Information processing in spiking networks: Converging assemblies

Tuesday, April 9, 2024
Hour: 12:30 - 13:30
Gerhard M.J. Schmidt Lecture Hall
Prof. Eran Stark
Sagol Department of Neurobiology Haifa University

How information is processed within the brain is a key question in systems neuroscience. We address the issue in spiking neuronal networks of freely moving mice. I will describe our recent findings and conclusions pertaining to three specific information processing steps: transmission, representation, and storage. First, using feedforward optogenetic injection of white noise input to a small group of adjacent neocortical excitatory cells, we find that spike transmission to a postsynaptic cell exhibits error correction, improved precision, and temporal coding. The results are consistent with a nonlinear coincidence detection model in the postsynaptic neuron. Second, by triggering input on animal kinematics, we create an artificial place field in an otherwise-silent pyramidal cell. In hippocampal region CA1 but not in the neocortex, artificial fields exhibit synthetic phase precession that persists for a full cycle. The local conversion of an induced rate code into an emerging phase code is compatible with a dual-oscillator interference model. Third, by triggering input on spontaneous spiking, we impose self-terminating spike patterns in a group of presynaptic excitatory neurons and a postsynaptic cell. The precise timing of all pre- and postsynaptic spikes has a more substantial impact on long-lasting effective connectivity than that of individual cell pairs, revealing an unexpected plasticity mechanism. We conclude that intrinsic properties of single neurons support millisecond-timescale operations, and that cortical networks are organized in functional modules which we refer to as “converging assemblies”.