Dynamic Regimes that are likely to support Probabilistic Inference
in Distributed Networks of Spiking Neurons
A new model for computation in networks of spiking neurons has recently emerged: spike-based Expectation Maximization (see http://www.igi.tugraz.at/maass/psfiles/191.pdf ). I will address in this talk the question which dynamical regimes would support distributed computations of this type in multiple brain areas. In addition I will discuss predictions of this model regarding the statistical structure of network activity.