研究人员利用脑机接口尝试解码内心独白

Core Insights - A research team has successfully decoded inner speech-related brain activity using brain-computer interface (BCI) technology, achieving an accuracy rate of up to 74% [1][2] Group 1: Research Findings - The study involved implanting microelectrodes in the motor cortex of four severely paralyzed patients to record neural activity while they attempted to vocalize specific words [1] - The BCI was able to decode sentences from a vocabulary of up to 125,000 words during a proof-of-concept experiment, with a maximum accuracy of 74% [2] - The neural activity patterns for attempted speech and inner speech were found to be highly overlapping, although the overall activation intensity for inner speech was weaker [1][2] Group 2: Technological Implications - The BCI can also capture spontaneous inner speech, such as when a subject mentally counts in response to visual stimuli [2] - Researchers noted that while the neural activity patterns for attempted speech and inner speech are similar, they can be reliably distinguished, allowing for potential improvements in BCI technology [2] - Current BCI systems face challenges in accurately decoding spontaneous inner speech without significant errors, but advancements in sensors and algorithms may enhance future capabilities [2]