Workflow
脑机接口技术新突破——科学家首次解码“内心言语”
Ke Ji Ri Bao·2025-08-15 00:49

Core Insights - Scientists have achieved a significant breakthrough in brain-computer interface technology by decoding "inner speech" brain activity with an accuracy of 74%, opening new avenues for helping patients with severe speech impairments to communicate naturally [1][2]. Group 1: Research Findings - The research team implanted microelectrodes in the motor cortex of participants suffering from severe paralysis due to amyotrophic lateral sclerosis or brainstem strokes [2]. - It was found that both attempting to speak and imagining speech activate overlapping areas of the brain, producing similar neural activity patterns, although the signal strength of "inner speech" is generally weaker [2]. - The team trained an AI model using data collected from "inner speech," enabling the brain-computer interface system to decode imagined sentences from a vocabulary of up to 125,000 words with a 74% accuracy rate [2]. Group 2: Communication Mechanism - The research team demonstrated a password control mechanism that allows the system to temporarily ignore "inner speech" and switch to a keyword password unlocking mode, achieving over 98% accuracy in recognizing and activating the decoding function when users silently think of specific phrases [2]. - Although the technology is not yet perfect in processing "inner speech," advancements in sensors and algorithms are expected to lead to more sophisticated devices capable of facilitating fluid communication akin to everyday conversations [2].