The Biophysical Basis of Firing Variability in Cortical Neurons - Computational Neuroscience: A Comprehensive Approach - page 158

1. Irregular presynaptic
spike trains

2. Unreliable transmitter
release

3. Stochastic ion
channel gating

4. Fluctuating V(t)

Figure 6.1
A summary of the noise sources contributing to the fluctuating membrane potential
of cortical neurons.

rates are highly voltage-dependent, open and close, generating postsynaptic action
potentials. The dynamics of synaptic integration are thus nonlinear, because of this
voltage-dependence, and are permeated with noise at each stage. In this chapter, I
will focus on the steps in the production of the noisy membrane potential which oc-
cur at the level of the single neuron, i.e., steps 2 through 4 in Figure 6.1. For a review
of ideas about step 1, see the chapter by Salinas and Sejnowski in this volume.
The biophysical mechanisms involved are central to understanding the reliability
of synaptic integration, and hence the strategies used to transmit and transform neural
information. What is encoded by the times at which spikes occur? The precision or
reliability of responses of individual cells is responsible for the degree of synchrony
in a connected population of neurons. How precise and how stable can coherent
firing amongst cells be? Does dynamical behaviour resulting from the interaction of
noise and nonlinearity, such as stochastic or coherence resonance [21], play a role in
cortical information processing? Being able to answer such questions will depend
on an understanding of the biophysics of firing variability.

6.2 Typical input is correlated and irregular
Because of the difficulty of recording from large numbers of neurons simultaneously
across the cortex, much of what we know about the synaptic input to cortical cells
is inferred from the firing of single cells. Firing patterns in the functioning cor-
tex are themselves highly variable. In some situations, firing resembles a Poisson
process, with an exponential distribution of interspike intervals [8, 12, 31]. Burst
firing is evident [39], and there is evidence for weakly periodic firing during certain
states of consciousness or sensory stimulation [54, 58]. Overall firing variability is
characterised by measures such as the coefficient of variation of interspike intervals,
CV(ISI), the ratio between the standard deviation and the mean of interspike inter-
vals. CV(ISI) can be high - higher than the value (1) expected for a Poisson point
process, a completely random point process of uniform rate.