To Improvise with a Machine: Composing for Interactive Dance

By Mimi Yin

Illustration by Itay Niv
Image Courtesy of the Author

Issue 2

Apophenia is the tendency to see relationships where there are none. Klaus Conrad first described it in a 1958 work on schizophrenia as “unmotivated seeing of connections accompanied by a specific feeling of abnormal meaningfulness.”1

Really, all of us are apopheniacs. Unrelated events that take place in the same place at the same time become related by virtue of proximity. Coincidence becomes collusion. Correlation becomes causality.

Apophenia is what allowed the choreographer Merce Cunningham and his collaborators (composer John Cage, visual artist Robert Rauschenberg, and others) to create independently of one another, yet still allow them to come together to occupy ‘the same time and space’ as a single work. Though the art was created separately, we as an audience perceive them as connected.

Which raises the question, why hook up performers to sensors to control sound and visuals if the relationships between the performers and other media already exist in our minds?

Interactive work must necessarily evolve beyond the novelty of composers, performers, and the audience bearing witness to cause and effect enabled by newfangled sensors. Interactivity offers up new surface area for exploring the psychology of human choice. As creators of interactive work, we should seize it as a means of upending how choices are made, who gets to make them, and for what purpose.

Is Interactive Composition a Contradiction in Terms?

To compose something is to craft it with care. A half-a-step to the left, a half-a-beat late makes all the difference to the detail-oriented composer. On the extreme end, artists such as Robert Wilson direct performers on stage with the same level of precision that architects exercise over their drawings (see video below).1

Interactivity on the other hand, necessarily means a transfer of decision-making power from composer to performers, or, in other words, from designer to users. Performers make the decisions in a performance that shape the overarching composition of the piece—if the composer of the interaction allows them this freedom.

Depending on how the composer creates the interactive system, however, she risks missing the opportunity to transfer power. This happens when user input only affects the output in a limited or superficial way so that no matter what the performer chooses to do, the character of the performance stays the same.

For example, if the height of the performer’s head affects light levels, some questions a composer might ask include: To what extent does the head affect the lighting? What’s done in complete darkness so that it is felt and heard but not seen? What’s done in blinding light? What’s done in twilight so it is only dimly perceived? By interrogating the system in this way, the composer can ensure that it allows enough room for the performer to play around, respond, and ultimately shape the composition.

The user also misses the opportunity if she never ventures beyond simply ‘seeing what happens’ when the system is provoked. Instead, ‘seeing what happens’ should provide input to the user that prompts a response, rather than functioning as the ultimate output of the system. It is only the first step toward understanding how the system behaves under a wide range of circumstances.

Performers can only achieve full understanding through a methodical exploration that stretches the system well beyond its breaking point. Only when the users grok the system can they begin to ‘play’ with it through conscientious decision-making. It’s not the sort of casual play that engages through immediate gratification. It’s the highly-skilled kind that you can do with an instrument, the kind that is capable of producing carefully crafted expressive output.

Workshopping Interactivity

To begin exploring the choices and relationships afforded by interactive technologies, I recently led a two-day workshop series at Danspace in St. Mark’s Church. I worked to compose sound through choreographic decision-making mediated by technology with a group of 28 dancers, sound artists, and those with no stake in what we were doing other than curiosity.

On one level we were playing with an interactive system that took in movement as input (cause) and generated sound as output (effect).

On another level, though, we were exploring a platform that enables individuals to come together as a group to compose something—in other words, make choices—that is simultaneously a dance and a piece of music.

Overall, we were attempting to turn dancers into composers and composers into both dancers and choreographers, all while trying to master the art of improvisation and the temperamental instrument, which was the interactive system I had created using the Kinect cameras.

The goal of our workshop was to explore a model of interactivity that was psychological rather than technological, playful rather than based on one-sided reactions.

Workshop Takeaways

Is it possible to compose through interaction? Yes.

Did we achieve the level of highly-skilled play required to get there? Sort of.

Though we didn’t even have enough time together for the novelty of playing with the instrument to entirely wear off, we still got remarkably far down the path towards collective composition. Across the five setups we explored, there were many, many moments of ‘compositional decision-making,’ even if a complete composition never emerged.

To reach such a goal, a dedicated group of practitioners would need to work together daily to find a collective voice and learn how to use it to communicate with the interactive system.

Above all, though, what the participants did learn in the process had more to do with us than our instrument. Like the stages of romantic love, interactive systems arouse in their human users a series of emotional states that together form a narrative that reveals much more about who we are than how the interactive system works.

The following are the Stages of Interactivity as I observed them in the workshop participants:

  1. Curiosity and the need to figure out how it works
  2. Followed by a need to be able to control it, however it works
  3. Followed by an urge to play with it
           a. Trick it
           b. Break it
           c. Make it do nothing
           d. Make it do something it’s not supposed to do
           e. Make it do something dramatic
           f. Make it do something ridiculous
  4. Followed by frustration at the inability to gain perfect control over the system
  5. Followed by acceptance of imperfect control, which finally opened the way for artful conversation with the interactive system (watch: Djassi Finale)

Another way to describe the ‘arc’ of participation would be in terms of focus:

  1. Everyone began with a steadfast, almost obsessive fixation on the cameras. The body always faced the camera, the eyes trained on the laptop screen that shows what the camera sees and the posture crouched as if to make eye contact with the apparatus that sat low on the floor.
  2. Gradually, focus shifted inward as participants grappled with the internal workings of the system, the logic behind the rules of interaction.
  3. Eventually, performers directed attention outward as they begin to hear and listen to the sounds, to the music being generated by the system.
  4. This was followed by a growing appreciation for the relationship between their own movements and the movements of others around them and how those relationships translated into sonic relationships (watch: Speed Duet).
  5. Last, but not least, there came an awareness of the gestalt emerging from individual choices and interactions.


I began my research asking the rather contrarian question: What’s the point of interactivity in performance if the performer’s choices do not dramatically shape the composition of the work?

The follow-up question to such a question is then, of course: What does it look like for performers to make such choices?

There is no silver bullet answer to this kind of question. Instead the workshop yielded a collection of techniques born of trial and error, in which we treated errors as opportunities rather than something to be avoided or stamped out.

Nevertheless, there are a set overarching principles that will inform the next phase of research:

  1. Work the system until everyone is over the novelty factor of being able to control something through thin air.
  2. Cultivate an expertise in the media you’re interacting with. In our case it was sound. (This sounds obvious but can easily get lost in the heat of getting technology to simply work.)
  3. Master the system warts and all. Treat the warts as opportunities for something unexpected to emerge.
  4. Play with it until the technology becomes transparent. You know this has happened when the focus is no longer on the relationship between the performers and the interactive media (sound) but instead all eyes are on the relationship between the performers and all ears are tuned into the relationship between the sounds.




Thanks to Tiriree Kananuruk for help with documentation and curation of sound. Special thanks also to Alexx Shilling and Kat Sullivan for helping me develop the workshop and Tiri’s documentation team: Mikey Asanin, Vijchika Udomsrianan, Woraya Boonyapanachoti, Wipawe Sirikolkarn, Ratawan Tanadumrongsak. Residency activity is supported through an ongoing partnership between Tisch Initiative for Creative Research and Danspace Project’s Community ACCESS program.

Mimi Yin (ITP 2012) is a designer, artist, and educator. Her work explores programmatic approaches to composition and improvisation across a wide range of mediums, from poetry to music, and from choreography to shaping conversation.

  • 1Sandra L. Hubscher, “Apophenia: Definition and Analysis,” Digital Bits Skeptic, November 4, 2007,
  • 2From "Einstein on the Beach," directed by Robert Wilson and composed by Philip Glass.