TL;DR
BCIs are best understood as constrained medical technologies rather than general-purpose enhancement tools… at least for the foreseeable future.
Background
Brain–computer interfaces (BCIs) represent a research area attracting significant attention across both academia and industry. Morgan Stanley’s 2024 report suggests a $400 billion market opportunity in the US alone (Oxley 2024). With hype comes a responsibility for medical practitioners, venture capitalists, and biotech engineers to understand both the genuine therapeutic potential and the fundamental limitations of the technology. The wider public is hearing companies like Neuralink and Australia’s own Synchron set ambitious goals, but how can we properly distinguish science fiction from clinical reality?
As a medical student and neurotechnology researcher, I aim to use my perspective to demystify the world of BCIs.
Invasive vs Non-Invasive
Perhaps the most critical distinction in understanding BCIs is whether a device is invasive — that is, whether it is implanted within the cranium or spinal canal, allowing direct access to neural tissue. This stands in contrast to non-invasive BCIs, which interface with the nervous system via sensors placed on the scalp or skin surface, without breaching the skull or spinal column.
The trade-off is relatively clear. By accessing the central nervous system (CNS) directly, invasive BCIs can record neural signals with higher spatial and temporal resolution, or stimulate grey matter with greater anatomical precision. However, the cost of implantation is high and almost always requires a neurosurgeon to perform an intracranial or intrathecal procedure. Beyond time, cost, and surgeon availability, developers of invasive BCIs must also justify the perioperative and long-term risks of implantation, and identify patients who meet the stringent clinical and ethical criteria required for such interventions.
While Neuralink’s flagship device does require a craniotomy for implantation, Synchron’s design is unique and considered only minimally invasive. Implantation of its device is performed via an endovascular approach, navigating through the vascular system to the brain, which is quicker, less resource-intensive, and can be performed by a broader range of appropriately trained clinicians (Mitchell et al. 2023).
Open vs Closed-Loop
Whether a BCI is designed to stimulate or read from the CNS depends on the intended therapeutic application. Broadly speaking, stimulation involves the delivery of electrical impulses to modulate neural activity, with the goal of inducing cognitive, restorative, or neuromuscular changes. Reading, by contrast, involves the measurement and decoding of the electrophysiological signals generated by neuronal populations.
In the context of BCIs, the terms open-loop and closed-loop describe how these processes are integrated. An open-loop BCI operates without real-time feedback: neural signals may be recorded or stimulation delivered according to pre-set parameters, independent of the brain’s ongoing response. Closed-loop systems, by contrast, continuously monitor neural activity and dynamically adjust stimulation in response, allowing the device to adapt to the user’s current neural state.
Closed-loop BCIs are often portrayed as more advanced, but this sophistication comes at the cost of greater algorithmic complexity. Ensuring that a system responds appropriately across patients and over time remains one of the field’s central technical challenges.
Clinical Indications vs Consumer Aspirations
One of the largest sources of hype distortion surrounding BCIs is a misunderstanding of how and for whom these technologies can currently be deployed. At present, the development and approval of invasive BCIs is effectively constrained to clinical or therapeutic justifications under existing regulatory frameworks. Cyborgs won’t be spawning around us anytime soon.
Regulators such as the FDA (US) and TGA (Australia) require that the risk–benefit ratio of an implanted medical device be clearly favourable. Because invasive BCIs carry non-trivial surgical and long-term risks, approval hinges on demonstrating potential clinical benefit for a defined patient population (e.g. paralysis, epilepsy, movement disorders) (U.S. Food and Drug Administration 2021).
The regulatory picture becomes more ambiguous with non-invasive devices, like EEG or fNIRS based BCIs which can be developed for wellness, research or general consumer use. Provided that no explicit therapeutic claims are made, such devices can avoid classification as regulated medical devices altogether, allowing companies to bypass FDA or TGA oversight entirely.
That said, the eventual commercial and enhancement-oriented potential of BCIs remains a major point of interest for investors and is frequently emphasised in public-facing company narratives. Leading Swiss neurotechnology group Mindmaze Therapeutics, for example, describes its mission as “unlocking the true power of humanity” (MindMaze 2024), while Elon Musk, founder of Neuralink, has speculated about the emergence of “some sort of futuristic cyborg” with “superhuman abilities” within the next 10–15 years (Fridman 2024). How such technologies could be scaled, marketed, or ethically deployed under current regulatory regimes remains unclear.
Patient Lived Experience
For now, companies are focusing on improving the lived experience of those who do have BCI implants. In practice, this means prioritising reliable day-to-day function, minimising cognitive and physical burden, and ensuring that interfaces are usable outside tightly controlled laboratory settings.
Actual users of clinical BCIs report that the lived experience of using these systems requires cognitive effort and psychological negotiation. Users in qualitative interview studies describe BCI operation as something that “challenges common experiences” even as it enables agency and increased social participation (Kögel et al. 2020).
Some participants in broader user studies raise concerns about the burden of device maintenance (repeat appointments, recalibration and the possibility of revision surgery) (Kögel et al. 2019). Another systematic review of patient preferences also notes that training and setup burdens can be significant, with accuracy improvements alone insufficient to justify the effort for many users (Brannigan et al. 2024).
Widespread uptake of BCIs will inevitably be met with reluctance from individuals unwilling to undergo invasive surgery for the implantation of a permanent neural device. Patient wariness and scepticism will be understandably high.
Outcomes: In Theory vs Practice
In theory, brain–computer interfaces promise high-fidelity decoding of neural intent and stable long-term integration with the nervous system. But in practice, the performance of BCIs and error-prone, even under tightly controlled experimental conditions. Neural signals are inherently noisy, non-stationary, and highly individual, meaning that decoding algorithms must continuously recalibrate to shifting signal characteristics rather than rely on fixed mappings between brain activity and device output.
For invasive BCIs in particular, long-term reliability is further constrained by the brain’s biological response to implanted hardware. The CNS’s Foreign body reaction (involving the inflammation, gliosis, and encapsulation of electrodes) can impair signal quality over time, reducing recording fidelity and stimulation precision. While advances in materials science and electrode design aim to mitigate these effects, notably by Stéphanie Lacour’s EPFL-affiliated lab (Lacour, n.d.), the risk cannot be fully eliminated and remains a fundamental limitation of chronic neural implants. How can companies properly quantify the repeat procedures in terms of patient risk and final cost?
Crucially, the locus of cutting-edge BCI development has increasingly shifted from academic laboratories to private industry. While this transition has accelerated engineering progress and access to capital, it has also reduced the openness that historically characterised neuroscience research. Proprietary hardware designs, closed-source decoding algorithms, and confidential clinical data now dominate the field and endangers the transparency of a technology which is, at its core, designed to improve the lives of the neurologically impaired.