Elsevier

Neurocomputing

Volumes 38–40, June 2001, Pages 153-157
Neurocomputing

Deciphering the neural code: neuronal discharge variability is preferentially controlled by the temporal distribution of afferent impulses

https://doi.org/10.1016/S0925-2312(01)00556-2Get rights and content

Abstract

Neuronal discharge variability has typically been studied as a function of stimulus rate in the context of a Poisson point process. Here we report that novel discharge behavior is elicited in a model cerebellar Purkinje cell when the timing of afferent impulses follows a pulsed Gaussian stimulation paradigm. We show that under these circumstances effective convergence superseded frequency of activation as the independent variable controlling cell discharge. This provides strong evidence that firing rate alone is an insufficient foundation for the neural code.

Introduction

We report the results of a detailed study that explored the discharge behavior of a realistic one-compartment Purkinje cell (PC) model when the properties of the activation process were varied, including the temporal distribution and frequency of afferent impulses, afferent fiber number (effective convergence), and presence of paired-pulse facilitation (PPF). We show that activation by a sequence of impulses displaying a non-Poissonian temporal pattern leads to novel cell discharge behavior that eliminates a requirement for temporal averaging that is present in rate-based models.

Section snippets

The model

Simulations employed a cell based on a previously reported composite model PC [3] that now incorporates a dynamic potassium equilibrium potential [6], [4], and PPF [1].

Rate-coded afferent excitation was modelled as a Poisson process. Alternatively, temporally coded activation was described by a Gaussian distribution of impulses in time such that the Gaussian envelope was symmetric about an assumed mean stimulus time [2]. Impulse timing within this “pulsed Gaussian” stimulation was controlled by

Results

A near logarithmic relation between the effective convergence and the mean PC discharge frequency was exhibited under both stimulation paradigms. Generally, the PC discharge frequency exhibited only weakly linear scaling with the product of impulse number and mean impulse frequency.

Under Poisson stimulation, facilitation had little influence on the PC current/frequency (f/i) relation. The slope was 135Hz/nA (−PPF condition) or 131Hz/nA (+PPF condition), cf. 141Hz/nA (calculated from Fig. 5B, [5]

Conclusions

The novel features of PC discharge behavior that occurred under pulsed Gaussian stimulation, but not Poissonian, stimulation, were associated with the timing of the afferent impulses rather than simply their rate. Only under pulsed Gaussian stimulation could convergence be tuned to achieve a 1:1 relation between the mean frequency of stimulation and the mean PC discharge frequency, a condition that was associated with a significantly reduced PC discharge CV.

The emergence of periodic discharge

References (6)

  • A.D. Coop et al.

    Simulating the temporal evolution of neuronal discharge

    Neurocomputing

    (2000)
  • P.P. Atluri et al.

    Determinants of the time course of facilitation at the granule cell to Purkinje cell synapse

    J. Neurosci.

    (1996)
  • A.D. Coop, S.J. Redman, The role of inhibitory interneurones in CA1 hippocampus studied by neural network simulation,...
There are more references available in the full text version of this article.

Cited by (5)

  • Temporal code versus rate code for binary Information Sources

    2016, Neurocomputing
    Citation Excerpt :

    Two non-mutually exclusive main theories are of special interest. The first theory is based on “temporal code” [3,7–9], which considers the structure of the spike trains while the second, referred to as “rate code” theory [1,3,10–12], assumes that the neural code is embedded in the spike frequency, defined as the number of spikes emitted per second. The temporal coding mechanism, which builds a temporal relationship between the output firing patterns and the inputs of the nervous system, has received significant attention [13–15].

This work was supported by Neurosciences Research Foundation.

View full text