EEG100 Manifesto

Toward a deontological framework for EEG science

Introduction

This document is a starting point for building a deontological framework for EEG science and upgrading our practices accordingly. It aims to outline the standards we believe are essential to collectively maximize the positive impact of EEG research for society. It is a call to action, to encourage critical reflection on the broader consequences of scientific research while sharing emerging alternatives.

In the following, we highlight what we perceive as the most pressing issues across three domains: (i) scientific integrity and epistemological rigor; (ii) Democratization; and (iii) technological and environmental responsibility.

More

Scientific integrity and epistemological rigor are two aspects of our practice that require continuous attention (Button et al., 2013; Ioannidis, 2005). As scientific practices evolve alongside new technologies, it is essential to ensure that existing methods remain valid while continuously improving them to avoid methodological flaws that can lead to biased, false, irreplicable findings. Our focus should be on fostering robust, reproducible, open, and participatory science.

Democratization of the scientific process: We need to embrace the diversity of agents and subjects at all stages of the research process, from scientists conducting experiments to study participants, as well as the people impacted by the findings. Diversity in the participant populations supports generalizable and ethical scientific findings, and diversity in applications and outreach makes our research accessible, relevant, and beneficial to a broader range of communities, engaging more diverse audiences. Thus, diversity enriches the scientific process by bringing in broader perspectives (Phillips et al., 2009), leading to more innovative, widely relevant, and robust outcomes, ensuring that scientific practices are more inclusive, ethical, trusted and understood by the communities impacted by the findings.

Finally, and perhaps most controversially, we pledge for technological and environmental responsibility, acknowledging the limitations of the hardware and processing methods, and their potential misuse. We need to acknowledge that not everything that is doable is desirable (Moor, 2005) and should carefully examine the ethical legitimacy of our research lines, considering the social and material impacts of any research project or discovery. These are key aspects for cognition research, not solely for ethical reasons, but because cognition is produced and shaped by biological, social, environmental, technological, and cultural dynamics (Newen et al., 2018).

Each section below includes a list of commitments that you, as an individual, lab group, association, or organization can support. This manifesto is addressed to the entire EEG community, including technicians, engineers, PhD students, post-docs, senior researchers, research project evaluators, research administrators, artists, collaborators, and scientific decision-makers. It invites each individual to reflect on the raised topics, tailored to their role and responsibilities within the community. There will be some general principles that we will all agree upon, but there will no doubt be some more controversial points that we hope will generate healthy discussion. Feel free to use the CuttingEEG forum for this purpose. We invite you to sign all or parts of this manifesto and share it with your community if you broadly agree with the initiative.

Context & Expectations

Electroencephalography

For a century, electroencephalography (EEG) has been a cornerstone of neuroscience (Mushtaq et al. 2024; Hari & Puce, 2023). Today, more than ever, we—scientists, engineers and clinicians—are witnessing unprecedented advances in recording technology and analysis methods. Modern EEG devices are increasingly affordable, robust, and portable, extending their reach well beyond clinical diagnostics and traditional research labs. At the same time, global computing power, cloud communication, and machine learning—now coupled with generative AI—promise ever more sophisticated analyses and real-time brain-computer interfaces (BCIs), making their widespread adoption seem not just possible, but inevitable.

However, these rapid technological advances bring significant challenges. A reproducibility crisis in science has eroded trust in established findings (Button et al., 2013; Collaboration, 2015; Höller et al., 2017; Ioannidis, 2005; Pavlov et al., 2021). Meanwhile, major societal concerns—including climate change, global inequality, growing obscurantism and escalating military conflicts—threaten research by constraining material resources and political priorities (Rae et al., 2022; Urai & Kelly, 2023). Taken together, these pressures highlight the need to take stock and reflect on our ethical standards and implement best practices for sustainable research involving human participants.

Grand ambitions inevitably bring significant challenges and responsibilities that must be navigated to ensure both scientific progress and societal benefit (Niso, Krol, et al., 2022). For instance, even the most advanced brain imaging methods cannot answer all questions in cognitive science (Forest 2013, 2022) and EEG of course also has its specific limits. In clinical contexts, while EEG might be able to inform diagnosis and guide treatment, it should never displace the primacy of human caregivers. Beyond research and medicine, the broader application of EEG technology introduces pressing ethical and practical dilemmas. Scaling EEG for mass-market use would demand high-volume manufacturing, amplifying environmental burdens. At the same time, over-dependence on EEG-based tools risks fostering cognitive complacency—outsourcing functions we should cultivate, rather than merely augmenting them.

This Manifesto

Ultimately, a manifesto’s value is measured not by its wording or by how many people sign it, but by the concrete actions it inspires. We urge you to consider these proposals in the context of your own work and to discuss them with your peers. Where these commitments resonate, incorporate them into your lab procedures and/or daily practice and share them with your colleagues. The extent of your individual commitment will naturally vary according to your interests, expertise, and available resources—but each positive step can help shape a more robust, responsible, and inclusive future for EEG science.

Although this manifesto centres on EEG, we expect that its core themes might resonate across neuroimaging, and perhaps even science at large. A key driving force behind the new potential and renewed excitement of EEG comes from the increase in the availability and robustness of portable systems (Hinrichs et al., 2020; Marini et al., 2019; Niso et al., 2023; Radüntz, 2018). Colleagues working with Functional Near Infrared Spectroscopy (fNIRS) (e.g. Burns et al., 2019) and perhaps soon magnetoencephalography (MEG) (e.g. Boto et al., 2017; Schofield et al., 2024) might also find themselves in a similar situation and may find these reflections relevant to their own fields. They are welcome to join this initiative or may prefer to develop similar approaches within their communities.

---

Validity and scientific integrity in times of rapidly evolving practices

Increased computational power and technology-assisted data mining allows massively testing data on each and every possible time x electrode x frequency (...) combination. Along with more and more flexible tools allowing to easily redefine statistical tests at the click of the mouse, these practice changes make it a real challenge to maintain valid standards and compromise the soundness and reproducibility of research (Onciul et al., 2025; Surianarayanan et al., 2023). As human neuroscience increasingly integrates AI and machine learning approaches for data processing and analyses (CITE), we have to be critical about the epistemological assumptions that underlie these technologies. Below, we advocate agreed-upon standard practices using open science tools and resources grounded in theoretically elaborated research.

Research practices and reproducible¹ science

Research practices are fundamental to scientific inquiry, and their proper application ensures reliable, reproducible, and openly extensible science.

1.1. Reproducible EEG science

Over the last 15 years, neuroscience has reckoned with how questionable research and publication practices inflate false positive rates and erode trust in scientific findings (Nosek et al., 2012). Community-driven efforts have highlighted specific challenges and offered some solutions (Niso, Botvinik-Nezer, et al., 2022; Niso, Krol, et al., 2022; Pernet et al., 2020; Simmons et al., 2012), general manifestos (Munafò et al., 2017) and pledges (see Commitment to Research Transparency). Despite the creation of national and international networks to initiate and support relevant communities (e.g. UK Reproducibility Network; ARIADNE; the Turing Way) fully reproducible science is far from being the norm in empirical science in general (Lakens et al., 2024) or EEG research in particular.

To make progress, the adoption of standardized protocols for data collection, processing, and statistical analysis is essential for ensuring robust, replicable findings in EEG. At the same time, keeping high standards and re-shaping publication practices—supporting non-profit and scientifically sound publishing initiatives (e.g. open access nonprofit platforms like Peer Community In), while avoiding opaque or predatory outlets—is critical to maintain the overall quality and credibility of EEG science.

Pledges

I commit to increasing robustness, replication, and standardization at each stage of my EEG work—from data collection to analysis to publishing results and data (see https://oreoni.github.io/). In doing so, I will consider:

• pre-registering my EEG studies, using dedicated platforms (e.g. OSF, Aspredicted, ClinicalTrials) and reporting templates (Govaart et al., 2022; Paul et al., 2021) where appropriate.
• clearly and systematically communicating all decisions made during data processing—adhering to guidelines such as COBIDAS and ARTEM-IS, recognizing that transparent reporting is critical to the integrity and interpretability of EEG research.
• sharing the code and analysis pipelines that can straightforwardly reproduce my original results.
• publicly releasing negative results.
• sharing raw data, employing general purpose solutions (e.g. gin.g-node, datalad for datasets) and EEG specific tools and standards (e.g., BIDS-EEG, ARTEM-IS see Niso, Krol, et al., 2022). See also Section 1.2: Standardization and Documentation of Data.

I commit to improve systemic reproducibility by:

• supporting initiatives focused on reproducing results from published articles (e.g. through initiatives like #EEGManyLabs, EEGManyPipelines, EEGManyAnalysts, TMS-EEG T4TE) either by participating in replication efforts or by supporting and communicating with researchers who are working to replicate my published results.
• improving the reproducibility and robustness of my own work by testing the effects of different recording setups and analysis pipelines (e.g. through specialist scripts, for example (Es et al., 2024) including “multiverse analysis” (Aczel et al., 2021; Del Giudice & Gangestad, 2021; Steegen et al., 2016), see example tools; boba, multiverse, multifear, multitool, comet, nipype, shiny app, Gorgolewski et al., 2011) and/or by validating findings across multiple recording devices and environments.
• supporting non-profit, scientifically rigorous publishing initiatives, and to refrain from participating in opaque or predatory publishing practices, whether as an author, reviewer, or editor.
• when acting as a reviewer, requesting that (where appropriate) authors follow standard, robust and replicable research protocols and make any pre-registered hypotheses, datasets and analysis scripts publicly available during the submission process to avoid access issues post-publication

1.2. Standardization and Documentation of Data

Generating large, annotated, and openly accessible EEG data repositories, including clinical and neuroimaging data, is crucial for advancing both basic and applied neuroscience. This facilitates discovery science, improves diagnostic applications, industry applications and educational settings, promoting learning and skill development among students and early-career researchers. By pooling resources and sharing data, we enable researchers to perform large-scale analyses that are otherwise impractical for individual laboratories. Standardising datasets, using the Brain Imaging Data Structure (BIDS), is a crucial stage that fosters FAIR science (Wilkinson et al., 2016). To promote transparency, we suggest sharing not only raw data but also analysis scripts and detailed metadata. Platforms such as OpenNeuro, g-node GIN, OSF, or Zenodo can be used to facilitate this process while ensuring adherence to FAIR principles. Such repositories provide diverse datasets that can help uncover new patterns, validate findings across different populations, and improve the generalizability of results.

Moreover, open data repositories are invaluable for developing and refining diagnostic applications. They provide the necessary volume and variety of data required to train machine learning algorithms and develop robust biomarkers for neurological and psychiatric conditions.

Pledges

I commit to making sure that the data I collect and work with uses best practices in data standardization and documentation. I will do so by:

• sharing curated EEG data as openly as possible (i.e. as institute agreements allow) along with detailed metadata, annotations, and tags, using well-established open repositories.
• using the FAIR principles for sharing data (Wilkinson et al., 2016, data should be Findable, Accessible, Interoperable, and Reusable).
• using standards such as the Brain Imaging Data Structure (BIDS) for my data to ensure interoperability and ease of use and sharing it as openly as institute agreements allow/institutional repositories.
• regularly update and enhance my shared data based on user feedback.

1.3. Open Science and Open Source

Using transparent approaches, making sure that results and data are accessible is an essential feature of science. Open science is a broad movement in contemporary science that promotes open access to all aspects of the scientific process. Provided that participants' privacy is protected, we generally advocate for open science practices.

Of particular interest to the EEG community, it is important to consider that EEG data requires elaborate software to perform more and more complex analyses. Open-Source (OS) software and hardware have become vital components in advancing EEG research. By supporting and contributing to Free and Open-Source Software (FOSS) and Open-Source Hardware (OSH), we encourage transparency, collaboration, and innovation within the community. Open-Source tools allow researchers to inspect, modify, and improve code and designs, leading to more robust and reliable scientific instruments and analyses. At the same time, OS fosters the development and sharing of cutting-edge methods and makes it easy to attribute developers for their contributions. This collective effort enhances the reproducibility of our work, accelerates scientific progress, and promotes a culture of openness and mutual support in the EEG community. Furthermore, open-source and open-sciences initiatives lower barriers to entry for researchers worldwide, especially those in resource-limited settings. By making software, hardware and knowledge freely available, we democratize access to cutting-edge tools, promoting inclusivity and diversity in EEG science (also see Democratization section below).

At the same time, high standards and best practice in developing and using OS scientific software must be maintained to harvest its full potential (see Westner et al., 2024)

Pledges

I commit to supporting open scientific practices to support transparency, collaboration, and innovation, in relation to source code in particular. I will do so by:

• integrating open-source tools, data management practices, and transparent workflows into my research whenever possible.
• citing open-source software properly, including specific version tags or DOIs, to 1) give full credit to all developers’ contributions and 2) enable full replication of analysis pipelines.
• following best practices in software use—such as version control, clear documentation, and code review—to enhance reliability and reproducibility.
• using open licenses for code and tools that I develop (e.g., BSD, MIT, GPL), creating readable documentation, and publishing my code on accessible platforms (e.g., GitHub, GitLab, OSF).
• contributing back, when I benefit from others’ tools, if I have the skills or resources—whether through bug reports, testing, documentation, tutorials, or feature development.
• recognizing and valuing software development as a genuine scientific contribution; where possible, I will encourage formal acknowledgment of developers’ work (e.g., in publications and grants).
• promoting open-source solutions in my institution, professional networks, and collaborations by highlighting their benefits for transparency, reproducibility, and inclusivity.
• encouraging administrators, funding bodies, and reviewers to recognize and reward open-science practices and open-source development as key indicators of scientific excellence.
• supporting, where relevant, mentor colleagues or trainees in adopting open-source tools and methods, lowering barriers to entry for all researchers—including those in resource-limited settings.

Interpretable and theoretically well-grounded research

1.4. Theory-driven and data-driven research

Recent years have seen an increasing emphasis on ever-larger datasets with insufficient attention to the theoretical underpinnings that might explain brain activity and its emergent properties (Borsboom et al., 2021; Eronen & Bringmann, 2021; Gerstner et al., 2012).

From a philosophical or epistemological standpoint, it is vital to recognize the limits of any single perspective. A purely data-driven understanding of neuroscience—particularly one focused on individual-brain analysis—can be limited in scope (Forest 2016). We therefore advocate for cultivating strong theoretical neuroscience foundations and for interdisciplinary exchanges that draw insights from fields such as philosophy, anthropology, psychology, the arts, and the humanities. These cross-pollinations could yield innovative frameworks that enrich our interpretations of EEG data.

In addition, the considerations outlined here link closely with our broader calls for epistemological rigor (see also Section 3). Effective science requires not just the “how” of sound methods and robust protocols, but also the “why”: the political, social, and ethical relevance behind our research questions.

By grounding EEG research in a well-articulated theoretical framework and seeking contributions from diverse disciplines, we can enhance the explanatory power of our findings, enrich interpretation, and promote more meaningful progress in understanding the human brain.

Pledges

I commit to balancing theory-driven and data-driven approaches in my research. I do so by:

• engaging in the development of strong preliminary theoretical reflections on the neurophysiological basis of the expected EEG results.
• adapting my experimental practices when performing data-driven experiments. I systematically adapt thresholds for multiple comparison correction, and validate conclusions with sound methodology.
• refraining from using theory-blind approaches to scientific investigation prone to questionable research practices—especially those consuming large computational resources—without a prior theoretical rationale (cf. pledges 3.4 on environmental responsibility).
• reading synthesis texts from other disciplines (e.g., philosophy, anthropology, history, political science). This is a worthwhile investment, helping align my research objectives with societal realities.
• embracing diverse perspectives e.g. integrating insights from the arts and humanities into my research and actively collaborating in interdisciplinary teams and consortia to foster innovation, cross-pollination of ideas, and a deeper theoretical grounding for EEG science.
• training students to engage with theoretical reflections involving interdisciplinary approaches.

1.5. Transparent, interpretable, and explainable modeling

A key challenge emerging for complex computational models—particularly those applied to data analysis and clinical decision-making—is ensuring their transparency, interpretability, and explainability. Modern ML/AI architectures often surpass our theoretical grasp (Holzinger et al., 2022), creating an epistemic gap where we rely on tools we cannot fully comprehend (Cichy & Kaiser, 2019). To address this, the EEG community must dedicate basic research and advanced computational tools (see the Local Interpretable Model-Agnostic or LIME approach as one such framework (Ribeiro et al., 2016) to uncovering how these models operate, while simultaneously leveraging their potential to advance knowledge and applications.

Fostering interpretability and explainability requires both a solid theoretical foundation and active collaboration across disciplines. The application of cross-validation with independent and diverse datasets will help mitigate overfitting and bias, increasing the likelihood that conclusions drawn from ML/AI models are reliable. We hence advocate for i) diverse and open datasets that will make sure AI models will be trained on neurophysiological data acquired from a wide range of population, cognitive states, and environmental contexts; ii) interpretable AI models that prioritise explainable AI/ML approaches rather than using black boxes; iii) integrative validation methods to acknowledge and incorporate in our models the fact that cognition cannot be meaningfully reduced to algorithmic patterns without considering the broader, complex organism-environment system.

Thus, by openly sharing parameters, feature-selection methods, and decision-making processes, we can build stronger trust within our community, mitigate algorithmic reductionism, and accelerate meaningful progress in ML/AI-driven EEG research.

Pledges

I commit to using transparent, interpretable and explainable modeling. I will do so by:

• supporting the development of more transparent, explainable, and interoperable computational models and tools (e.g., LIME), including the use of independent databases for cross-validation.
• publicly disclosing ML/AI scripts and documenting all relevant decisions in my model designs—including feature-selection methods, parameter settings, and validation steps—to foster transparency and interpretability in ML/AI-driven neuroscience.
• promoting best practices for AI transparency when training students and early career researchers, reviewing grants and evaluating manuscripts.
• contributing to counteracting the risks of algorithmic reductionism, which mistakenly reduces cognition to an isolated signal that can be decoded.

Democratization: the importance of diversity and inclusion to support the development of EEG science

A key challenge for science is to ensure that research primarily benefits everyone rather than a privileged few, and that it is never weaponized to cause intentional harm (Carroll et al., 2020). Achieving this goal requires democratizing both research outcomes and the people involved—whether they are scientists, participants, or patients. This means creating an environment where research tools, data, and findings are readily accessible and representative of all, free from gender and geopolitical discrimination, and open to international collaborations. It also entails extending opportunities to institutions and researchers in regions experiencing conflict or crises, especially where advanced platforms for basic and clinical research are present.

Diversity in the scientific community is very important, we must include individuals from different backgrounds, disciplines, and perspectives to facilitate epistemological diversity (i.e., Who does the science?, La Scala et al., 2022; Parada et al., 2024), methodological innovation (i.e., How do we do science?, Choy et al., 2022), and the mitigation of risks such as groupthink (Szanto, 2016). Diversity in research samples is key to produce generalizable and ethical scientific findings. Biased participant selection leads to misleading conclusions that, in the worst scenario, might reinforce structural inequalities and outdated beliefs. It is essential to ensure inclusivity in participants, be it age, gender, cultural background, socioeconomic status, and more. This will improve the validity, replicability, and applicability of our research (Tzovara et al., 2021). Furthermore, diversity in applications and outreach will make our research accessible, relevant, and beneficial to a broader range of communities. Scientific knowledge should be open, but also presented in ways that facilitate the engagement of diverse audiences. If our research cannot be read, understood, and used by non-scientific audiences, does it really exist? Scientific communities should avoid exclusionary language, consider the needs of underrepresented groups, and actively involve other actors such as artists, designers, journalists, and beyond to create bidirectional links involving the communities impacted by our research (Ishaq et al., 2021).

2.1. Diversity in the scientific community

Despite gradual progress, workplace discrimination remains a major issue in science (Woolston, 2021). Enhancing diversity—whether in terms of gender or cultural background—can be both challenging and transformative. Research shows that inclusive teams promote creativity and may even yield stronger scientific outcomes (Achieving Diversity in Research, 2020; AlShebli et al., 2018; Powell, 2018). Building such teams means not only nurturing an inclusive lab culture, but also ensuring conference participation and networking opportunities reflect a broad range of voices (Corneyllie et al., 2024).

At the same time, well-intentioned diversity initiatives can unintentionally widen global disparities. When institutions and labs in wealthier countries recruit talented researchers from low- and middle-income regions, they may exacerbate “brain drain,” weakening the scientific capacity of those researchers’ home institutions. The pledges below seek to balance benefits for both parties, encouraging ethically responsible collaboration and mindful engagement with underrepresented communities.

Pledges

I commit to support increasing the diversity of my scientific community. I will do so by:

• continually learning how to recognize and challenge systemic inequities and proactively addressing social dynamics that may exclude or marginalize colleagues.
• cultivating a welcoming and inclusive environment where all team members—irrespective of their professional status, neurodiversity, physical abilities, gender, or ethnicity—feel supported and represented.
• following the TRUST code - a global code of conduct for equitable research partnerships - engaging fairly with researchers with fewer resources for research, adapting research methodologies to be culturally appropriate and respectful, ensuring studies are designed with input from the communities involved, avoiding “ethics dumping” (Schroeder et al., 2021), providing training and appropriate authorship credit to collaborators.
• promoting a more democratic global academic sector by advocating for funding and ethics-review criteria and engaging in practices that expand opportunities for underrepresented and underserved groups—including but not limited to those from low- and middle-income countries, rural areas, indigenous communities, and regions affected by conflict.
• I commit to engaging in practices such as distributed and online conferences that foster accessible academic networking and prevent barriers to participation (visa, time commitment, costs, etc.)
• engaging in practices such as distributed and online conferences that foster accessible academic networking and prevent barriers to participation (visa, time commitment, costs, etc.)

2.2. Diversity in study participants and sample populations

Mining homogenous, diversity-less datasets -whether it is demographics, experimental paradigms, recording environments- might amplify any existing biases in those datasets. The field of neuroscience has long relied disproportionately on individuals from Western, Educated, Industrialized, Rich, and Democratic populations (Caspar, 2024; Dotson & Duarte, 2020; Henrich, 2020; Muthukrishna et al., 2020). This overreliance narrows the scope of our findings, undermining the goal of identifying universal principles of brain function. Moreover, systematic exclusion of people —whether due to technical constraints in EEG caps (Webb et al., 2022) (Girolamo et al., 2022) entrenched lab practices, or broader structural racism (Choy et al., 2022; Parker & Ricard, 2022)—can foster distrust among underrepresented communities and limit the generalizability of research. Crucially, participant diversity should be embraced as a scientifically valuable feature rather than dismissed as “noise.” Recent evidence shows that more inclusive recruitment is not only necessary but also feasible in varied environments, including rural settings in different countries (Caspar, 2024; Pech et al., 2024), highlighting the potential for creating truly representative and equitable research populations.

When AI models are trained on datasets that do not reflect the full variability of human cognitive and physiological states, they will encode a narrow, static representation of cognition–one that is detached from the ecological, cultural, and biological diversity shaping real-world neurobehavioral dynamics.

Pledges

I commit to support including more diverse populations in study samples. I will do this by:

• attempting to recruit as diverse samples of experiment participants as possible given the study goals, with special attention to gender and ethnoracial representation
• training myself and my team to include everyone, regardless of the hair styles and to never use hair styles or other physical aspects as a direct exclusion criteria.
• engaging with minority communities actively
• paying attention to cultural sensitivities, addressing stigma in brain research², and histories of marginalization
• adapting my experimental equipment (EEG head caps…) to account for potential individual differences, obtaining informed consent in ways that are understandable and meaningful to participants, respecting local customs, languages and ethics
• including a Constraints of Generalizability (CoG) section to provide a clearer understanding of the findings’ generalizability, which is crucial for interpretation and replication (Caspar, 2024, Simons et al., 2017)
• exploring and sharing qualitative epistemology to establish barriers and facilitators to taking part in under-represented groups
• using or developing templates that ensure consistent reporting of demographic information and keep records of sample diversity across studies to assess progress and identify areas for improvement.

2.3. Accessibility of research outcomes

Ensuring that research outcomes—whether new knowledge or potential applications—remain accessible to both the scientific community and the general public is a core ethical responsibility for public funded research. The rise of the open science movement has substantially advanced this goal, aided by open-source software and hardware, DIY (“do-it-yourself”) materials, open knowledge platforms (e.g., Wikipedia), open data repositories, open-access publications, and a wealth of instructional content on video platforms. Underpinned by legal frameworks (e.g., copyleft) and data use agreements that prevent misappropriation or misuse, these initiatives foster transparency and collaboration within clear regulatory boundaries.

However, meaningful access must also address the inverse care law, by which those who most need healthcare—including improved brain health—are often least likely to receive it (World Health Statistics). Expanding access to research tools, knowledge, and resources can help reduce inequalities in care, as well as bolster participant diversity (see pledges 2.2) and promote equitable cross-community collaborations (see pledges 2.1). By prioritizing widespread availability, we can ensure that EEG research benefits the broadest spectrum of society, particularly those who stand to gain the most.

Pledges

I commit to promoting a wide use of my research for the scientific community, and beyond, if applicable. I will do so by:

• using open and transparent approaches (e.g., preprints, preregistration) responsibly, and to withdrawing deposits that do not yield validated findings or conclusive outcomes.
• developing publicly accessible summaries or visuals of my findings to make the results (and their limitations) understandable and beneficial to broader, non-specialist audiences.
• using research results to inform policies and investments that target health inequalities and under-resourced areas, ensuring that EEG applications reach communities where they are needed most.

Responsibility: considering societal impacts, issues of equity, and sustainability.

EEG's non-invasive nature makes it a seemingly harmless technology, but beneath this benign facade lie complex questions that warrant careful consideration. The steady progress in neurotechnology makes it important to carefully examine the implications of foreseeable uses. Further, although technological advancement itself continually unlocks new possibilities, this very fact simultaneously shapes how we understand and inquire about the world, and should not obscure the original purpose of science.

EEG research typically uses relatively few consumables and less resource-intensive electronics than many other domains, making it seem innocuous by comparison. However, the growing affordability of EEG devices and their expansion into consumer markets could turn this advantage into a challenge if we overlook its broader impact and possible rebound effects. We share the responsibility of examining these wider implications—beyond our immediate scientific interests and practices—to encourage responsible use.

Societal and technological responsibility

Applications of EEG research are potentially far reaching, and potential misuses should not be overlooked.

3.1. Ethical use of EEG

Conversations around ethical accountability for scientific discoveries and their real-world applications have a long history, extending beyond the scientific community itself. Each research field faces distinct risks and benefits, and EEG is no exception. In particular, EEG has been explored for sensitive or potentially sensitive uses, such as detecting “concealed information” (Farwell, 2012; Rosenfeld, 2005; Wolpe et al., 2005), enabling implicit interaction outside the participant’s awareness (Fairclough, 2017; Krol et al., 2020; Zander et al., 2016), and launching so-called “side-channel attacks” that decode private data like PIN codes (Lange et al., 2018). Together, these applications demonstrate the capacity of EEG to decode—and possibly influence—a person’s brain activity, heightening concerns about individual privacy (Mecacci & Haselager, 2019), mental integrity (Ienca & Andorno, 2017), and broader social equity (Hyman, 2011).

Researchers must carefully evaluate their work’s potential societal impacts, including how future developments may build on today’s findings. Although many EEG studies cite adherence to the Declaration of Helsinki, that document largely focuses on medical research, leaving gaps for more general EEG practices. Current legal discussions (e.g. Muhl, 2024; Yuste, 2023) and best-practice recommendations (Niso, Krol, et al., 2022) underscore the importance of addressing EEG-specific ethical considerations. Meanwhile, the increasing integration of ML/AI in EEG (e.g., automated signal classification, neural decoding) amplifies risks of inaccurate interpretations, overgeneralizations, and bias (Birhane et al., 2024; Bolte & van Wynsberghe, 2024). Beyond the Declaration of Helsinki, therefore, researchers and developers of EEG-based neurotechnology should look both at more generic frameworks concerning the use of personal data, automated decision-making, and AI, and at more specific guidelines concerning aspects directly related to neurotechnology, like neurorights. The former can guide the researcher, for example, in specifying verifiable quality criteria ahead of time and ensuring proper oversight, whereas the latter can aid in the evaluation of specific solutions, among other things.

A variety of organisations offer further guidance, both generic and specific. UNESCO's reports on neurotechnology ethics and the OECD’s recommendations on responsible innovation in neurotechnology provide foundational strategies to safeguard privacy and autonomy, including in areas such as brain fingerprinting or generative AI (Knoechel et al., 2024). The NeuroRights Foundation likewise advocates treating neural data as highly sensitive and recommends technical protections like encryption and differential privacy to avert misuse (Yuste, 2023). Taken together, these measures illustrate the growing consensus that EEG researchers must proactively address both the near-term ethical challenges and the longer-term implications of their work.

Pledges

I commit to carefully consider the ethical implications of my research practice, and the data that I collect. I will do so by:

• respecting each individual’s right to self-determination and always prioritize the welfare of participants and (future) users over personal, scientific, or commercial interests.
• adhering to the CARE framework (Collective Benefit, Authority to Control, Responsibility, Ethics) (Carroll et al., 2020), integrating these values in all relevant processes and collaborations.
• conducting thorough ethical evaluations before pursuing sensitive EEG applications (e.g., lie or intention detection), ensuring they are backed by sound scientific evidence and embedded in an up-to-date discussion on ethical, legal, and societal implications.
• refraining from marketing or deploying neurotechnological products or services unless they are supported by robust, peer-reviewed evidence of safety and efficacy.
• making responsible and ethical use of generative AI in accordance with UNESCO guidelines, remaining vigilant about potential misuse.
• educating colleagues and trainees about risks in hardware, software, modeling, and data interpretation, advocating for safe, fair, and sustainable neurotechnology.
• disclosing and minimizing any conflicts of interest, and will not endorse or profit from EEG products lacking evidence-based grounding.
• not collecting EEG or neural data without explicit consent, nor pass them on to third parties or merge them with other data sources without proper authorization.
• not contributing to the development of, or market any, neurotechnology whose main purpose is to circumvent a user’s control over their own information.
• making no scientific claims without reasonable empirical support, and will acknowledge uncertainty when disseminating EEG findings.

3.2. Overestimating technology

New research technologies and analysis methods can correct shortcomings of previous tools, enable new observations and crucially progress our understanding of the brain; hence, tool development is a cornerstone of contemporary neuroscience.

However, there can also be hype dynamics, and possibly cruel optimism (Berlant, 2011), as our fascination with advanced, cutting-edge instrumentation can make projects that feature such tools seem inherently more compelling and worthwhile (Nielsen & Andersen, 2022). Similar to the current fascination of the general public with neuroscientific explanations (Weisberg, 2008) and brain imaging (McCabe & Castel, 2018), we should also see the risks of such biases, though - hypes fade, and only time will tell how sustainable a progress was made.

New tools should be developed, and when they are they must be validated and tested out. The governing thought when deciding for a particular tool in empirical research, however, should be scientific progress rather than mere aesthetics or business.

Blindly adapting new tools and abandoning established techniques might even have negative consequences: For example, powerful ML tools can progress a relative devaluation of theory (see Section 1). Focusing on new recording devices may counteract the importance of fair career development and cultivating expertise in data quality control and open science. Furthermore, adopting new tools is resource intensive, which might reproduce global inequalities (see Democratization) and create problems in terms of ecological sustainability (see following Section).

Given its relatively low-tech nature, EEG research can particularly benefit from easy-to-repair, shareable, or open hardware, if we as a field resist this urge. By prioritizing resilience over novelty—such as defining long-term sustainability plans, pooling resource-intensive tools, and collaborating with hardware manufacturers—we can strike a better balance between investing in human capital and in equipment. This strategy not only encourages more equitable and innovative research but can also contribute to ensuring a sustainable future for EEG science.

Pledges

I commit to critically evaluate new technologies through the lens of their contribution to science. I will do so by:

• explicitly weighing investments in new technology against the development of in-house expertise, fair compensation, and the effective use of the technologies already in place.
• developing long-term operational and sustainability plans for the technological systems I am responsible for
• critically assessing potential benefits of new technologies against costs
• considering old methods as an asset that can be exploited, rather than replaced
• always looking at papers that did it in the 60s
• valuing simplicity over complexity
• never applying new tools without validation (and benchmarking)
• moderating my use, design and the conclusions obtained with more complex analyses
• developing projects that incorporate other disciplinary approaches.
• supporting the development and adoption of easy-to-repair, modular hardware with replaceable components, alongside open-source tools that adhere to open data and hardware standards, fostering greater accessibility, interoperability, and sustainability.
• fostering collaboration by sharing high-cost EEG tools, across labs and institutions, reducing waste and increasing efficiency in research.

3.3. Research evaluation

There are growing concerns about the criteria used to evaluate scientific projects and the promises made in relation to them. Many assessment bodies still rely on narrow metrics—such as journal impact factor or h-index—that fail to capture the breadth and nuance of academic careers. This focus discourages interdisciplinary, collaborative, and open-science approaches.

Adopting broader research evaluation procedures and metrics, as recommended by the Coalition for Advancing Research Assessment (CoARA), is key to creating a more open, equitable, and ethically grounded research culture. Criteria that value transparency, data sharing, and open-source development can strengthen the robustness, reliability, and social relevance of scientific findings. Embracing these more comprehensive measures can help foster integrity, encourage collaboration, and ultimately benefit both the scientific community and society at large.

Pledges

I commit to participate in changing research evaluation to avoid focusing on single criteria. I will contribute to broadening research evaluation by:

• promoting evaluation processes that credit a wide range of research contributions—such as datasets, software, outreach activities, and open-access publications—rather than focusing solely on narrow bibliometric indicators.
• encouraging assessment criteria that prioritize open practices (data sharing, preregistration, code release) and value multidisciplinary teamwork
• advocating for and implementing assessment frameworks (e.g., those proposed by CoARA) that emphasize integrity, inclusivity, and social impact alongside scientific excellence.
• continually reviewing and refining evaluation metrics within my sphere of influence—whether as a grant reviewer, hiring committee member, or collaborator—and to pushing for broader acceptance of open, ethical, and diversified research practices.

Environmental responsibility

3.4 Reconsidering Environmental Impact

The scientific consensus is clear: human activity, including scientific research, profoundly affects Earth’s climate and resources—posing a serious threat to humanity’s future. Scientists, positioned at the forefront of knowledge, have a duty not just to document this crisis but also to alter their research practices accordingly (Aron et al., 2020; Racimo et al., 2022; Rae et al., 2022; Urai & Kelly, 2023). Although EEG and related neuroscience fields may appear less resource-intensive than many other domains, we are still part of a broader infrastructure that depends on finite materials, substantial energy, and complex global supply chains.

Some lines of EEG research offer potential long-term benefits for human health and well-being, but they can also require high-performance computing, extensive travel, or large-scale distribution of equipment (e.g. in population neuroscience). Pursuing such work without critically evaluating, or even considering, environmental costs risks perpetuating a “do it because we can” mentality and may be untenable in the long run. Wider collaborations and resource sharing can lead to synergy effects that reduce the ecological footprint of large scale projects. Moreover, systemic pressures in academia, such as publication demands and grant requirements, may push researchers to pursue expansions in scope without equivalent scrutiny of environmental impact. Achieving genuine sustainability will require confronting these norms, ensuring that EEG studies are not only scientifically rigorous but also responsive to the realities of climate change and resource scarcity.

Pledges

I commit to consider the environmental impact of my lab and my research. I will do this by:

• advocating for ethical regulations and policies that require serious consideration of environmental impact—especially for high-resource or energy-intensive studies.
• weighing potential short- and long-term environmental costs against any projected benefits when starting or collecting funds for a project.
• remaining mindful of how academia’s incentive structures can undermine sustainability goals. I will strive to foster discussions, both within my lab and in broader networks, about rethinking these systemic norms.
• prioritizing or designing research programs that have a clearly defined path toward tangible, long-term societal or environmental benefits rather than short-lived or insular academic gains.
• not considering EEG or my specific field as inherently exempt from sustainability concerns. My research is part of a larger, resource-intensive system; I acknowledge my role within that system and strive to minimize its negative impacts.
• cultivating a “climate handshake,” using my position to influence others positively—whether that means introducing sustainable measures in my department, mentoring students on responsible research practices, or partnering with like-minded groups to effect broader change.

3.5 Measuring and reducing the environmental footprint of our work

While EEG itself generally has a moderate ecological impact, the datasets involved can be very large, and the required storage and computational power used to support EEG sciences continue to expand. Moreover, we need to account for the broader environmental footprint of our professional activities (e.g., conference travel, institutional purchases, and overall infrastructure; see Souter et al., 2024, for similar initiative in fMRI).

A key first step is to quantify this impact (see tools examples below). With that knowledge in hand, researchers can identify effective measures to reduce negative outcomes while preserving core scientific aims. However, rebound effects—where well-intentioned actions produce unintended harm—must be carefully monitored to ensure that measures to reduce our carbon footprint do not inadvertently lead to other adverse consequences

Pledges

I pledge to minimize the environmental impact of my research through sustainable practices across multiple domains. I will do this by:

• using responsible data management, such as archiving datasets efficiently to reduce redundancy and prioritizing cloud-based services powered by renewable energy.
• advocating for virtual conferences, workshops, and meetings to reduce travel-related emissions while promoting hybrid formats for inclusivity and accessibility.
• reserving air travel for young researchers who need it the most to launch their careers and build their network.
• using energy-efficient algorithms, optimizing computational methods to reduce runtime and energy consumption, and scheduling resource-intensive analyses during off-peak energy hours where feasible.
• evaluating the environmental impact of my laboratory and research projects by quantifying emissions, energy usage, and waste production through available tools, and use these evaluations to guide more sustainable practices in daily operations and long-term planning.
• critically evaluating my own research projects and considering renouncing to run a project if it does not lead to foreseeable future benefits.

3.6 Using research resources, equipment and data thoughtfully

While non-invasive EEG often appears low-impact compared to more resource-intensive methods, individual projects can still consume substantial time, funding, and materials—particularly if their design is inefficient or redundant. Consequently, reviewing existing datasets and clarifying genuine methodological advances before embarking on new recordings are essential steps to ensure our collective use of participant time and laboratory resources remains both ethical and environmentally sound. By being judicious about data collection, we both respect our participants and optimize the use of laboratory, computational, and environmental resources.

Further, the production and disposal of scientific equipment contribute significantly to pollution and waste. Encouraging equipment pooling, and sharing, or applying the 3Rs approach (“Reduce, Reuse, Recycle” Marques & Fritzen Gomes, 2020) in the lab can help reduce the demand for new devices and extend the lifespan of existing ones. By reducing resource usage and waste production, we not only lessen our environmental impact but also encourage a more collaborative and cost-effective research environment.

Likewise, our field generates vast volumes of data, and we bear a responsibility to make sure they are fully utilized. Meta-analyses—where multiple teams aggregate already-collected datasets—also decrease resource usage, strengthen collaboration, and foster more reliable findings (Koile & Cristia, 2021). To facilitate sharing and reuse, it is essential to standardize data acquisition and analysis and to manage data according to the FAIR principles (Wilkinson et al., 2016), ensuring that it remains Findable, Accessible, Interoperable, and Reusable (See also Section 1.2).

Pledges

I commit to foster a thoughtful use of resources, equipment and data, by:

• reviewing existing datasets and reusing previously collected or open data whenever feasible and only collecting new EEG data if it provides novel insights, significantly improves upon current methodologies, or addresses a well-defined research gap—thereby respecting participants’ time, minimizing resource use, and fostering more sustainable, ethical EEG science.
• contributing to collaborations across labs to share facilities and optimize resource utilization - from headsets to high performance clusters.
• establishing and participating in equipment-sharing networks or consortia that facilitate access to specialized instruments.
• advocating for manufacturers to support sustainable practices by providing long-term support, repair services, and recycling programs for their equipment.
• selecting EEG equipment based on durability and ability to repair, promoting practices that extend the life of equipment, such as regular maintenance, repairs, and upgrades, rather than immediate replacement.

Conclusion

Outline

The challenges outlined here underscore that EEG science must evolve responsibly, balancing innovation with ethical, social, and environmental considerations. As individuals and as a community, we share the duty to anticipate potential misuses, curb unsustainable practices, and embrace virtuous approaches. Yet even with the best intentions, unintended or adverse outcomes—sometimes called rebound effects—remain a constant possibility. Staying alert to these risks, seeking continual improvement, and adapting based on new insights are all crucial steps in ensuring that our efforts genuinely ‘do good.’

A significant part of meeting these responsibilities lies in embracing Open Science and FAIR principles. By pre-registering studies, openly sharing data and code, and using transparent publication outlets, we can foster reproducibility, collaboration, and equitable access to scientific knowledge. This inclusivity will strengthen the entire field, broadening impact and accelerating discovery.

Finally, we believe that fostering an international, ethical, and collaborative approach through science is essential to promoting cross-cultural exchange and contributing to human well-being and unity. This text is an initiative reflecting that commitment within the field of EEG. By signing onto this manifesto and adopting its principles in your daily practice, you contribute to a future where EEG research is not only more robust and sustainable but also firmly rooted in ethical and equitable engagement. Each pledge is a step toward maximizing EEG’s societal benefits—ensuring it advances knowledge, improves well-being, and respects the planet and its people.

Sign the Pledge

First name(s)*

Last name(s)*

Affiliation*

Email*

Afghanistan Albania Algeria American Samoa Andorra Angola Anguilla Antarctica Antigua and Barbuda Argentina Armenia Aruba Australia Austria Azerbaijan Bahamas Bahrain Bangladesh Barbados Belarus Belgium Belize Benin Bermuda Bhutan Bolivia Bosnia and Herzegovina Botswana Bouvet Island Brazil British Indian Ocean Territory Brunei Darussalam Bulgaria Burkina Faso Burundi Cambodia Cameroon Canada Cape Verde Cayman Islands Central African Republic Chad Chile China Christmas Island Cocos (Keeling) Islands Colombia Comoros Congo Congo Cook Islands Costa Rica Côte d'Ivoire Croatia Cuba Cyprus Czechia Denmark Djibouti Dominica Dominican Republic Ecuador Egypt Salvador Equatorial Guinea Eritrea Estonia Ethiopia Falkland Islands (Malvinas) Faroe Islands Fiji Finland France French Guiana French Polynesia French Southern Territories Gabon The Gambia Georgia Germany Ghana Gibraltar Greece Greenland Grenada Guadeloupe Guam Guatemala Guinea Guinea-Bissau Guyana Haiti Heard Island and McDonald Islands Holy See (Vatican City State) Honduras Hong Kong Hungary Iceland India Indonesia Iran Iraq Ireland Israel Italy Jamaica Japan Jordan Kazakhstan Kenya Kiribati North Korea Korea, Republic of Kuwait Kyrgyzstan Lao People's Democratic Republic Latvia Lebanon Lesotho Liberia Libya Liechtenstein Lithuania Luxembourg Macao Madagascar Malawi Malaysia Maldives Mali Malta Marshall Islands Martinique Mauritania Mauritius Mayotte Mexico Micronesia, Federated States of Moldova, Republic of Monaco Mongolia Montserrat Morocco Mozambique Myanmar Namibia Nauru Nepal The Netherlands New Caledonia New Zealand Nicaragua Niger Nigeria Niue Norfolk Island North Macedonia Northern Mariana Islands Norway Oman Pakistan Palau Palestine Panama Papua New Guinea Paraguay Peru Philippines Pitcairn Islands Poland Portugal Puerto Rico Qatar Reunion Romania Russia Rwanda Saint Helena Saint Kitts and Nevis Saint Lucia Saint Pierre and Miquelon Saint Vincent and the Grenadines Samoa San Marino Sao Tome and Principe Saudi Arabia Senegal Seychelles Sierra Leone Singapore Slovakia Slovenia Solomon Islands Somalia South Africa South Georgia and the South Sandwich Islands Spain Sri Lanka Sudan Suriname Svalbard and Jan Mayen Eswatini Sweden Switzerland Syrian Arab Republic Taiwan Tajikistan Tanzania Thailand Timor-Leste Togo Tokelau Tonga Trinidad and Tobago Tunisia Turkey Turkmenistan Turks and Caicos Islands Tuvalu Uganda Ukraine UAE UK United States United States Minor Outlying Islands Uruguay Uzbekistan Vanuatu Venezuela Vietnam Virgin Islands, British Virgin Islands, U.S. Wallis and Futuna Western Sahara Yemen Zambia Zimbabwe Aland Islands Bonaire, Sint Eustatius and Saba Curaçao Guernsey Isle of Man Jersey Montenegro Saint Barthélemy Saint Martin (French part) Serbia Sint Maarten (Dutch part) South Sudan Kosovo Other Country of Residence*

Male Female Other Prefer not to say Gender

Emeritus Full professor Faculty/Staff Scientist Lab Manager PostDoc Researcher PhD Completed PhD Student Student Company/Start-up Executive R&D Scientist Other Prefer not to say Career Stage

Afghanistan Albania Algeria American Samoa Andorra Angola Anguilla Antarctica Antigua and Barbuda Argentina Armenia Aruba Australia Austria Azerbaijan Bahamas Bahrain Bangladesh Barbados Belarus Belgium Belize Benin Bermuda Bhutan Bolivia Bosnia and Herzegovina Botswana Bouvet Island Brazil British Indian Ocean Territory Brunei Darussalam Bulgaria Burkina Faso Burundi Cambodia Cameroon Canada Cape Verde Cayman Islands Central African Republic Chad Chile China Christmas Island Cocos (Keeling) Islands Colombia Comoros Congo Congo Cook Islands Costa Rica Côte d'Ivoire Croatia Cuba Cyprus Czechia Denmark Djibouti Dominica Dominican Republic Ecuador Egypt Salvador Equatorial Guinea Eritrea Estonia Ethiopia Falkland Islands (Malvinas) Faroe Islands Fiji Finland France French Guiana French Polynesia French Southern Territories Gabon The Gambia Georgia Germany Ghana Gibraltar Greece Greenland Grenada Guadeloupe Guam Guatemala Guinea Guinea-Bissau Guyana Haiti Heard Island and McDonald Islands Holy See (Vatican City State) Honduras Hong Kong Hungary Iceland India Indonesia Iran Iraq Ireland Israel Italy Jamaica Japan Jordan Kazakhstan Kenya Kiribati North Korea Korea, Republic of Kuwait Kyrgyzstan Lao People's Democratic Republic Latvia Lebanon Lesotho Liberia Libya Liechtenstein Lithuania Luxembourg Macao Madagascar Malawi Malaysia Maldives Mali Malta Marshall Islands Martinique Mauritania Mauritius Mayotte Mexico Micronesia, Federated States of Moldova, Republic of Monaco Mongolia Montserrat Morocco Mozambique Myanmar Namibia Nauru Nepal The Netherlands New Caledonia New Zealand Nicaragua Niger Nigeria Niue Norfolk Island North Macedonia Northern Mariana Islands Norway Oman Pakistan Palau Palestine Panama Papua New Guinea Paraguay Peru Philippines Pitcairn Islands Poland Portugal Puerto Rico Qatar Reunion Romania Russia Rwanda Saint Helena Saint Kitts and Nevis Saint Lucia Saint Pierre and Miquelon Saint Vincent and the Grenadines Samoa San Marino Sao Tome and Principe Saudi Arabia Senegal Seychelles Sierra Leone Singapore Slovakia Slovenia Solomon Islands Somalia South Africa South Georgia and the South Sandwich Islands Spain Sri Lanka Sudan Suriname Svalbard and Jan Mayen Eswatini Sweden Switzerland Syrian Arab Republic Taiwan Tajikistan Tanzania Thailand Timor-Leste Togo Tokelau Tonga Trinidad and Tobago Tunisia Turkey Turkmenistan Turks and Caicos Islands Tuvalu Uganda Ukraine UAE UK United States United States Minor Outlying Islands Uruguay Uzbekistan Vanuatu Venezuela Vietnam Virgin Islands, British Virgin Islands, U.S. Wallis and Futuna Western Sahara Yemen Zambia Zimbabwe Aland Islands Bonaire, Sint Eustatius and Saba Curaçao Guernsey Isle of Man Jersey Montenegro Saint Barthélemy Saint Martin (French part) Serbia Sint Maarten (Dutch part) South Sudan Kosovo Other Country of Origin

Age

ORCiD

Comment

I'm happy for my name to be displayed publicy

---

1. ‘Reproducible’ is used here as an umbrella term, encompassing all aspects of recreating scientific results (aka replicable, generalisable, robust, ..) as described in Niso, Botvinik-Nezer, et al., 2022; Niso, Krol, et al., 2022b ↩

2. Stigma around brain research participation can affect a person's willingness to participate. People may be afraid of public stigma or their own internalized stereotypes about a condition, which can make them less likely to learn about research studies. Also, people may be afraid of discrimination and stigmatization, which can make them less likely to engage with healthcare services or research institutes. ↩

Unfold all Fold all Jump to Sign