Generating a formal set of collaborative standards for sharing behavioral data and task designs to enable reproducibility in neuroscience

点击翻译按钮获取中文摘要

Abstract The goal of this project is to develop an archival data format and a formal task specification language to serve as standards for describing behavioral experiments. Because different laboratories use different behavioral systems, hardware, and software, it has been difficult to communicate behavioral task design, share data, or reproduce experiments. To fill this gap, we have assembled a team that includes neuroscientists with extensive behavioral experience, a computer science expert in real-time formal language design, and scientific software engineers. To meet the needs of the broader community, we have formed a large consortium of laboratories, including BRAIN-initiative grantees that have supplied numerous use cases spanning the range of behavioral tasks. In Aim 1, we will identify requirements for behavioral data formats by convening a large consortium of experimentalists. To represent behavioral data, we will create an extension to the NWB:N format to facilitate analysis alongside other behavior types. To specify behavioral contingencies, we will develop a formal BEhAvioral task Description Language (BEADL) using the latest computer science techniques for real-time language design. In Aim 2, we will develop an open-source software suite for editing, executing and visualizing behavioral tasks. The software will interpret the same behavior into machine code for hardware controllers, or publication figures for humans. To ensure the standards address the needs of the community, we will collaborate with users and consortium labs to implement an extensive battery of published tasks. In Aim 3, we will expand our outreach to broaden adoption of behavioral task description and data formats. We will engage the broader community through web-based tutorials, hackathons, and workshops at major conferences, and will encourage feedback. A publication-ready, graphical BEADL format will be developed to generate summary figures for behavioral tasks that are formal, yet easy to understand.