Noisy heteroclinic networks and sequential decision making
Yuri Bakhtin, Georgia Tech

I will talk about sequential decision making models based on diffusion  along heteroclinic networks of dynamical systems, i.e., multiple  saddle-type equilibrium points connected by heteroclinic orbits. The  goal is to give a precise description of the asymptotic behavior in  the limit of vanishing noise. In particular, I will interpret exit  times for stochastic dynamics as decision making times and give a  result on their asymptotic behavior. I will report on extensive data  on decision making in no a priori bias setting obtained in a  psychology experiment (joint with Joshua Correll, University of  Chicago), and compare the data with the theoretical results. I will  also show that the same behavior of exit times appears innonequilibrium models of statistical mechanics.