FAIR Principles in Neuroscience and Psychology

1. Introduction to FAIR in Neuroscience

The transformation of neuroscience from a closed to an open science has accelerated in recent years. The FAIR data principles, introduced in 2016, represent a pivotal framework that establishes minimum requirements for scientific data to be useful: data should be Findable, Accessible, Interoperable, and Reusable.

Neuroscience presents unique challenges for FAIR implementation due to:

• Diversity of data types (imaging, electrophysiology, behavioral, genetic, etc.)
• Multiple scales of investigation (molecular to systems level)
• Variety of model systems and experimental paradigms
• Complexity of data acquisition and analysis workflows
• Large volumes of data generated by modern techniques

Despite early resistance to data sharing in neuroscience compared to fields like genomics and structural biology, the field has made significant progress in the past decade. This has been driven by:

• Large prospective data sharing initiatives (ADNI, Allen Brain Atlas, Human Connectome Project)
• National and international brain projects with mandated open data policies
• Increasing concern about reproducibility in neuroscience research
• Funder requirements and journal policies mandating data sharing

The International Neuroinformatics Coordinating Facility (INCF) has played a crucial role in promoting FAIR practices in neuroscience, providing training, standards development, and infrastructure coordination.

2. The FAIR Principles Defined

The FAIR principles were formulated during a workshop in Leiden in 2014 and subsequently published by Wilkinson et al. (2016). Below is an explanation of each principle as it applies to neuroscience data:

3. The FAIR Partnership: Key Stakeholders

FAIR implementation in neuroscience requires collaboration among multiple stakeholders:

Key Stakeholders and Their Roles

4. FAIR Data Management in Laboratories

Effective FAIR implementation begins at the laboratory level. Below are key laboratory data management practices based on the FAIR principles:

FAIR Goal	Principle	Laboratory Practices	References
Findable	Unique identifiers	Create globally unique identifiers within the lab for all key entities (subjects, experiments, reagents). Implement a central registry or use existing systems (e.g., RRIDs for reagents and tools).	Fouad et al. (2023)
	Rich metadata	Accompany each identifier with detailed metadata (e.g., dates, experimenter, description for experiments; species/strain, age, weight for subjects). Use identifiers consistently in file names, folder names, and physical labels.	Fouad et al. (2023)
Accessible	Authentication and authorization	Create a centralized, accessible store for data and code under a lab-wide account to prevent data from being scattered or accessible only via personal accounts.
Interoperable	FAIR vocabularies	Replace idiosyncratic naming with community standards like Common Data Elements and community-based ontologies. Create a lab-wide data dictionary where all variables are clearly defined.	Bush et al. (2022), Fouad et al. (2023)
Reusable	Documentation	Create a "Read me" file for each dataset with notes and information for reuse.
	Community Standards	Store files in well-supported open formats. Adopt community standards within the lab, especially those required by target repositories.	Bush et al. (2022), INCF Standards Portfolio
	Provenance	Version datasets clearly and document differences. Keep a stable "version of record." Include detailed experimental protocols and computational workflows. Use dedicated tools like protocols.io, NeuroShapes, and ReproNIM.	Kennedy et al. (2019)
	Licenses	Ensure data sharing agreements are in place with all collaborators. For clinical datasets, verify that consents permit sharing of de-identified data.

As noted in the NASEM workshop on "Changing the Culture on Data Management and Sharing" (Martone & Nakamura, 2022): "If you can share data with people in your lab, you are much more likely to have something worthwhile to share outside the lab."

5. Choosing a Repository

Selecting an appropriate repository is critical for ensuring long-term FAIR compliance. The neuroscience repository landscape is organized primarily by data type, with additional specialization by domain or region.

Types of Neuroscience Repositories

Repositories generally fall into these categories:

• Specialized by data type (e.g., neuroimaging, electrophysiology)
• Domain-specific (e.g., SPARC for peripheral nervous system)
• Region or institution-specific (e.g., CONP, BrainCode, Donders Repository)
• Generalist repositories that span disciplines (e.g., FigShare, Zenodo)

KiltHub, CMU's Institutional Repository is an instance of FigShare, and aligns with FAIR principles. 

Selection Criteria

When selecting a repository, consider:

• Alignment with data type and domain
• Tool support and integration
• Curation services
• Support for data citation
• License options
• Data size limitations
• Help with data management plans
• Cost structure
• Long-term sustainability plan

Repository finder tools include:

• INCF Infrastructure Catalog
• NITRC for neuroimaging repositories
• re3data catalog
• NLM listing of repositories
• NIF listing of BRAIN Initiative Repositories

Repository FAIR Implementation

Table below compares FAIR features across major neuroscience repositories:

Principle	Function	EBRAINS	SPARC	DANDI	CONP Portal	OpenNeuro
F1: Globally unique identifier	Basic core	DOI	DOI	DOI	ARK, DOI	DOI
F2: Rich metadata		Y	DataCite	Y	DATS	Y
A1: Retrievable by identifier		Y	Y	Y	Y	Y
A1.1: Free, open, universal retrieval protocol	Enhanced access	Y	Y	Y	Y	Y
F4: Registered in a searchable resource		KS, GDS	KS, GDS	KS, GDS	KS	KS, GDS
A1.2: Authentication and authorization		Y	Y	Y	Y	Y
R1.1: Clear data usage license		Y	CC-BY	CC-BY, CC0	Y	CC0
R1.3: Community standards	Use of standards	Multiple	SDS, MIS	NWB, BIDS	Y*	BIDS
F3: Metadata contains identifier		Y	Y	Y	Y	Y
I1: Formal knowledge representation language		Y	Y	N	Y
R1: Plurality of relevant attributes	Rich(er) metadata	OpenMinds	OpenMinds, MIS	NWB	DATS	Y
I2: FAIR vocabularies		Y	Y	Y	Y	N
I3: Qualified references to other metadata		Y	Y	Y	Y	Y
R1.2: Provenance	Provenance and context		Exp Protocol		Y	N
Data citation	Additional features	Y	Y	Y	Y	Y
Curation		Y	Y	N	Y	N

KS: INCF Knowledge Space; GDS: Google Dataset Search; DOI: Digital object identifier; NWB: NeuroData Without Borders; BIDS: Brain Imaging Data Structure; DATS: Data tag suite

6. Standards in Neuroscience

Standards are critical for ensuring interoperability and reusability of neuroscience data. Major standards include:

Data Format Standards

• Brain Imaging Data Structure (BIDS): Standard for organizing and describing neuroimaging data
• NeuroData Without Borders (NWB): Standard for neurophysiology data
• 3D-MMS: Metadata standard for 3D microscopy
• SPARC Data Structure (SDS): For peripheral nervous system data

Common Coordinate Frameworks (CCFs)

• Waxholm Space: Registration standard for mouse and rat brain data
• Allen Institute CCF v3: Widely used for mouse brain data across international projects

Vocabularies and Ontologies

• UBERON: Cross-species anatomy ontology
• Brain Standards Data Ontology: Model for data-driven definitions of taxonomic classes
• Cell Cards: Tool for exploring BICCN taxonomic cell types

Standards Adoption

Repositories play a key role in standards adoption by:

• Supporting or requiring standards for data submission
• Providing tools to facilitate standards compliance
• Contributing to standards development

The INCF has implemented an open community review and endorsement process for neuroscience standards and maintains a searchable Standards and Best Practices Portfolio.

7. Neuroscience Infrastructure for FAIR

Neuroscience is served by a globally distributed ecosystem of data repositories and computational platforms that work together to facilitate FAIR data practices.

Search and Discovery Infrastructure

• INCF Knowledge Space: Searches across 16 major neuroscience databases
• Google Dataset Search: Many repositories now mark up content with schema.org for discovery
• Knowledge Graphs: Used by EBRAINS, CONP, and SPARC to link neuroscience concepts

Data Federation Approaches

• Canadian Open Neuroscience Portal: Allows search across multiple repositories with unified metadata (DATS standard)
• Data access tools: DataLad for download, Boutiques for containerized workflows

Sustainability Strategies

• Repository exit plans: Clear persistence policies
• Institutional partnerships: Repositories partnering with university libraries
• Infrastructure reuse: Sharing infrastructure components across projects (e.g., Brainlife and NEMAR using OpenNeuro)

8. IRB Considerations and Ethical Aspects

Implementing FAIR principles in neuroscience and psychology must be balanced with ethical considerations, particularly for human subjects research. This requires careful navigation of Institutional Review Board (IRB) requirements and data privacy regulations.

Contact CMU's Institutional Review Board if your research involves human subject data.

IRB and Consent Considerations

Prospective Planning for Data Sharing

• Informed Consent: Design consent forms that explicitly address data sharing intentions
• Tiered Consent Options: Consider providing options for different levels of data sharing (e.g., within institution, with specific collaborators, or public repositories)
• Future Use Language: Include language about potential future research uses of the data
• Re-contact Options: When appropriate, include provisions for re-contacting subjects for additional permissions

Retrospective Data Sharing

• Waiver of Consent: Understand when IRBs might grant waivers for sharing previously collected data
• Data Use Agreements: Utilize institutional agreements to enable sharing of sensitive data
• Secondary Use Determinations: Consult with IRB about requirements for secondary use of existing datasets

De-identification Standards and Practices

Proper de-identification is critical for sharing neuroscience data while protecting subject privacy:

Data Type	De-identification Considerations	Best Practices
Neuroimaging Data	Facial features may allow re-identification	Use established defacing algorithms (e.g., pydeface, mri_deface); verify results visually
Genetic Data	Potentially identifiable even when "anonymized"	Consider controlled access models; use data use agreements
Demographics/Phenotypic	Unique combinations of variables can lead to re-identification	Remove or bin variables that could allow re-identification; keep only necessary variables
Video/Audio Recordings	Direct identifiers	Blur faces, alter voices; consider secure processing with extracted features only

Special Populations Considerations

Extra protections are required for vulnerable populations:

• Children: Additional parental/guardian consent requirements
• Clinical Populations: Sensitivity to stigmatization concerns
• Small or Unique Cohorts: Higher risk of re-identification requiring additional safeguards
• Indigenous Communities: Respect for data sovereignty and community-level permissions

Regulatory Frameworks

Various regulations affect FAIR implementation across different jurisdictions:

• GDPR (European Union): Emphasizes data minimization and purpose limitation principles
• HIPAA (United States): Establishes standards for Protected Health Information (PHI)
• 21st Century Cures Act (United States): Promotes data sharing while maintaining privacy
• National/Regional Regulations: May impose additional requirements or limitations

Balancing FAIR with Privacy

Strategies to maximize both FAIR principles and privacy protection:

• Controlled Access Models: Apply different access levels based on data sensitivity
• Safe Havens/Secure Environments: Allow analysis without direct data access
• Privacy-Preserving Analysis Methods: Consider federated learning approaches
• Metadata-Only Sharing: Share rich metadata while restricting access to raw data
• Synthetic Data Generation: Create artificial datasets that preserve statistical properties

IRB Engagement Strategies

Proactive approaches for working with IRBs on FAIR data practices:

1. Education: Provide IRB members with information about FAIR principles and their benefits
2. Template Language: Develop institutionally approved language for consent forms
3. Data Management Plans: Share comprehensive plans addressing both FAIR and privacy
4. Staged Approach: Propose gradual implementation with appropriate safeguards
5. Collaborative Development: Involve IRB members in developing institutional policies

International Collaboration Considerations

For multinational projects, additional considerations apply:

• Harmonization of Requirements: Address different regulatory frameworks
• Data Localization Laws: Be aware of restrictions on cross-border data transfers
• Regional Standards: Adapt approaches to meet local ethical requirements

9. Implementation Challenges and Solutions

Despite progress, several challenges remain for achieving fully FAIR neuroscience:

Findability Challenges

• Problem: No comprehensive search system across all neuroscience data repositories
• Solution: Support efforts by IBI, INCF, and CONP to create federated search systems

Accessibility Challenges

• Problem: Multimodal datasets require submission to multiple repositories
• Solution: Develop unified workflows and centralized registries; implement ORCID across repositories

Interoperability Challenges

• Problem: Different repository interfaces and access policies
• Solution: Adopt "coopetition" model where repositories compete on innovation but share core functionality

Reusability Challenges

• Problem: Variable data quality and documentation across repositories
• Solution: Support curation services and improve customer service to help researchers comply with standards

International Data Governance

• Problem: Barriers to transferring data across national borders
• Solution: Federated approaches where data remains in place while computation is brought to the data

10. Resources and Tools

INCF Resources

• INCF Standards and Best Practices Portfolio: https://www.incf.org/resources/sbps
• INCF Training Space: https://training.incf.org/
• INCF Infrastructure Catalog: https://www.incf.org/infrastructure-portfolio

Standards Resources

• BIDS Website: https://bids.neuroimaging.io/
• NeuroData Without Borders: https://www.nwb.org/
• FAIRsharing.org: Catalog of standards across biomedicine

Tools for FAIR Implementation

• SODA Tool: Helps researchers organize and upload files according to SPARC standards
• DataLad: Distributed data management system
• Boutiques: Framework for creating and using containerized applications
• Protocols.io: Platform for sharing detailed experimental protocols

11. FAIR Implementation Checklist

This checklist provides a step-by-step guide to implementing FAIR principles throughout the research lifecycle in neuroscience and psychology.

This checklist can be adapted based on specific research protocols, data types, and institutional requirements. Regular review of FAIR practices should be integrated into project workflows to ensure continuous improvement in data management and sharing.

12. References

Abrams, M. B., Bjaalie, J. G., Das, S., Egan, G. F., Ghosh, S. S., Goscinski, W. J., et al. (2021). A standards Organization for Open and FAIR neuroscience: the International Neuroinformatics Coordinating Facility. Neuroinformatics, 20, 25–36.

Bush, K. A., Calvert, M. L., & Kilts, C. D. (2022). Lessons learned: a neuroimaging research Center's transition to open and reproducible science. Frontiers in Big Data, 5, 988084.

Ferguson, A. R., Nielson, J. L., Cragin, M. H., Bandrowski, A. E., & Martone, M. E. (2014). Big data from small data: data-sharing in the 'long tail' of neuroscience. Nature Neuroscience, 17, 1442–1447.

Fouad, K., Vavrek, R., Surles-Zeigler, M. C., Huie, J. R., Radabaugh, H., Gurkoff, G. G., et al. (2023). A practical guide to data management and sharing for biomedical laboratory researchers. Zenodo.

Gorgolewski, K. J., Auer, T., Calhoun, V. D., Craddock, R. C., Das, S., Duff, E. P., et al. (2016). The brain imaging data structure, a format for organizing and describing outputs of neuroimaging experiments. Scientific Data, 3, 160044.

Kennedy, D. N., Abraham, S. A., Bates, J. F., Crowley, A., Ghosh, S., Gillespie, T., et al. (2019). Everything matters: the ReproNim perspective on reproducible neuroimaging. Frontiers in Neuroinformatics, 13, 1.

Martone, M. E. (2024). The past, present and future of neuroscience data sharing: a perspective on the state of practices and infrastructure for FAIR. Frontiers in Neuroinformatics, 17, 1276407.

Martone, M. E., & Nakamura, R. (2022). Changing the culture on data management and sharing: overview and highlights from a workshop held by the National Academies of Sciences, Engineering, and Medicine. Harvard Data Science Review.

Poline, J.-B., Kennedy, D. N., Sommer, F. T., Ascoli, G. A., Van Essen, D. C., Ferguson, A. R., et al. (2022). Is Neuroscience FAIR? A Call for Collaborative Standardisation of Neuroscience Data. Neuroinformatics, 20, 507–512.

Wilkinson, M. D., Dumontier, M., Aalbersberg, I. J., Appleton, G., Axton, M., Baak, A., et al. (2016). The FAIR guiding principles for scientific data management and stewardship. Scientific Data, 3, 160018.