Machines are more capable of reading human thoughts when users are specifically trained to do so, a new Swiss study has found.
The research offers a significant breakthrough in brain-machine interface (BMI) technology, with potential applications for individuals who lose the ability to speak due to strokes or other medical conditions.
In the study, published in the journal Communications Biology, researchers at the University of Geneva worked with 15 volunteers to test how training affected a machine’s ability to interpret human brain signals.
The participants were connected to electrodes that recorded brain activity as they silently imagined saying the syllables “fo” and “gi.”
Over five consecutive days, the volunteers received real-time visual feedback on how accurately the system was interpreting their internal speech.
The better the system understood the syllables, the more a visual display on a screen filled up — effectively guiding participants to improve their communication with the machine.
Despite individual learning differences, trained participants showed marked improvement in making the machine understand their thoughts, compared to a control group that received irregular or inconsistent feedback and showed no such gains.
“This research underlines the previously underestimated importance of training when using brain-machine interfaces,” the University of Geneva said in a statement on Monday.
The team believes their findings could pave the way for the development of new forms of communication for people with severe speech impairments, such as those recovering from strokes or neurological disorders. The ability to mentally articulate syllables that machines can recognise opens the door to more effective non-verbal communication technologies.
Brain-machine interfaces work by detecting voltage fluctuations in the brain through electrodes placed on the scalp. These signals are then processed by computer systems, often enhanced by artificial intelligence, to decode thoughts or intentions into text, sound, or action.
The Geneva researchers say their study adds to a growing body of work showing that consistent, feedback-driven training is key to improving the performance and reliability of BMI systems.
As global interest in neurotechnology accelerates, the findings offer a timely insight into how human and machine cooperation can be enhanced with behavioural reinforcement — making thought-based communication one step closer to reality.
