A new non-invasive decoder has been introduced to reconstruct continuous language from cortical semantic representations, recorded through functional magnetic resonance imaging. The decoder is capable of generating intelligible word sequences to recover the meaning of perceived speech, imagined speech, and silent videos. The decoder can be applied to a range of tasks and can be separately decoded from multiple regions across cortex. The study tested whether subject cooperation is required for successful decoding, and it was found that the decoder requires subject cooperation to be trained and applied. Additionally, the study emphasizes respecting mental privacy as brain-computer interfaces become more common.
