Mathematical Biology Seminar|
University of Pittsburgh, School of Medicine
3:05PM, Wednesday, April 13, 2011
Nonlinear dynamics of single neuron coding properties
Neurons transmit information using action potential, or spikes. When
given the same input, different neurons (or even the same neuron under
different conditions) can produce very different output spike trains.
This has important implications for neural information processing.
Based on simulations in reduced conductance-based models and on
experimental data, I will demonstrate that neurons have a limited
repertoire of basic nonlinear dynamical mechanisms by which they can
generate spikes. These nonlinear mechanisms reflect time- and
voltage-dependent competition between membrane currents. Although
different types of neurons may express very different membrane
currents, even subtle shifts in the balance between membrane currents,
such as might occur during normal modulation or under pathological
conditions, can lead to qualitative changes in threshold mechanism.
From an encoding perspective, the threshold mechanism bears directly
on what sort of information a neuron’s output spike train can convey
about the neuron’s input. I will argue that neurons can preferentially
encode the integral or derivative of their input depending on the
magnitude and direction of the neuron’s subthreshold current. These
encoding properties in turn affect what sort of coding schemes (e.g.
rate coding vs. temporal coding) can be used by single neurons and by
larger neuronal ensembles.