The Cybersecurity, Privacy, and Ethics of EEG-BCI Systems

The Cybersecurity, Privacy, and Ethics of EEG-based BCI Systems

By Emma Szczesniak

Abstract

Neurotechnology, the integration of technology with the nervous system, is an advancing field with potential medical applications by offering individuals with neuromuscular disorders or physical injuries an alternative means of communication beyond traditional verbal speech. One such advancement is the development of an electroencephalography-based brain-computer interface (EEG-based BCI) system, which functions to record and decode electrical signals from the brain into intelligible speech. While advancements in neurotechnology prove promising for assisting patients, this technology calls into question the ethical implications of cognitive translation that have yet to be explored in depth. As such, relevant publications investigating the efficacy and applicability of EEG-based BCI systems for brain-to-speech communication will be examined to determine the attitude of the scientific field regarding alternative communication technology and to assess the impact of this technology on cognitive privacy.

Understanding EEG-BCI Systems

In broad terms, a brain-computer interface (BCI) is a communication pathway between the brain and a computing device.[1] In technical terms, it is a computer-based system that obtains, processes, and converts brain signals into commands that are then sent to a device to carry out the signaled action. To be considered a BCI, the system must directly acquire brain signals from the central nervous system rather than the traditional output pathways of the peripheral nerves and muscles.[2] BCI systems are comprised of four components that occur in sequence: signal acquisition, feature extraction, feature translation, and device output.[3] In signal acquisition, brain signals are measured using a form of sensor modality. BCIs mainly utilize electrophysiological signals recorded either on the scalp surface by electroencephalography (EEG), beneath the scalp by electrocorticography (ECoG), or within the brain by intracortical microelectrode arrays (MEAs).

EEG is the brain recording technique that has been most widely used to detect patterns of brain activity for use in BCI systems. Alone, an EEG is not a BCI as it records rather than processes brain signals; therefore, an EEG-based BCI system utilizes EEG for signal acquisition in which the collected data is amplified, digitized, and transmitted to a computer.[4] The computational aspect of a BCI system involves signal processing in which algorithms are applied to the acquired brain signals to undergo feature extraction and feature translation. In feature extraction, the digitized brain signals are filtered to obtain the relevant data, while in feature translation, the filtered data is interpreted to formulate specific commands.[5] Most EEG-based BCI systems employ preprocessing techniques for noise reduction, signal enhancement, and artifact removal, which strengthens signal processing. As the final stage of the BCI system, output involves command execution through a device such as a robotic arm or communication aid.[6] In this regard, the user can perform an action while bypassing any required motor movement.

Current Capabilities and Applications

EEG-based BCI systems are a commonly studied technique for decoding brain activity due to their noninvasive procedure. Noninvasive BCIs pose several advantages over invasive BCIs including their simplicity, minimal risk, low cost, ease of use, and high temporal resolution.[7] As they do not require a surgical procedure, the risk of brain damage or scarring is absent. Although limitations of EEG-based BCI systems remain, they are applicable to a variety of fields.

These systems have transformed the medical device industry through their range of applications in assistive devices and disease diagnosis. The main function of EEG-based BCIs in clinical settings has been to assist individuals with damaged cognitive or sensory-motor functions.[8] Through the recording and processing capabilities of these systems, rehabilitation of patients with severe neuromuscular disorders is possible. This technology enables patients to initiate commands to wheelchairs, prostheses, and communication devices to restore mobility lost due to neurological dysfunction, assisting individuals suffering from amyotrophic lateral sclerosis (ALS), spinal cord injuries, stroke, and other neuromuscular conditions.[9] Additionally, EEG-based BCI systems currently play a role in diagnosis and detection of conditions such as brain tumors and epilepsy.The automated signal processing analysis performed by a BCI enables more accurate readings of EEG data than when performed manually, allowing for a more precise diagnosis of a brain abnormality or disease. EEG-based BCI systems have a classification accuracy of 87% for a brain tumor, 93% for epilepsy, and 98% for no abnormality.[10] Although it continues to be enhanced and developed, EEG-based BCI technology has already provided many advancements to diagnostic tests and assistive devices in the field of medicine.

In addition to healthcare, EEG-based BCI systems have applications in neuromarketing, workplace efficiency, and entertainment. Neuromarketing covers a section of market research that examines cognitive responses to products and services.[11] Traditional market research techniques provide inaccurate data on consumer preferences as individuals lack sincerity when completing surveys, interviews, or focus group discussions.[12] Implementation of EEG-based BCI systems aims to overcome conventional marketing constraints and collect accurate consumer data. A study assessing the application of a machine learning framework to classify EEG data in response to consumers’ attitude towards various products and services concluded that the proposed model effectively replicates real-world reported results.[13] This study represents an initial step towards industry use of EEG-based BCI technology for market research.

EEG-based BCI systems can monitor employee productivity and reduce safety risks by measuring cognitive load, attention, and drowsiness states.[14] Alpha and theta bands are amplified when an individual experiences fatigue, thereby acting as an indicator of a person’s arousal state.[15] Additionally, alpha and theta rhythm changes in parietal and frontal brain regions, respectively, signify cognitive overload.[16] Fatigue and cognitive overload lead to decreased performance of complex tasks; therefore, implementation of EEG-based BCI technology to record and process alpha and theta brain waves effectively determines safety risk so that it may be minimized.

The entertainment industry is currently finding ways to implement EEG-based BCI systems into the gaming sector. BCI games utilizing EEG data monitor a player’s enthusiasm level and dynamically alter game difficulty levels, in turn augmenting user engagement.[17] The availability of user-friendly BCI games is growing, but significant development of these games is needed before they surpass standard game control methods. EEG-based BCI technology is present in both the medical and commercial spheres and will continue to impact these areas as the technology develops.

Limitations and Challenges of EEG-Based BCI Systems

Challenges limiting the efficacy and application of EEG-based BCI systems can be categorized into two groups: technical constraints and BCI system usability. EEG-based BCI systems experience reduced reliability in noisy environments due to signal-to-noise ratio reduction, which lends itself to less effective signal acquisition.[18] External noise more significantly limits commercial applications of EEG-based BCI systems as they are typically applied in less controlled environments, leading to greater noise disturbances. EEG-based BCI technology is also limited by the capabilities of computational language processing models.[19] The low efficacy of signal processing can hinder accuracy of command execution, as limitations in machine learning may prevent intended commands from being properly carried out. Additionally, inter-subject variability in the amplitude and character of brain signals limits generalizability of EEG-based BCI systems as signal classification must be adapted to a specific person.[20] Development of a subject-independent system would eliminate challenges in the generalized application of EEG-based BCI systems.

The extent to which EEG-based BCI systems are user-friendly significantly impacts their application in both medical and commercial settings. BCIs require a degree of concentration and awareness that is difficult to maintain over an extended period. An EEG-based BCI system requires a user to focus on an input and output, resulting in a forced signal acquisition dynamic rather than an organic one.[21] Furthermore, the complexity of using an EEG-based BCI system presents a challenge in training users, with the added burden of needing recalibration prior to each session, prolonging training and diminishing concentration and productivity.[22] As these systems must be operable by BCI technology novices, simplifying control mechanisms and reducing training times are essential for achieving proficiency.

The Future of EEG-Based BCI Brain-to-Speech Technology

Current use and development of EEG-based BCIs are a foundation for advanced application of this technology. One area of advancement is the use of EEG-based BCI technology for brain-to-speech communication. Invasive procedures for a cognitive speech interface such as Neuralink’s brain chip, Telepathy, have made significant headway in recent years, encouraging continued advancement of noninvasive methods for alternative communication.[23] Studies have assessed a variety of sensors and data processing methods to enhance capabilities of EEG-based BCIs for brain-to-speech communication. Commonly adopted models employed for communication between a user and device include P300, steady-state visual evoked potentials, and motor imagery.[24] These paradigms are limited by their prolonged response time; therefore, researchers have pivoted to imagined speech for communication between the brain and a device. Imagined speech is the internal production of speech without articulatory movement.[25] Recording imagined speech with EEG reduces response duration, allowing for more efficient communication between the user and device. Signal processing of EEG data has also made progress in accurately classifying words from imagined speech.[26]Additionally, preprocessing has been found to reduce the signal-to-noise ratio and refine data quality for feature extraction and classification, leading to more accurate results.[27] Due to speech production complexity, continued research on machine learning algorithms for imagined speech classification is required.

Collectively, these studies provide insight into the future of EEG-based BCI systems for brain-to-speech communication. Integration of the imagined speech paradigm with enhanced machine learning algorithms will make a more robust cognitive speech interface. This integration promises faster decoding speeds and higher classification accuracy, making EEG-based BCI systems more viable for extended user application, particularly for individuals with speech conditions. As advancements continue, people with severe neuromuscular conditions such as near-complete paralysis, could regain the ability to communicate. Continued development in this domain has the potential to transform interactions for individuals with speech impairments, improving the quality of life of thousands of individuals.

Privacy Implications and Ethical Considerations of a Cognitive Speech Interface

Despite the benefits offered by EEG-based BCI technology, there remains a need to consider the impact of implementing this technology on human rights. Advancement in information and communications technology has raised concerns of digital privacy rights for individuals on a global scale.[28] The development of neurotechnology should raise similar concerns, but with a focus on cognitive privacy. Cognitive privacy is the concept that people should have control over access to their neural data and the information obtained from analyzing that data.[29] Cognitive privacy refers to the protection of neural rights. However, debates continue to question the extent to which an individual’s thoughts are truly their own.

Naturally, humans attempt to interpret other individuals’ thoughts and emotions during interactions through facial responses, body language, and verbal context.[30] Thus, neurotechnology capable of interpreting brain waves as speech is simply an advancement in one’s ability to infer a person’s state of being. Additionally, thoughts developed in the brain are built from a collection of past experiences whether that be from education, relationships, or media, none of which are unique to a person’s cognition. Therefore, the degree to which a thought is unique to an individual must be addressed and explored prior to formulating legislation on cognitive privacy.

A second consideration is the ownership and accessibility of neural data. Without legislation concerning consumer data obtained from neurotechnology products, companies have the right to provide a consumer’s use history to the government, law enforcement, or other companies in industry. In 2017, data from a Fitbit was used as evidence during a criminal investigation to determine a timeline of events more accurately.[31] Data from EEG-based BCI technology would be valuable evidence in court cases. As such, privacy regulations for consumers need to be considered as this technology continues to progress.

Legislation relating to cognitive privacy and neural rights remains limited; however, certain countries and international organizations have taken steps towards enacting regulations on neurotechnology. The first set of international guidelines regarding ethical use of neurotechnology in clinical and research practices was created by The Organization for Economic Cooperation and Development (OECD) in 2019.[32] These guidelines cover prioritizing patients’ well-being, fostering responsible management of neuroscience research, and promoting transparent public engagement. Furthermore, after revising their constitution in 2021, Chili became the first country to constitutionally address the ethical concerns of neurotechnology.[33] The question remains on whether other countries will follow Chili’s example in preemptively addressing privacy rights surrounding developing neurotechnology. Current available EEG-based BCI systems serve as the foundation for advancing neurotechnology systems, thus it is imperative to gain a more comprehensive understanding of the potential impact of these technologies and how to implement them in an ethical manner to safeguard the cognitive privacy of all individuals.

[1] Ruida Zeng, Ajay Bandi, and Abdelaziz Fellah, “Designing a Brain Computer Interface Using EMOTIV Headset and Programming Languages,” 2018 Second International Conference on Computing Methodologies and Communication (ICCMC) 00 (2018): 908–13, https://doi.org/10.1109/iccmc.2018.8487684.

[2] Jerry J. Shih, Dean J. Krusienski, and Jonathan R. Wolpaw, “Brain-Computer Interfaces in Medicine,” Mayo Clinic Proceedings87, no. 3 (2012): 268–79, https://doi.org/10.1016/j.mayocp.2011.12.008.

[3] Shih, Krusienski, and Wolpaw.

[4] Shih, Krusienski, and Wolpaw.

[5] Shih, Krusienski, and Wolpaw.

[6] Uzair Shah et al., “The Role of Artificial Intelligence in Decoding Speech from EEG Signals: A Scoping Review,” Sensors 22, no. 18 (2022): 6975, https://doi.org/10.3390/s22186975.

[7] Zahrah Alwi Alkaff et al., “Applications of Brain Computer Interface in Present Healthcare Setting,” IntechOpen, January 5, 2024, https://doi.org/10.5772/intechopen.112353.

[8] Alkaff et al.

[9] Alkaff et al.

[10] Alkaff et al.

[11] Fazla Rabbi Mashrur et al., “BCI-Based Consumers’ Choice Prediction From EEG Signals: An Intelligent Neuromarketing Framework,” Frontiers in Human Neuroscience 16 (2022): 861270, https://doi.org/10.3389/fnhum.2022.861270.

[12] Mashrur et al.

[13] Mashrur et al.

[14] Kaido Värbu, Naveed Muhammad, and Yar Muhammad, “Past, Present, and Future of EEG-Based BCI Applications,” Sensors22, no. 9 (2021): 3331, https://doi.org/10.3390/s22093331.

[15] Khalida Douibi et al., “Toward EEG-Based BCI Applications for Industry 4.0: Challenges and Possible Applications,” Frontiers in Human Neuroscience 15 (2021): 705064, https://doi.org/10.3389/fnhum.2021.705064.

[16] Douibi et al.

[17] M. F. Mridha et al., “Brain-Computer Interface: Advancement and Challenges,” Sensors 21, no. 17 (2021): 5746, https://doi.org/10.3390/s21175746.

[18] Mridha et al.

[19] Mridha et al.

[20] Mridha et al.

[21] Reza Abiri et al., “A Comprehensive Review of EEG-Based Brain–Computer Interface Paradigms,” Journal of Neural Engineering 16, no. 1 (2019): 011001, https://doi.org/10.1088/1741-2552/aaf12e.

[22] Abiri et al.

[23] Omar Fares, “Neuralink’s Brain Chip Implant Marks New Era in Human-Computer Synergy,” The Conversation, February 4, 2024, https://neurosciencenews.com/neuralink-bci-neuroethics-255555/.

[24] Nicolás Nieto et al., “Thinking out Loud, an Open-Access EEG-Based BCI Dataset for Inner Speech Recognition,” Scientific Data 9, no. 1 (2022): 52, https://doi.org/10.1038/s41597-022-01147-2.

[25] Nieto et al.

[26] Ciaran Cooney et al., “Evaluation of Hyperparameter Optimization in Machine and Deep Learning Methods for Decoding Imagined Speech EEG,” Sensors 20, no. 16 (2020): 4629, https://doi.org/10.3390/s20164629.

[27] Yogesh Paul, Ram Avtar Jaswal, and Sanjay Kajal, “Classification of EEG Based Imagine Speech Using Time Domain Features,” 2018 International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering (ICRIEECE) 00 (2018): 2921–24, https://doi.org/10.1109/icrieece44171.2018.9008572.

[28] Amaris Rancy, “Big Question: How Does Digital Privacy Matter for Democracy and Its Advocates?,” ed. Maya Recanati and Beth Kerley, January 22, 2024, https://www.ned.org/big-question-how-does-digital-privacy-matter-for-democracy-and-its-advocates/.

[29] Abel Wajnerman Paz, “Is Mental Privacy a Component of Personal Identity?,” Frontiers in Human Neuroscience 15 (2021): 773441, https://doi.org/10.3389/fnhum.2021.773441.

[30] Laura Cabrera, “New Neurotechnology Is Blurring the Lines around Mental Privacy – but Are New Human Rights the Answer?,” The Conversation, August 7, 2023, https://theconversation.com/new-neurotechnology-is-blurring-the-lines-around-mental-privacy-but-are-new-human-rights-the-answer-205446.

[31] Christine Hauser, “In Connecticut Murder Case, a Fitbit Is a Silent Witness,” New York Times, April 7, 2017, https://www.nytimes.com/2017/04/27/nyregion/in-connecticut-murder-case-a-fitbit-is-a-silent-witness.html.

[32] Shravishtha Ajaykumar, “Ethics of Neurotechnology The Intersection of Neuroscience and Military Applications.Pdf,” Observer Research Foundation, September 5, 2023, http://20.244.136.131/expert-speak/ethics-of-neurotechnology.

[33] Cabrera, “New Neurotechnology Is Blurring the Lines around Mental Privacy – but Are New Human Rights the Answer?”

Bibliography

Abiri, Reza, Soheil Borhani, Eric W Sellers, Yang Jiang, and Xiaopeng Zhao. “A Comprehensive Review of EEG-Based Brain–Computer Interface Paradigms.” Journal of Neural Engineering 16, no. 1 (2019): 011001. https://doi.org/10.1088/1741-2552/aaf12e.

Ajaykumar, Shravishtha. “Ethics of Neurotechnology The Intersection of Neuroscience and Military Applications.Pdf.” Observer Research Foundation, September 5, 2023. http://20.244.136.131/expert-speak/ethics-of-neurotechnology.

Alkaff, Zahrah Alwi, Nurul Hashimah Ahamed Hassain Malim, Putra Sumari, and Jafri Malin Abdullah. “Applications of Brain Computer Interface in Present Healthcare Setting.” IntechOpen, January 5, 2024. https://doi.org/10.5772/intechopen.112353.

Cabrera, Laura. “New Neurotechnology Is Blurring the Lines around Mental Privacy – but Are New Human Rights the Answer?” The Conversation, August 7, 2023. https://theconversation.com/new-neurotechnology-is-blurring-the-lines-around-mental-privacy-but-are-new-human-rights-the-answer-205446.

Cooney, Ciaran, Attila Korik, Raffaella Folli, and Damien Coyle. “Evaluation of Hyperparameter Optimization in Machine and Deep Learning Methods for Decoding Imagined Speech EEG.” Sensors 20, no. 16 (2020): 4629. https://doi.org/10.3390/s20164629.

Douibi, Khalida, Solène Le Bars, Alice Lemontey, Lipsa Nag, Rodrigo Balp, and Gabrièle Breda. “Toward EEG-Based BCI Applications for Industry 4.0: Challenges and Possible Applications.” Frontiers in Human Neuroscience 15 (2021): 705064. https://doi.org/10.3389/fnhum.2021.705064.

Fares, Omar. “Neuralink’s Brain Chip Implant Marks New Era in Human-Computer Synergy.” The Conversation, February 4, 2024. https://neurosciencenews.com/neuralink-bci-neuroethics-255555/.

Hauser, Christine. “In Connecticut Murder Case, a Fitbit Is a Silent Witness.” New York Times, April 7, 2017. https://www.nytimes.com/2017/04/27/nyregion/in-connecticut-murder-case-a-fitbit-is-a-silent-witness.html.

Mashrur, Fazla Rabbi, Khandoker Mahmudur Rahman, Mohammad Tohidul Islam Miya, Ravi Vaidyanathan, Syed Ferhat Anwar, Farhana Sarker, and Khondaker A. Mamun. “BCI-Based Consumers’ Choice Prediction From EEG Signals: An Intelligent Neuromarketing Framework.” Frontiers in Human Neuroscience 16 (2022): 861270. https://doi.org/10.3389/fnhum.2022.861270.

Mridha, M. F., Sujoy Chandra Das, Muhammad Mohsin Kabir, Aklima Akter Lima, Md. Rashedul Islam, and Yutaka Watanobe. “Brain-Computer Interface: Advancement and Challenges.” Sensors 21, no. 17 (2021): 5746. https://doi.org/10.3390/s21175746.

Nieto, Nicolás, Victoria Peterson, Hugo Leonardo Rufiner, Juan Esteban Kamienkowski, and Ruben Spies. “Thinking out Loud, an Open-Access EEG-Based BCI Dataset for Inner Speech Recognition.” Scientific Data 9, no. 1 (2022): 52. https://doi.org/10.1038/s41597-022-01147-2.

Paul, Yogesh, Ram Avtar Jaswal, and Sanjay Kajal. “Classification of EEG Based Imagine Speech Using Time Domain Features.” 2018 International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering (ICRIEECE) 00 (2018): 2921–24. https://doi.org/10.1109/icrieece44171.2018.9008572.

Paz, Abel Wajnerman. “Is Mental Privacy a Component of Personal Identity?” Frontiers in Human Neuroscience 15 (2021): 773441. https://doi.org/10.3389/fnhum.2021.773441.

Rancy, Amaris. “Big Question: How Does Digital Privacy Matter for Democracy and Its Advocates?” Edited by Maya Recanati and Beth Kerley, January 22, 2024. https://www.ned.org/big-question-how-does-digital-privacy-matter-for-democracy-and-its-advocates/.

Shah, Uzair, Mahmood Alzubaidi, Farida Mohsen, Alaa Abd-Alrazaq, Tanvir Alam, and Mowafa Househ. “The Role of Artificial Intelligence in Decoding Speech from EEG Signals: A Scoping Review.” Sensors 22, no. 18 (2022): 6975. https://doi.org/10.3390/s22186975.

Shih, Jerry J., Dean J. Krusienski, and Jonathan R. Wolpaw. “Brain-Computer Interfaces in Medicine.” Mayo Clinic Proceedings 87, no. 3 (2012): 268–79. https://doi.org/10.1016/j.mayocp.2011.12.008.

Värbu, Kaido, Naveed Muhammad, and Yar Muhammad. “Past, Present, and Future of EEG-Based BCI Applications.” Sensors 22, no. 9 (2021): 3331. https://doi.org/10.3390/s22093331.

Zeng, Ruida, Ajay Bandi, and Abdelaziz Fellah. “Designing a Brain Computer Interface Using EMOTIV Headset and Programming Languages.” 2018 Second International Conference on Computing Methodologies and Communication (ICCMC) 00 (2018): 908–13. https://doi.org/10.1109/iccmc.2018.8487684.