Data Sharing Practices In A Collaborative Research Environment: The Case of The Brain-Computer Interface Field

Open data sharing is promoted by the Open Science movement and is often required by many research funders and journal publishers. Open data facilitates verification of research procedures and findings, safeguards against research misconduct, enables replication studies, and contributes to scientific progress. However, data sharing is complex due to technical and ethical challenges and varies across disciplines. Data sharing, especially in collaborative research environments, can take many forms and degrees of openness and may not always be fully open.

This presentation will report the findings of a study investigating researchers' data-sharing practices in the field of Brain-Computer Interface (BCI). BCI scientists study and measure brain activity to understand and extend its functionality in applied systems. Potential applications of BCI systems include assistive technologies, rehabilitation, gaming, human-computer interaction, education, and many others. BCI research is highly collaborative and interdisciplinary. This study interviewed 11 BCI researchers in four European countries. The participants were all involved in collaborative research projects with multiple national and European partners. The study identified a range of data-sharing approaches, from fully open to selective. Data-sharing practices among the BCI researchers vary, depending on the type of data, the stakeholders involved, and the nature of the sharing request. Data is routinely shared with team members within the same institution and with project collaborators at partner institutions. Sharing data with collaborators at other institutions and in different countries requires different levels of permission. While BCI researchers share data with other researchers and peer reviewers, they are often reluctant to make it openly accessible. In many cases, data sharing is limited and selective due to security and privacy concerns. Only certain datasets are released openly, and typically only after thorough anonymization.