Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Mar 2;42(9):1791-1803.
doi: 10.1523/JNEUROSCI.2524-20.2021. Epub 2022 Jan 17.

The Brain Selectively Tunes to Unfamiliar Voices during Sleep

Mohamed S Ameen  1   2 , Dominik P J Heib  1   2 , Christine Blume  3   4 , Manuel Schabus  5   2
Affiliations

The Brain Selectively Tunes to Unfamiliar Voices during Sleep

Mohamed S Ameen et al. J Neurosci. .

Abstract

The brain continues to respond selectively to environmental stimuli during sleep. However, the functional role of such responses, and whether they reflect information processing or rather sensory inhibition, is not fully understood. Here, we present 17 human sleepers (14 females) with their own name and two unfamiliar first names, spoken by either a familiar voice (FV) or an unfamiliar voice (UFV), while recording polysomnography during a full night of sleep. We detect K-complexes, sleep spindles, and microarousals, and assess event-related and frequency responses as well as intertrial phase synchronization to the different stimuli presented during nonrapid eye movement (NREM) sleep. We show that UFVs evoke more K-complexes and microarousals than FVs. When both stimuli evoke a K-complex, we observe larger evoked potentials, more precise time-locking of brain responses in the delta band (1-4 Hz), and stronger activity in the high frequency (>16 Hz) range, in response to UFVs relative to FVs. Crucially, these differences in brain responses disappear completely when no K-complexes are evoked by the auditory stimuli. Our findings highlight discrepancies in brain responses to auditory stimuli based on their relevance to the sleeper and propose a key role for K-complexes in the modulation of sensory processing during sleep. We argue that such content-specific, dynamic reactivity to external sensory information enables the brain to enter a sentinel processing mode in which it engages in the important internal processes that are ongoing during sleep while still maintaining the ability to process vital external sensory information.SIGNIFICANCE STATEMENT Previous research has shown that sensory processing continues during sleep. Here, we studied the capacity of the sleeping brain to extract and process relevant sensory information. We presented sleepers with their own names and unfamiliar names spoken by either an FV or a UFV. During NREM sleep, UFVs elicited more K-complexes and microarousals than FVs. By contrasting stimuli that evoked K-complexes, we demonstrate that UFVs evoked larger, more synchronized brain responses as well as stronger power at high frequencies (>16 Hz) relative to FVs. These differences in brain responses disappeared when no K-complexes were evoked. Our results suggest a pivotal role for K-complexes in the selective processing of relevant information during NREM sleep.

Keywords: EEG; K-complexes; auditory stimulation; information processing; microarousals; sleep.

PubMed Disclaimer

Figures

Figure 1.
Figure 1.
Experimental design. A, Protocol. Participants were invited for an initial screening interview during which they were given the wrist-worn actigraphy and advised to keep a regular sleep-wake cycle. Participants slept in the sleep laboratory during two nights, an adaptation night during which polysomnography (PSG) was recorded but no stimuli were presented, and an experimental night during which we recorded PSG and presented auditory stimuli throughout the night. B, Procedure of the experimental night. Participants slept for ∼8 h with PSG and auditory stimulation. All participants went through one stimulation session in the evening before sleep (Awake - pre) and one in the morning after waking up (Awake- post); however, these sessions are irrelevant to the current analysis. The auditory stimuli started directly after participants went to bed and continued for 90 min, then paused for ∼30 min to allow for a period of undisturbed sleep. This cycle was repeated four times for the whole duration of the night. C, Top, Stimuli. We presented the subject's own name (SON) and two unfamiliar names (UN1 and UN2) spoken by either a familiar voice (FV) or an unfamiliar voice (UFV). Bottom, An exemplary sequence of stimulus presentation. Stimuli were presented in a pseudorandom order; each stimulus was presented 690 times, and the interstimulus intervals ranged between 2800 and 7800 ms.
Figure 2.
Figure 2.
K-complex detection and examples. A, LDA scores for all detected K-complexes for all subjects. The dashed line represents our minimum cutoff at an LDA value of 0.8, which is the threshold used for the selection of reliable K-complex events. Bottom, The percentage of the events used in our analyses from all the detected events for each subject are indicated in red. B, Examples of the detected K-complexes at channel C3 referenced to the contralateral mastoid. We show the standard EEG montage that we used for sleep staging as well as event detections. Specifically, we used the following channels (from top to bottom): F3, F4, C3, C4, O1, and O2 referenced to the contralateral mastoid. Moreover, we show one EMG and two EOG channels (hEOG_R and hEOG_L) channels as well as the average of both mastoids (A1-A2). Examples 1 and 2 show K-complexes detected in N2 sleep from one subject, whereas examples 3 and 4 are K-complexes detected during N3 in a different subject.
Figure 3.
Figure 3.
Examples of the detected microarousals. We show the standard EEG montage that is used for sleep staging as well as event detections. Specifically, we used the following channels (from top to bottom): F3, F4, C3, C4, O1, and O2 referenced to the contralateral mastoid. Moreover, we show one EMG and two EOG channels (hEOG_R and hEOG_L) as well as the average of both mastoids (A1-A2). Examples 1 and 2 are events detected in N2 sleep from one subject. Example 2 shows a microarousal that is preceded by a K-complex. Example 3 is from an N3 epoch from a different subject.
Figure 4.
Figure 4.
Auditory stimulation influenced sleep microstructure but not macrostructure. In each participant, we selected equal numbers of epochs for all conditions. A, Difference in sleep architecture, that is, the distribution of sleep stages, between the adaptation and the experimental night. To account for the effect of sleep-onset latency, we discarded all epochs that preceded the first N1 epoch. We found no difference between the architectures of both nights. B, Difference in sleep architecture between stimulation and no-stimulation periods during the experimental night. The experimental night consisted of periods of continuous auditory stimulation and periods of no stimulation. We found no difference in the sleep architecture between those periods. C, Microstate differences between the stimulation and the no-stimulation periods. To circumvent the poor temporal resolution of the classical 30-second sleep stages, we opted for a more time-resolved analysis of sleep stages based on the Hori scoring system. We found a higher number of sleep epochs and lower number of drowsy epochs during the stimulation periods, suggesting deeper sleep during auditory stimulation. D–F, A comparison of the densities of sleep microstructure between stimulation and no-stimulation periods. K-complex density during the stimulation periods is higher than that in the no-stimulation periods (D). Slow and Fast spindles densities are higher during the stimulation periods (E). Microarousal density, however, did not differ between stimulation and no-stimulation periods (F). Box plots show the median and the whiskers depict the 25% and the 75% quartiles. Each dot/triangle represents one participant in one condition. *p < 0.05, **p < 0.001.
Figure 5.
Figure 5.
Auditory stimulation increased the occurrence sleep microstructures. A, A graphical illustration of the stimulus-ON and stimulus-OFF windows used for the analysis. We detected events in the 2000 ms poststimulus window (Stimulus-ON, green), as well as during 2000 ms prestimulus windows that are at least 2000 ms after the last stimulus (Stimulus-OFF, orange). B–E, The numbers of (B) K-complexes, (C) slow spindles, (D) fast spindles, as well as (E) microarousals were significantly higher during the stimulus-ON than the stimulus-OFF periods. The y-axis depicts the percentage of epochs in which events were detected. Box plots show the median, and the whiskers depict the 25% and the 75% quartiles. Each dot/triangle represents one participant in one condition. *p < 0.05, **p < 0.001.
Figure 6.
Figure 6.
Selective sleep-specific responses to unfamiliar voices during NREM sleep. A, Differences in the triggered K-complexes between the familiar voice (FV) and the unfamiliar voice (UFV) in the 2000 ms poststimulus-onset window. UFVs triggered more K-complexes than FVs. B, C, However, the numbers of (B) slow and (C) fast spindles did not differ between FV and UFVs. D, Differences in triggered microarousals between FVs and UFVs demonstrating the higher number of microarousals triggered by UFVs. E, The amount of triggered K-complexes and microarousals by all our stimulus categories and compared with those detected in the same number of 2000 ms no-stimulation epochs. F, G, Temporal aspects of the difference in the triggered K-complexes and microarousals. The difference between UFV and FV in the number of triggered K-complexes was significant from 100 ms to 800 ms (indicated by bar and asterisks; F). The difference in the number of microarousals between FVs and UFVs was significant in the periods from 200 to 400 ms, and from 500 to 700 ms (G). Box plots show the median, and the whiskers depict the 25% and the 75% quartiles. Each dot/triangle represents one participant in one condition. Top, The dashed horizontal line represents the mean duration of our stimuli (808 ms), the lines depicts the means of the bins, and the shadings indicate the standard error of the mean (F, G). *p < 0.05, **p < 0.001. FVSON, Familiar voice speaking the subject's own name; UFVSON, unfamiliar voice speaking the subject's own name; FVUNs, familiar voice speaking two unfamiliar names, UFVUNs, unfamiliar voice speaking two unfamiliar names.
Figure 7.
Figure 7.
The effect of time on sleep-specific responses to unfamiliar voices during NREM sleep. A, The difference in the numbers of triggered K-complexes to the familiar voice (FV) and unfamiliar voice (UFV) from the first to the second half of the night. A Generalized Linear Mixed Model (GLMM) using Poisson distribution revealed a significant interaction, Time × Voice, as only the amount of UFV-triggered K-complexes decreased in the second half. B, A GLMM using zero-inflated Poisson distribution showed that the number of triggered microarousals did not change from the first to the second half of the night. C, A statistical report of the GLMM model of K-complexes. We added Time (first half and second half), and Voices (familiar and unfamiliar voices) as fixed effects. Moreover, we assigned random intercepts and slopes for each subject. We found a significant main effect of voice, no effect of time, and a significant interaction, Time × Voice. D, Statistical report of the zero-inflated Poisson GLMM of microarousals. Similar to K-complexes, we added Time (first half and second half), and Voices (familiar and unfamiliar voices) as fixed effects and random intercepts and slopes for each subject. No main effect of time and no interaction, Time × Voice, indicating that the amount of triggered microarousals did not change from the first to the second half of the night. *p < 0.05. Box plots show the median, and the whiskers depict the 25% and the 75% quartiles. Each dot/triangle represents one participant in one condition.
Figure 8.
Figure 8.
Unfamiliar voices elicited stronger brain responses in the presence of the evoked K-complex. A, ERP contrast between familiar voice (FV) and unfamiliar voice (UFV) in the presence of the auditory-evoked K-complexes. UFVs triggered a larger amplitude of the evoked response between 510 ms and 1400 ms as shown by the gray shadings. Extended Data Figures 8-1 and 8-2 illustrate the methodological underpinnings of the discrepancies between the amplitude and latency of the negative component in A as compared with the conventional amplitude and latency of the N550 in the literature. B, Comparison of the peak-to-peak amplitude (P2P; N550 to P900) of the evoked K-complexes showing no difference in the amplitudes of the evoked K-complexes between FVs and UFVs. C, ERP responses to FVs and UFVs in the absence of evoked K-complexes. There was no difference in the ERP amplitudes between FVs and UFVs when no K-complexes were evoked. The solid blue and red lines and the shadings represent the mean and the standard error of the mean, respectively. Top, The dashed horizontal line represent the mean duration of our stimuli (808 ms). Vertical dashed lines (at x = 0) represent stimulus onset. Bottom left, The red dots on the topographical plots indicate the locations of the channels used for the analysis. Box plots show the median, and the whiskers depict the 25% and the 75% quartiles. Each dot/triangle represents one participant in one condition.
Figure 9.
Figure 9.
Brain responses to unfamiliar voices in the presence of the evoked K-complex reflect further processing. A, The difference in intertrial phase coherence (ITPC) values between the familiar voice (FV) and the unfamiliar voice (UFV) in the presence of the evoked K-complex. UFVs evoked significantly higher ITPC than FV in the delta (1–4 Hz) frequency band. Note that the largest difference in phase locking overlaps with the difference between the ERPs in Figure 8A. B, Separate ITPC plots for FVs (top) and UFVs (bottom) showing stronger ITPC values following UFVs and indicating that the difference in ITPC values is because of an increase in ITPC following UFVs. C, ITPC difference between UFVs and FVs in the absence of the evoked K-complexes. There is no difference in ITPC between FVs and UFVs in the absence of evoked K-complexes. D, E, Spectral power maps of the power differences between FVs and UFVs. We demonstrate stronger responses to UFV in a broad frequency range (∼1–10 Hz) regardless of the presence (D) or the absence (E) of K-complexes. However, stronger high-frequency (>16 Hz) responses to UFVs appeared only in the presence of K-complexes. Top, The dashed horizontal line represents the mean duration of our stimuli (808 ms). Vertical dashed lines (at x = 0) represent stimulus onset. Bottom left, The red dots on the topographical plots indicate the locations of the channels used for the analysis.
See this image and copyright information in PMC

References

    1. Ameen MS, Cheung LM, Hauser T, Hahn MA, Schabus M (2019) About the accuracy and problems of consumer devices in the assessment of sleep. Sensors 19:4160. 10.3390/s19194160 - DOI - PMC - PubMed
    1. American Sleep Disorders Association (1992) EEG arousals: scoring rules and examples: a preliminary report from sleep disorders atlas task force of the American Sleep Disorders Association. Sleep 15:173–184. - PubMed
    1. Anderer P, Gruber G, Parapatics S, Woertz M, Miazhynskaia T, Klosch G, Saletu B, Zeitlhofer J, Barbanoj MJ, Danker-Hopfe H, Himanen S-L, Kemp B, Penzel T, Grozinger M, Kunz D, Rappelsberger P, Schlogl A, Dorffner G (2005) An E-health solution for automatic sleep classification according to Rechtschaffen and Kales: validation study of the Somnolyzer 24 x 7 utilizing the Siesta database. Neuropsychobiology 51:115–133. 10.1159/000085205 - DOI - PubMed
    1. Anderer P, Moreau A, Woertz M, Ross M, Gruber G, Parapatics S, Loretz E, Heller E, Schmidt A, Boeck M, Moser D, Kloesch G, Saletu B, Saletu-Zyhlarz GM, Danker-Hopfe H, Zeitlhofer J, Dorffner G (2010) Computer-assisted sleep classification according to the standard of the American Academy of Sleep Medicine: validation study of the AASM version of the Somnolyzer 24 × 7. Neuropsychobiology 62:250–264. 10.1159/000320864 - DOI - PubMed
    1. Andrillon T, Poulsen AT, Hansen LK, Léger D, Kouider S (2016) Neural markers of responsiveness to the environment in human sleep. J Neurosci 36:6583–6596. 10.1523/JNEUROSCI.0902-16.2016 - DOI - PMC - PubMed

Publication types

LinkOut - more resources