A novel method to analyse response patterns of taste neurons by artificial neural networks

Takatoshi Nagai, Takashi Yamamoto, Hiroshi Katayama, Masaharu Adachi, Kazuyuki Aihara

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

Despite several notions on the gustatory code proposed over three decades, investigators have not yet reached a consensus. This paper describes a new approach to analyse gustatory neural activities. Three-layer neural networks were trained by the back-propagation learning algorithm, to classify the neural response patterns to four basic taste qualities. The discrimination by the trained networks on taste qualities in the response patterns of rat chorda tympani fibres (CT) and cortical taste neurons (CN) was consistent both with the correlation analysis and with behavioural experiments. By examining the connection weights of each neuron, some input neurons representing CN were 'pruned' without deteriorating the ability of the network to discriminate taste. This characteristic of the network is contrary to a previous hypothesis, that taste neurons are of equal importance in the neural coding.

Original languageEnglish
Pages (from-to)745-748
Number of pages4
JournalNeuroReport
Volume3
Issue number9
DOIs
Publication statusPublished - 1992 Sep

    Fingerprint

Keywords

  • Algorithms
  • Back-propagation
  • Cerebral cortex
  • Chorda tympani nerve
  • Information processing
  • Learning
  • Neural networks
  • Rats
  • Taste

ASJC Scopus subject areas

  • Neuroscience(all)

Cite this