Creativity in Orchestras

Client

Graeme Bailey, Computer Science Department
<bailey@cs.cornell.edu>

Project

A particular context has arisen in the general area of understanding human interactions. In a study of how 'creativity' is mediated within orchestras (no, it's not a top-down approach from conductor to section leaders to the rest, in spite of naive appearances!!) the question has arisen of how to investigate this. In a real sense, these 'bursts of creativity' appear to happen spontaneously and propagate through the group in intriguing fashions.

Some work has been done via surveys and interviews, but all this depends on the recall of the participants and is often masked by quasi-political factors relating to perceived reputations. There are differences between 'community orchestras' and the various varieties of professional orchestras, and between groups which have a fairly stable membership and those for whom the members are far more likely to vary over time. However, there appear to be some quite significant commonalities. The issue is how to investigate these interactions in a more precise manner, in a way which might shed light on the propagation of 'events' within the group.

One approach could be to place a pair of inexpensive webcams on every music stand with one pointing to the player(s) and the other trying to capture their normal field of view (placing headcams would be too awareness-raising to the players and hence likely to maintain an influence on their behavior). A difficulty with that approach (apart from cost) is the relatively strong sense of 'invasiveness' which is likely to invoke sensitivities of privacy-invasion and hence non-participation. So a preferred approach is to have four regular videocams sited to capture the group (high-mounted to catch the left, the right, the front-facing players, and the conductor). By synchronizing these four video-feeds one can create an environment where study of players' physically-demonstrated interactions can be integrated with their own perceived interactions.

In particular, consider the case of player X. In interview post-rehearsal (or post-concert), X might recall certain 'events' -- those 'events' can then be found and viewed from X's perspective to allow X to confirm and potentially elaborate on the context, and then tagged for review by those other players seemingly caught up in that sequence of propagated 'events'.

Clearly, since interviewing takes time, and memories fade relatively quickly, a significant improvement on this would be to have these synchronized videos viewable via a web interface so that all the players can perform their memory-dumps of the rehearsal/concert soon afterwards, with automated bracketed (e.g., 10 seconds before/after) tagging of the relevant video segments once identified as prompts to the relevant other players. There would need to be some care in respecting privacy (details available from me), but the potential of this approach to reveal how ideas propagate through groups is highly non-trivial.

Although this is being proposed within the context of orchestras (primarily because this can be put into action with a variety of orchestras around the world as soon as it's built and tested), it's very obvious how this approach can extend to other groups of people attempting to work together to achieve various goals.