Parts | Days | Selection | Search | Downloads | Help
DY: Fachverband Dynamik und Statistische Physik
DY 28: Nonlinear stochastic systems
DY 28.2: Talk
Thursday, March 29, 2007, 14:15–14:30, H2
Self-organized critical control in human balance behaviour — •Markus Riegel, Christian Eurich, and Klaus Pawelzik — Institut für Theoretische Physik, Universität Bremen, Otto-Hahn Allee 1, D-28334 Bremen
Humans are known to exhibit power law distributed fluctuations in their behaviour e.g. when trying to stand still. Here we show, that power law distributed fluctuations generically arise in control of unstable dynamical systems driven with gaussian noise when an optimal controller can use observations only from the immediate past to estimate the parameters of the controlled subsystem. With increasing memory the exponent of the distributions grows and the auto-correlations of fluctuation amplitude decays faster. We tested the predictions of this self-organized critical control (SOCC) in a simple task where humans where required to stabilize an unstable target on a computer screen with the computer mouse. We found that also here the resulting dynamics are close to Levy-flights and exhibit power law distributed fluctuations. In stationary tasks the exponent of the distributions and the decay rates of auto-correlations increased with duration indicating that in this case the memory span utilized by humans expands. No such changes were seen in experiments where the systems parameters where constantly changed. Taken together our results indicate that the nervous system indeed employs self-organzed critical control and furthermore adapts its memory span depending on the stationarity of the task.