Brain–Computer Interface: Toward an Emancipatory Neurotechnology from the Global South

1. Introduction

Brain–computer interfaces (BCIs) have consolidated as a convergence field in neuroscience, engineering, and machine learning (a process by which machines learn data patterns), translating brain activity into commands for devices or communication signals [1, 2]. This technological field has the potential to transform rehabilitation, assisted communication, and learning by enabling direct interaction between the brain and the digital world. Technical progress in recent decades has reduced classic barriers of control, robustness, and latency (response delay), opening the door to applications outside the laboratory environment [3].

However, this dynamism coexists with a landscape of profound inequality, which the authors call the “technological abyss” in neurotechnology (technologies that interact with the nervous system) [4]. Research, development, patents, and advanced hardware and software infrastructure are overwhelmingly concentrated in a handful of power centers in the Global North (developed countries such as the United States, Europe, and China) [5]. This imbalance is evident when examining quantitative data: a bibliometric analysis (quantitative study of scientific publications) from 2010 to 2021 revealed that only 2.7% of scientific publications in neurotechnology journals come from Latin America [6]. Similarly, private investment in artificial intelligence (AI, systems that simulate human intelligence) across the Latin American region does not exceed 1.8% of US investment or 21% of China’s in 2024 [7]. This pattern extends to the development of software tools, such as CiteSpace II (citation analysis tool) and Statistical Parametric Mapping (SPM, software for statistical parametric mapping), as well as pioneering hardware companies like g.tec medical engineering (Austria) and OpenBCI (USA), which are headquartered in developed countries [8].

This concentration of resources and knowledge creates a landscape of inequality that limits independent validation and replication of studies in low- and middle-income settings [9]. The Global South, therefore, faces a dual challenge: leveraging the opportunity offered by new platforms and models while reconfiguring the rules governing the production, circulation, and use of knowledge. Failure to do so could deepen patterns of dependency or cognitive extractivism (exploitation of mental data without local benefits), where neural data from Global South populations are used to train models that provide no direct or proportional benefit to those communities [10]. This article advances a conceptual, technical, and governance architecture that seeks to make neurotechnology a socially anchored and culturally relevant tool, subject to evaluation under public criteria of efficacy, safety, equity, and sustainability [11].

2. Conceptual framework

To guide the analysis and proposed agenda, a conceptual framework of three axes is used, which transcends the technicist view of BCIs, anchoring them in their socio-political context.

First, cognitive sovereignty is defined as an extension of data sovereignty to the neural realm [12]. This principle establishes that individuals and communities retain rights over the capture, processing, inferences, and subsequent uses of their brain activity [13]. Cognitive sovereignty goes beyond a minimalist understanding of consent, often reduced to a simple checkbox on a form. Instead, it incorporates principles of reciprocity, traceability, and benefit-sharing, fundamental elements when research or service provision is based on public institutions or vulnerable user cohorts [14].

Second, epistemic justice addresses the centrality of datasets and standards built from Global North populations and laboratories [15]. Evidence shows that decoders trained in homogeneous contexts tend to generalize poorly in culturally and physiologically diverse environments, with accuracy losses that entail direct clinical and educational consequences [16]. The response to this problem is not merely technical; it requires a fundamental change in research and innovation design, co-designing protocols and tools with local users, therapists, and educators. In this way, external validity ceases to be an occasional byproduct and becomes an intrinsic design criterion from the start of the process [17].

Finally, the concept of neurodigital complexity underscores that BCIs are not simple artifacts but sociotechnical systems whose real impact depends on “ecologies of use” [18]. This includes a complex network of factors, such as the availability of trained professionals, data infrastructure, institutional organizational culture, and regulatory frameworks, among others [19]. Understanding this complexity is crucial for moving from laboratory demonstrations to sustained technology adoption in the real world. These three concepts (sovereignty, justice, and complexity) act as the analytical keys guiding the critical diagnosis and programmatic agenda developed in this article.

The deep interconnection of these concepts reveals why the technological gap is so harmful. It is not simply a matter of the Global South lacking technology; the existing technology, developed in specific Northern contexts, often functions ineffectively or inequitably for Southern populations [9]. The technical failure of poor decoder generalization thus becomes a matter of social and ethical justice. When the dominant knowledge base (datasets) is not representative, the resulting technology perpetuates existing inequalities. The solution, therefore, does not lie in passive access to technology but in active co-design and localized validation.

3. Methodology

4. Critical diagnosis

The view of technology as inherently neutral has been an obstacle to a justice-oriented perspective in the BCI field. The analysis shows that biases are not limited to the algorithm; they begin in data sampling, continue in signal acquisition, and become rooted in evaluation, especially when laboratory metrics are prioritized over sustained use indicators in real environments [26]. The technological gap is not an accident but a consequence of inherent biases in the ecosystem, perpetuated by the concentration of patents, infrastructure, and talent [27].

The recent decline in open access to AI data repositories (“data commons,” shared data resources) illustrates how the privatization of these resources reduces transparency and raises verification costs, shifting the burden of proof to actors with fewer resources and limiting independent replication in the Global South [28]. This phenomenon is a direct consequence of investment disparity, where large Global North corporations have the capital to acquire and privatize the most valuable datasets. In this context, the hasty implementation of commercial solutions without domain-specific protections (e.g., in education or welfare) creates risks of neuro-privacy and “brainhacking” (brain hacking, unauthorized manipulation of neural data), ranging from sensitive data leaks to attention profile manipulation [29]. Therefore, the field not only needs engineering standards but also clear rules for responsible experimentation and effective redress mechanisms, aligned with best practices in cybersecurity and research ethics [11].

5. Technical state of the art (emphasis: 2023–2025)

The field of noninvasive BCIs has experienced significant advances in recent years, focusing on improving performance, usability, and accessibility. In signal processing, transformer-type models pre-trained for EEG (electroencephalogram: recording of brain electrical activity) have produced brain activity representations that are more invariant to device differences, sampling rates, and noise levels [30]. This improvement facilitates knowledge transfer between subjects and tasks, reducing dependence on extensive and prolonged calibration [16]. In parallel, online adaptation has shown that decoders can be adjusted during use without interrupting user interaction [31]. The effectiveness of machine learning models used in brain–computer interfaces (BCIs) is often limited by the availability of training data [32]. This capability is crucial for scenarios like rehabilitation and classroom learning, where operation time is limited, and continuity is essential for progress [33]. Diffusion-based noise removal (a probabilistic filtering technique) has improved signal-to-noise ratios without erasing relevant neurophysiological components, translating into performance gains for motor, attentional, and affective tasks [34].

Combined with techniques such as neural symbolic regression (a method combining neural networks with symbolic expressions for interpretability) and recurrent networks (models that process temporal sequences), these advances have unified interpretability and dynamic modeling [35]. This is of great value when the goal is not only to classify a mental state but also to sustain continuous control or synthesize speech with clinically significant latencies [36].

Regarding sensors and materials, the development of electrodes designed for prolonged use has stabilized impedance (electrical resistance) and increased comfort – necessary conditions for studies outside the laboratory and everyday applications [37]. Injectable hydrogels (biocompatible gels) and smart gels have expanded inclusion for pediatric populations and users with skin hypersensitivity, while high-density surface EEG has narrowed the gap with invasive recordings for specific use cases [38]. In Latin America, research has predominantly focused on noninvasive and low-cost technologies, such as EEG, which represented 85% to 94.68% of studies in a recent analysis [5]. This focus is due to advantages like low cost, ease of use, and no need for special facilities, unlike more expensive and complex modalities such as functional magnetic resonance imaging (fMRI), positron emission tomography (PET), or magnetoencephalography (MEG: recording of brain magnetic fields), which are less common in the region [39].

This focus on EEG is not a sign of technical inferiority but a pragmatic and situated response to infrastructure and resource limitations. The region is innovating within its possibilities, developing inherently more accessible and scalable solutions. This situated design process fosters an open-source hardware and software ecosystem that reduces operation and maintenance costs, making local manufacturing viable [40].

Current BCI applications show differential maturity. Assisted communication is in the validation phase, with speech neuroprostheses (devices that convert thoughts into voice) beginning to meet clinical utility criteria, both with implants and surface recordings [41]. Rehabilitation programs based on motor imagery (mental visualization of movements) continue to report functional gains and measurable changes in cortical activation [33]. In education, EEG integration with cultural practices like music and rhythm has shown positive effects on attention and fatigue, with implications for curriculum design and formative assessment strategies [42]. Recent advances include Neuralink’s N1 implant, which enables cursor control for games and digital tasks in volunteer trials, and new noninvasive approaches like micro-sensors between hair for continuous monitoring [10, 43].

6. Geopolitics, political economy, and data governance

BCIs have become a terrain of strategic competition [44]. Whoever controls technical standards, data flows, and talent training will establish reference frameworks and capture the surplus. The concentration of patents and capital, dependence on chip and sensor supply chains, and the growing value of neural repositories draw a map of power with persistent asymmetries [4]. Quantitative data reaffirm this reality: joint private investment in AI across all Latin American countries does not exceed 1.8% of US investment or 21% of China’s in 2024, severely limiting the Global South’s capacity to compete in the technological race [7].

In this context, the South’s response cannot be limited to demanding access; it must build institutional alternatives [10]. A layered governance architecture (data, infrastructure, algorithms, and talent) helps delineate responsibilities and control mechanisms [45]. This includes complementing FAIR principles (Findable, Accessible, Interoperable, Reusable) with CARE principles (Collective Benefit, Authority to Control, Responsibility, Ethics) for communities [46]. The agenda is compatible with OECD regulatory regimes and Europe’s digital rights trajectory, and can leverage South–South agreements to reduce risks posed by sanctions and export controls on critical inputs [47].

7. Regulatory frameworks and neuro-rights

Regulatory progress has moved from general principles toward more operational instruments [45]. The constitutional protection of brain information and psychic integrity in Chile has established a regional benchmark, while Colorado’s law on consumer brainwaves has opened the debate on non-clinical commercial uses and the need for domain-specific consent and limits [48]. This Chilean initiative, as the first country in the world to legislate and protect the human mind at a constitutional level, seeks to create a “legal wall” to safeguard psychic integrity, mental privacy, and free will [49]. In collaboration with neuroscientists like Rafael Yuste, five fundamental neuro-rights have been proposed for incorporation into the Universal Declaration of Human Rights: the right to personal identity, the right to free will, the right to mental privacy, the right to equitable access, and the right to protection against biases and discrimination [11]. By 2025, states like California and Montana have followed with similar laws requiring express consent for neural data, and UNESCO has advanced toward a global ethical standard for neurotechnology [50].

This contribution demonstrates that leadership in neurotechnology is not defined solely by technical production. The Chilean initiative is a clear example of how the Global South can lead the ethical debate and the construction of a global legal framework, inverting the traditional dynamic where norms are imposed by the North [51]. The absence of commercial pressure existing in the Global North grants the South a unique opportunity for proactive and preventive governance. The Global South can establish the ethical and legal foundations that will guide technology development worldwide, contributing significantly to the dignity and human rights of future generations [18].

8. Education, literacy, and epistemic change

Without critical education, neurotechnology remains a topic for specialists; with literacy, it can become a care policy [52]. The learning pathway begins in school with basic neuroscience, notions of privacy and biases, and responsible practices with portable devices [53]. It continues in technical education with instrumentation, assembly, and data flow management [51]. The path culminates at the university level with subjects on signals and systems, AI applied to EEG, neuroethics, and digital law [12]. Continuing education for clinicians, therapists, and teachers must integrate applied BCI modules and impact assessment, in line with UNESCO guidance and approaches inspired by FATE (Fairness, Accountability, Transparency, and Ethics) [46]. Evaluation must go beyond “does it work?” to include measures of sustained adoption, satisfaction, and comfort, so that technical success aligns with user experience and clinically and educationally relevant outcomes [54].

9. A sovereign cognitive strategy: Five articulated pillars

The strategy seeks to turn criticism into installed capacity. Pillar 1: Research and data foster regional networks; build benchmarks in motor, speech, and affective domains; and develop repositories with dataset cards, reciprocity-based licenses, and traceability. By prioritizing underrepresented populations and pediatric stages, this strategy improves generalization and corrects biases from the source [38].

Pillar 2: Local design and manufacturing promote open-source hardware and software, low-code libraries for clinical and classroom professionals, auditable firmware, and nearby maintenance services that reduce downtime and operating costs [40]. Pillar 3: Neurodigital training and literacy combine MOOCs (Massive Open Online Courses), intensive courses, and technical programs with trainer training pathways and inclusion objectives [52]. Pillar 4: Regulation and cooperation align standardized impact assessments, risk-class certification authorities, South–South data and technology exchanges, and supervised controlled environments with output metrics and regulatory feedback circuits [45]. Pillar 5: Pilot projects with social return focus on rehabilitation, assisted communication, and EEG-assisted learning, with efficacy evaluation and longitudinal follow-up [50]. The decisive step is converting pilot projects into sustainable services within health and education systems, with stable funding and continuous improvement mechanisms [55].

10. 2025–2030 roadmap and risk management

11. Conclusions

The question underlying this work is not merely technical but profoundly political: What kind of neurotechnology do we want, and for whom? If the Global South limits itself to accepting the terms of the race for patents and closed repositories, BCIs will reproduce hierarchies of extraction and dependency [9]. However, if it advances with an agenda of sovereign data, context-sensitive design, and critical education, the conversation will shift from spectacular demonstrations to services that genuinely improve lives without compromising dignity or cognitive freedom [11]. The Global South has the opportunity not to lead in the technological race, but in building governance frameworks that ensure these advances serve the well-being of all humanity [14].

The 2025–2030 window is narrow but sufficient. There is already accumulated knowledge, an emerging ethical framework, and institutional will to build fairer governance [46]. Success will depend on universities, hospitals, schools, regulators, and communities acting as an interconnected system rather than as isolated entities. The bar for success is demanding and clear: measurable efficacy, verifiable safety, demonstrable equity, and sustained public trust.

Acknowledgments

We gratefully acknowledge the support of our respective universities and the Benchmark Accrediting Solutions Group (BASG). We extend special thanks to Dr. Juan Tapia, President of BASG, for his leadership and encouragement in supporting this work.

Nomenclature and abbreviations

Declaration of no competing interests

Both authors declare that they have no competing interests related to this research. Neither author holds financial or non-financial relationships, advisory roles, equity, patents, royalties, or personal affiliations that could be perceived as influencing the work.

Institutional support from our two universities and from the Benchmark Accrediting Solutions Group (BASG) created no competing interests. BASG had no role in the study design, data collection, analysis, interpretation, manuscript preparation, or the decision to submit. Both authors approved the final manuscript and accept full responsibility for its content.