Kresimir Josic

The paper develops an ideal observer model to infer environmental state and change rate in dynamic environments, using a moment closure approximation and neural network models with plasticity rules.

The paper tests if Levelt's Propositions apply to interocular grouping perceptual multistability and finds they do, with a hierarchical model explaining the results.

This paper introduces a recurrent network model that learns both the order and precise timing of event sequences through long-term plasticity and short-term facilitation.