participants extracting observations for discussion of how the system affects the performers and generates performance. Students were tasked with posting a minimum of four daily observations drawn from their readings, group discussions, the contributions of other participants, or their experience of self and others in performance. They were instructed that the postings should engage with the overall inquiry of how the systems work. To develop this line of inquiry, a set of key concepts were introduced on the first day and returned to over the week. These concepts were “presence, reconstructive memory, perceptual attention, and awareness”. On the fifth day students created their own systems and, in the weeks that followed, they submitted more developed written reflections on how their systems worked along with performance recipes. The latter is a both graphic and written presentation of the components of a system (source materials, tasks, rules, etc.) that generate performance along with instructions for how to engage them through rehearsal and performance. It is meant as an open source recipe, a method for sharing with other artists the generative architecture of these otherwise elusive creations across time and space.

Following a protocol approved by the Conjoint Faculties Research and Ethics Board, 13 of these students participated as research subjects and a comparable group were recruited as control subjects. Both groups had equal gender representation and balanced distribution of participants from the disciplines of dance, theatre, and music. The first pilot was designed to test the effect of the performance generating systems on working memory capacity and selected executive functions through cognitive measures. We used Engle’s Automated Operation Span (AOS) test to measure working memory capacity in the context of a dual task (Unsworth, Heitz, Schrock, & Engle, 2005). In this test, participants were first asked to retain numbers, then they were presented with math problems to solve, and finally they were asked to recall their retained numbers in accurate order. This procedure was repeated with increasingly longer numbers until the participant reached his or her capacity limit and no longer could recall numbers while solving problems correctly. Delis–Kaplan Executive Function System (DKEFS) colour-word interference and design fluency tests were administered to measure executive inhibition and cognitive flexibility. The DKEFS design fluency also measured generation of designs, thought to index creativity (Delis, Kaplan, & Kramer, 2001). In the colour-word interference tests participants were presented with a test pad showing a variety of colour blocks within which the word of a different colour is written. In various test conditions the participant was asked to complete tasks like naming the colour, or naming the written text, each of which involves inhibiting their processing of irrelevant colour information. The design fluency test sheet displays squares with either filled or empty dots. The participants were asked to create novel designs tracing lines between dots under a series of different conditions, such as “use empty dots only” or “switch between filled and empty dots”. These tests were completed before and after the intervention, alongside questionnaires collecting demographic information, motivation, and expectations of the types of changes this class would bring. More specifically, on an information questionnaire participants were asked to indicate the following variables: age; gender; cultural background(s); whether or not they had a cognitive health condition; the amounts of course credits completed in respectively dance, drama, and music; amount of courses taken involving improvisation; years of training in these disciplines; and years of improvisation practice. On pre- and post-motivation questionnaires administered before and after the intervention, participants were asked to indicate on a scale from 1–7 (1 being the lowest, 4 average or neither/nor, and 6 being the highest or significant) how motivated they were to take the course prior to the first day and and how motivated they where while taking the course; how much new insight and skill they expected to take away and did take away from the course; how challenging they expected to find or did find the course; whether any challenges experienced might become easier or did become easier over time; and whether their “way of experiencing (incl. perceiving, thinking, doing)” would change or did change over the course? The control group was tested at the same time interval, but without participating in any of the performance interventions and, of course, without filling in the pre- and post-questionnaires. These tests produced numeric performance scores that, given the small sample size, were analysed statistically using non-parametric texts (Wilcoxon rank-sum tests).

The second pilot looked into learning capacity and embodiment. The 13 participants’ daily observations, their final written reflections, and video recordings from class were coded by two research assistants in the qualitative research software NVivo. The assistants used two kinds of codes: (1) pre-established codes that were derived from our hypothetical inferences and examined the inferences deductively and (2) new grounded codes that were inductively generated from the data, allowing for the discovery of novel factors and relationships. Among the pre-established codes were learning categories, which research assistants used to identify participant observations that were engaging with respectively semantic, practical, conceptual, or experiential knowledge. To relate developments in participant engagement across these categories to both the practical workshops and the analytical discussions of each course day, we also chose codes of “how the system works”, “system properties” (tasks, rules, sources), and a series of abstract key concepts (presence, perceptual attention, memory, and awareness) of relevance to the systems. More specifically, the video recordings and text that form our qualitative data were uploaded as sources to Nvivo and the pre-determined codes were established in the software. Initially working together, and later working individually but reconvening regularly to test and align their coding practices and interpretation of emerging codes, two of the research assistants went through these sources one by one marking utterances that matched one or several of the codes, linking them to the specific codes in the software. When new concerns, issues, or concepts emerged as potential factors to the research assistants they discussed them with the full group, arrived at conceptual clarity, and then implemented them as new codes working backwards through the data. Emerging subcategories were considered and implemented using the same procedure. The research assistants finished processing the daily observations and generated the following grounded codes: challenges, comparison to conventions, individual focus, and group orientation. When analysing the coded data, Hansen compared changes in the participants’ engagement with the different types of knowledge to their comprehension of the key notions and their focus on the remaining subjects coded. The goal of this qualitative analysis was to uncover the relationships and patterns that most significantly affected the participants’ learning process and achievements and then compare them to our hypothesis. Central to this analysis was thus the question of whether or not a relationship could be detected between attention to experiential, sensory-motor learning and advances in participants’ ability to comprehend abstract concepts, as well as the identification of factors that affected such a relationship.

Preliminary results: quantitative measures

Overall, the result of the cognitive capacity tests are neutral: the group that took the course scored neither better nor worse on post-tests relative to pre-tests than the control group. Across all AOS and DKEFS tests we find no differences in either composites or the individual items (Wilcoxon p > .30 in all areas). Further, we did not identify any significant changes (pre-/post-test differences statistically different from zero) in composites or individual items within the treatment and control groups. While this result neither supports nor provide new insight in regards to out hypothesis, it is consistent with the argument since put forth by Redick (2015) that training interventions are unlikely to produce measurable changes in individual higher-order cognitive functioning.

However, in two specific areas we do observe statistically identifiable differences between pre- and post-test performance by participants across the treatment and control groups that can motivate further studies with longer interventions. Specifically, participants in the treatment group showed moderate improvements in design fluency relative to those in the control group (Wilcoxon p = .015). This difference was detected within the data from a test condition measuring cognitive flexibility (set-shifting ability) and non-verbal creativity (ability to develop novel designs). Thus we can reject the null hypothesis that changes from pre- to post-test in design fluency across the treatment and control group were drawn from the same distribution. This is suggestive of participation in the performance generating system interventions reducing the errors in design fluency tests by 2 points relative to the control group. Although it also is broadly consistent with our initial hypothesis that the simultaneity of various processing functions required by a performance generating system may affect cognitive processing in some dimension, further steps are needed to identify such changes. We furthermore identified a marginal improvement of processing speed and a reduction in errors in the colour-word testing in the treatment group relative to the control group. However, while this improvement is not statistically significant (Wilcoxon p = .10), it is illustrative of the manner in which engagement in a performance generating system aims to sustain focus and continuous cognitive processing while inhibiting certain responses. This result provides a motivation for further research considering how a longer (i.e. beyond five days) intervention in a performance generating system may affect standard measures of working memory.

In conclusion, it appears our implemented training in performance generating systems did not produce fundamental changes in objective measures of cognition. This could either be (1) because there is no effect (leading to the rejection of our hypothesis) or (2) because our pilot with its 13 participants and five full days of intervention was underpowered to produce an objectively measurable effect. This interpretation is supported both by Redick’s findings and the fact that we were unable to analyse our data using matrixes of the surveyed variables (e.g. a matrix of years of improvisation practice, discipline, and motivation) to reveal more defined differences, should they exist, because such subdivision resulted in sample groups that were too small (i.e. three or less). We submit that our objective measures suggest a small effect that is insufficient to adequately identify the actual changes provided by performance generating systems. However, a larger sample is unlikely to address this problem alone. In a current discussion of working memory tests, Foroughi, Monfort, Paczynski, McKnight, and Greenwood (2016) and others, argue that cognitive capacity is heritable and fairly stable. In a study that administered a popular working memory training intervention after an either suggestive or non-suggestive training invitation, they were able to produce a placebo effect equating to a 5–10 point improvement on IQ tests. This leads them to question the reading of the typically small effects on intelligence measured in experiments involving working memory training interventions.

A third interpretation of the cognitive capacity pilot results, which we cautiously support, is thus that there may have been a cognitive effect which could be expressed with greater statistical significance in a larger study if, and only if, new test measures that are better equipped to measure the subtle changes related to our intervention are developped.

This interpretation is supported by a puzzling finding that is drawn from the motivation questionnaires. On a scale from 1–7 (1 being “not at all”, 5 “somehwhat” and 7 “yes, significantly”), the average indication of whether the participants felt their way of experiencing had changed was 6 on their post questionnaires. This score was merely a 7.57% increase from the average on the pre questionnaries. Like the small changes measured in the cognitive tests, the participants could be reporting a placebo effect wherein participants qualitatively perceived a change because they felt they “should” benefit from their classes and training. However open comments that the participants added on their surveys in extension of this question indicated that the expectations marked on the pre-motivation questionnaire were in regards to acquiring new skills or learning about new practices (i.e. expectations that match other types of BFA and MFA courses) while many of the post-test scores referred to experiences that were more closely reflecting perceptual and cognitive processes. For example, compare the following pre- and post-questionnaire comments:

I think that having the use of 3 disciplines will change the way I interact with creative processes. As well, performance generating systems is a new idea to me so I will hopefully take something away form that which will change how I perceive things. (Participant #9, pre-motivation questionnaire)

There was a moment while working with Christopher where I felt like I had almost unlocked the secret to utilizing the frantic way my brain operates. I tend to experience hyper focus on tasks and subjects but I get distracted easily and the focus changes to something else. During one excercise I was shifting focus so rapidly that it was almost to the point of focusing on all aspects of the exercise at once. (participant #9, post-motivation questionnaire)

Interpreting the quantitative survey in relation to the qualitative differences exemplified above, our results do indicate an important discrepancy between the objective measures and participant experiences that cannot be explained as a placebo effect alone. This discrepancy suggests that our objective test measures (AOS, DKEFS) were unable to identify the reported effect and that different test measures need to be designed. Thus our results suggest several directions for further research into the relationships between performance, cognition, and learning.

Preliminary results: qualitative measures

The qualitative results from the learning capacity pilot yielded clearer, interpretive results. However, as relationships between multiple factors are drawn into matrices for analysis, the overlap between observations coded becomes smaller. Such matrixes thus provide weaker qualification of our interpretations than relationships that are drawn between fewer codes. The clearest result is demonstrated in the observations regarding the four knowledge and learning categories over the five days of the course. Participants started with a predominant focus on semantic and practical knowledge and ended with a significantly increased focus on experiential knowledge. This is in spite of the fact that the course facilitation was designed to equally teach all four categories every day. As indicated in the chart below, engagement with semantic knowledge and attempts to arrive at definitions decreased by 100%, engagement with practical knowledge of how to execute tasks decreased by 50%, and engagement with experiential knowledge increased by 50% (Figure 2).

Figure 2. Matrix coding query – Results preview.

Only the conceptual knowledge category remained consistent throughout the week, indicating a relationship between experiential and conceptual advancement. We did find a larger engagement of key concepts that are associated with experiential knowledge (awareness and perceptual attention) than the ones that initially were understood by participants as being theoretical and thus more abstract (memory and presence).

Furthermore, the participants achieving the most advanced understanding of these key concepts and how the systems work were all among the individuals who contributed the most experiential observations. However, the inverse is not true: one participant who posted a very high number of experiential observations did not advance much analytically. This individual stood out in two ways: the advancing participants sustained engagement with conceptual knowledge throughout the week and they transitioned from a focus on the individual to a group focus within the first 2–3 days; the individual did neither. The only participant who neither increased her experiential nor conceptual observations but advanced significantly in the practical category was unable to participate physically and had to mostly observe after the second day. This suggests that increased engagement with experiential knowledge does depend on fully embodied practice and it provides further support for the conclusion that experiential and conceptual advancements are closely connected.

The participants posted more observations on “challenges” than on any other code subject. Initially these observations were practical and conceptual in focus, included negative comparisons to prior knowledge, and intersected with discussions of system tasks. Over the week, observations about challenges became practical and experiential in focus and shifted to primarily intersect with discussions of system rules and of how the systems work. In other words, the participants’ early attempts to understand “what” a performance generating task is came up against a boundary of prior knowledge in practical and conceptual terms. The task did not match the foundation ideas of practice that participants were familiar with, and they were troubled by the need to step outside of their field of competence. Further into the course, they reflected on challenges experienced while attempting to remain aware of rules during task execution. Instead of comparing the experience to prior knowledge, they turned to discussions of habits, momentary loss of concentration, exhaustion, and difficulty distinguishing between different ways of attending, as they related to their growing understanding of how the system works and affects them. This was also the point at which participants began to associate their experience from within the sensory-motor praxis with the concepts that initially were considered the most abstract and theoretical.

I found today’s system extremely challenging. It wasn’t so much that the required physicality was draining; rather, the level of concentration was. I was surprised at how “in my head” I had to be in order to keep track of all the points of attention I had to balance. …  There were many points today throughout the exercise where my concentration slipped. … It would always take me a few seconds to register what was going on and then figure out how to solve the problem. (Participant # 6, observations posted on teaching portal, day 4)

This example was coded at “experiential” engagement with knowledge and the concept of “presence” and speaks to the challenge of concentration and attention that sustained performance presence requires while working physically in Deborah Hay and Christopher House’s praxis. This praxis requires that the dancer responds to information from the surroundings and the body without planning or building sequences and without holding on to the information. In practical terms, the participants were trained and asked to continuously “turn their head” (i.e. shift point of visual and proprioceptive attention) in order to respond to new information while working on their interpretation of a written dance score.

For the section when we were looking at the audience, we needed to have contact with the audience but not a connection with them. You need to be present in the moment to be able to differentiate between those two ideas. (Participant # 10, observations posted on teaching portal, day 4)

The example above is an observation on “presence” that was coded at both “experiential” and “conceptual”, reflecting emerging overlap between these two forms of engagement. The following example was coded as a “conceptual” engagement with the concept of “memory”. However at this stage of the course, knowledge that derives from the sensory-motor experience of working through challenges in practice informs the participant’s conceptual reflection.

The idea of surrendering myself to the system by unlearning what I knew and was comfortable with. Not to hold on to a moment, experience or idea, to get away from old habits, to disengage from what is not needed in the moment …  These are all parameters that combined bring me to a place of intense vulnerability. It is extremely scary to go there. It is unknown and therefore, frightening. Without a probable ending nor possibility of achieving a goal. (Participant # 8, observations posted on teaching portal, day 4)

As mentioned, the literature on embodied learning describes the employment of movement in the service of semantic and conceptual understanding. If the movement-based and conceptual learning categories run parallel to each other in such a hierarchical relationship there should be no increase or decrease in engagement with either over time. In our pilot, participants’ engagement with learning categories did change over time while they were actively “unlearning” prior knowledge as individuals and reflecting upon the challenges experienced around learning new cognitive practices as a group. Discrepancies between prior learning and the task at hand demanded conscious attention, so did moments when a rule became difficult to implement because of habits. It was partly through the participants’ work on these barriers that they heightened their cognitive ability to remain attentive, inhibit, and multi-task while reporting on how the activity affected them as performers; and it is within this enhanced self-reflexivity that the key concepts gained operational meaning. In terms of embodied learning, the performance generating systems helped participants push past a hierarchical relationship between conceptual and experiential learning and instead establish a reciprocal relationship between the embodied understanding of concepts and conceptual engagement with movement. As previously discussed, empirical studies of embodiment in both cognitive linguistics and kinesiology are limited to measuring the embodiment effect of highly motor action-salient concepts. It is likely that the reciprocal relationship described above was instrumental for our participants’ ability to achieve understandings of abstract concepts that were advanced through and embedded in self-reflexive experiences of complex sensory-motor praxis.

We are tempted to describe the learning principles discovered in this pilot as a form of fully embodied critical reflection. By taking on the challenge of performing a different cognitive practice, the participants became empowered to reflect upon and push beyond otherwise implicit past learning that informs their embodied conceptualisation of and responses to artistic tasks. In doing so they moved beyond physical engagement of action-salient concepts and began to develop new and more conceptually abstract phenomenological knowledge.

Recommendations: measures and transferability

What are the implications of the quantitative and qualitative findings of engagement with performance generating systems for measures of cognition? In reflection, performance generating praxis may be too complex to produce an objectively measurable effect on individual cognitive skills (cognitive flexibility, working memory capacity, etc.). Thus, while we do observe marginal changes in some cognitive performance skills (as measured by the DKEFS tests), a larger and more relevant effect might be on the integration of such skills, including modalities from both bottom-up and top-down cognition, kinaesthetic perception and conscious choice-making (i.e. embodied cognition). In other words, participants may have perceived changes at a more global level, rendering the cognitive tasks we introduced as test measures less relevant to the changes they experienced. Alternatively, it could be that the changes experienced after the performance generating systems class do not improve specific cognitive abilities, even though they are relevant to gestalt processing. Finally, it may be that the types of cognitive abilities that are affected in a performance generating system are not adequately identifiable given the measures we used, or even the set of existing measures in the field. Our pilots are illustrative of the type of study that crosses boundaries between embodied cognition, existing cognitive abilities, learning, and performance. The mere nature of performance generating systems may require alternate measures for the identification of the relationships between cognition and performance.

Generalisation from one form of learning to another form of learning has always been difficult to produce in cognitive science studies that apply far less complex learning interventions than Performance Generating Systems (Green & Bavelier, 2008; Melby-Lervåg & Hulme, 2013). Because of introducing more complex interventions, this challenge has become larger and is, effectively, forcing us to look for unconventional solutions. We are not alone in arriving at this conclusion in response to research processes that cross the performing arts and cognitive science (e.g. Hansen, 2017; Hansen & Barton, 2009, pp. 130–135; Jola, 2010, p. 211; Stevens, 2005, pp. 167–168), but while the solutions proposed and advanced by the field have produced methods for testing audience responses, understanding cognitive processes involved in expert performance, and setting up feedback loops between experimental research and creative practice, they do not, to our knowledge, attempt to measure transferable cognitive effects of performance practice and training. New solutions are needed to advance the embodied, cognitive psychology of learning into complex, intersubjective, and creative contexts.

How do we move forward? Our cognitive capacity pilot was measuring effects on transfer of performance generating skills to specific higher-level cognitive tasks. In the cognitive science literature, training studies often refer to three types of transfer: target, near, and far. Target transfer refers to improvement in the specific skills or tasks practiced. Near transfer refers to improvement in skills similar (i.e. in the same domain) to the ones trained (Morrison & Chein, 2011). For example, if we trained working memory, we would expect improvement on other working memory tasks that were not trained. Far transfer refers to improvement in skills that are in a different domain than those trained. For example, if we trained working memory, we would expect improvement in cognitive flexibility. In cognitive science, literature of most significance is typically demonstrating far transfer – the generalisability of training specific skills. In our qualitative pilot, we saw improvement in the specific skills practiced by the class from the first session to the last session, demonstrating targeted transfer. We only looked at the effect on academic performance within the intervention itself. Our quantitative cognitive pilot focused on far transfer. It measured whether the intervention had an effect on individual cognitive modalities when performing very different dual tasks. Perhaps in a new area that was a too ambitious choice. A focus on demonstrating near transfer would be beneficial, as far transfer without documented near transfer is difficult to interpret. It would also help us bypass the previously mentioned issues regarding far transfer readings of the effect of working memory training on higher cognitive functions that Foroughi et al. (2016) bring up. To arrive at more suitable, interdisciplinary ways of studying the cognitive and learning effects of performance generation, we propose the measurement and development of near transfer tests of integrated cognitive skills that match the relationships between learning categories and the reciprocal processes discovered in the learning capacity pilot. Such a test could include observation of the participants’ skill integration when engaged in a complex, embodied multi-task, like a simulation of left side driving or an unfamiliar performance generating system. Further, methodological suggestions for such studies include larger sample sizes in order to allow for the investigation of factors that affect training outcomes (e.g. age, gender, training), and a better understanding of the relationship between subjective cognitive changes experienced and more objective cognitive tests.

Moreover, it is worth exploring whether engagement with a performance generating system has longer-term effects on both cognitive skills and embodied critical reflection. Our intervention focused on a one-week experience using differences in pre- and post-tests as measures of cognitive change and coding of participant observations as measures of embodied learning. Indeed, participation and use of performance generating systems may fundamentally change an individual’s artistic practice. As this practice changes, it is likely that the cognitive skills and learning modalities employed will change and that the individual’s embodied critical agency and practice will evolve. As such, this suggests approaching the relationship between cognitive skills, learning categories, and the use of performance generating systems as they form a longer arc over a performance practice.

Disclosure statement

No potential conflict of interest was reported by the author(s).