| Sumario: | Speech is a complex cognitive process that begins with conceptualization, proceeds to word-level processing, and ends with articulation. Neural decoding of speech (i.e., using neural activity to decode the content of language production) has been mostly conducted by mapping neural activities in the later part of language production (i.e., phonological and motor processing). Here we show that neural decoding of speech can also be performed by mapping neural activities associated with semantic representations that occur in the early part of language production. Furthermore, we demonstrated that the classifier trained using the neural activity patterns of language perception was able to decode the content of language production, indicating a cross-modality similarity between language perception and language production in semantic representations.
|
|---|