A comment on a retinal neuron model.
A comment on a retinal neuron model.
复制标题
对视网膜神经元模型的评论。
DOI：
10.1016/s0006-3495(79)85301-1
复制
发表时间：
1979
影响因子：
3.4
通讯作者：
C. E. Smith
中科院分区：
文献类型：
--
作者：
C. E. Smith
尊敬的先生：在本杂志最近发表的一篇论文中，Levine 和 Shefner (1) 提出了金鱼视网膜神经节细胞持续放电期间尖峰间隔变化的模型。该模型假设了一个噪声尖峰启动过程，其中两次发射之间的时间是一个随机变量，等于正态分布变量的倒数。据报道，该模型重现了实验结果，即峰间间隔的变异系数（CV）不随平均放电率的变化而变化。当尖峰间隔的模型概率密度函数 (PDF) 的前两个矩不存在时，很难看出他们的模型如何拥有此属性。对底层 PDF 的修改将纠正这个问题并产生预期的结果。Levine 和 Shefner 的“超正态”PDF 矩的不可积性源于其 I/l2 尾部，因为尖峰时间 t 接近无穷大。虽然尾部可以在某个有限的时间被截断，但任何这样的截断都是完全任意的。尽管存在这样的慢尾似乎合理的实验示例 (2)，但解决此问题的一个更合理的替代方案是选择一个其倒数变换接近无穷大的速度比 1/t2 快的基础 PDF。伽玛 PDF 就是这样的选择
Dear Sir: In a recent paper published in this journal, Levine and Shefner (1) presented a model for the variability of interspike intervals during the sustained firing of goldfish retinal ganglion cells. The model assumed a noisy spike initiation process in which the time between firings is a random variable equal to the reciprocal of a normally distributed variable. The model was reported to reproduce the experimental result that the coefficient of variation (CV) of the interspike intervals did not change as the mean firing rate changed. It is difficult to see how their model could possess this property when the first two moments of the model's probability density function (PDF) for interspike intervals do not exist. A modification of the underlying PDF will rectify this problem and produce the intended results.The nonintegrability of the moments of Levine and Shefner's" hypernormal" PDF stems from its I/l2 tail as the interspike time, t, approaches infinity. While the tail could be truncated at some fi-nite time, any such truncation would be totally arbitrary. Although there exist experimental examples where such a slow tail seems justified (2), a more reasonable alternative for this problem would be to choose an underlying PDF whose reciprocal transformation approaches infinity faster than 1/t2. Onesuch choice is the gamma PDF