Abstract
Intracranial recordings from the human amygdala and the hippocampus during an emotional memory encoding and discrimination task reveal increased awake sharp-wave/ripples (aSWR) after encoding of emotional compared to neutral stimuli. Further, post-encoding aSWR-locked memory reinstatement in the amygdala and the hippocampus was predictive of later memory discrimination. These findings provide electrophysiological evidence that post-encoding aSWRs enhance memory for emotional events.
Main
Multiple mechanisms have been proposed to explain the prioritized encoding of emotional experiences1–3, including the neuromodulatory effects on plasticity and the interplay between the amygdala and the hippocampus1,4,5. Several studies have found memory reinstatement during the immediate post-encoding period to be predictive of later memory performance 6,7. Sharp-wave/ripples (SWRs) are transient hippocampal oscillations (80-150 Hz), associated with synchronous neural activation in the hippocampus and the amygdala8,9, and are implicated in the binding of anatomically distributed memory traces10. Behaviorally relevant reactivation of emotional memory occurs during aSWRs 11, and disruptions of post-experience aSWR interfere with memory utilization12. Based on these findings, we hypothesized that aSWRs occurring immediately after stimulus encoding (post-encoding) facilitate emotional memory discrimination through the coordinated hippocampal-amygdala memory reinstatement. Using intracranial electroencephalographic (iEEG) recordings in epilepsy patients during the performance of an emotional encoding and discrimination task, we first confirm reports of better discrimination memory for arousing stimuli3. Next, we demonstrate that the number of aSWR events immediately after encoding is associated with both stimulus-induced arousal and the accuracy of later discrimination. Finally, the coordinated memory reinstatement between the amygdala and the hippocampus during post-encoding aSWRs is predictive of later memory discrimination performance, with the amygdala reinstatement showing a directional influence on the hippocampal reinstatement. Together, these findings provide evidence for aSWRs-mediated memory reinstatement in the amygdala and hippocampus as a mechanism accounting for better remembering of emotional experiences.
We performed simultaneous iEEG recordings from the amygdala (nelectrode = 20) and the hippocampus (nelectrode = 17, Fig. 2a) in 7 human subjects, while performing an emotional memory encoding and discrimination task13,14 (Methods, Fig. 1a). During the encoding stage, subjects were presented with a stimulus (image; stimulus encoding) and asked to rate the stimulus valence as negative, neutral, or positive (post-encoding/response). During the retrieval stage, subjects were presented with one of the 3 types of stimuli - Repeats (identical), Lure (slightly different) or Novel (stimuli not seen during encoding) - and classified each stimulus as “New” or “Old.”
Memory discrimination is defined as the correct classification of: 1) Repeat stimuli as Old, 2) Novel stimuli as New, or 3) Lure stimuli as New. Subjects classified Repeat stimuli and Novel stimuli with high accuracy (Repeat: 89.4 ± 2.4%, Novel: 93.9 ± 1.4%; Fig. 1b). Memory discrimination accuracy was lower for Lure stimuli, relative to both Repeat or Novel stimuli (Lure: 61.5 ± 3.7 %; pNovel vs Lure < 0.001, t = 8.36; pRepeat vs Lure < 0.001, t = 6.13, paired t-test), reflecting similarity-induced memory interference. Indeed, there was a strong negative association between subjects’ stimulus discrimination ability and stimulus similarity rating (p = 0.039, t = -2.06, see Methods, Fig. 1c-d). Stimulus-induced arousal (irrespective of valence) was associated with better memory discrimination, confirming previous reports1–3 (p = 0.047, t = 1.98, Fig. 1c-d, Extended Data Fig. 1).
We defined the post-encoding period as the interval between stimulus offset and subjects’ stimulus valence rating response (Fig. 1a). We tested the association of post-encoding aSWR occurrence (i.e., the number of aSWRs) with the stimulus emotional content (stimulus-induced arousal and valence) and correct discrimination during retrieval. Higher post-encoding aSWR occurrence was associated with stimulus-induced arousal (p = 0.03, z = -2.2, Wilcoxon signed-rank test, Fig. 2c) and also predicted correct discrimination during retrieval (p = 0.03, z = -2.2, Wilcoxon signed-rank test, Fig. 2c), but was not associated with stimulus valence (p = 0.77, F(2, 15) = 0.25, one-way ANOVA; Extended Data Fig. 3). Taken together, these results provide the first report of post-encoding aSWRs as a potential electrophysiological mechanism for enhanced memory discrimination of arousing stimuli, previously characterized at behavioral level2,3,15. Furthermore, the positive associations between aSWRs and stimulus-induced arousal/later discrimination were present in all individual subjects (Fig. 2c). The post-encoding response time (RT) did not differ based on stimulus-induced arousal (p = 0.2, z = 0.7, RThigh-arousal = 0.8 ± 0.1 sec; RTlow-arousal = 0.6 ± 0.2 sec) or later discrimination (p = 0.25, z = 0.6, RTcorrect = 0.7 ± 0.2 sec, RTincorrect = 0.7 ± 0.3, Wilcoxon signed-rank test). Therefore, the associations between stimulus-induced arousal or correct discrimination and post-encoding aSWR occurrence were unrelated to post-encoding duration. Associations between aSWR and stimulus-induced arousal/later correct discrimination accuracy were selective for the post-encoding time window. These relationships were absent for the stimulus encoding or the retrieval task stage (p > 0.05, Wilcoxon signed-rank test; Fig. 2c, Extended Data Fig. 3, 4). The aSWRs probability was significantly higher during low theta power periods (Extended Data Fig. 5), consistent with observations that cholinergic tone promotes theta oscillations and suppresses SWRs10,12. In addition, aSWRs did not overlap with increased broadband gamma power, suggesting that aSWRs are distinct from non-specific broadband power fluctuations16 (Extended Data Fig. 5).
Recent studies suggest that post-encoding memory reinstatement supports successful subsequent memory retrieval 6,7. Meanwhile SWR is associated with reactivation of pre-established neuronal patterns17. We hypothesized that memory reinstatement during the post-encoding aSWR window could enhance later memory discrimination. Distinct neural populations have been proposed to represent individual stimuli, resulting in stimulus-specific high-frequency activity (HFA) patterns18,19. We, thus, quantified memory reinstatement as the Spearman correlation between HFA power spectral vectors (PSVs), for each combination of the encoding-response time bins from the same trial (Extended Data Fig. 6). Next, we computed the average reinstatement activity during ± 250 msec around post-encoding aSWR peaks. The reinstatement significance was determined relative to a null distribution, obtained by circular jittering of aSWR timestamps. The post-encoding aSWR-locked memory reinstatement was stronger for arousing and correctly discriminated stimuli (Extended Data Fig. 7). To assess specific contributions of the amygdala and the hippocampus to this phenomenon, we calculated post-encoding memory reinstatement for each region, relative to aSWR peak (Fig. 3a). The significant reinstatement period in the amygdala consisted of two intervals, the first starting slightly earlier and overlapping with the hippocampal reinstatement (−105 to -50 msec), and a second period following the hippocampal reinstatement (40 to 200 msec). The significant reinstatement period in the hippocampus lasted from -100 to 50 msec (Fig. 3b). These results demonstrated region-specific timing of the post-encoding aSWR-locked memory reinstatement in the amygdala and the hippocampus. Next, we tested for the temporal compression17 of post-encoding aSWR-locked reinstatement (no compression, 2x, 4x, and 6x compression) and showed the strongest aSWR-locked reinstatement with no compression (Extended Data Fig. 8). We then analyzed the association of the post-encoding memory reinstatement with the stimulus-induced arousal and later discrimination. Remarkably, we observed a region-specific double dissociation. Specifically, the amygdala, not the hippocampus, showed a positive association between aSWR-locked memory reinstatements and the stimulus-induced arousal (AMY: -80 to -10 msec, p = 0.035; HPC: p > 0.05, see Methods; Fig. 3c). In contrast, the hippocampus, but not the amygdala, revealed a positive association between aSWR-locked memory reinstatement and later correct discrimination (AMY: p > 0.05; HPC: -15 to 90 msec, p = 0.008, see Methods; Fig. 3c). To summarize, post-encoding aSWR-locked memory reinstatements in the amygdala and the hippocampus followed distinct temporal dynamics and were associated with reactivation of distinct aspects of encoded stimuli (i.e., the amygdala for stimulus-induced arousal and the hippocampus for later discrimination accuracy).
In rodents, the coordinated memory reactivation in the amygdala and hippocampus during sleep SWRs is proposed to bind neuronal ensembles encoding emotional and spatial information, respectively20. We reasoned that a similar interaction between the amygdala and the hippocampus exists in which cross-regional post-encoding aSWR-locked memory reinstatement facilitates later discrimination. We hypothesized that the reinstatement in both structures co-occurs during the same aSWR events and follows a consistent temporal dynamic. To test this, we separately computed aSWR-locked joint memory reinstatement for the correctly and incorrectly discriminated stimuli (Methods). A significant joint aSWR-locked memory reinstatement in the amygdala and hippocampus was present during the post-encoding period only for correctly discriminated stimuli (Fig. 3d; Extended Data Fig. 9). Specifically, the amygdala reinstatement preceded the hippocampal reinstatement by ∼100 msec. Further, mutual information analysis showed a significant unidirectional influence from the amygdala to the hippocampus before aSWR peak (−70 to -30 msec, p = 0.038; see Methods; Fig. 3e). To conclude, aSWR-mediated coordination of memory reinstatement in the amygdala and the hippocampus promotes later successful discrimination.
Rodent studies have implicated the SWRs in the retrieval and consolidation of emotional memory. However, it is unclear whether it supports the memory benefits of emotional experience21. Our study reveals an association of higher aSWR occurrence with stimulus-induced arousal and subsequent correct stimulus discrimination, providing direct evidence for aSWR-mediated strengthening of emotional memory. Interestingly, the higher aSWRs occurrence has been shown in rodents, after exposure to a novel or reward-associated context22. Together, this suggests that aSWRs may play a general role in the selective enhancement of salient experiences23.
Notably, such association is specific to the post-encoding period that starts immediately after memory encoding, when memory retrieval is essential to rate the emotional content of the stimuli. This finding supports theoretical assumptions that SWRs mediate both the retrieval of stored representation utilized in decision-making, and the strengthening of the same representation, contributing to memory consolidation22.
Next, we aimed to discern the link between the aSWR-associated interaction between the amygdala and hippocampus during post-encoding and subsequent memory effect. We found the aSWRs were accompanied by memory reinstatement during the post-encoding period. Specifically, the reinstatement in the amygdala appears shortly before the aSWR peak and shows association with arousing stimuli, while the hippocampal reinstatement appears around the aSWR peak and shows associations with correct subsequent memory discrimination. Moreover, the co-occurrence of the amygdala and the hippocampal reinstatement during the same post-encoding aSWR events - with the amygdala reinstatement leading hippocampal by ∼100 msec - is predictive of subsequent correct memory discrimination. This finding suggests that the coordinated reinstatement in the amygdala and hippocampus during aSWR is responsible for combining emotional and contextual aspects of the memory20,21.
Both the joint-reinstatement and mutual information analyses further confirm the predictive validity of directional influence from the amygdala to the hippocampus before aSWRs on correct discrimination, establishing a link between the amygdala reinstatement and memory discrimination as a physiological mechanism of emotional memory enhancement. Together, our data support a model wherein the memory reinstatement in the amygdala, triggered by emotional stimuli, elicits amygdala-hippocampal aSWR-associated memory reinstatement, enabling the coordinated joint-reinstatement, which facilitates subsequent memory performance.
Competing Interests statement
The authors declare no competing interest.
Extended data
Methods
Subjects
Intracranial electroencephalography (iEEG) recordings were obtained from 7 subjects (3 females; mean age ± SD = 33 ± 16), undergoing presurgical monitoring of epileptic foci at the University of California Irvine Medical Center (UCIMC) Epilepsy Monitoring Unit. The individual subject demographic information is shown in Table 1. Only the subjects with the correct discrimination rate of Novel trials >= 85% (see Emotional memory encoding and discrimination task) were included in the analysis. Electrode placements were determined entirely based on clinical considerations. All the research procedures were approved by the UCI Institutional Review Board and data was collected following informed consent.
Statistics
All the statistical tests were performed with the individual subject as the unit of analysis. Unless stated otherwise, all the parametric statistical tests (e.g., Wilcoxon signed-rank test, t-test) were two-tailed. The effects of valence, stimulus-induced arousal and similarity on stimulus discrimination (Fig. 1c) were assessed using the logistic linear mixed-effect model (for details, see Behavioral Analysis). Conditional comparisons of aSWR occurrence (correct/incorrect discrimination or high/low arousal; Fig. 2c) were done using the Wilcoxon signed rank test (p < 0.05). Statistical significance of aSWR-locked memory reinstatement strength (Fig. 3b) was assessed by comparing the real test statistics with empirical null distribution, obtained using Monte Carlo method (for details, see Representational Similarity Analysis). We implemented the cluster-based nonparametric permutation test1 to assess the conditional differences (correct/incorrect discrimination or high/low arousal) of memory reinstatement strength (Fig. 3c), mutual information (Fig. 3e), by randomly shuffling the conditional trial labels 1000 times (for details, see Representational Similarity Analysis).Similarly, the significant temporal windows for the cross structure aSWR-locked joint memory reinstatement (Fig. 3d) were assessed by comparing to empirical null distribution (for details, see Joint-reinstatement Analysis).
Emotional memory encoding and discrimination task
The emotional memory encoding and discrimination (EMOP) task consists of encoding and discrimination blocks. During the encoding block (148 trials), each trial consists of a cross fixation (1000 msec), followed by stimulus encoding (2000 msec) and self-paced post-encoding response period (up to 2000 msec). During the post-encoding response period, subjects are asked to classify the stimulus emotional valence as either negative, neutral or positive, using the corresponding laptop key. During the retrieval block (290 trials), trial time structure is identical to encoding phase. Following the cross fixation (1000 msec), the subjects are presented for 2000 msec with a stimulus identical (Repeat, 54 trials), slightly different (Lure, 97 trials) or unrelated (Novel, 139 trials) to previously encoded stimuli. Next, during the self-paced memory discrimination epoch (up to 2000 msec), subjects are asked to discriminate if the presented stimulus was seen during encoding (Old) or not (New). Correct discrimination is defined as classifying the Repeat stimuli as Old and Lure or Novel stimuli as New. The stimuli were selected from the continuous distributions across the valence and stimulus-induced arousal axes (Extended Data Fig. 1). The same set of stimuli was used across subjects. In addition, the valence, arousal and similarity of each stimulus were rated by separate cohorts of healthy subjects. Specifically, a first cohort (N = 50, 32 females; age mean ± SD = 22 ± 5) rated the stimulus emotional valence on a continuous scale (range 1-9, with 1 denoting the most negative, 9 the most positive, and 5 neutral valence). Stimuli were assigned in Negative (valence <= 3.5), Neutral (3.5 > valence < 6) or Positive (valence >= 6) groups. Another cohort of healthy subjects (N = 16, 4 females; age mean ± SD = 23 ± 5) rated the stimulus-induced emotional arousal on a scale 1 - 9 (1 being the least and 9 being the most arousing). Finally, a third cohort (N = 17, 11 females; age mean ± SD = 20 ± 1) examined relative similarity on the scale 1-8 2. The high correspondence of stimulus valence ratings obtained from study subjects and healthy population (match rate = 85.3 ± 1.3%) suggests the intact emotional processing in study subjects (Extended Data Fig. 1).
Behavioral Analyses
To assess the effects of valence, stimulus-induced arousal and similarity on Lure stimulus discrimination, we implemented the logistic linear mixed-effect model In this model, y indicates the responses across the individual Lure discrimination trials (0-Old; 1-New), X = [x1 x2 x3]T denotes three fixed effect regressors (encoded stimulus valence and arousal as well as similarity between the encoded and Lure stimulus), Z = [z1]T denotes random effect regressor (subject identity), βand u denote the fixed and random-effect regression coefficients, and ε denotes the error term. The model includes random intercept to incorporate individual subject differences. We normalized the valence, stimulus-induced arousal and similarity values relative to the scale of 0 to 1. The statistics reported in Fig. 1c corresponds to the fixed-effect coefficients β.
Data collection
The behavioral experiment was administered using the PsychoPy2 software3 (Version 1.82.01). The laptop was placed at a comfortable distance in front of the subject. The iEEG signal was recorded using a Nihon Kohen system (256 channel amplifier, model JE120A), with an analog high-pass filter (0.01 Hz cutoff frequency) and sampling frequency 5000 Hz.
Electrode localization
We localized each electrode using pre-implantation structural T1-weighted MRI scans (pre-MRI) and post-implantation MRI scans (post-MRI) or CT scans (post-CT). Specifically, we co-registered pre-MRI and post-MRI (or post-CT) scans by means of a rigid body transformation parametrized with three translation in x,y,z directions as well as three rotations using Advanced Normalization Tools (ANTs https://stnava.github.io/ANTs/). We implemented a high-resolution anatomical template with the label of medial temporal lobe subfields2 to guide the localization for individual electrodes. We resampled the template with 1mm isotropic, and aligned it to pre-MRI by ANTs Symmetric Normalization4 to produce a subject-specific template. The electrode localization was identified by comparing the subject-specific template subfield area with electrode artifacts.(Fig. 2a) The localization results were further reviewed by the neurologist (J.J.L.).
Preprocessing
The signal preprocessing was done using the custom-written MATLAB code (Version 9.7) and Fieldtrip Toolbox5. The 60 Hz line noise and its harmonics were removed using a finite impulse response (FIR) notch filter (ft_preprocessing.m function in FieldTrip). The EEG signal was down-sampled to 2000 Hz, demeaned and high-passed filtered (cutoff frequency 0.3 Hz). The power spectrum density (PSD) was computed using the multitaper method with the Hanning window (ft_freqanalysis.m function in FieldTrip). All the channels were re-referenced to the nearest white matter channel from the same depth electrode, based on the electrode localization results. The interictal epilectic discharges were manually marked by an epileptologist (J.J.L.), using the ft_databrowser.m function in FieldTrip. The channels with severe contamination and trials containing epileptiform discharges were excluded from further analyses.
Awake sharp-wave/ripple detection
Following the removal of channels with excessive epileptic activity and individual trials containing visually identified interictal epilectic discharges, awake sharp-wave/ripples (aSWRs) were detected on the remaining hippocampal channels, using the Freely Moving Animal Toolbox (FMA; http://fmatoolbox.sourceforge.net/). First, the iEEG traces from the trials used in the analysis were concatenated. Next, concatenated traces were bandpass-filtered (80 - 150 Hz, Chebyshev 4th order filter, function filtfilt.m in Matlab) and the voltage values during periods ± 75 msec around the trial onsets/offsets were set to zero, to avoid the edge effects resulting from filtering discontinuous traces. The analytical amplitude was obtained by computing the absolute value of Hilbert-transformed filtered trace (function hilbert.m in Matlab) and z-scored (Extended Data Fig. 2a). Detected events were considered aSWRs if the z-scored analytical amplitude remained above the lower threshold (z = 2) for 20 - 100 msec and if the peak value during this period exceeded higher threshold (z = 5). Only the channels with >150 detected aSWR events were used in the analysis. If the multiple channels from a single subject passed this criteria, a channel with highest number of detected aSWRs was selected for further aSWR-related analysis. Due to the low number of detected aSWRs, one subject was eliminated from the aSWR-related analysis.
Unsupervised decomposition of iEEG signal
To assess the memory reinstatement, high-frequency activity (HFA; 30-280 Hz) was used as an indirect measure of local populational activity6–9. To avoid the effect of low-frequency harmonics on the HFA estimate, we applied the Ensemble Empirical Mode Decomposition7,10 (EEMD; https://github.com/leeneil/eemd-matlab.git). Briefly, the EEMD decomposes a non-stationary signal into its elementary components, referred to as intrinsic mode functions10 (IMFs; Extended Data Fig. 6). The procedure iteratively applies an empirical mode decomposition algorithm, while adding white noise to prevent the mode mixing10,11. Using this approach, decomposition output entirely depends on the signal’s intrinsic properties, avoiding prior assumptions7,10,11. The resulting IMFs captured several canonical spectral features consistently across subjects and anatomical structures (Extended Data Table 2). Finally, the HFA time-series on individual channels were reconstructed by summing the channel-specific IMFs with center frequencies > 30 Hz7.
Time-frequency representation of the HFA
The instantaneous spectral power at each time-frequency bin was derived from the reconstructed HFA time series (x), using a wavelet transform12,13. This approach consists of convolving the time series xwith a set of Morlet wavelets, parametrized by a range of cycle numbers (n = 2, 3, …, 10) at a given frequency f, and computing the geometric average of resulting spectral power at each time-frequency bin: This approach results in a high temporal and frequency resolution, facilitating the detection of narrow-band, transient oscillatory events12,13. The wavelet center frequencies were within 30 - 280 Hz range, with 1 Hz increments. The wavelet cycle number range (2-10) is commonly used14. To avoid the edge effects, this procedure was applied on the entire individual recording sessions, and the resulting time-frequency response matrices were segmented into trial epochs (starting -1000 msec prior to stimulus onset and ending 1000 msec after the response time). The power within each trial epoch was then normalized by z-transforming each frequency bin and subtracting the average pre-trial baseline (−1000 - 0 msec, relative to stimulus onset14).
Representational Similarity Analysis (RSA)
The representational similarity was quantified as the Spearman correlation between the HFA power spectral vectors (PSVs), for each combination of the encoding-response time bins from the same trial15–18 (Extended Data Fig. 6). Specifically, the instantaneous spectral power at each frequency was estimated for 100 msec time bins (10 msec step size, 90% overlap), producing the time bin - specific power spectrum vectors (PSV), spanning the encoding (2 sec time window after stimulus onset) and post-encoding response (time window after stimulus offset and before button press) periods: Similar to previous studies15–20, we computed Spearman’s correlation as a measure of PSV similarity between the encoding time t1 and response time t2 for each encoded stimulus, , with rg representing the ranking operator on the vector , and σ the variance of the vector. This produced a trial-specific two-dimensional similarity matrices, containing all the combinations of encoding (t1) and response (t2) time bins (Extended Data Figure 6d). The correlation coefficients r were then Fisher transformed, with the resulting coefficients following Gaussian distribution. The region-specific (amygdala and hippocampus) similarity matrices were averaged across trials within individual subjects, and used for group-level statistical analysis.
aSWR-locked memory reinstatement
Memory reinstatement during individual post-encoding time bins was computed by averaging the bin-specific similarity with the encoding period (200 time bins over 2 sec), resulting in a memory reinstatement time series. To obtain the aSWR-locked memory reinstatement, we averaged the memory reinstatement within ± 250 msec around the individual aSWR peak times, separately for amygdala and hippocampus (Fig. 3a). We next tested whether the memory reinstatement is locked to aSWRs (Fig. 3b), by comparing the grand-average aSWR-locked reinstatement trace with an empirical null distribution obtained from Monte Carlo simulation. Specifically, we circularly randomly jittered the aSWR peak times within ± 500 msec window for 1000 times, obtaining an empirical null distribution of memory reinstatement strength.
To test whether the aSWR-locked reinstatement is associated with stimulus-induced arousal and later discrimination (Fig. 3c), we first derived the aSWR-triggered reinstatement, a metric taking the time-locked specificity relative to aSWR peak time into account. For every per-aSWR reinstatement trace around aSWR peak time, we circularly jittered the time as the procedure described above. This results in an empirical null distribution of reinstatement (i.e., correlation coefficient) for every time point around aSWR. We normalized the real reinstatement by z-scoring with mean and standard deviation of the null distribution. We referred to the resulting z-value as aSWR triggered reinstatement and it follows Gaussian distribution. We quantified the aSWR-locked reinstatement difference between the high/low arousal and between correct/incorrect discrimination at every time point by t-test, and corrected for the multiple comparisons using cluster-based nonparametric permutation test. Specifically, we performed the group-level comparisons using paired t-test and identified contiguous time bins with the p < 0.05, defined as clusters. The t-values within each cluster were summed as the cluster statistics. We created an empirical null distribution by shuffling the conditional trial labels 1000 times where the maximum cluster statistics was identified for each permutation. It is considered as statistically significant if the real t-sum cluster statistics exceeded the 95% percentile of the null distribution.
Cross-structure joint aSWR-locked memory reinstatement
The cross-structure joint aSWR-locked memory reinstatement was obtained by calculating the outer product between the structure-specific reinstatement traces (hippocampus and amygdala) during post-encoding aSWR windows. The resulting joint reinstatement matrices were averaged across the individual aSWRs for each subject, separately for later correctly or incorrectly discriminated trials. To assess the statistical significance of joint cross-structure memory reinstatement, we performed a Monte Carlo simulation to generate an empirical null distribution by circularly jittering the aSWR peak times. The reinstatement significance was defined as exceeding the 95% percentile of null distribution (Fig. 3d).
Dual states analyses
Recorded periods were divided into low-and high-theta (3 - 10 Hz) or gamma (30 - 250 Hz) periods, based on the subject-specific power median split. The aSWR occurrences are defined as the proportions of aSWRs occurring during each period. The aSWR occurrence comparisons between the low- and high-theta or gamma periods were performed using one-tailed Wilcoxon signed-rank test (p < 0.05; Extended Data Figure 9).
Mutual information
Mutual information (MI)14,21 is a method for quantifying the amount of information shared between the variables of interest. In electrophysiology, MI is applied to test for the presence and directionality of information flow between the multiple time-series. We applied MI to assess the directional influence between the memory reinstatement in amygdala and hippocampus during the post-encoding aSWR windows (Fig. 3e). First, the structure-specific memory reinstatement traces from the amygdala and hippocampus were obtained around each aSWR event (± 250 msec; see aSWR-locked memory reinstatement). Next, we calculated the MI between the amygdala and hippocampal memory reinstatement traces, using the 200 msec bin size (10 msec step size), covering the ± 250 msec window around aSWR peaks. For each time bin, the reinstatement strength was binned into 10 bins (with uniform bin count), consistently across the subjects and conditions. The MI between the time series X and Y was defined as , where p(xi) and p(yj) represented the marginal probability of signals X and Y, p(xi, yj) indicated their joint probability, while m and n represented the numbers of reinstatement strength bins for time series X and Y14,21. To test the directionality of information flow, we calculated the time-lagged MI by shifting one time series relative to another across all the time bin combinations. The MIAMY → HPC and MIHPC → AMY at individual time bins were defined as the mean of all the subsequent time-lagged MI bins in the other region14,22. We defined the MI directional influence as the significant difference between the MIAMY → HPC and MIHPC → AMY, assessed using Wilcoxon signed-rank test for each time bin. Correction for multiple comparisons was performed using the cluster-based nonparametric permutation test.
Acknowledgement
The authors thank all the participants for taking part in the study, as well as the nurses, technicians, and physicians at the UCI Epilepsy Unit. This work was supported by NIH Grant 1U19NS107609-01 to R.T.K. (subcontract to J.J.L).