Master of Science
Many cognitive processes are surprisingly preserved during sleep, including the processing of basic language stimuli. However, whether the sleeping brain can process complex, natural speech is not yet known. The present study used regularized linear regression to understand which features of narrative speech, ranging from low-level acoustic information to higher-level linguistic information, are processed during sleep. Participants were exposed to an intact and scrambled narrative story while they were napping or lying awake. Temporal response functions (TRFs) mapped the relationship between participants’ EEG neural responses and the (1) auditory envelope, (2) word onsets and (3) semantic dissimilarity of words. For all three analyses, delayed but statistically similar TRF components were observed during sleep and wake. These findings suggest that the sleeping brain is capable of low-level auditory processing, speech segmentation and semantic processing of narrative speech. These findings highlight that natural language processing remains remarkably intact during sleep.
Summary for Lay Audience
The brain is known to monitor its surroundings for important and dangerous stimuli during sleep. Previous research has shown that the sleeping brain is able to process some aspects of language during sleep. However, it is still not yet known which features of natural, continuous speech are processed during sleep. First, we aimed to examine if the brain can process the acoustic information of a natural speech source during sleep. Furthermore, we aimed to examine if the sleeping brain can segment the words uttered during natural speech. Finally, we aimed to determine if the brain can extract and understand the meanings of words in natural speech during sleep.
Participants were exposed to an excerpt from an audiobook called J.D. Salinger’s Pretty Mouth and Green My Eyes and their brain activity was recorded, while they were either napping or lying awake in a bed. They were also exposed to a scrambled version of the excerpt that acted as a means of comparison when analyzing the participants’ brain activity. A relatively new analytical method was used to associate the participants’ brain activity with the acoustic audio information, the beginnings of words, and the meanings of the individual words in the natural speech stream. The results indicated that sleeping and awake participants exhibited similar brain activity patterns in association with the acoustic information, the beginnings of words, and the meanings of words conveyed in the audio excerpt. However, the key neural components associated with these processes occurred later for sleeping participants, as compared to wake. This indicates that the sleeping brain not only processes the low-level acoustic information of a natural speech stream during sleep but can also segment and extract the meanings of words in natural speech. These results are among the first to display that the brain can understand key high-level conceptual information in natural speech during sleep, demonstrating the advanced capabilities of the sleeping brain.
Harimohan, Ashwin, "Understanding Neural Signals related to Speech Processing in Humans During Sleep" (2021). Electronic Thesis and Dissertation Repository. 8168.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.