In a dance class, everyone follows the instructor. The opposite situation would be if everyone in the class performed without a designated leader — an activity known as joint improvisation. In a paper published in Proceedings of the National Academy of Sciences, Noy and colleagues investigate which of two such situations is the more effective (L. Noy et al. Proc. Natl Acad. Sci. USA http://doi.org/hbz; 2011).

Day-to-day examples of joint improvisation include two toddlers playing together. A rather structured example is improvisation in artistic performances (pictured). By the improvisers' own admission, there are 'moments of togetherness' when the level of performance is high but no one knows who is leading. But how does this work? The lack of an experimental paradigm means that this question has not been studied extensively — at least not for open-ended actions.

Credit: J. RENSTEN/GETTY

Noy and colleagues designed an experiment based on the mirror game, a widely used theatrical practice. Specifically, they asked two players each to move a handle along one of two parallel tracks in one dimension. The instruction was: 'Imitate each other, create synchronized and interesting motion, and enjoy playing together.' The nine one-minute rounds in each game were of two types — leader–follower rounds and joint-improvisation rounds. The authors measured the players' movements with high resolution in time (20 milliseconds) and space (1 millimetre).

They investigated the movements of expert players — artists with more than 10 years of experience in joint improvisation. In the leader–follower rounds, the follower showed jittery motion, which oscillated around the leader's confident movement. By contrast, with no designated leader the players performed better, reaching lower errors in velocity of movement and stopping times. In fact, the players jointly showed confident motion 12% of time, compared with 2% in the leader–follower situation.

So, is having a leader really counterproductive? It depends. With novice players, Noy et al. obtained opposite results: these players performed much better with a designated leader. As moments of togetherness are rare in day-to-day life, having a leader is perhaps beneficial for most of us, at least while we learn a new skill.