O'Connor K N, Barruel P, Hajalilou R, Sutter M L
Center for Neuroscience, University of California, Davis 95616, USA.
J Acoust Soc Am. 1999 Aug;106(2):954-65. doi: 10.1121/1.427108.
Temporal integration for pure tones was examined in two rhesus macaques. The subjects were required to respond to a brief sound (a tone burst) that deviated from a previous series of sounds (noise bursts) on a trial (a deviant-stimulus detection paradigm). Psychometric functions and thresholds were determined from correct detections (hit proportions) alone, and from d' scores. Two models describing the decline in threshold as a function of stimulus duration, one a power function the other an exponential, were tested against the data. When the decline (slope) in threshold per log stimulus duration is used as a rate measure, our results yield a lower estimate of temporal integration rate in rhesus than did a previous study [Clack, J. Acoust. Soc. Am. 40, 1140-1146 (1966)]. Both studies, however, gave slope estimates of integration rate that were higher than in most other species. Comparison of the models using data from several species, revealed that the exponential, but not the power model, could account for two sources of variation in threshold measurement. One source is due to the range across threshold as a function of duration (the linear rate component), and is described by the constant of proportionality Ik in the model. The other source of variation arises from the rate of decline within this range (the nonlinear rate component), and is described by the time constant tau. In terms of this model, differences in rate estimates between Clack's study and ours (and between rhesus and other species) are primarily due to the linear component. The nonlinear rate component was about equal for our study and Clack's (tau = approximately 150 ms): a time constant that is just slightly larger (indicating a rate of temporal integration slightly slower) than for most other species examined.
在两只恒河猴身上研究了纯音的时间整合。实验要求受试者对一个短暂声音（一个音爆）做出反应，该声音在一次试验中与先前的一系列声音（噪声爆）不同（异常刺激检测范式）。心理测量函数和阈值仅根据正确检测（命中比例）以及d'分数来确定。测试了两个将阈值下降描述为刺激持续时间函数的模型，一个是幂函数，另一个是指数函数，并与数据进行对比。当将每对数刺激持续时间的阈值下降（斜率）用作速率度量时，我们的结果得出恒河猴的时间整合速率估计值低于先前的一项研究[克拉克，《美国声学学会杂志》40, 1140 - 1146 (1966)]。然而，两项研究给出的整合速率斜率估计值都高于大多数其他物种。使用来自几个物种的数据对模型进行比较，发现指数模型而非幂函数模型能够解释阈值测量中的两个变异来源。一个来源是阈值随持续时间变化的范围（线性速率成分），由模型中的比例常数Ik描述。另一个变异来源是该范围内的下降速率（非线性速率成分），由时间常数tau描述。根据这个模型，克拉克的研究与我们的研究之间（以及恒河猴与其他物种之间）速率估计的差异主要归因于线性成分。我们的研究和克拉克的研究中非线性速率成分大致相等（tau约为150毫秒）：这个时间常数仅略大于大多数其他被研究物种（表明时间整合速率略慢）。
