Recent advances in “developmental” approach (combining experimental study with computational modeling) of neural networks produce increasingly large data sets, in both complexity and size. This poses a significant challenge in analyzing, visualizing and understanding not only the spatial structure but also the behavior of such networks. This paper describes a virtual reality application for visualization of two biologically accurate computational models that model the anatomical structure of a neural network comprised of 1500 neurons and over 80,000 connections. The visualization enables a user to observe the complex spatiotemporal interplay between seven unique types of neurons culminating in an observable swimming pattern. We present a detailed description of the design approach for the virtual environment, based on a set of initial requirements, followed up by the implementation and optimization steps. Lastly, the results of a pilot usability study are being presented on how confident participants are in their ability to understand how the alternating firing pattern between the two sides of the tadpole’s body generates swimming motion.

神经网络的“发展”方法（将实验研究与计算模型相结合）的最新进展产生了越来越大的数据集，无论是复杂性还是规模。这不仅在分析，可视化和理解这种网络的空间结构而且在行为方面也构成了重大挑战。本文介绍了一种虚拟现实应用程序，用于可视化两个生物学上准确的计算模型，该模型对包含1500个神经元和80,000多个连接的神经网络的解剖结构进行建模。可视化使用户可以观察到以观察到的游泳模式达到顶点的七个独特类型的神经元之间复杂的时空相互作用。我们基于一系列初始要求，对虚拟环境的设计方法进行了详细说明，随后是实施和优化步骤。最后，正在提供一项试验性可用性研究的结果，说明参与者对如何理解the身体两侧交替射击方式如何产生游泳动作的信心。