Sussman H M
Brain Lang. 1986 May;28(1):12-23. doi: 10.1016/0093-934x(86)90087-8.
A speculative neuronal model for vowel normalization and representation is offered. The neurophysiological basis for the premise is the "combination-sensitive" neuron recently documented in the auditory cortex of the mustached bat (N. Suga, W. E. O'Neill, K. Kujirai, and T. Manabe, 1983, Journal of Neurophysiology, 49, 1573-1627). These neurons are specialized to respond to either precise frequency, amplitude, or time differentials between specific harmonic components of the pulse-echo pair comprising the biosonar signal of the bat. Such multiple frequency comparisons lie at the heart of human vowel perception and categorization. A representative vowel normalization algorithm is used to illustrate the operational principles of the neuronal model in accomplishing both normalization and categorization in early infancy. The neurological precursors to a phonemic vocalic system is described based on the neurobiological events characterizing regressive neurogenesis.
本文提出了一种用于元音归一化和表征的推测性神经元模型。该前提的神经生理学基础是最近在长须蝙蝠听觉皮层中记录到的“组合敏感”神经元（N. 须贺、W. E. 奥尼尔、K. 久慈来、T. 真锅，1983年，《神经生理学杂志》，49卷，1573 - 1627页）。这些神经元专门对构成蝙蝠生物声纳信号的脉冲 - 回波对的特定谐波分量之间的精确频率、幅度或时间差异做出反应。这种多频率比较是人类元音感知和分类的核心。一个具有代表性的元音归一化算法被用来阐明神经元模型在婴儿早期完成归一化和分类的操作原理。基于表征退行性神经发生的神经生物学事件，描述了音素元音系统的神经学前身。
