Stanford study decodes imagined speech from brain signals

In short: Stanford researchers reported a brain-computer interface that can turn imagined speech into words, but it works better when someone tries to speak than when they only think the words.

What happened

A brain-computer interface, or BCI, is a system that reads brain activity and uses it to control something, like text on a screen. In this work, Stanford researchers tested a BCI that aims to decode imagined speech, meaning silently thinking of words and sentences without moving the mouth or using the vocal cords.

The team reported that imagined speech created clear patterns in parts of the brain linked to speech. These patterns looked similar to the ones seen during attempted speech, when a person tries to speak but may not be able to produce sound. The imagined speech signals were smaller, like a quieter version of the same “song,” and the system decoded them with lower accuracy than attempted speech.

The researchers also showed a privacy feature. The system only starts decoding after the user mentally imagines a chosen password phrase. You can think of it like a lock that only opens when you “say” a specific phrase in your head.

Why it matters

This research is aimed at helping people with paralysis or severe speech impairment communicate faster and more naturally. It also matters for public understanding. Headlines can make this sound like general mind reading, but the study is about a specialized system that recognizes certain imagined speech patterns under controlled conditions, not unrestricted access to private thoughts.

Source: NYTimes