New Breakthrough in Brain-Computer Interface Technology—Scientists Decode "Inner Speech" for the First Time

　　Imagine being able to communicate just by thinking about speaking? Scientists have made a major breakthrough in the field of brain-computer interfaces (BCIs), successfully identifying and decoding brain activity related to "inner speech"—the silent monologue in people's minds—with an accuracy rate of up to 74%. This groundbreaking study, led by a team from Stanford University in the U.S., was published on the 14th in the journal Cell, opening a new pathway to help individuals with severe speech disorders regain natural communication abilities.

　　For a long time, BCIs have been seen as a bridge connecting the brain to the external world. Previous technologies primarily relied on decoding neural signals generated when users attempted to vocalize or converting eye movements into text. However, the former requires high muscle control and is exhausting, while the latter is slow in conversion speed. To pursue a more efficient and effortless communication method, the research team turned their focus to "inner speech"—silent, imagined speech that occurs purely in the mind without any muscle involvement.

　　The team implanted microelectrodes in the motor cortex (the language-related area) of participants with severe paralysis caused by amyotrophic lateral sclerosis (ALS) or brainstem strokes. They found that both attempted speech and imagined speech activated overlapping brain regions and produced similar neural activity patterns, though the signal strength of "inner speech" was generally weaker. Based on this discovery, the team used the collected "inner speech" data to train an artificial intelligence model. In proof-of-concept tests, the BCI system could decode imagined sentences from a vast vocabulary of up to 125,000 words with 74% accuracy, and even captured numbers silently recited by users to complete calculation tasks.

　　Additionally, for users seeking faster or more effortless communication methods, the team demonstrated a password-controlled mechanism. This allows the system to be trained to temporarily ignore "inner speech" and switch to a keyword password unlock mode—users only need to mentally recite a specific phrase, and the system can recognize and activate the decoding function with over 98% accuracy.

　　Although the technology is not yet perfect at processing "inner speech," the research team believes that with advancements in sensors and algorithms, more advanced devices in the future could achieve fluid mind-to-mind communication akin to everyday conversation. This work truly brings hope for restoring natural and comfortable communication.