Noise induced switching between attractors can be an important contribution to complex neural dynamics such as those, which occur in the pupil light reflex (PLR). Fluctuations in pupil size are measured as a function of the force length curve of the iris and its musculature and the feedback in the reflex arc. These fluctuations reflect the interplay between stochastic and deterministic forces. Depending on the balance between the antagonistic muscle pair, the force length curve can have different shapes. Considerations of mathematical models for the PLR lead to the hypothesis that the properties of pupil fluctuations will change as the force length curve and feedback change. Possibilities include noisy perturbations about a single stable fixed point, noisy limit cycle oscillations and noise induced switching between two, or more, co-existent attractors (multi-stability). This hypothesis is tested by measuring pupil size fluctuations as the force length curve of the iris and the feedback in the reflex arc are manipulated experimentally. The statistical properties of the fluctuations in pupil size are studied in the context of stochastic delay (normal `closed loop' illumination) and stochastic ordinary (`open loop' illumination) differential equations. Time series methods are developed to distinguish a noisy attractor from noise induced switching between attractors. These methods are validated through studies of three experimental paradigms: pupil cycling in the clamped PLR; the recurrently clamped neuron and the human postural sway. In each case the parameter ranges from which, noise induced switching is identified can be fed into a mathematical model to confirm the presence of multi-stability. Adaptive control strategies are developed to confirm the presence of multi-stability by using perturbations to trap the dynamics successively into each of the basins of attraction. The identification of noise induced switching between attractors in complex neural dynamics may lead to therapeutic strategies based on the manipulation of multistable dynamical systems.
| Wu, Jianhong; Zivari-Piran, Hossein; Hunter, John D et al. (2011) Projective clustering using neural networks with adaptive delay and signal transmission loss. Neural Comput 23:1568-604 |
| Milton, John; Solodkin, Ana; Hlustik, Petr et al. (2007) The mind of expert motor performance is cool and focused. Neuroimage 35:804-13 |
