Oxenham A J
Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge 02139, USA.
J Acoust Soc Am. 2001 Feb;109(2):732-41. doi: 10.1121/1.1336501.
The aim of this study was to attempt to distinguish between neural adaptation and persistence (or temporal integration) as possible explanations of forward masking. Thresholds were measured for a sinusoidal signal as a function of signal duration for conditions where the delay between the masker offset and the signal offset (the offset-offset interval) was fixed. The masker was a 200-ms broadband noise, presented at a spectrum level of 40 dB (re: 20 microPa), and the signal was a 4-kHz sinusoid, gated with 2-ms ramps. The offset-offset interval was fixed at various durations between 4 and 102 ms and signal thresholds were measured for a range of signal durations at each interval. A substantial decrease in thresholds was observed with increasing duration for signal durations up to about 20 ms. At short offset-offset intervals, the amount of temporal integration exceeded that normally found in quiet. The results were simulated using models of temporal integration (the temporal-window model) and adaptation. For both models, the inclusion of a peripheral nonlinearity, similar to that observed physiologically in studies of the basilar membrane, was essential in producing a good fit to the data. Both models were about equally successful in accounting for the present data. However, the temporal-window model provided a somewhat better account of similar data from a simultaneous-masking experiment, using the same parameters. This suggests that the linear, time-invariant properties of the temporal-window approach are appropriate for modeling forward masking. Overall the results confirm that forward masking can be described in terms of peripheral nonlinearity followed by linear temporal integration at higher levels in the auditory system. However, the difference in predictions between the adaptation and integration models is relatively small, meaning that influence of adaptation cannot be ruled out.
本研究的目的是试图区分神经适应和持续性（或时间整合），将其作为前掩蔽可能的解释。在掩蔽声偏移与信号偏移之间的延迟（偏移 - 偏移间隔）固定的条件下，测量正弦信号的阈值作为信号持续时间的函数。掩蔽声为200毫秒的宽带噪声，以40分贝（相对于20微帕）的频谱水平呈现，信号为4千赫的正弦波，用2毫秒的斜坡进行门控。偏移 - 偏移间隔固定在4至102毫秒之间的各种持续时间，并且在每个间隔处针对一系列信号持续时间测量信号阈值。对于长达约20毫秒的信号持续时间，观察到阈值随着持续时间增加而大幅下降。在短的偏移 - 偏移间隔下，时间整合量超过了在安静环境中通常发现的量。使用时间整合模型（时间窗口模型）和适应模型对结果进行了模拟。对于这两种模型，纳入类似于在基底膜研究中生理上观察到的外周非线性对于很好地拟合数据至关重要。两种模型在解释当前数据方面都同样成功。然而，使用相同参数时，时间窗口模型对来自同时掩蔽实验的类似数据的解释稍好一些。这表明时间窗口方法的线性、时不变特性适合于对前掩蔽进行建模。总体而言，结果证实前掩蔽可以用外周非线性随后在听觉系统较高水平的线性时间整合来描述。然而，适应模型和整合模型之间的预测差异相对较小，这意味着不能排除适应的影响。
