Decoding Thoughts: Non-Invasive Brain-Machine Interface Translates Mental Activity into Text
Researchers at the University of Texas at Austin have made a significant breakthrough in the field of brain decoding technology. Using a combination of functional magnetic resonance imaging (fMRI) scanners and artificial intelligence (AI), they have developed a non-invasive method to translate thoughts into text. Unlike previous attempts, this approach holds promise for patients who are unable to communicate due to conditions such as paralysis. The technology requires participants to undergo AI training while inside the scanner, and the AI system can then approximate their thoughts in text form.
The researchers behind this technology believe it is a major advancement in non-invasive methods for decoding thoughts. Previous methods typically focused on single words or short sentences, but this new approach enables the decoding of continuous language and complex ideas for extended periods of time.
To train the AI system, participants spend several hours in the fMRI scanner listening to podcasts. This training allows the AI to recognize the specific fMRI signatures associated with certain thoughts. Once the training is complete, participants can choose to have their thoughts “read” by the system while sitting in the scanner, and the technology will analyze their brain activity to approximate the conveyed thoughts.
It is important to note that the system does not provide an exact word-for-word transcript of thoughts but rather offers an approximation. For example, the original thought “I don’t have my driver’s license yet” might be translated as “She has not even started to learn to drive yet.”
Considering concerns about privacy and the potential misuse of this technology, the researchers have taken precautions. The system cannot decode thoughts from unwilling participants, as individuals must actively participate in training the AI. Additionally, the system can be easily confused if individuals intentionally think about unrelated topics, adding a layer of protection against unwarranted intrusions.
The researchers acknowledge the need to address ethical considerations and ensure that such technologies are used responsibly. Their intention is for people to have control over the use of these technologies and to use them only when they willingly choose to do so, ensuring that the technology serves as a beneficial tool for individuals.
In the future, the system may not exclusively rely on fMRI scanners but could also be compatible with more portable imaging modalities. Functional near-infrared spectroscopy (fNIRS) is mentioned as a potential alternative. This technology measures changes in blood flow in the brain, similar to fMRI, and may offer a lower resolution but still a viable option for decoding thoughts.
The study detailing this breakthrough, titled “Semantic reconstruction of continuous language from non-invasive brain recordings,” was published in the journal Nature Neuroscience. The research paves the way for further advancements in non-invasive brain-machine interfaces, potentially enhancing communication and improving the quality of life for individuals who face severe communication challenges.
Study in journal Nature Neuroscience: Semantic reconstruction of continuous language from non-invasive brain recordings https://www.nature.com/articles/s41593-023-01304-9
Zen KOH
Founder & Executive Director, MotusAcademy
Incoming President, IISART
Co-Founder & Global CEO, Fourier Intelligence
Zen is a visionary and pioneering influencer in the field of MedTech and Robotics for Rehabilitation. He has been involved in several successful start-ups and businesses in Singapore, Switzerland, and China for over two decades, providing medical devices, healthcare solutions, and services for people with disabilities and neurological patients. He was nominated as one of the 40 under 40 most influential industry leaders in MedTech in 2012.
Zen holds multiple leadership roles, including the incoming president and ambassador of the International Industry Society for Advanced Rehabilitation Technology (IISART), General Chair for RehabWeek 2023, co-founder and Executive Director of the Swiss-based MotusAcademy Association, Managing Editor of the Journal of Rehabilitation Methods and Technologies (JRMT), and co-founder and Global CEO of Fourier Intelligence Group. Under his leadership, Fourier has raised over USD100 million, notably in 2022, from renowned investors such as Saudi Aramco Prosperity 7 venture and Softbank Vision Fund 2.
