Utilizing recent advances in machine learning, we introduce a systematic approach to characterize neurons’ input/output (I/O) mapping complexity. Deep neural networks (DNNs) were trained to faithfully replicate the I/O function of various biophysical models of cortical neurons at millisecond (spiking) resolution. A temporally convolutional DNN with five to eight layers was required to capture the I/O mapping of a realistic model of a layer 5 cortical pyramidal cell (L5PC). This DNN generalized well when presented with inputs widely outside the training distribution. When NMDA receptors were removed, a much simpler network (fully connected neural network with one hidden layer) was sufficient to fit the model. Analysis of the DNNs’ weight matrices revealed that synaptic integration in dendritic branches could be conceptualized as pattern matching from a set of spatiotemporal templates. This study provides a unified characterization of the computational complexity of single neurons and suggests that cortical networks therefore have a unique architecture, potentially supporting their computational power.

利用机器学习的最新进展，我们引入了一种系统方法来表征神经元的输入/输出 (I/O) 映射复杂性。深度神经网络 (DNN) 经过训练，能够以毫秒（尖峰）分辨率忠实地复制皮质神经元的各种生物物理模型的 I/O 功能。需要具有五到八层的时间卷积 DNN 来捕获第 5 层皮质锥体细胞 (L5PC) 的真实模型的 I/O 映射。当输入广泛地超出训练分布时，这个 DNN 泛化得很好。当去除 NMDA 受体时，一个更简单的网络（具有一个隐藏层的全连接神经网络）足以拟合模型。对 DNN 权重矩阵的分析表明，树突分支中的突触整合可以概念化为来自一组时空模板的模式匹配。这项研究提供了单个神经元计算复杂性的统一特征，并表明皮层网络因此具有独特的架构，可能支持其计算能力。