Acoustic feature extraction by cross-correlation in crickets?

J Comp Physiol A Neuroethol Sens Neural Behav Physiol. 2003 Aug;189(8):589-98. doi: 10.1007/s00359-003-0438-7. Epub 2003 Jul 19.

Abstract

Common concepts of acoustic feature extraction within the auditory pathway of vertebrates and insects assume temporal filters tuned to particular periodicities. Crickets respond selectively to the conspecific song pattern and reveal a bandpass characteristic, which is thought to arise from a matched filter for a restricted range of periods. Unexpectedly, females of the two sibling species Teleogryllus oceanicus and T. commodus differed in fundamental filter properties. While T. oceanicus revealed a period filter, T. commodus exhibited a pulse duration filter. This finding raises the question of how so distinct properties of homologous neuronal circuits for pattern analysis have evolved during speciation. Evidence is presented that signal analysis by cross-correlation offers a simple explanation for differences in pattern selectivity as well as for the evolutionary transition. By cross-correlation the similarity of an external pattern with an internal template is determined over a certain time window. A comparison of behavioural data and cross-correlation values suggested that both species have similar templates. However, time windows were significantly different between species (T. oceanicus: 180-400 ms, T. commodus: 90-160 ms). Consequently, solely a change in the evaluation time window is required to account for the observed differences in feature extraction that serve to maintain species isolation.

Publication types

  • Comparative Study
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Acoustic Stimulation / methods*
  • Animal Communication
  • Animals
  • Auditory Perception / physiology*
  • Female
  • Gryllidae / classification
  • Gryllidae / physiology*
  • Models, Neurological
  • Pattern Recognition, Automated
  • Periodicity*
  • Species Specificity
  • Speech Production Measurement / methods*
  • Statistics as Topic*
  • Vocalization, Animal / physiology*