Background

For the first few years of my life, I had a minor speech disability classified as “PDD-NOS.” Personally, the disability did not inhibit most activities but made normal communication difficult. I was able to improve my speaking skills after receiving occupational and speech therapy. During my time in therapy, I saw many of my special-ed classmates, some of whom had more severe speech disabilities, struggling to  use their voice to express themselves despite therapy intervention. When I became a student in the neuroscience class in the Academic Talent Development Program (ATDP) at UC Berkeley, I researched how brainwave feedback improves attention-span in children with Aspergers’. Novel research into Brain-Computer Interfaces (BCI) is  a promising field of study. Currently, I am part of the Jadoo Technologies’s Exoskeleton Team at  Khalid Lab, where we are creating a  robot suit for stroke patients that will utilize EEG encoding signals to facilitate interpretation of and interaction with their environment.⁷ The BCIs have the potential to  fill the communication gap between neurodivergent and neurotypical people, especially those with speech disabilities. For many, BCIs may be a better-suited treatment than traditional speech therapy. 

Introduction into BCI and its mechanisms

BCIs have the ability to alleviate severe symptoms associated with certain neurological disorders, merging many fields of science such as programming, neuroengineering, mathematics, and neurosurgery. BCIs connect the central nervous system (CNS) to computers, which connect to primary sensory organs such as the ears and eyes, or primary motor organs like vocal chords, arms, and legs. BCI technology translates  neural signals into speech. First the artificial sensor and controller attached to (where are these controllers and who is in charge of them, the patient or doctor?) take information from nerve signals. Then, the BCI acts as a “decoder,” uncovering one’s intended speech and/or environmental interaction. 

The BCI utilizes a variety of signals such as EEG and fNRIS to perform decoding of brain signals.⁸ Electroencephalography(EEG) encodes brainwaves and trains our brain to identify emotional and social cues. Functional near-infrared spectroscopy (fNIRS) is  used to signal body parts to move. The scalp being thick can also lessen the precision of data transfer. However, EEG captures the local field potential, the mean activity of a group of neurons rather than each signal of the neuron. For fNIRS, the scalp being thick can also lessen the precision of data transfer. These errors in estimating signals could lead to inaccuracies.  To combat these challenges, intracranial electrodes are implanted into the brain, allowing for higher precision given they are closer to neurons and actually touch the brain.. These electrodes can reach the cell bodies of neurons and record action potentials (neuronal spikes). Then, the electrodes act as capacitors, decoding and interpreting, with high precision, as actions or speech.

BCIs are trained on signal processing and machine learning algorithms.⁴ For example, our exoskeleton group uses neural networks and scikit learn. There is a complex process of calibrating, optimizing, and designing various types of BCIs.  Researchers have multiple questions to consider  before a BCI is made and used by patients. How real can it mimic speech and at what speed? Can BCIs adopt different languages, speech patterns, and age ranges? Does the BCI need instructions  or are they already encoded in the brain? How much is the cost for the funding, materials, and packaging? Most importantly,  compatibility—whether it is biocompatibility, safety, or portability—is what sets various BCIs apart and gets them into the medical market. Socioeconomic status and geographic location could affect access to BCI technology.

Real World Applications

In our society, communication is a social and political inclusion tool. The inability for neurodivergent people, specifically those with speech disabilities,  to communicate in a typical manner has resulted in their frequent maltreatment. Those with serious neurological injuries require surgical intervention or necessity to manipulate information. In the BCI market, the Speech Generating Devices (SGD) and prediction/word completion systems have recently been developed.

Efforts to close these communication gaps have been around for quite some time given the importance of this problem. For example, the late renowned physicist Stephen Hawking “used a series of computer programs through which he could operate switches to select phrases from predictive word generating software first using his hand, then a sensor on his cheek.”⁵ In fact, he collaborated with Intel to develop eye tracking, signing, and finger spelling or electroencephalography (EEG) and event-related potentials (ERPs) signals to select phrases. This could be seen as the first version of BCIs!

Implementing BCIs as an aid for bilateral communication, the ability to receive and express cues, allows for individuals to have better sensory-motor skills (sensing and moving different objects),  cognitive  function (abstraction or interpretation of information), and linguistic capabilities (grammar, syntax, and punctuation).  In fact, the communication aspects present in up-and-coming BCI are  being used  in legal settings.  Information by and from the BCI can be used as evidence in a legal trial and as a way for one to communicate their consent and assert their agency as a participant. In 2008, the UN Convention on the Rights of Persons with Disabilities (CRPD)  made it an obligation to use Augmentative and Alternative Communication (AAC) technologies to  ensure more equitable spaces in places like universities, increasing university and social accessibility.⁶ So far, this act protects  disabilities and developmental disorders such as cerebral palsy, severe autism spectrum disorder, intellectual disability, and PDD-NOS.  However, neurological inflictions like traumatic brain injury, high-level spinal cord injury, stroke leading to aphasia, and dysarthria are not included in this act. 

Finally, there is the degenerative phase of clinical BCI, such as amyotrophic lateral sclerosis (ALS) and multiple sclerosis.

Ethical Undertones + Advice/Considerations

Some argue that using BCI technology to restore “normal function” in disabled individuals could take away their unique abilities. To prevent the undertones present in implicit ableism, the BCI needs to maintain the integrity of the brain and provide a sense of autonomy. Generally, people confuse communication and autonomy for decision making. In the case of Byndom v. State (2001), a woman with cerebral palsy used a Dynavox computer to type the words and icons based on “yes” or “no” questions.⁵ She was refused by the defense, who claimed pre-programmed hearsay. In a usual court proceeding, statements are under the oath, which are in real-time. The statement was offered outside of court proceedings and deemed as outsider evidence that cannot be cross-examined by the opposing party. In another example, R v WH, the accused could not have a trial as he had aphasia affecting comprehension, and  communication is seen as the typical way to establish credibility and restate one’s side.⁵  There is a difference between understanding and manipulating vs speaking or communicating. Communication does not relate to the autonomy of the accused person. For example, a person with a disability is capable of understanding the court proceedings, but their mode of expression is not construed as the norm. Often, the silence to court might negatively impact the testimony. Now, courts try to accommodate this by allowing aiding devices, but actually saying it in real time under oath/evidence is still viewed as crucial and the most valuable type of testimony. 

Although BCIs do not require external treatment methods, it is recommended to have therapists, neurologists, psychiatrists, etc. to monitor healthy progress in the long term. If the BCI is invasive, children should fully understand how it works and manipulate their brainwaves. In other words, unless there is an underlying injury or fatality, until frontal lobes still develop to some extent, BCI use should be carefully monitored and used as a last resort. AACs also pick up on implicit or imaginary thoughts that the person might have observed themselves, creating a sense of ambiguity and an invasion of privacy.⁶ There is fear in tracking or hacking brains, especially with monetization of online activity.. The interference of commercial BCIs like Elon Musk’s Neuralink could compromise mental health and, by extension, physical health by warping a sense of autonomy, but with consent, the biodata could improve clinical studies.

References

1. Abigail DeCook, M. (2024, June 28). Setting the record straight on AAC and speech. The Informed SLP. https://www.theinformedslp.com/review/setting-the-record-straight-on-aac-and-speech 
2. Brahambhatt, R. (2024, May 5). Non-invasive brain-computer interface to help control objects by thought. Interesting Engineering. https://interestingengineering.com/science/ai-powered-non-invasive-bci 
3. Brain-implant companies create groups to support nascent industries. (n.d.-b). https://www.bloomberg.com/news/articles/2024-03-11/brain-implant-companies-create-group-to-support-nascent-industry
4. Lebedev, M. A., Shaderkin, I. A., Ryabkov, I. V., & Lebedev, G. S. (1970, January 1). Augmentation through interconnection: Brain-nets and telemedicine. SpringerLink. https://link.springer.com/chapter/10.1007/978-3-030-54564-2_16 
5. Chandler, J. A., Van der Loos, K. I., Boehnke, S., Beaudry, J. S., Buchman, D. Z., & Illes, J. (2022, March 3). Brain Computer Interfaces and Communication Disabilities: Ethical, legal, and social aspects of decoding speech from the brain. Frontiers. https://www.frontiersin.org/journals/human-neuroscience/articles/10.3389/fnhum.2022.841035/full 
6. Blank, R. H. (1970, January 1). United States policy on bcis: Funding research, regulating therapies, and commercializing consumer technology. SpringerLink. https://link.springer.com/chapter/10.1007/978-3-031-26801-4_11 
7. UC Berkeley Nanotechnology Lab. (n.d.). https://nanotech.berkeley.edu/areas 
8. Orban, M., Elsamanty, M., Guo, K., Zhang, S., & Yang, H. (2022, December 5). A review of brain activity and EEG-based brain-computer interfaces for rehabilitation application. Bioengineering (Basel, Switzerland). https://pmc.ncbi.nlm.nih.gov/articles/PMC9774292/