Threshold fatigue and information transfer

J Comput Neurosci. 2007 Dec;23(3):301-11. doi: 10.1007/s10827-007-0033-y. Epub 2007 Apr 14.

Abstract

Neurons in vivo must process sensory information in the presence of significant noise. It is thus plausible to assume that neural systems have developed mechanisms to reduce this noise. Theoretical studies have shown that threshold fatigue (i.e. cumulative increases in the threshold during repetitive firing) could lead to noise reduction at certain frequencies bands and thus improved signal transmission as well as noise increases and decreased signal transmission at other frequencies: a phenomenon called noise shaping. There is, however, no experimental evidence that threshold fatigue actually occurs and, if so, that it will actually lead to noise shaping. We analyzed action potential threshold variability in intracellular recordings in vivo from pyramidal neurons in weakly electric fish and found experimental evidence for threshold fatigue: an increase in instantaneous firing rate was on average accompanied by an increase in action potential threshold. We show that, with a minor modification, the standard Hodgkin-Huxley model can reproduce this phenomenon. We next compared the performance of models with and without threshold fatigue. Our results show that threshold fatigue will lead to a more regular spike train as well as robustness to intrinsic noise via noise shaping. We finally show that the increased/reduced noise levels due to threshold fatigue correspond to decreased/increased information transmission at different frequencies.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Action Potentials / physiology*
  • Algorithms
  • Animals
  • Data Interpretation, Statistical
  • Electric Fish / physiology
  • Electrophysiology
  • Information Theory
  • Models, Neurological
  • Neurons / physiology*
  • Pyramidal Cells / physiology
  • Refractory Period, Electrophysiological / physiology
  • Sodium Channels / physiology

Substances

  • Sodium Channels