Self-organized microcircuit structure in networks of spiking neurons with plastic synapses.

Meeting Details

For more information about this meeting, contact Carina Curto, Timothy Reluga.

Speaker: Brent Doiron, University of Pittsburgh
(Host: Carina Curto)

Abstract: The wiring of cortex is overwhelmingly complex. In particular, the microcircuit structure of cortical networks has both an over and under representation of certain motifs, compared to unstructured networks. This structure is not hardwired, yet rather emerges from experience-dependent plasticity rules that govern the synaptic connections between cells. A majority of theoretical studies of plastic networks consider the effect of external training on shaping network architecture. We study the complementary situation of how a recurrent network with plastic synapses interacts with spontaneous spiking dynamics to create self-organized network structure. We use a self-consistent theory that combines fast spike-time correlations computed using a linear response framework, with an adiabatic theory for synaptic weight dynamics, to derive a theory for the evolutions of network architecture. With a finite size expansion of network dynamics we obtain a closed set of nonlinear differential equations for the evolution of three-cell motif structure within the network. With this theory in hand we give a principled exploration of how the details of the plasticity rule drives the evolution of microstructure in cortical networks.

---

Room Reservation Information

Room Number: 106 McAllister

Date: 09/09/2014

Time: 1:00pm - 1:50pm