THE STOCHASTIC NATURE OF SINGLE NEURONS

近年来，许多研究都集中在神经元对突触输入或直流电反应的时间精度上。这种精确度涉及到用于表示和传递信息的神经元代码的本质这一关键问题。我们建议通过测量和分析各种噪声源来量化噪声对时间精度的基本限制，这些噪声源包括(i)由于电压依赖离子通道的随机性质而产生的噪声，（ii）“自发”突触背景放电的影响，以及（iii）由突触传输的不可靠的概率性质引入的噪声。1. 我们打算通过实验表征这些噪声源在新皮质神经元中使用药理学操作，这使我们能够分离单个噪声源如Na+， K+或Ca2+通道的贡献。我们将在单个神经元的体细胞和树突上使用全细胞贴片记录，并在体外制备的成对连接的神经元中使用红外光学进行可视化。2. 我们将反复将冻结电流噪声注入体细胞、树突和突触前神经元（使用双电极），并在突触后电位（亚阈值PSPs）中记录噪声，并记录由此产生的尖峰序列的计时滤波器。这将允许我们计算注入电流和输出尖峰串之间的互信息。3. 我们将把这些测量结果与弱活性线性电缆的热、通道和突触噪声的分析和数值模型进行比较，这些模型是通过结合这些细胞的详细形态（我们将在细胞下注射生物细胞素后重建）和电生理特性获得的。4. 我们将使用理论技术推导出尖峰滤波和尖峰序列之间噪声源相似性的度量，并将它们与步骤2中获得的实验数据进行比较。我们的研究计划将导致神经元噪声源的特性及其对单个皮质神经元信息容量的影响的定量图像。

Much effort has focused in recent years on the temporal precision with which neurons respond to synaptic input or to direct current injection. This precision relates to the crucial issue of the nature of the neuronal code used to represent and transmit information. We propose to quantify the fundamental limits that noise places on temporal precision by measuring and analyzing the various noise sources, including (i) noise due to the stochastic nature of voltage-dependent ionic channels, (ii) the effect of "spontaneous" synaptic background firing and (iii) the noise introduced by the unreliable, probabilistic nature of synaptic transmission. 1. We intend to experimentally characterize these noise sources in neocortical neurons using pharmacological manipulations which allow us to isolate the contribution of individual noise sources like Na+, K+ or Ca2+ channels. We will use whole-cell patch recordings at the soma and dendrites of single neurons, and in pairs of connected neurons of an in vitro preparation, visualized using infrared optics. 2. We will repeatedly inject frozen current noise into the soma, the dendrite and the presynaptic neuron (using dual electrodes), and record the noise in the post-synaptic potential (sub-threshold PSPs) as well as the fitter in timing of the resulting spike train. This will allow us to compute the mutual information between the injected current and the output spike train. 3. We will compare these measurements with analytical and numerical models of thermal, channel and synaptic noise in weakly- active linear cables, obtained by incorporating the detailed morphology (that we will reconstruct following infra-cellular injection of biocytin) and the electrophysiological properties of these cells. 4. We will use theoretical techniques to derive measures of spike fitter and similarity between spike trains in terms of the noise sources and compare them against the experimental data obtained in step 2. Our research plan will lead to a quantitative picture of the properties of neuronal noise sources and their effect on the information capacity of individual cortical neurons.