Skip to content

When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.

PLOS BLOGS The Official PLOS Blog

Brain-computer interface restores communication to the completely locked-in

Imagine an existence without the ability to communicate. You cannot speak, gesture or even blink an eye. All means to share your thoughts, emotions and basic needs are entirely lost. For those in the advanced stages of amyotrophic lateral sclerosis (ALS), a debilitating and eventually fatal degenerative motor neuron disease, this nightmare can become reality. At the most severe disease stages when all motor control is lost, the individual enters a completely locked-in state (CLIS), in which they retain full consciousness and cognitive function but are unable to communicate with the outside world. Prior attempts to develop tools to enable communication in those suffering from CLIS have failed, but a new study published in PLOS Biology used a novel brain-computer interface (BCI) that allows, for the first time, accurate communication in CLIS patients.

Probing communication with a fNIRS BCI

Past efforts to communicate with CLIS patients employed BCIs using electrical brain signal from EEG or electrocorticography that depended upon voluntary control by patients over their neural responses. The authors of the new PLOS Biology study speculate that this requirement of at least partially intact somato-motor function for attention and learning may explain the failure of these BCIs in CLIS. They therefore tested whether a BCI based on functional near-infrared spectroscopy (fNIRS), which measures brain oxygenation levels and does not involve voluntary modulation of neuroelectric responses during BCI training, may avoid these pitfalls and be a more effective communication tool.

To test this hypothesis, the team enrolled four advanced-stage ALS patients aged 24-76 who had entered a CLIS (or nearly-CLIS) 3-4 years after diagnosis. The patients had lost all ability to communicate and all motor function including eye movement, and were supported by ventilation and tube feeding. With the exception of one patient whose unique condition (a rare genetic mutation causing 1-2% of ALS cases) prevented her from completing training, all other participants performed a series of 46-60 sessions. During BCI training, patients were asked questions with known answers (i.e., Paris is the capital of France) and were prompted to mentally respond “yes” or “no.” A classifier was trained to recognize fNIRS brain signals associated with responses to positive versus negative questions. Training was followed by feedback sessions, in which patients responded to personal questions (i.e., You were born in Berlin) and were given feedback as to whether the BCI interpreted their response as “yes” or “no.” Finally, open sessions asked questions probing the patient’s quality of life (i.e., You have back pain).

Breaking communication barriers

The change in oxygenation during the response periods significantly differed between positive and negative questions. For comparison, neither the EEG signal nor eye movements reliably distinguished “yes” and “no” conditions, confirming that fNIRS was able to detect brain representations of the patients’ answers not measurable through neuroelectric or motor signals. Accuracy of the fNIRS classifier exceeded chance over 70% of the time, and was significantly more accurate than either the EEG or eye movement classifiers. Furthermore, BCI-determined answers between semantically paired questions were concordant for 67% or more of cases, highlighting consistency of the classifier. Why a change in brain oxygenation better informs about cognitive state than an electrical brain response isn’t entirely clear. A study in animals by the research team showed that increased oxygenation is associated with multiunit neural activity, suggesting that oxygenation measurable with fNIRS may reflect increased neural activation or coherence.

fNIRS signal during response periods for YES and NO questions for each patient, at 20 brain regions (each line depicts a different site). (From Chaudhary et al., 2017)
fNIRS signal during response periods for YES and NO questions for each patient, at 20 brain regions (each line depicts a different site). (From Chaudhary et al., 2017)

The fNIRS classifier was impressively accurate at decoding the patients’ responses, but what might explain its failure at some questions? Looking to the EEG for further insight, the researchers noted differences in middle frequency (theta/alpha) oscillations between the question and response periods, confirming that the patients are in distinct brain states when listening to and answering questions. Critically, theta power–a rhythm that increases during states of low arousal–was reduced when BCI classification of the patients’ responses was successful. This suggests that when low frequency activity dominates, the brain response is more obscure due to the patient’s low vigilance, and the BCI therefore fails to decode the patient’s answer. To ensure more accurate readings, an fNIRS classifier could theoretically be paired with EEG to identify when a patient is sufficiently alert to communicate.

A bright future for the CLIS

Before an fNIRS-based BCI will become commonplace for enabling communication in CLIS patients, there are important practical considerations. An fNIRS-EEG system would be financially costly and would require either extensive training of caretakers or the presence of a trained technician. Study coauthor Niels Birbaumer admits that the BCI may be prohibitively costly, but they recently fought and won for insurance coverage in Germany. He says that although “we have to make it more simple to avoid need of a technician,” with assistance from first author Ujwal Chaudhary, “one husband of a patient is using it already.” Furthermore, the cost/benefit ratio depends on how accurately the BCI can read the mental state of the CLIS individual. Birbaumer shares that they are refining their BCI to decode more complex brain signals, and explains that “EEG and invasive recordings have to be exploited in the future. EEG and fNIRS will be compared and if useful combined, we hope that other laboratories will join in.”

Despite these obstacles to implementing fNIRS for everyday use in non-communicative individuals, these successful case studies offer hope not only for restoring communication in advanced ALS, but also for understanding the cognitive and emotional states of those bound by a CLIS. Although the authors warn that they must “remain cautious about our judgments to open questions’ answers, particularly if it comes to quality of life,” the patients responded overwhelming positively to such questions. Birbaumer is optimistic about the potential of their findings to improve quality of life:

“We hope that the attitude of patients, doctors and families may change and that more patients will decide for life when breathing becomes difficult and invasive ventilation necessary.”


Chaudhary U, Xia B, Silvoni S, Cohen LG, Birbaumer N. (2017). Brain-Computer Interface-Based Communicaiton in the Completely Locked-In State. PLOS Biology. doi: 10.1371/journal.pbio.1002593

Kubler A, Birbaumer N. (2008). Brain-computer interfaces and communication in paralysis: extinction of goal directed thinking in completely paralysed patients? Clin Neurophysiol. 119(11):2658-66. doi: 10.1016/j.clinph.2008.06.019

Zaidi AD et al. (2015). Simultaneous epidural functional near-infrared spectroscopy and cortical electrophysiology as a tool for studying local neurovascular coupling in primates. Neuroimage. 120: 394-9. doi: 10.1016/j.neuroimage.2015.07.019

Image credit

Any views expressed are those of the author, and do not necessarily reflect those of PLOS.

Emilie Reas received her PhD in Neuroscience from UC San Diego, where she used fMRI to study memory. As a postdoc at UCSD, she currently studies how the brain changes with aging and disease. In addition to her tweets for @PLOSNeuro she is @etreas.

Back to top