Networks of biological neurons (e.g. in cerebral cortex) consist of a complex directed graph. Communication
in these networks occurs through discrete messages (spikes) travelling along the directed edges. There is a
growing experimental evidence from neurophysiological data that the dynamics of spiking neurons could
implement some form of Bayesian inference which could account for the response of the network to external stimuli,
e.g. in shape perception or in visuomotor control. It is therefore a challenging issue to understand
quantitatively which are the potentialities of probabilistic computational schemes and how networks of
spiking neurons can compute the posterior distribution of one or more random variables of interest given
input data.
Research in the theoretical neuroscience group is currently focused on four major directions:
- Statistics of firing of neurons in persistent activity states
- Learning and distribution of synaptic weights
- Number of fixed-point attractors in recurent neural network models
- Learning in networks with binary synapses
