The Future of Thought: Neuralink and Brain Computer Interfaces
Daniel Wunschel
Illustrations by JD Jarolimek
You have a seemingly endless list of work to complete: an essay to write, a presentation to make, and a book to read, but you haven't even started. Lying in bed all day, you hope your work will get done with no physical effort. But what if you could write your essay, complete your presentation, and read your book by just thinking about it? You stare at the ceiling and think about turning on the lights, and suddenly, your room brightens. On your way to the kitchen, you think about turning on the coffee machine, which subsequently stirs to life. With a cup of coffee in hand, you sit down, open your laptop, and begin to think about writing your essay as the words you envision fill the screen. While this may seem like a world of science fiction, certain biomedical companies are attempting to bring these ideas closer to reality by developing brain-computer interfaces (BCIs). BCIs are hardware and software systems that capture the user’s brain activity and translate it into commands that can control devices such as a computer cursor, robotic limb, or electric wheelchair [1, 2]. Neuralink is a company that intends to create a commercialized, surgically implanted BCI that can translate thoughts into actions [3]. The main stated goals of the company are to use their device to restore mobility in paralyzed individuals via controlled prosthetics, improve communication for non-verbal people, treat neurological conditions, and enhance cognitive functions such as memory. But while Neuralink’s proposal sounds revolutionary, the mechanisms of Neuralink’s technology are ambiguous and pose substantial ethical issues [3, 4].
Thought into Action: How BCIs Work
BCIs work by measuring the electrical activity of neurons in the brain [2]. Neurons play a crucial role in communication within the brain and can either be on, actively transmitting signals, or off, at rest and transmitting no signals [5]. While at rest, neurons maintain a constant voltage known as the resting membrane potential [6]. The voltage of a membrane is determined by the concentration of charged particles, called ions, inside the neuron compared to the ion concentration directly outside the neuron. Neurons generally maintain a negative resting potential, meaning there are more positively charged ions outside the neuron than inside. When a neuron receives a signal from another neuron, channels open that allow ions to flow in and out of the neuron, changing the neuron’s voltage. The flow of many positive ions into the neuron can trigger an action potential, a rapid increase in the cell's voltage that allows for communication between neurons. Every bodily function or activity results in specific neurons firing action potentials, which a sensor can record as ‘spikes’ to create a unique pattern [6, 7]. For example, moving your fingers to type on a keyboard generates a unique set of spikes [7]. Interestingly, performing an action and thinking about performing that same action produces comparable spike patterns [8]. BCIs record these spikes and convert the data into information that an external computer can process [9]. The external computer can then control connected devices, such as prosthetics, based on the specific spikes [9].
A multi-step process allows BCIs to record spikes and subsequently control devices [10]. As an example, someone who is paralyzed in all four limbs and would like to use a BCI to control their wheelchair would first have their brain activity recorded to determine the spiking pattern associated with wheelchair movement. Most BCIs record spikes using techniques such as electroencephalography (EEG), which monitors neuronal electrical activity via sensors placed on the scalp [2, 10]. The EEG would record the activity as the person imagines rolling their wheelchair back and forth, resulting in a spike pattern. Since any and all movements generate spikes, subtle actions like blinking or coughing create irrelevant signals — known as ‘noise’ — which have to be filtered out to focus on the activity of interest [11]. Artificial intelligence software helps to discard and filter out the noise and extract the relevant signal to be analyzed [10]. Next, processing algorithms pair the spikes to their corresponding action. The BCI’s refinement of data can be likened to editing an essay; you start with a rough draft, then remove any unnecessary words or phrases to clarify meaning, culminating in a final essay that is ready to submit. Once refined, the data would be sent from the connected computer to the wheelchair, activating a motor to move the wheelchair in the desired direction [2, 10, 12]. Since spike patterns are unique to each action and person, BCIs must be calibrated specifically to the user [13, 14]. Calibration typically involves exercises such as moving an on-screen cursor, moving one’s arm, or merely imagining such movements, allowing the BCI to recognize their specific spike patterns and link them to corresponding actions [14]. The calibration process can be taxing for users, as sessions can last between 15 to 30 minutes, involve a varying number of activities depending on the user, and may require up to 80 trials per activity [15]. While BCIs aim to assist individuals with disabilities, for some people, the calibration process can ultimately detract from their treatment rather than aid it due to the mental energy required [16]. Despite their drawbacks, BCIs have successfully enabled people to control cursors, robotic arms, and wheelchairs, offering greater independence and improved quality of life for those with disabilities [17].
Taking a Step Further: How Does Neuralink Work?
Most BCIs are non-invasive, meaning they are not surgically implanted into the brain [18]. In contrast, Neuralink claims that their device, ‘the Link,’ detects spikes more accurately than other BCIs because it is directly implanted in the brain [19]. Non-invasive EEG-based devices are less sensitive than invasive EEGs, such as the Link, since they sit on the scalp and are physically farther from the neurons generating the spikes [20]. Distance from recorded neurons may make non-invasive devices more prone to contamination of noise, such as electrical signals from the body and even household electronics [21]. Neuralink claims that their device detects spikes more accurately than EEG-based BCIs because of the Link’s direct implantation in the brain [22]. Specifically, the Link is implanted into the sensorimotor cortex — the brain region responsible for controlling movement — so that the device can precisely record spikes related to movement in individuals using the device for motor disabilities or needs [22, 23]. The Link is designed to achieve high precision by inserting an intricate system of threads into the brain [24]. In comparison to the rigid threads found in older BCIs, Neuralink’s threads are upgraded to be flexible, microscopic strands of synthetic material that contain electrodes. The flexible material in the Link’s threads allows them to move with the brain, which prevents damage to the surrounding brain tissue [25]. The electrodes contained in these threads are delicate sensors that record neuronal activity [22, 24]. The invasive design of the Link also allows the implant to record groups of neurons [20]. When Neuralink's threads are implanted, they are placed on a region of interest, such as specific neurons in the sensorimotor cortex [26]. By placing threads directly on top of certain neural regions, the electrodes can record signals from distinct populations and avoid picking up on background noise that EEG-based BCIs encounter. Neuralink can then record voltage changes in targeted neuronal groups and detect the activity of those neurons with higher clarity [20]. Neuralink’s invasive approach enables highly accurate spike recording, which may lead to clearer signals being sent to connected devices [27].
Another important factor that theoretically enhances the Link’s precision is the large number of sensors that the device uses [26]. The Link is equipped with 32 threads, each containing 96 electrodes, totaling 3,072 electrodes [26]. Neuralink’s work reflects a significant increase in electrodes compared to previous BCIs, which had no more than 256 electrodes [28]. A higher electrode count allows for greater precision by detecting more spikes, ultimately improving the device’s functionality [22]. Past BCIs were limited by the difficult and time-consuming process of manually implanting numerous threads into the brain by hand [22]. Neuralink has circumvented such technical constraints by developing a robotic system to insert the threads [26, 29]. Automation of the implantation process makes it possible to insert more electrodes precisely, thus improving the Link’s spike detection and, ultimately, its functionality [22]. While Neuralink's technology already presents significant strides in technological development, the company aims to create a new type of BCI that would restore mobility in users [22]. The Link currently functions as a read-out BCI, meaning it receives and records neural signals that are often used to control a device, as with most BCIs [30]. In comparison, write-in BCIs contain devices that send signals to neural tissue. However, Neuralink’s future goal to eventually restore mobility in users would likely cause the device to act as both a write-in and read-out BCI. Write-in BCIs are widely used as therapeutic devices that work by stimulating the neural tissue in certain brain regions with electricity or light. Cochlear implants are a type of write-in BCI device that functions therapeutically in the ear. A cochlear implant restores some auditory function in individuals with hearing impairments by stimulating auditory nerves [30]. Methods similar to the one utilized by the cochlear implant could theoretically be applied to other losses of function, like movement or vision, by utilizing the Link as a write-in BCI [5, 31].
Sci or Sci-Fi? Neuralink’s Technological Limitations
Neuralink has set ambitious goals but has not yet published a practical explanation of how its goals will be achieved [32, 33]. Although the Link offers a functional replacement for movement, the company also claims that its technology could restore mobility in paralyzed individuals [24]. However, Neuralink does not specify how the technology will accomplish this or acknowledge potential limitations to their goal. For example, injuries such as spinal cord trauma may impede neurons’ ability to successfully generate action potentials, which would hinder the BCI’s effectiveness in restoring mobility. If neurons cannot generate action potentials, the Link would be unable to detect spikes and thus be unable to control an external device. While Neuralink may plan to restore mobility in paralyzed individuals, their technology cannot currently regenerate activity in degraded neurons [24].
In addition to restoring mobility, Neuralink aims to use the Link for nerve and brain stimulation to treat conditions like blindness [5, 31]. Electrically stimulating parts of the visual pathway — such as the retina, optic nerve, thalamus, or visual cortex — may aid in restoring aspects of visual perception. However, because vision processing relies on complex networks across multiple brain regions that develop over time, the efficacy of electrical stimulation is yet to be assessed [5, 31]. Similarly, electrical stimulation of motor-related brain regions may potentially facilitate movement through the reactivation of dormant motor pathways, though such treatment is currently unstudied [30, 34]. While stimulation-related advancements seem promising, the long-term effects of neural stimulation are largely unknown [25, 30, 34]. The Link is a permanent implant, so if it loses functionality after being implanted, one would need to undergo a second surgery to have it removed [36]. However, the removal process carries its own risks. Surrounding tissue that adheres to the device would need to be removed along with the implant, which could potentially cause brain tissue damage and carries a great risk of infection [36]. Since the long-term effects of invasive BCIs are not publicly available or even well-studied, is it worth the surgical and health risks associated with the procedure to get a device that has uncertain durability and longevity [37]? Moreover, despite Neuralink’s claims of starting human trials with three participants, they have not yet published any supporting data in a peer-reviewed scientific journal [32, 33]. Sharing scientific research is crucial to the advancement of science, as it allows others to review and replicate findings as well as conduct further studies [38]. The only peer-reviewed publication — as evaluated by qualified members outside of the original research group — from Neuralink thus far is based on animal experiments [26]. Animal models can be useful in predicting outcomes in humans, but the results are not directly translatable [39]. The lack of peer-reviewed publications makes it difficult for the scientific community to assess the validity of Neuralink’s claims [32, 33]. Someone interested in using the Link may be unable to find trustworthy, peer-reviewed information to corroborate the data on Neuralink’s website [32, 33]. Additionally, the U.S. Food and Drug Administration (FDA) initially rejected Neuralink’s device in 2022, citing concerns about the device’s implantation, potential thread migration to other parts of the brain, and device removal [40]. While the device is now approved for clinical trials, it is unclear what changes were made to address the FDA’s concerns, as the company has not publicly disclosed this information [40].
Privacy and Neuralink: What’s the Link?
In addition to the technological limitations of Neuralink, the company’s ambitious goals are constrained by ethical concerns. One key concern is the protection of personal data [41]. As data privacy becomes increasingly important in safeguarding personal information, protecting neuroscience data — data related to brain activity — has become crucial. Currently, neuroscience data is used to create large datasets that may enable new insights into brain function in both healthy and diseased states. However, as with other types of personal data, issues of privacy arise. Accessing or analyzing neuroscience datasets that contain imaging data, such as those used by Neuralink, may reveal an individual’s cognitive and emotional states, such as moods, feelings, and thought patterns. Furthermore, analysis of neuroscience datasets may be used to predict an individual’s future outcomes, such as the likelihood of developing neurological disorders or even the potential risk of criminal behavior, which may lead to the unwanted disclosure of personal information [41]. Since BCIs record and potentially store neuroscience data, one’s data could be accessed and analyzed by various companies or researchers [42]. Despite the personal nature of neuroscience data, there are relatively few regulations and laws regarding how the data is shared, stored, and used [41, 43]. For example, the California Consumer Privacy Act [44] provides regulations for biometric data, such as fingerprints and DNA, but does not clarify whether neuroscience data is a part of this category [43]. While the CCPA restricts the use of biometric data for identification purposes, it does not impose restrictions on its use to infer an individual’s mental state [43]. The aforementioned example highlights the need for clearer regulations that distinguish neuroscience data from other forms of biometric data and limit how neural data can be utilized, particularly by large corporations.
Letting Neuralink Sink In: What’s Next?
By translating thought into action, BCIs allow people to regain a connection between brain and body, and the Link is one of many BCIs contributing to this restoration [37]. However, unlike other BCIs, the Link’s unique methods and ambitious goals raise serious scientific and ethical concerns [24, 41]. The Link’s potential to revolutionize treatments for various neurological disorders, such as paralysis, could significantly improve the quality of life for millions [24]. However, the complexity of the technology and the risks associated with invasive brain surgery, such as infection or brain damage, cannot be overlooked [37]. Moreover, the company has yet to produce peer-reviewed literature demonstrating their results, making it difficult for others to assess the validity of their claims through reexamination and replication [32, 33]. The lack of published research and adequate regulatory oversight raises serious concerns regarding the efficacy of Neuralink [32, 33]. Thus far, Neuralink’s sample size is very small, consisting of only three individuals who have received the implant [45, 46]. The lack of verified information surrounding the technology may discourage additional participants [33]. Such a limited sample size makes it difficult to generalize the results and draw definite conclusions to further progress toward the commercialized use the company envisions [47]. While the promise of Neuralink is compelling, it requires much more rigorous testing, thoughtful regulation, and broader societal discussion before it can be deemed safe and beneficial for widespread use [24, 48].
References
Vilela, M. & Hochberg, L. R. (2020). Applications of brain-computer interfaces to the control of robotic and prosthetic arms. Handbook of Clinical Neurology, 186, 87-99. doi:10.1016/B978-0-444-63934-9.00008-1.
Mridha, M. F., Das, S. C., Kabir, M. M., Lima, A. A., Islam, Md. R., & Watanobe, Y. (2021). Brain-computer interface: Advancement and challenges. Sensors, 21(17). doi:10.3390/s21175746
Adeoye, S., & Adams, R. (2025). Neuralink’s brain-machine interfaces: A new frontier in healthcare transformation. Cognizance Journal of Multidisciplinary Studies, 5(1), 105–124. doi:10.47760/cognizance.2025.v05i01.009
Dadia, T., & Greenbaum, D. (2019). Neuralink: The ethical ‘rithmatic of reading and writing to the brain. AJOB Neuroscience, 10(4), 187–189.doi:10.1080/21507740.2019.1665129
Zhao, W., Johnston, K. G., Ren, H. (2023). Inferring neuron-neuron communications from single-cell transcriptomics through NeuronChat. Nature Communications, 14. doi:10.1038/s41467-023-36800-w
Drukarch, B., & Wilhelmus, M. M. (2023). Thinking about the action potential: The nerve signal as a window to the physical principles guiding neuronal excitability. Frontiers in Cellular Neuroscience, 17. doi:10.3389/fncel.2023.1232020
Brandman, D. M., Cash, S. S., & Hochberg, L. R. (2017). Review: Human intracortical recording and neural decoding for brain-computer interfaces. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 25(10), 1687–1696. doi:10.1109/TNSRE.2017.2677443
Bonini, L., Rotunno, C., Arcuri, E., & Gallese, V. (2022). Mirror neurons 30 years later: Implications and applications. Trends in Cognitive Sciences, 26(9), 767–781. doi:10.1016/j.tics.2022.06.003
Ajrawi, S., Rao, R., & Sarkar, M. (2021). Cybersecurity in brain-computer interfaces: RFID-based design-theoretical framework. Informatics in Medicine Unlocked, 22. doi:10.1016/j.imu.2020.100489
dos Santos, E. M., Cassani, R., Falk, T. H., & Fraga, F. J. (2020). Improved motor imagery brain-computer interface performance via adaptive modulation filtering and two-stage classification. Biomedical Signal Processing and Control, 57. doi:10.1016/j.bspc.2019.101812
Ahmed, S. F. B., Amin, M. R., & Islam, M. K. (2024). Motion artifact contaminated multichannel EEG dataset. Data in brief, 57. doi:10.1016/j.dib.2024.110994
Gao, X., Wang, Y., Chen, X., & Gao, S. (2021). Interface, interaction, and intelligence in generalized brain–computer interfaces. Trends in Cognitive Sciences, 25(8), 671–684. doi:10.1016/j.tics.2021.04.003
Lotte, F., Bougrain, L., Cichocki, A., Clerc, M., Congedo, M., Rakotomamonjy, A., & Yger, F. (2018). A review of classification algorithms for EEG-based brain–computer interfaces: A 10 year update. Journal of Neural Engineering, 15(3). doi:10.1088/1741-2552/aab2f2
Rao, Z., Zhang, R., He, S., Zhou, Y., Lu, Z., Li, K., & Li, Y. (2025). A once-calibration brain-computer interface to enhance convenience for continuous BCI interventions in stroke patients. IEEE Sensors Journal, 25(2), 3949–3963. doi:10.1109/jsen.2024.3510059
Giles, J., Ang, K. K., Phua, K. S., & Arvaneh, M. (2022). A transfer learning algorithm to reduce brain-computer interface calibration time for long-term users. Frontiers in Neuroergonomics, 3. doi:10.3389/fnrgo.2022.837307
Pasqualotto, E., Matuz, T., Federici, S., Ruf, C. A., Bartl, M., Olivetti Belardinelli, M., Birbaumer, N., & Halder, S. (2015). Usability and workload of access technology for people with severe motor impairment: A comparison of brain-computer interfacing and eye tracking. Neurorehabilitation and Neural Repair, 29(10), 950–957. doi:10.1177/1545968315575611
Maiseli, B., Abdalla, A. T., Massawe, L. V., Mbise, M., Mkocha, K., Nassor, N. A., Ismail, M., Michael, J., & Kimambo, S. (2023). Brain-computer interface: Trend, challenges, and threats. Brain Informatics, 10(1), 20. doi:10.1186/s40708-023-00199-3
Veena, N., & Anitha, N. (2020). A review of non-invasive BCI devices. International Journal of Biomedical Engineering and Technology, 34(3), 205. doi:10.1504/ijbet.2020.111471
Armstrong, W., & Michael, K. (2020). The implications of Neuralink and brain machine interface technologies. 2020 IEEE International Symposium on Technology and Society (ISTAS), 201–203. doi:10.1109/istas50296.2020.9462223
Ramadan, R. A., & Vasilakos, A. V. (2017). Brain computer interface: Control signals review. Neurocomputing, 223, 26–44. doi:10.1016/j.neucom.2016.10.024
Caiado, F., & Ukolov, A. (2025). The history, current state and future possibilities of the non-invasive brain computer interfaces. Medicine in Novel Technology and Devices, 25, 100353. doi:10.1016/j.medntd.2025.100353
Rouanne, V., Costecalde, T., Benabid, A. L. (2022) . Unsupervised adaptation of an ECoG based brain–computer interface using neural correlates of task performance. Scientific Reports, 12. doi:10.1038/s41598-022-25049-w
Umeda, T., Isa, T., & Nishimura, Y. (2022). Temporal dynamics of the sensorimotor convergence underlying voluntary limb movement. Proceedings of the National Academy of Sciences, 119(48). doi:10.1073/pnas.2208353119
Fiani, B., Reardon, T., Ayres, B., Cline, D., & Sitto, S. R. (2021). An examination of prospective uses and future directions of Neuralink: The brain-machine interface. Cureus, 13(3). doi:10.7759/cureus.14192
Yang, D., Tian, G., Chen, J., Liu, Y., Fatima, E., Qiu, J., Malek, N. A., & Qi, D. (2025). Neural electrodes for brain‐computer interface system: From rigid to soft. Biomedical Engineering Materials. doi:10.1002/bmm2.12130
Musk, E. (2019). An integrated brain-machine interface platform with thousands of channels. Journal of Medical Internet Research, 21(10).
doi:10.2196/16194
Pisarchik, A. N., Maksimenko, V. A., & Hramov, A. E. (2019). From novel technology to novel applications: Comment on “An integrated brain-machine interface platform with thousands of channels” by Elon Musk and Neuralink. Journal of Medical Internet Research, 21(10). doi:10.2196/16356
Valle, G. (2019). The connection between the nervous system and machines: Commentary. Journal of Medical Internet Research, 21(11). doi:10.2196/16344
Aler, R., & Galván, I. M. (2015). Optimizing the number of electrodes and spatial filters for brain–computer interfaces by means of an evolutionary multi-objective approach. Expert Systems with Applications, 42(15–16), 6215–6223. doi:10.1016/j.eswa.2015.03.008
Sun, X. Y., & Ye, B. (2023). The functional differentiation of brain–computer interfaces (BCIs) and its ethical implications. Humanities and Social Sciences Communications, 10. doi:10.1057/s41599-023-02419-x
Niketeghad, S., Pouratian, N. (2019). Brain machine interfaces for vision restoration: The current state of cortical visual prosthetics. Neurotherapeutics, 16, 134–143. doi:10.1007/s13311-018-0660-1
Armocida, D., Garbossa, D. & Cofano, F. (2024). Letter: Ethical concerns and scientific communication on neuralink device. Neurosurgical Review, 47. doi:10.1007/s10143-024-02432-x
Waisberg, E., Ong, J. & Lee, A.G. (2024). Ethical considerations of Neuralink and brain-computer interfaces. Annals of Biomedical Engineering, 52, 1937–1939. doi:10.1007/s10439-024-03511-2
Bai, Z., Fong, K. N. K., Zhang, J. J. (2020) . Immediate and long-term effects of BCI-based rehabilitation of the upper extremity after stroke: A systematic review and meta-analysis. Journal of NeuroEngineering and Rehabilitation, 17. doi:10.1186/s12984-020-00686-2
Bjånes, D. A., Kellis, S., Nickl, R., Baker, B., Aflalo, T., Bashford, L., Chivukula, S., Fifer, M. S., Osborn, L. E., Christie, B., Wester, B. A., Celnik, P. A., Kramer, D., Pejsa, K., Crone, N. E., Anderson, W. S., Pouratian, N., Lee, B., Liu, C. Y., Tenore, F., Rieth, L., & Andersen, R. A. (2024). Quantifying physical degradation alongside recording and stimulation performance of 980 intracortical microelectrodes chronically implanted in three humans for 956-2246 days. doi:10.1016/j.actbio.2025.02.030
Leuthardt, E. C., Moran, D. W., & Mullen, T. R. (2021). Defining surgical terminology and risk for brain computer interface technologies. Frontiers in Neuroscience, 15. doi:10.3389/fnins.2021.599549
Maiseli, B., Abdalla, A. T., Massawe, L. V., Mbise, M., Mkocha, K., Nassor, N. A., Ismail, M., Michael, J., & Kimambo, S. (2023). Brain-computer interface: Trend, challenges, and threats. Brain informatics, 10(1). doi:10.1186/s40708-023-00199-3
Hancock, P. A. (2022). Science in peril: The crumbling pillar of peer review. Theoretical Issues in Ergonomics Science, 25(2), 187–191. doi:10.1080/1463922x.2022.2157066
Neziri, S., Köseoğlu, A. E., Deniz Köseoğlu, G., Özgültekin, B., & Özgentürk, N. Ö. (2024). Animal models in neuroscience with alternative approaches: Evolutionary, biomedical, and ethical perspectives. Animal Models and Experimental Medicine, 7(6), 868–880. doi:10.1002/ame2.12487
Parikh, P. M., & Venniyoor, A. (2024). Neuralink and brain-computer interface-exciting times for artificial intelligence. South Asian journal of cancer, 13(1), 63–65. doi:10.1055/s-0043-1774729
Jwa, A. S., & Poldrack, R. A. (2022). Addressing privacy risk in neuroscience data: From data protection to harm prevention. Journal of law and the Biosciences, 9(2). doi:10.1093/jlb/lsac025
Ochang, P., Stahl, B. C., & Eke, D. (2022). The ethical and legal landscape of brain data governance. PLOS One, 17(12). doi:10.1371/journal.pone.0273473
Magee, P., Ienca, M., & Farahany, N. (2024). Beyond neural data: Cognitive biometrics and mental privacy. Neuron, 112(18), 3017–3028. doi:10.1016/j.neuron.2024.09.004
California Consumer Privacy Act of 2018, Cal. Civ. Code § 1798.100 et. seq.
Isbrücker, A. (2024). There’s a basilisk in the bathwater: AI and the apocalyptic imagination. Religions, 15(5). doi:10.3390/rel15050560
Lavazza, A., Balconi, M., Ienca, M., Minerva, F., Pizzetti, F. G., Reichlin, M., Samorè, F., Sironi, V. A., Sosa Navarro, M., & Songhorian, S. (2025). Neuralink’s brain-computer interfaces: Medical innovations and ethical challenges. Frontiers in Human Dynamics, 7. doi:10.3389/fhumd.2025.1553905
Edelman, B. J., Zhang, S., Schalk, G., Brunner, P., Müller-Putz, G., Guan, C., & He, B. (2025). Non-invasive brain-computer interfaces: State of the art and trends. IEEE Reviews in Biomedical Engineering, 18, 26–49. doi:10.1109/rbme.2024.3449790
Maynard, A. D., & Scragg, M. (2019). The ethical and responsible development and application of advanced brain machine interfaces. Journal of Medical Internet Research, 21(10). doi:10.2196/16321