Epping W J, Eggermont J J
Hear Res. 1986;24(1):55-72. doi: 10.1016/0378-5955(86)90005-5.
The coding of fine-temporal structure of sound, especially of frequency of amplitude modulation, was investigated on the single-unit level in the auditory midbrain of the grassfrog. As stimuli sinusoidally amplitude modulated sound bursts and continuous sound with low-pass Gaussian noise amplitude modulation have been used. Both tonal and wideband noise carriers have been applied. The response to sinusoidally amplitude modulated sound bursts was studied in two aspects focussing on two types of possible codes: a rate code and a synchrony code. From the iso-intensity rate histogram five basic average response characteristics as function of modulation frequency have been observed: low-pass, band-pass, high-pass, bimodal and non-selective types. The synchronization capability, expressed in a synchronization index, was non-significant for 38% of the units and a low-pass function of modulation frequency for most of the other units. The stimulus-response relation to noise amplitude modulated sound was investigated by a non-linear system theoretical approach. On the basis of first- and second-order Wiener-Volterra kernels possible neural mechanisms accounting for temporal selectivity were obtained. About one quarter of the units had response characteristics that were invariant to changes in sound pressure level and spectral content of the carrier. These units may function as feature detectors of fine-temporal structure of sound. The spectro-temporal sensitivity range of the auditory midbrain of the grassfrog appeared not to be restricted to and showed no preference for the spectro-temporal characteristics of the ensemble of conspecific calls. Comparison of response characteristics to periodic click trains as studied in the companion paper (Epping and Eggermont, 1986) and sinusoidally amplitude modulated sound bursts revealed that the observed temporal sensitivity is due to a combination of sensitivities to sound periodicity and pulse duration. It was found that for most units the first-order kernels for Gaussian amplitude modulated stimuli and Poisson distributed click stimuli were alike. In contrast second-order kernels for the Gaussian amplitude modulated stimuli often represented only static non-linearities, while second-order kernels for Poisson distributed clicks (Epping and Eggermont, 1986) mostly revealed dynamic non-linearities.
在沼蛙的听觉中脑单神经元水平上，研究了声音精细时间结构的编码，特别是调幅频率的编码。使用了正弦调幅声脉冲和具有低通高斯噪声调幅的连续声音作为刺激。同时应用了纯音和宽带噪声载波。从两个方面研究了对正弦调幅声脉冲的反应，重点关注两种可能的编码类型：速率编码和同步编码。从等强度速率直方图中，观察到了作为调制频率函数的五种基本平均反应特性：低通、带通、高通、双峰和非选择性类型。用同步指数表示的同步能力，对于38%的神经元不显著，而对于大多数其他神经元是调制频率的低通函数。通过非线性系统理论方法研究了对噪声调幅声音的刺激 - 反应关系。基于一阶和二阶维纳 - 沃尔泰拉核，获得了可能解释时间选择性的神经机制。约四分之一的神经元具有对载波声压级和频谱内容变化不变的反应特性。这些神经元可能作为声音精细时间结构的特征检测器。沼蛙听觉中脑的频谱 - 时间敏感范围似乎不限于同种叫声的频谱 - 时间特征，也没有表现出偏好。在配套论文（Epping和Eggermont，1986）中研究的对周期性点击序列的反应特性与正弦调幅声脉冲的比较表明，观察到的时间敏感性是对声音周期性和脉冲持续时间敏感性的组合。发现对于大多数神经元，高斯调幅刺激和泊松分布点击刺激的一阶核相似。相比之下，高斯调幅刺激的二阶核通常仅代表静态非线性，而泊松分布点击的二阶核（Epping和Eggermont，1986）大多显示动态非线性。
