Biological Cybernetics

, Volume 86, Issue 2, pp 111–116

A stochastic model of multistable visual perception

Authors

  • I. Merk
    • Gerling Global Reinsurance Company, Cologne, Germany
  • J. Schnakenberg
    • Institut für Theoretische Physik D, Sommerfeldstraße 26, RWTH Aachen, 52074 Aachen, Germany

DOI: 10.1007/s004220100274

Cite this article as:
Merk, I. & Schnakenberg, J. Biol Cybern (2002) 86: 111. doi:10.1007/s004220100274

Abstract.

 Multistability in vision is an intriguing phenomenon that is currently not well understood. In this paper, we present a new, stochastic model for multistable visual perception. It is based on results of time series analysis of experimental data, yielding evidence for it being a linear, stochastic process. This is the outcome of testing for unstable periodic orbits and comparing the correlation dimension of the data to that of white noise. In the model, all degrees of freedom but one can be determined by general knowledge, thus resulting in a high degree of parsimony. The remaining parameter is used to model the individual characteristics that vary between subjects. Fitting simulations to the experimental data proves the parameter to be in a physiologically highly plausible range.

Download to read the full article text

Copyright information

© Springer-Verlag Berlin Heidelberg 2002