CoChat

Bidirectional prefrontal-hippocampal dynamics organize information transfer during sleep in humans

Open access

The DOI matches its Crossref record; the references it cites are not checked yet.

Abstract

How are memories transferred from short-term to long-term storage? Systems-level memory consolidation is thought to be dependent on the coordinated interplay of cortical slow waves, thalamo-cortical sleep spindles and hippocampal ripple oscillations. However, it is currently unclear how the selective interaction of these cardinal sleep oscillations is organized to support information reactivation and transfer. Here, using human intracranial recordings, we demonstrate that the prefrontal cortex plays a key role in organizing the ripple-mediated information transfer during non-rapid eye movement (NREM) sleep. We reveal a temporally precise form of coupling between prefrontal slow-wave and spindle oscillations, which actively dictates the hippocampal-neocortical dialogue and information transfer. Our results suggest a model of the human sleeping brain in which rapid bidirectional interactions, triggered by the prefrontal cortex, mediate hippocampal activation to optimally time subsequent information transfer to the neocortex during NREM sleep.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41467-019-11444-x

Cites

Statistics

Figures

Fig. 1 Sleep architecture and coupled NREM oscillations. a Top: hypnogram of a single subject (MT = movement time). Bottom: corresponding multi-taper spectrogram with superimposed number of detected SO and sleep spindle events (white solid lines; 5 min averages). b Top: normalized circular histogram of detected spindle events relative to the SO phase. Note the peak around 0° (circular mean = 15.27°, i.e. after the SO peak). Bottom: peak-locked sleep spindle average across all detected events in NREM sleep (black). Low-pass filtered events (red) highlight that the sleep spindles preferentially peaked during the SO ‘up-state’. c Peak-locked spindle grand average (black; N = 18; scalp EEG) superimposed with the SO low-pass filtered signal (red). Inset: spindles were significantly nested in the SO peak (V-test: p = 0.0050; red dots depict individual subjects). d Directional SO-spindle coupling as measured by the phase-slope index (PSI; mean ± SEM). SO phases significantly predicted spindle amplitudes (one-tailed t-test: p = 0.0436). Red dots depict individual subjects
Fig. 1 Sleep architecture and coupled NREM oscillations. a Top: hypnogram of a single subject (MT = movement time). Bottom: corresponding multi-taper spectrogram with superimposed number of detected SO and sleep spindle events (white solid lines; 5 min averages). b Top: normalized circular histogram of detected spindle events relative to the SO phase. Note the peak around 0° (circular mean = 15.27°, i.e. after the SO peak). Bottom: peak-locked sleep spindle average across all detected events in NREM sleep (black). Low-pass filtered events (red) highlight that the sleep spindles preferentially peaked during the SO ‘up-state’. c Peak-locked spindle grand average (black; N = 18; scalp EEG) superimposed with the SO low-pass filtered signal (red). Inset: spindles were significantly nested in the SO peak (V-test: p = 0.0050; red dots depict individual subjects). d Directional SO-spindle coupling as measured by the phase-slope index (PSI; mean ± SEM). SO phases significantly predicted spindle amplitudes (one-tailed t-test: p = 0.0436). Red dots depict individual subjectsCC BY · PubMed Central
Fig. 2 Intracranial electrodes and hierarchical nesting of sleep oscillations. a Schematic depiction in MNI space of the intracranial electrode placement in the medial temporal lobe (MTL; in black) and three PFC regions-of-interest (ROI) across the subset of subjects (N = 15) with simultaneous coverage of all three prefrontal ROIs (dlPFC: n = 167; OFC: n = 62; mPFC: n = 116). b Schematic depiction of MTL contacts (n = 184 electrodes in 18 subjects) on a hippocampal mesh in MNI space. Note that electrode size has been rescaled for display purposes and does not reflect the original electrode size. Precise electrode localizations were determined in native space prior to warping to MNI space. c Single subject time-frequency spectrogram from one MTL channel relative to a rescaled trough-locked cortical average SO (black). Note the circumscribed increase in high-frequency-band activity (~80–100 Hz) just prior to the SO peak. d Single subject time-frequency spectrogram from one MTL channel relative to a rescaled peak-locked cortical spindle average (black). High-frequency band activity exhibited several bursts during spindle events. e Single subject, single electrode example of simultaneous nesting of HFB power in the cortical SO peak (x-axis) and the cortical spindle trough (y-axis). f Cross-frequency coupling analysis indicated increased SO-Spindle, SO-HFB, and Spindle-HFB coupling during NREM sleep (gray dots depict individual subjects; mean ± SEM). g Full comodulogram (wake vs. NREM) depicting the effect size of the difference (Cohen’s d)
Fig. 2 Intracranial electrodes and hierarchical nesting of sleep oscillations. a Schematic depiction in MNI space of the intracranial electrode placement in the medial temporal lobe (MTL; in black) and three PFC regions-of-interest (ROI) across the subset of subjects (N = 15) with simultaneous coverage of all three prefrontal ROIs (dlPFC: n = 167; OFC: n = 62; mPFC: n = 116). b Schematic depiction of MTL contacts (n = 184 electrodes in 18 subjects) on a hippocampal mesh in MNI space. Note that electrode size has been rescaled for display purposes and does not reflect the original electrode size. Precise electrode localizations were determined in native space prior to warping to MNI space. c Single subject time-frequency spectrogram from one MTL channel relative to a rescaled trough-locked cortical average SO (black). Note the circumscribed increase in high-frequency-band activity (~80–100 Hz) just prior to the SO peak. d Single subject time-frequency spectrogram from one MTL channel relative to a rescaled peak-locked cortical spindle average (black). High-frequency band activity exhibited several bursts during spindle events. e Single subject, single electrode example of simultaneous nesting of HFB power in the cortical SO peak (x-axis) and the cortical spindle trough (y-axis). f Cross-frequency coupling analysis indicated increased SO-Spindle, SO-HFB, and Spindle-HFB coupling during NREM sleep (gray dots depict individual subjects; mean ± SEM). g Full comodulogram (wake vs. NREM) depicting the effect size of the difference (Cohen’s d)CC BY · PubMed Central
Fig. 3 Precise cortical SO-spindle coupling phase predicts MTL HFB activity. a Single subject average of all coupled spindles that were nested in the SO peak (in red) and trough (in blue). b All uncoupled spindles, i.e. that did not coincide with a separate SO detection. c Cortical spindle to MTL HFB phase amplitude coupling for coupled and uncoupled spindles as well as event-free random NREM intervals (normalized by the individual mean to reduce inter-subject variability). d Single subject HFB activity in the MTL as a function of the precise cortical SO-Spindle phase at the time of the spindle peak. Bars depict the average across electrodes; gray lines depict single electrodes. The red arrow indicates the SO-spindle coupling phase where the highest HFB amplitude was observed (~0°). e Different single subject example. Same conventions as in panel d. Note that HFB peaked after the SO peak (>0°). f Coupling strength: in every subject, we assessed the coupling strength by comparing the observed circular-linear correlation between SO-spindle coupling phase and HFB amplitude to a surrogate distribution. The correlation coefficient was then z-scored relative to a phase-shuffled distribution of correlation coefficients and considered significant at z = 1.96 in individual subjects (corresponding to a two-tailed p < 0.05; red solid line). The gray bar depicts the mean across subjects; black dots depict individual subjects. We observed significant HFB modulations in 18/18 subjects (mean z = 3.40, pgroup = 0.0007). g Preferred coupling phase: circular histogram of the best phase (panels d and e) where HFB peaked in individual subjects. At the group level, HFB amplitude was highest during the spindle when the spindle was nested in the SO peak (mvl = 0.44). h Preferred coupling phase across time (circular mean ± circular SEM). We repeated the analysis depicted in panel f for all time points (±1.25 s around the spindle peak). Shaded bars depict cluster-corrected significant Rayleigh tests (cluster p < 0.001) for circular non-uniformity (test for phase consistency at the group level), indicating that a preferred phase across subjects emerged around the spindle peak (cluster 1: −0.3 to 0.04 s; cluster 2: 0.12 to 0.60 s); hence, the effect was transient and not sustained. i Rayleigh Z and significant clusters as described in panel h. j Directional CFC analysis (gray dots depict individual subjects; mean ± SEM) indicating that the PFC spindle phase was predictive of hippocampal HFB, but not vice versa (rank sum test: p = 0.0005). All error bars indicate the mean ± SEM
Fig. 3 Precise cortical SO-spindle coupling phase predicts MTL HFB activity. a Single subject average of all coupled spindles that were nested in the SO peak (in red) and trough (in blue). b All uncoupled spindles, i.e. that did not coincide with a separate SO detection. c Cortical spindle to MTL HFB phase amplitude coupling for coupled and uncoupled spindles as well as event-free random NREM intervals (normalized by the individual mean to reduce inter-subject variability). d Single subject HFB activity in the MTL as a function of the precise cortical SO-Spindle phase at the time of the spindle peak. Bars depict the average across electrodes; gray lines depict single electrodes. The red arrow indicates the SO-spindle coupling phase where the highest HFB amplitude was observed (~0°). e Different single subject example. Same conventions as in panel d. Note that HFB peaked after the SO peak (>0°). f Coupling strength: in every subject, we assessed the coupling strength by comparing the observed circular-linear correlation between SO-spindle coupling phase and HFB amplitude to a surrogate distribution. The correlation coefficient was then z-scored relative to a phase-shuffled distribution of correlation coefficients and considered significant at z = 1.96 in individual subjects (corresponding to a two-tailed p < 0.05; red solid line). The gray bar depicts the mean across subjects; black dots depict individual subjects. We observed significant HFB modulations in 18/18 subjects (mean z = 3.40, pgroup = 0.0007). g Preferred coupling phase: circular histogram of the best phase (panels d and e) where HFB peaked in individual subjects. At the group level, HFB amplitude was highest during the spindle when the spindle was nested in the SO peak (mvl = 0.44). h Preferred coupling phase across time (circular mean ± circular SEM). We repeated the analysis depicted in panel f for all time points (±1.25 s around the spindle peak). Shaded bars depict cluster-corrected significant Rayleigh tests (cluster p < 0.001) for circular non-uniformity (test for phase consistency at the group level), indicating that a preferred phase across subjects emerged around the spindle peak (cluster 1: −0.3 to 0.04 s; cluster 2: 0.12 to 0.60 s); hence, the effect was transient and not sustained. i Rayleigh Z and significant clusters as described in panel h. j Directional CFC analysis (gray dots depict individual subjects; mean ± SEM) indicating that the PFC spindle phase was predictive of hippocampal HFB, but not vice versa (rank sum test: p = 0.0005). All error bars indicate the mean ± SEMCC BY · PubMed Central
Fig. 4 Physiological and artifactual ripple activity in the MTL. a Ripple-locked grand average (blue; N = 18) and the band-bass filtered signal (black; mean ± SEM). b Artifactual ripple events grand average (red) that are characterized by sharp transients that are artificially rendered sinusoidal by band-pass filtering (black). c 1/f-corrected power spectrum (mean ± SEM; irregular resampling) in the MTL during a ripple event exhibits two distinct oscillatory peaks: in the delta and spindle frequency range. Gray shaded areas indicate significant oscillatory activity (cluster test: p = 0.001). d Histogram of ripple occurrence relative to the cortical SO as detected at scalp level. Note ripple occurrence is enhanced during the cortical upstate (i.e., the SO peak, first cluster: p = 0.0010, d = 2.82; second cluster: p = 0.0040, d = 1.42), but significant more ripples occur prior to the downstate (in gray; 400 ms window around the SO peaks, paired t-test: t17 = 5.01, p < 0.0001, d = 1.88). e This relationship was not present for artifactual ripple events, which occurred at random (smallest cluster p = 0.450; no peak differences: t17 = −0.63, p = 0.5381, d = 0.27). f Physiologic ripples were preferentially nested in the trough of the cortical spindle oscillation (V-test against ± pi: v = 5.13, p = 0.0434, mvl = 0.29). g This tight temporal relationship was again not present for artifactual ripple events (v = 1.46, p = 0.3135, mvl = 0.08). h Single subject ripple-locked EEG spectrogram indicates the presence of cortical spindle oscillations during a hippocampal ripple. Note that activity in the spindle range can be observed before and after the ripple (every ~3 s; asterisks indicate elevate spindle power). i Left Spectral analysis of the spindle amplitude (green), its fractal (1/f estimate; black dashed line) as obtained from irregular resampling as well as the wake condition as a control (absence of spindle activity; cluster test: p = 0.0490, d = 0.49). Right: after subtraction of the fractal component from the NREM spectrum, we observed a distinct oscillatory peak indicating a rhythmic comodulation at ~0.37 ± 0.03 Hz (median ± SEM; sign test against zero: p < 0.0001; d = 0.70). All error bars indicate the mean ± SEM
Fig. 4 Physiological and artifactual ripple activity in the MTL. a Ripple-locked grand average (blue; N = 18) and the band-bass filtered signal (black; mean ± SEM). b Artifactual ripple events grand average (red) that are characterized by sharp transients that are artificially rendered sinusoidal by band-pass filtering (black). c 1/f-corrected power spectrum (mean ± SEM; irregular resampling) in the MTL during a ripple event exhibits two distinct oscillatory peaks: in the delta and spindle frequency range. Gray shaded areas indicate significant oscillatory activity (cluster test: p = 0.001). d Histogram of ripple occurrence relative to the cortical SO as detected at scalp level. Note ripple occurrence is enhanced during the cortical upstate (i.e., the SO peak, first cluster: p = 0.0010, d = 2.82; second cluster: p = 0.0040, d = 1.42), but significant more ripples occur prior to the downstate (in gray; 400 ms window around the SO peaks, paired t-test: t17 = 5.01, p < 0.0001, d = 1.88). e This relationship was not present for artifactual ripple events, which occurred at random (smallest cluster p = 0.450; no peak differences: t17 = −0.63, p = 0.5381, d = 0.27). f Physiologic ripples were preferentially nested in the trough of the cortical spindle oscillation (V-test against ± pi: v = 5.13, p = 0.0434, mvl = 0.29). g This tight temporal relationship was again not present for artifactual ripple events (v = 1.46, p = 0.3135, mvl = 0.08). h Single subject ripple-locked EEG spectrogram indicates the presence of cortical spindle oscillations during a hippocampal ripple. Note that activity in the spindle range can be observed before and after the ripple (every ~3 s; asterisks indicate elevate spindle power). i Left Spectral analysis of the spindle amplitude (green), its fractal (1/f estimate; black dashed line) as obtained from irregular resampling as well as the wake condition as a control (absence of spindle activity; cluster test: p = 0.0490, d = 0.49). Right: after subtraction of the fractal component from the NREM spectrum, we observed a distinct oscillatory peak indicating a rhythmic comodulation at ~0.37 ± 0.03 Hz (median ± SEM; sign test against zero: p < 0.0001; d = 0.70). All error bars indicate the mean ± SEMCC BY · PubMed Central
Fig. 5 PFC-MTL connectivity and directionality. a Phase locking analyses revealed increased spindle phase coupling during NREM sleep (cluster tests: p = 0.0120, d = 1.37), but no elevated SO synchrony (p = 0.3087, d = 0.69). During wakefulness, we observed enhanced alpha-band coupling (p = 0.005, d = 0.83). b Power correlations (mean ± SEM) between scalp EEG and hippocampus revealed differences in amplitude couplings during NREM sleep (SO cluster: p = 0.0080, d = 1.07; spindle cluster: p = 0.0070, d = 1.20). c State-dependent directionality analysis indicates significant EEG to MTL influences (phase-slope index; mean ± SEM) in the spindle-band during NREM sleep (cluster test: p = 0.0410, d = 0.84). Inset: given the inter-subject variability in the precise spindle peak frequency, we also compared the peak-aligned (14.9 Hz ± 2.7 Hz; mean ± SD) PSI estimates to a surrogate distribution (gray), which confirmed the original observation (t-test: t17 = 2.16, p = 0.0452, d = 0.69). d Directional spindle coupling for every PFC ROI (lower panel; RM-ANOVA: F1.57,21.92 = 9.33, p = 0.002, η2 = 0.25) indicating a stronger modulatory influence from both the dlPFC and the mPFC but not the OFC. All error bars indicate the mean ± SEM
Fig. 5 PFC-MTL connectivity and directionality. a Phase locking analyses revealed increased spindle phase coupling during NREM sleep (cluster tests: p = 0.0120, d = 1.37), but no elevated SO synchrony (p = 0.3087, d = 0.69). During wakefulness, we observed enhanced alpha-band coupling (p = 0.005, d = 0.83). b Power correlations (mean ± SEM) between scalp EEG and hippocampus revealed differences in amplitude couplings during NREM sleep (SO cluster: p = 0.0080, d = 1.07; spindle cluster: p = 0.0070, d = 1.20). c State-dependent directionality analysis indicates significant EEG to MTL influences (phase-slope index; mean ± SEM) in the spindle-band during NREM sleep (cluster test: p = 0.0410, d = 0.84). Inset: given the inter-subject variability in the precise spindle peak frequency, we also compared the peak-aligned (14.9 Hz ± 2.7 Hz; mean ± SD) PSI estimates to a surrogate distribution (gray), which confirmed the original observation (t-test: t17 = 2.16, p = 0.0452, d = 0.69). d Directional spindle coupling for every PFC ROI (lower panel; RM-ANOVA: F1.57,21.92 = 9.33, p = 0.002, η2 = 0.25) indicating a stronger modulatory influence from both the dlPFC and the mPFC but not the OFC. All error bars indicate the mean ± SEMCC BY · PubMed Central
Fig. 6 Precise SO-spindle coupling phase modulates MTL-PFC connectivity. a Single subject example of EEG-MTL connectivity calculated separately for coupled spindles (black), uncoupled/isolated spindles (purple), and random event-free intervals (gray). Event numbers were equated by a bootstrapping procedure. Note the distinct peak in the spindle band. b We re-aligned power spectra to the individual spindle peak (14.5 ± 2.6 Hz; mean ± SD) and mean-normalized the aligned spectra to minimize inter-subject variability. c We then extracted the normalized EEG-MTL iPLV peak values for all events. A repeated measures ANOVA revealed a significant main effect (F1.49,25.27 = 7.93, p = 0.0043, η2 = 0.21; p-values from post-hoc pair-wise comparisons are depicted). We found significantly more inter-areal connectivity for coupled than for uncoupled spindles or random event-free intervals. d Directionality analysis during spindle events revealed a significant influence from prefrontal EEG spindles to the MTL (dark gray) as compared to a surrogate distribution (light gray). e Two single subject examples of event-locked spindle connectivity as a function of the precise coupling phase (mean ± SEM). Upper: the spindle band exhibits a strong peak in the spectrum and also peaks in the standard deviation spectrum (inset in red), thus, indicating that the spindle band exhibits the largest variance. f Spindle-band connectivity binned as a function of the precise coupling phase revealed strong non-uniform distributions. We highlighted the worst (lowest synchrony; red) and best (highest synchrony; blue) phase. g Anti-phasic relationship between best and worst connectivity phase (EEG-MTL: V-test against ±π: v = 7.33, p = 0.0073, mvl = 0.49). h Realigned (to the best phase) grand average spindle (12–16 Hz) iPLV histogram reveals a non-uniform relationship (mean ± SEM) between SO-Spindle coupling and EEG-MTL phase locking. Note we excluded the central aligned phase bin from subsequent testing (RM-ANOVA: F5.23,88.83 = 3.13, p = 0.011, η2 = 0.15). All error bars indicate the mean ± SEM
Fig. 6 Precise SO-spindle coupling phase modulates MTL-PFC connectivity. a Single subject example of EEG-MTL connectivity calculated separately for coupled spindles (black), uncoupled/isolated spindles (purple), and random event-free intervals (gray). Event numbers were equated by a bootstrapping procedure. Note the distinct peak in the spindle band. b We re-aligned power spectra to the individual spindle peak (14.5 ± 2.6 Hz; mean ± SD) and mean-normalized the aligned spectra to minimize inter-subject variability. c We then extracted the normalized EEG-MTL iPLV peak values for all events. A repeated measures ANOVA revealed a significant main effect (F1.49,25.27 = 7.93, p = 0.0043, η2 = 0.21; p-values from post-hoc pair-wise comparisons are depicted). We found significantly more inter-areal connectivity for coupled than for uncoupled spindles or random event-free intervals. d Directionality analysis during spindle events revealed a significant influence from prefrontal EEG spindles to the MTL (dark gray) as compared to a surrogate distribution (light gray). e Two single subject examples of event-locked spindle connectivity as a function of the precise coupling phase (mean ± SEM). Upper: the spindle band exhibits a strong peak in the spectrum and also peaks in the standard deviation spectrum (inset in red), thus, indicating that the spindle band exhibits the largest variance. f Spindle-band connectivity binned as a function of the precise coupling phase revealed strong non-uniform distributions. We highlighted the worst (lowest synchrony; red) and best (highest synchrony; blue) phase. g Anti-phasic relationship between best and worst connectivity phase (EEG-MTL: V-test against ±π: v = 7.33, p = 0.0073, mvl = 0.49). h Realigned (to the best phase) grand average spindle (12–16 Hz) iPLV histogram reveals a non-uniform relationship (mean ± SEM) between SO-Spindle coupling and EEG-MTL phase locking. Note we excluded the central aligned phase bin from subsequent testing (RM-ANOVA: F5.23,88.83 = 3.13, p = 0.011, η2 = 0.15). All error bars indicate the mean ± SEMCC BY · PubMed Central
Fig. 7 Ripple-mediated information transfer. a Illustration of the information analysis strategy. A single EEG trace featuring SOs and spindles (in blue) and the corresponding MTL trace (in red) are depicted relative to a ripple event (see close-up; in black). Undirectional information was calculated by a moving time-window approach (center left), while directional information was calculated by keeping one window centered on the MTL ripple time point (center right; here fixed MTL window to moving EEG window; EEG to MTL information was calculated accordingly). b Left: time-resolved Mutual Information (MI) relative to the hippocampal ripple event (mean ± SEM). We detected enhanced MI between the hippocampus and prefrontal cortex ~1 s after the ripple event (1.3–2.0 s; cluster test: p = 0.0040, d = 1.06), but also observed two non-significant clusters (positive cluster from 0.25 to 0.35 s, p = 0.0849; negative cluster from −0.15 to −0.10 s, p = 0.1349). Right: spectrally resolved undirected mutual information. Significant (two-tailed p < 0.05) deviations from the average MI. Prior to ripple onset, mutual information was enhanced in the spindle and SO/delta range, which after 1 s dropped. c Left: time-resolved MI relative to the ripple event. We disentangled the information flow from prefrontal EEG to the MTL (purple) and vice versa (cyan). A significant cluster was observed after the ripple (cluster test: negative cluster from 0.15 to 0.50 s; p = 0.0130, d = 1.31), indicating increased information flow from the prefrontal EEG to the MTL. We found a second significant cluster where MI increased after ~1 s, which mainly reflected MTL to prefrontal EEG information flow (positive cluster from 1.1 to 1.85 s; p = 0.001, d = 1.08). Center/right: frequency- and directionality-resolved information flow. Outlined areas (black) reflect significant clusters (p < 0.05), where information flow was enhanced relative to the mean. Center: frequency-specific information flow from the MTL to neocortex was primarily increased in the SO (<2 Hz) and spindle-bands (~16 Hz). Right: information flow from the prefrontal EEG to MTL was not frequency-specific. d Upper panel: phase transfer entropy between the prefrontal EEG and the MTL (>1) is stronger than in the opposite direction (<1), in particular in the spindle-band (from −1–1.4 s; cluster test: p < 0.0010, d = 2.02). Lower left: excerpt during the ripple (t = 0) of the normalized PTE spectrum highlights the peak at ~16 Hz. Lower right: single subject observations in the spindle-band during the ripple. e Normalized spindle occurrence relative to the information peak. Spindle occurrence was significantly enhanced prior to the information peak (−2 to −0.5 s; peak time −0.91 s; cluster test: p = 0.001, d = 3.37). f Left: spindle-locked time-frequency representation of a single subject over 30 s highlights the slow (~0.4 Hz) pattern in spindle power. Superimposed is the rescaled, time-resolved undirected MI between the MTL and scalp EEG (black). Right: average spindle power (red; demeaned) and undirected MI (black; z-scored) indicating an anti-phasic relationship (gray arrows indicate MI peaks that coincide with spindle troughs). Note that the y-axis is truncated at around 0 to highlight the side lobes. Inset: group-level results revealing a statistically significant anti-phasic relationship between spindle power and MI between the MTL and frontal EEG sensors. g MI was enhanced only after a physiologic ripple as compared to an artifactual ripple event. Given that artifactual ripples were noisy and hence, variance was increased, we averaged in the time domain (cluster in panel a) prior to statistical testing. h MI was enhanced only after a physiologic ripple as compared to an artifactual ripple event (paired t-test: t17 = 2.82, p = 0.0117, d = 0.70). Note that artifactual ripples were also not significantly different from the baseline (paired t-test: t17 = −0.04, p = 0.9699, d = 0.01), thus, indicating that no information was transferred. The asterisk indicates that physiologic ripples were different from baseline as tested in panel a. i Topographical depiction of significant MI increases. All electrodes that showed a significant MI increase (z > 1.96, uncorrected p < 0.05) are color-coded according to the z-score, while electrodes without significant modulation are depicted in white. Note, the increase in MI was widespread and not restricted to a circumscribed cortical region. j MI increases per prefrontal sub-region (RM-ANOVA: F1.64, 23.01 = 9.13, p = 0.0020, η2 = 0.05). The strongest increase was observed in the dlPFC. All error bars indicate the mean ± SEM
Fig. 7 Ripple-mediated information transfer. a Illustration of the information analysis strategy. A single EEG trace featuring SOs and spindles (in blue) and the corresponding MTL trace (in red) are depicted relative to a ripple event (see close-up; in black). Undirectional information was calculated by a moving time-window approach (center left), while directional information was calculated by keeping one window centered on the MTL ripple time point (center right; here fixed MTL window to moving EEG window; EEG to MTL information was calculated accordingly). b Left: time-resolved Mutual Information (MI) relative to the hippocampal ripple event (mean ± SEM). We detected enhanced MI between the hippocampus and prefrontal cortex ~1 s after the ripple event (1.3–2.0 s; cluster test: p = 0.0040, d = 1.06), but also observed two non-significant clusters (positive cluster from 0.25 to 0.35 s, p = 0.0849; negative cluster from −0.15 to −0.10 s, p = 0.1349). Right: spectrally resolved undirected mutual information. Significant (two-tailed p < 0.05) deviations from the average MI. Prior to ripple onset, mutual information was enhanced in the spindle and SO/delta range, which after 1 s dropped. c Left: time-resolved MI relative to the ripple event. We disentangled the information flow from prefrontal EEG to the MTL (purple) and vice versa (cyan). A significant cluster was observed after the ripple (cluster test: negative cluster from 0.15 to 0.50 s; p = 0.0130, d = 1.31), indicating increased information flow from the prefrontal EEG to the MTL. We found a second significant cluster where MI increased after ~1 s, which mainly reflected MTL to prefrontal EEG information flow (positive cluster from 1.1 to 1.85 s; p = 0.001, d = 1.08). Center/right: frequency- and directionality-resolved information flow. Outlined areas (black) reflect significant clusters (p < 0.05), where information flow was enhanced relative to the mean. Center: frequency-specific information flow from the MTL to neocortex was primarily increased in the SO (<2 Hz) and spindle-bands (~16 Hz). Right: information flow from the prefrontal EEG to MTL was not frequency-specific. d Upper panel: phase transfer entropy between the prefrontal EEG and the MTL (>1) is stronger than in the opposite direction (<1), in particular in the spindle-band (from −1–1.4 s; cluster test: p < 0.0010, d = 2.02). Lower left: excerpt during the ripple (t = 0) of the normalized PTE spectrum highlights the peak at ~16 Hz. Lower right: single subject observations in the spindle-band during the ripple. e Normalized spindle occurrence relative to the information peak. Spindle occurrence was significantly enhanced prior to the information peak (−2 to −0.5 s; peak time −0.91 s; cluster test: p = 0.001, d = 3.37). f Left: spindle-locked time-frequency representation of a single subject over 30 s highlights the slow (~0.4 Hz) pattern in spindle power. Superimposed is the rescaled, time-resolved undirected MI between the MTL and scalp EEG (black). Right: average spindle power (red; demeaned) and undirected MI (black; z-scored) indicating an anti-phasic relationship (gray arrows indicate MI peaks that coincide with spindle troughs). Note that the y-axis is truncated at around 0 to highlight the side lobes. Inset: group-level results revealing a statistically significant anti-phasic relationship between spindle power and MI between the MTL and frontal EEG sensors. g MI was enhanced only after a physiologic ripple as compared to an artifactual ripple event. Given that artifactual ripples were noisy and hence, variance was increased, we averaged in the time domain (cluster in panel a) prior to statistical testing. h MI was enhanced only after a physiologic ripple as compared to an artifactual ripple event (paired t-test: t17 = 2.82, p = 0.0117, d = 0.70). Note that artifactual ripples were also not significantly different from the baseline (paired t-test: t17 = −0.04, p = 0.9699, d = 0.01), thus, indicating that no information was transferred. The asterisk indicates that physiologic ripples were different from baseline as tested in panel a. i Topographical depiction of significant MI increases. All electrodes that showed a significant MI increase (z > 1.96, uncorrected p < 0.05) are color-coded according to the z-score, while electrodes without significant modulation are depicted in white. Note, the increase in MI was widespread and not restricted to a circumscribed cortical region. j MI increases per prefrontal sub-region (RM-ANOVA: F1.64, 23.01 = 9.13, p = 0.0020, η2 = 0.05). The strongest increase was observed in the dlPFC. All error bars indicate the mean ± SEMCC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

The citation lists are queued

They are fetched later and appear here then.

References 79 · Cited by 222

Metadata

Authors
Helfrich, R. F., Lendner, J. D., Mander, B. A., Guillen, H., Paff, M., Mnatsakanyan, L., Vadera, S., Walker, M. P., Lin, J. J., & Knight, R. T.
Affiliations
  • University of California, Berkeley (US)
  • University Medical Center Hamburg-Eppendorf (DE)
  • University of California, Irvine (US)
Corresponding author
Helfrich, Randolph F.
Venue
Nature Communications
Published
2019-08-08
Volume
10
Issue
1
Pages
3572-3572
ISSN
2041-1723 (ISSN-L)
OpenAlex
W2967042025
Cited by
222
References
79
Citations by year
2026: 21 · 2025: 32 · 2024: 40 · 2023: 39 · 2022: 38 · 2021: 32 · 2020: 19 · 2019: 1
FWCI
12.99
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • eScholarship (California Digital Library) (submittedVersion, cc-by)
Funders
U.S. Department of Health & Human Services | NIH | National Institute of Neurological Disorders and Stroke (1U19NS107609-01)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Sleep and Wakefulness Research
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Memory and Neural Mechanisms
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neural dynamics and brain function
Keywords
memory consolidation, prefrontal cortex, hippocampus, NREM sleep, slow waves, sleep spindles, hippocampal ripples, information transmission, intracranial recordings
MeSH terms
Memory Consolidation, Drug Resistant Epilepsy, Sleep, Slow-Wave, Adult, Electrodes, Implanted, Electroencephalography, Female, Hippocampus, Humans, Male, Middle Aged, Models, Psychological, Polysomnography, Prefrontal Cortex, Young Adult

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Bidirectional prefrontal-hippocampal dynamics organize information transfer during sleep in humans. https://app.cochat.ai/papers/10.1038/s41467-019-11444-x
The paper
Helfrich, R. F., Lendner, J. D., Mander, B. A., Guillen, H., Paff, M., Mnatsakanyan, L., Vadera, S., Walker, M. P., Lin, J. J., & Knight, R. T. (2019). Bidirectional prefrontal-hippocampal dynamics organize information transfer during sleep in humans. Nature Communications. https://doi.org/10.1038/s41467-019-11444-x

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.