Bayesian meta-analysis reveals the mechanistic role of slow oscillation-spindle coupling in sleep-dependent memory consolidation

  1. Thea Ng
  2. Eunsol Noh
  3. Rebecca MC Spencer  Is a corresponding author
  1. Neuroscience & Behavior Program, Mount Holyoke College, United States
  2. Department of Mathematics & Statistics, Mount Holyoke College, United States
  3. Neuroscience & Behavior Program, University of Massachusetts, United States
  4. Department of Psychological & Brain Sciences, University of Massachusetts, United States
  5. Institute of Applied Life Sciences, University of Massachusetts, United States
14 figures, 15 tables and 8 additional files

Figures

Measurement of phase-amplitude coupling (PAC) in slow oscillation and spindle events.

(A) The origin of neural oscillations during sleep. SO slow oscillation, SP sleep spindle, SWR sharp wave ripple. In each subgraph, the vertical line indicates the typical amplitude of that sleep wave, while the horizontal line indicates the typical frequency and duration. Note that SPs also propagate along the cortex, and the figure only displays the origin. (B) Electrophysiology representation diagram of the SO-SP coupling. SO and SP amplitudes are normalized. The phase of SOs when SPs are at their maximum instantaneous amplitude is recorded as the coupling phase. The occurrence of SPs and SO-SP coupling is not necessarily continuous as shown in the diagram. (C) Coupling preferred phase and strength diagram. (Left) The phase and strength used in the circular plot are simulated data from existing dataset for visualization purposes only. At the group level, the mean circular direction shows the preferred SO phase, while the mean vector length shows the strength of the precise coupling. (Right) The phase of SO peaks is noted as 0, while the phase of SO troughs is noted as ±π.

Hierarchical diagram of coupling measures.

PETH peri-event time histogram, MVL mean vector length, MI modulation index. SPcSO Percentage of SPs coupled with SOs in all SP events. In contrast to the term ‘frequency’ used throughout the text in reference to the neural oscillation, the coupling prevalence in the diagram indicates the occurrence of SO–SP coupling events.

Risk of bias assessment summary plot adapted from ROBINS-I (Sterne et al., 2016).

The most significant heterogeneity is revealed in the measurement of outcome, while the overall assessment indicated a moderate risk of bias across studies after requesting unreported results and data transformation. Based on the complexity of the type of measures involved in phase-amplitude coupling analysis, we believe that this degree of risk of bias is acceptable. Specific evaluations for each study were reported in Appendix 1.

Forest and regression plots for the association between SO-SP coupling phase and memory retention.

(A) Overall model forest plot at study-level. Dashed lines indicate the 95% credible interval (CrI) of the pooled effect size. The black point and error bar for each study show the adjusted estimation of effect size and 95% CrI combining data and prior information. The gray dots under each distribution show raw effect sizes of each study. Effect size-level plots can be found in Supplementary file 3. (B) Meta regression plot with age as moderator. Blue lines represent 200 overplotted spaghetti fit lines to visualize predictions. (C) Moderator-level forest plot. Each box represents a type of moderator. Mixed effect sizes with mixed conditions from different factor levels listed above. Weight Stacked weight of each moderation model in the paired model performance comparison between the moderator and overall (intercept-only) model. The stacked weight of the overall model in each pair of comparisons can be calculated as 1 - weight of the moderation model.

Figure 4—source data 1

Subtable of effect size-level metadata included in the coupling phase–memory analysis.

https://cdn.elifesciences.org/articles/101992/elife-101992-fig4-data1-v1.zip
Preferred slow oscillation-fast spindle coupling phase and its association with memory retention.

(A) Quadratic regression of the phase-memory association under different regions of PSG channels aggregated from studies included in the meta-analysis. 0 peak of SO upstate; ±π trough of SO downstate; r circular-linear correlation coefficient; rz standardized circular-linear correlation coefficient. Bars represent the mean memory retention scores per π/4 radian (45°). The dashed vertical line represents the mean preferred phase across studies. The colored quadratic fit line represents the direction of the relationship. The direction of their relationship gradually flips as the PSG channel moves from the front to the posterior area. Only under the frontal and central channels do fit lines display a quadratic relationship, with a peak of memory score near the up-state peak of SOs. In posterior channels, in contrast, the relationship is convex, although it is not significant. Non-significant quadratic regressions between SO-slow SP coupling and memory are reported in Appendix 6. (B) Posterior distributions of mean preferred phases from the Bayesian circular mixed-effect model. The circular posterior distribution is shown in the top-right corner, and the area between two black lines is projected on a linear scale in the main graph. The vertical line reflects the up-state peak of SOs. Points and error bars denote the mean and 95% credible intervals of phases detected from each channel cluster. Phase values are reported as radians. (C) Circular plot of the preferred coupling phase. From top to bottom, frontal, central, and posterior. The direction of each colored dot represents the preferred coupling phase of each subject recorded from PSG channels in each cluster. The direction of the mean resultant vector indicates the mean preferred coupling phase across subjects, the width indicates the 95% credible interval of the mean coupling phase, the length from 0 (center) to 1 (circumference) indicates the consistency of coupling phase across subjects.

Forest and regression plots for the association between SP amplitude and memory retention.

(A) Overall model forest plot at study-level. The solid vertical line represents the mean Pearson correlation coefficient under the null hypothesis. Dashed lines indicate the 95% credible interval (CrI) of the pooled effect size. The black point and error bar for each study show the adjusted estimation of effect size and 95% CrI combining data and prior information. The gray dots under each distribution show raw effect sizes of each study. Effect size-level plots can be found in Supplementary file 3. (B) Meta regression plot with age as moderator. Blue lines represent 200 overplotted spaghetti fit lines to visualize predictions. (C) Moderator-level forest plot. Each box represents a type of moderator. Mixed effect sizes with mixed conditions from different factor levels listed above. Weight Stacked weight of each moderation model in the paired model performance comparison between the moderator and overall (intercept-only) model. The stacked weight of the overall model in each pair of comparisons can be calculated as 1 weight of the moderation model.

Figure 6—source data 1

Subtable of effect size-level metadata included in the spindle amplitude memory analysis.

https://cdn.elifesciences.org/articles/101992/elife-101992-fig6-data1-v1.csv
Forest and regression plots for the association between coupling strength and memory retention.

(A) Overall model forest plot at study-level. The solid vertical line represents the mean Pearson correlation coefficient under the null hypothesis. Dashed lines indicate the 95% credible interval (CrI) of the pooled effect size. The black point and error bar for each study show the adjusted estimation of effect size and 95% CrI combining data and prior information. The gray dots under each distribution show raw effect sizes of each study. Effect size-level plots can be found in Supplementary file 3. (B) Meta regression plot with age as moderator. Blue lines represent 200 overplotted spaghetti fit lines to visualize predictions. (C) Moderator-level forest plot. Each box represents a type of moderator. Mixed effect sizes with mixed conditions from different factor levels listed above. Weight Stacked weight of each moderation model in the paired model performance comparison between the moderator and overall (intercept-only) model. The stacked weight of the overall model in each pair of comparisons can be calculated as 1 weight of the moderation model.

Figure 7—source data 1

Subtable of effect size-level metadata included in the coupling strength–memory analysis.

https://cdn.elifesciences.org/articles/101992/elife-101992-fig7-data1-v1.zip
Forest and regression plots for the association between coupling percentage and memory retention.

(A) Overall model forest plot at study-level. The solid vertical line represents the mean Pearson correlation coefficient under the null hypothesis. Dashed lines indicate the 95% credible interval (CrI) of the pooled effect size. The black point and error bar for each study show the adjusted estimation of effect size and 95% CrI combining data and prior information. The gray dots under each distribution show raw effect sizes of each study. Effect size-level plots can be found in Supplementary file 3. (B) Meta regression plot with age as moderator. Blue lines represent 200 overplotted spaghetti fit lines to visualize predictions. (C) Moderator-level forest plot. Each box represents a type of moderator. Mixed effect sizes with mixed conditions from different factor levels listed above. Weight Stacked weight of each moderation model in the paired model performance comparison between the moderator and overall (intercept-only) model. The stacked weight of the overall model in each pair of comparisons can be calculated as 1 weight of the moderation model.

Figure 8—source data 1

Subtable of effect size-level metadata included in the coupling percentage–memory analysis.

https://cdn.elifesciences.org/articles/101992/elife-101992-fig8-data1-v1.zip
PRISMA flow diagram of literature search, screening, and inclusion for systematic review and meta-analysis.
Appendix 1—figure 1
Risk of bias (ROB) assessment for individual studies.
Appendix 4—figure 1
Funnel Plot and Egger regression for assessing publication bias.

Each colored dot represents one effect size and corresponding standard error. The outer border of the transparent triangle indicates the area where it is anticipated that 95% of the included studies would fall if there were no publication biases present. The red dashed line represents the superimposed Egger’s regression line. We noted that with sufficient sample sizes, the phase-memory association and strength-memory association tend towards a moderate magnitude.

Appendix 5—figure 1
Comparison of sampling distributions of standardized Circlin rz under null.
Appendix 5—figure 2
Sampling distributions of standardized Circlin rz drawn from population correlations.

(A) Sampling distributions of standardized Circlin drawn from populations with moderate correlations (rz = 0.3–0.4). Vertical dashed lines indicate the true population values from which samples were generated. (B) Sampling distributions of standardized Circlin drawn from populations with strong correlations (rz = 0.6–0.7).

Appendix 6—figure 1
Preferred slow oscillation-slow spindle coupling phase and its association with memory retention.

(A) Quadratic regression of the phase-memory association under different regions of PSG channels aggregated from studies included in the meta-analysis. 0 peak of SO upstate; ±π trough of SO downstate; r circular-linear correlation coefficient; rz standardized circular-linear correlation coefficient. Bars represent the mean memory retention scores per π/4 radian (45°). The dashed vertical line represents the mean preferred phase across studies. The colored quadratic fit line represents the direction of the relationship. None of the PSG channels displays a typical quadratic relationship around the down-state trough of SOs. (B) Posterior distributions of mean preferred phases from the Bayesian circular mixed-effect model. The circular posterior distribution is shown in the top right corner, and the area between two black lines is projected on a linear scale in the main graph. The vertical line reflects the down-state trough of SOs. Dots and error bars denote the mean and 95% credible intervals of phases detected from each channel cluster. Phase values are reported in radians. (C) Circular plot of the preferred coupling phase. From top to bottom, frontal, central, and posterior. The direction of each colored dot represents the preferred coupling phase of each subject recorded from PSG channels in each cluster. The direction of the mean resultant vector indicates the mean preferred coupling phase across subjects, the width indicates the 95% credible interval of the mean coupling phase, the length from 0 (center) to 1 (circumference) indicates the consistency of coupling phase across subjects.

Tables

Table 1
Result of directional hypothesis tests for each pair of factor levels (conditions) in overall and each moderation model of the coupling phase-memory association.
ModeratorOverallMemory TaskAgeSpindlePSG ChannelStageBout
ConditionH1VerbalEmotionalSpatialYoungerFastFrontalFrontalCentralN2Night
ControlH0MotorMotorMotorOlderSlowCentralPosteriorPosteriorSWSNap
BF1058.351.301.090.07160.9411.3913.586.130.861.742.23
Probability0.980.570.520.060.990.920.930.860.460.630.69
  1. Condition conditions hypothesized to be associated with stronger phase-memory association than other factor levels; Control Variables hypothesized to be associated with weaker phase-memory association; BF10 Bayes factor in favor of H1 over H0; N2 nREM2 stage; SWS slow-wave sleep.

Table 2
Result of directional hypothesis tests for each pair of factor levels (conditions) in overall and each moderation model of the SP amplitude-memory association.
ModeratorOverallMemory TaskAgeSpindlePSG ChannelStageBout
ConditionH1VerbalEmotionalSpatialYoungerFastFrontalFrontalCentralN2Night
ControlH0MotorMotorMotorOlderSlowCentralPosteriorPosteriorSWSNap
BF108.2822.191.509.303.7012.211.8824.6713.655.372.54
Probability0.890.960.600.900.790.920.650.960.930.840.72
  1. Condition conditions hypothesized to be associated with stronger amplitude-memory association than other factor levels; Control Variables hypothesized to be associated with weaker amplitude-memory association; BF10 Bayes factor in favor of H1 over H0; N2 nREM2 stage; SWS slow-wave sleep.

Table 3
Result of directional hypothesis tests for each pair of factor levels (conditions) in overall and each moderation model of the coupling strength-memory association.
ModeratorOverallMemory TaskAgeSpindlePSG ChannelStageBout
ConditionH1VerbalEmotionalSpatialYoungerFastFrontalFrontalCentralN2Night
ControlH0MotorMotorMotorOlderSlowCentralPosteriorPosteriorSWSNap
BF10111.042.7710.621.109.971.912.811.570.700.332.66
Probability0.990.730.910.520.910.660.740.610.410.250.73
  1. Condition conditions hypothesized to be associated with stronger strength-memory association than other factor levels; Control Variables hypothesized to be associated with weaker strength-memory association; BF10 Bayes factor in favor of H1 over H0; N2 nREM2 stage; SWS slow-wave sleep.

Table 4
Results of directional hypothesis tests for each pair of factor levels (conditions) in overall and each moderation model of the coupling percentage–memory association.
ModeratorOverallMemory TaskAgeSpindlePSG ChannelStageBout
ConditionH1VerbalEmotionalSpatialYoungerFastFrontalFrontalCentralN2Night
ControlH0MotorMotorMotorOlderSlowCentralPosteriorPosteriorSWSNap
BF100.380.050.170.222.160.285.153.381.242.020.44
Probability0.280.050.140.180.680.220.840.770.550.670.31
  1. Condition conditions hypothesized to be associated with stronger percentage-memory association than other factor levels; Control Variables hypothesized to be associated with weaker percentage memory association; BF10 Bayes factor in favor of H1 over H0; N2 nREM2 stage; SWS slow-wave sleep.

Table 5
Descriptive table of the meta-analysis result of measures of SO-SP coupling characteristics.
MeasurenkPooled Effect SizeBF10Influential Moderators
Coupling Phase23900.07 [0.01, 0.13]58.35Memory Type, Age, Channel, Spindle
Spindle Amplitude18780.07 [−0.04, 0.18]8.28Memory Type, Age, Channel, Spindle
Coupling Strength22860.08 [0.02, 0.15]111.04Age
Coupling Percentage1143−0.03 [−0.15, 0.07]0.38None
  1. n number of studies included; k number of effect sizes included; BF10 Bayes factor in favor of H1 over H0 (see Table 8 for the interpretation).

Table 6
Interpretation of the standardized circular-linear correlation coefficient.
rzStrength
<0No Effect
0Null
0.1Small
0.3Moderate
0.5Strong
Table 7
Summary of memory tasks.
Memory Task DomainMemory Task Modality
Verbal Tasks (n=12)Word-Pair (WP)
Word List
Novel Metaphor
Word-Image Pair (WIP)
Emotional Task (n=3)Picture-recognition (IMG)
Spatial Tasks (n=4)Spatial Memory
2D Object Location (2DL)
Visuo-spatial (VS)
Motor Tasks (n=6)Motor Sequence (MST)
Gross-motor Juggle
Mirror-tracing Task (MTT)
Visuomotor Adaptation (VMA)
Table 8
Interpretation of Bayes Factor (BF10) for the strength of evidence.
BF10Direction
<0.1Strong, Favor alternative
0.1–0.33Moderate, Favor alternative
0.33–1Weak, Favor alternative
1–3Weak, Favor hypothesis
3–10Moderate, Favor hypothesis
>10Strong, Favor hypothesis
Appendix 1—table 1
Risk of bias assessment supplemental criteria.
DomainSupplemental Signaling Questions
Bias due to confounding variables1.1 Was sufficient information provided to assess the presence of major potential confounding variables?
1.2 Were major potential confounding variables not relevant to study controlled during the data collection and analysis? Were influences from other experimental tasks or stimuli existing?
1.3 Were confounding factors (such as gender, age, etc.) added to models to calculate and interpret as the “corrected” effect size?
Bias due to subject selection2.1 Was subject selection representative? Can subjects represent the population or community targeted by the experiment?
2.2 Was a random sampling method applied during data collection? Are subjects recruited mostly from a single source (e.g., university) or at different times and caused biases?
2.3 Are the subjects independent from each other? Are the subjects socially connected (e.g., patients and their relatives, between groups if there are multiple groups in the original study)? Have the same subjects been measured repeatedly in pretest-posttest designs?
Bias due to classification of groups3.1 For studies with multiple groups, can group(s) containing only healthy human subjects without intervention be clearly classified?
3.2 When reporting the effect size, did the authors report the effect size separately for different groups?
Bias due to missing outcome data4.1 Were only scatterplots reported in the article, necessitating the use of graph tools to estimate the effect size?
4.2 Was data including t statistics, p-values, β statistics, and η² the only data provided that could be used to estimate the correlation, or only imprecise data provided for non-significant correlation?
4.3 Have pre-sleep/sleep/post-sleep memories and/or sleep data for individual participants been lost? If true, were missing outcome data interpolated, averaged, simulated, or deleted?
Bias in measurement of the outcome5.1 Were measurements, units, and signal processing approaches (including slow oscillillation, spindle, and coupling detection) used by the paper reliable and consistent with others?
5.2 Were only nonparametric effect sizes including Spearman’s rho (ρ), Kendall’s tau (τ), or subjective estimation reported instead of Pearson’s r, Fisher’s z, or circular linear r?
5.3 Did the authors analyze, transform, or clarify non-normal data, introduce resampling techniques and/or exclude outliers?
Bias in selection of the reported result6.1 For multiple groups measured, if only group(s) with the largest effect size, or supported their hypotheses were reported, while non-significant or contradictory results were omitted?
6.2 Have the authors declared their research proposal and expected outcomes within the framework of pre-registration, or declared any conflict of interest in the paper?
Overall Bias7.1 Does the article not fully meet the applicable expectations of the Risk Of Bias In Non-randomized Studies of Interventions (Robins-I) and the above supplementary criteria in multiple domains listed?
7.2 Alternatively, does the article significantly conflict with the criteria in one of these domains?
Appendix 2—table 1
Summary of models for SO–SP coupling–memory association measures.
ModelModeratorModel Purposes
MNoneStudy associations between SP amplitude, coupling phase, coupling strength, and coupling percentage, each in relation to memory consolidation.
M1Memory TaskInvestigate potential distinctions in coupling and memory association mechanisms between declarative memory—including verbal, spatial, and emotional memory—as well as procedural memory.
M2Mean AgeUnderstand the potential impact of development and aging in coupling and memory associations.
M3Spindle TypeExplore how the coupling between SOs and fast or slow SPs predicts memory retention performance differently.
M4PSG ChannelStudy the relation between sleep brain oscillations and memory in different cortical regions, as the frontal, central, and parietal areas were reported to be the most active area for SO–SP coupling but might play different roles.
M5Sleep StageExamine the impact of sleep stage on the relationship between SO–SP coupling and memory, considering that SPs are most active during N2 sleep, while SOs dominate cortical oscillation during SWS.
M6Sleep BoutInvestigate the potential impact of sleep timing and circadian rhythms on the relationship between SO–SP coupling and memory.
M7Age × ChannelInvestigate interactions between age differences and PSG channels in the memory consolidation mechanism, considering the frontal lobe is the latest area of the brain to develop.
M8Age × TaskStudy whether age increase implies a difference in predictive power of SO–SP coupling in the development of declarative and procedural memory consolidation.
M9Channel × SpindleExamine interactions between SP types and cortical areas in models.
MfAll predictorsInclude all pre-specified moderators to the model as fixed effects to explore their explanatory power and potential collinearity.
McNoneFitted controlled model for focal and sensitivity analysis.
Appendix 2—table 2
Main characteristics for each study included in the meta-analysis.
Author (Year)NMAgeStageConditionPSG ChannelsSpindlesTaskMemory TypeMeasures Included
Bastian et al., 20221523.3N2, SWSnapfrontal, centralfast, slowSpatialdeclarativephase, amp, str, pct
Cox et al., 20182430.2N2, SWSovernightfrontal, central, posteriorfast, slowMSTmotorphase, str
Denis et al., 20223122.3N2, SWSovernightfrontal, centralfastIMGdeclarativephase, str, pct
Denis et al., 20213422N2, SWSnapfrontal, central, posteriorfast, slowWPdeclarativephase, amp, str, pct
Donnelly et al., 20221614.1N2, SWSovernightfrontal, central, posteriorfast, slow2 DLdeclarativephase, amp, str
Hahn et al., 202033CH: 9.5, AD: 16SWSovernightfrontal, central, posteriorfast, slowWPdeclarativephase, amp, str, pct
Hahn et al., 202242AD: 12.9, YA: 22.0SWSovernightfrontal, central, posteriorfast, slowJugglemotorphase, amp, str
Halonen et al., 20212722N2, SWSovernightfrontal, centralfastMetaphordeclarativephase, str, pct
Halonen et al., 202215117N2, SWSovernightfrontal, centralfast, slowIMGdeclarativephase, amp, str, pct
Helfrich et al., 201852YA: 20.4, OA: 73.8SWSovernightfrontalfastWPdeclarativephase, str
Kurz et al., 20211911.24N2, SWSovernightfrontal, central, posteriorfast, slowIMGdeclarativephase, amp, str, pct
Kurz et al., 20233011.43N2, SWSovernightfrontal, central, posteriorfast, slowWord Listdeclarativephase, amp, str
Ladenbauer et al., 202143YA: 23, OA: 66N2, SWSnapfrontal, centralfastWP, VSdeclarativephase, amp, str
Mikutta et al., 20192027.1N2, SWSovernightcentralfast, slowWord List, MTTdeclarative, motorphase, amp, str
Mylonas et al., 20202830N2overnightcentralfastMSTmotorphase, amp, str, pct
Mylonas et al., 20221413N2overnightcentralfast, slowSpatialdeclarativephase, amp, str, pct
Nicolas et al., 20222421.9N2, SWSnapfrontal, central, posteriorfastMSTmotorphase, amp
Niknazar et al., 20152822N2napcentralfastWPdeclarativephase, amp, str
Perrault et al., 20191623.4N2, SWSovernightfrontal, posteriorfast, slowWPdeclarativephase, amp
Schreiner et al., 20212020.8N2, SWSnapcentralfastWIPdeclarativephase, amp, str, pct
Solano et al., 20221024.3N2, SWSovernightfrontal, central, posteriorfast, slowVMAmotorphase, amp, str, pct
Weiner et al., 20242569.1N2, SWSovernightfrontal, centralfast, slowWPdeclarativephase, amp, str
Zhang et al., 20202820.6N2, SWSovernightfrontal, central, posteriorfast, slowWPdeclarativephase, str
  1. Notes. N sample size, only includes groups in the meta-analysis; N2 Stage 2 nREM sleep; SWS slow wave sleep; amp spindle amplitude; str coupling strength; pct oupling percentage; CH children; AD adolescents; YA young adults; OA older adults; PSG posterior channels include both parietal and occipital electrodes. Details about memory task types are listed in Table 7; additional detailed information and data are listed in Source data 1 (study-level characteristics) and source data of Figures 4 and 68 (effect size-level characteristics of each coupling metric).

Appendix 3—table 1
Summary of interaction and sensitivity models for the coupling phase-memory association.
ModelsWeightFactorsEstimate (95% CrI)Age/Year slope (95% CrI)
Age× Channel0.17Age × Frontal0.11(−0.05, 0.27)0.001(−0.005, 0.007)
Age × Central0.19 (0.04, 0.33)−0.007 (−0.012, −0.001)
Age × Posterior0.12 (−0.18, 0.43)−0.004(−0.019, 0.012)
Age× Task0.06Age × Verbal0.10 (−0.01, 0.25)−0.001 (−0.006, 0.004)
Age × Spatial0.16 (−0.21, 0.56)−0.010 (−0.022, 0.002)
Age × Emotional0.22 (−0.33, 0.71)0.010 (−0.048, 0.028)
Age × Motor0.27 (−0.14, 0.68)0.009 (−0.028, 0.010)
Channel× Spindle0.16Frontal × Fast SP0.18 (0.06, 0.29)
Central × Fast SP0.06 (−0.05,0.16)
Posterior × Fast SP−0.01 (−0.18, 0.16)
Frontal × Slow SP0.02 (−0.11, 0.14)
Central × Slow SP0.04 (−0.09, 0.17)
Posterior × Slow SP−0.05 (−0.23, 0.44)
Time-lag0.00Time-lag bias−0.010 (−0.050, 0.026)
Prior sensitivity0.02N(0, 2.5), InvGamma(2, 0.5)0.07 (−0.01, 0.14)
1.00Non-informative0.07 (0.01, 0.13)
Main model1 − WeightNone0.07 (0.01, 0.13)
Appendix 3—table 2
Summary of interaction and sensitivity models for the SP amplitude–memory association.
ModelsWeightFactorsEstimate (95% CrI)Age/Year slope (95% CrI)
Age× Channel0.00Age × Frontal0.17 (−0.02, 0.36)−0.002 (−0.008, 0.003)
Age × Central0.15 (−0.05, 0.34)−0.002 (−0.009, 0.004)
Age × Posterior0.13 (−0.28, 0.55)−0.008 (−0.032, 0.016)
Age× Task0.73Age × Verbal0.10 (−0.09, 0.31)0.001 (−0.005, 0.007)
Age × Spatial0.38 (0.08, 0.67)−0.009 (−0.016, −0.001)
Age × Emotional0.20 (−0.46, 0.73)−0.015 (−0.058, 0.028)
Age × Motor−0.26 (−0.80, 0.35)0.013 (−0.018, 0.044)
Channel× Spindle0.00Frontal × Fast SP0.14 (-0.03, 0.32)
Central × Fast SP0.13 (−0.03, 0.23)
Posterior × Fast SP0.05 (−0.17, 0.28)
Frontal × Slow SP0.11 (−0.07, 0.28)
Central × Slow SP−0.01 (−0.20, 0.19)
Posterior × Slow SP−0.18 (−0.46, 0.10)
Time-lag0.00Time-lag bias0.00 (−0.070, 0.070)
Prior sensitivity0.00N(0, 2.5), InvGamma(2, 0.5)0.07 (−0.04, 0.18)
1.00Non-informative0.07 (−0.04, 0.19)
Main model1 − WeightNone0.07 (−0.04, 0.18)
Appendix 3—table 3
Summary of interaction and sensitivity models for the coupling strength–memory association.
ModelsWeightFactorsEstimate (95% CrI)Age/Year slope (95% CrI)
Age× Channel0.00Age × Frontal0.14 (−0.03, 0.30)−0.001 (−0.007, 0.004)
Age × Central0.18 (0.01, 0.33)−0.005 (−0.011, 0.001)
Age × Posterior−0.06 (−0.38, 0.27)0.009 (−0.009, 0.026)
Age× Task0.00Age × Verbal0.11 (−0.04, 0.27)−0.002 (−0.007, 0.003)
Age × Spatial0.13 (−0.18, 0.45)−0.004 (−0.015, 0.007)
Age × Emotional0.17 (−0.39, 0.64)−0.001 (-0.035, 0.035)
Age × Motor0.14 (−0.45, 0.67)−0.004(−0.028, 0.023)
Channel× Spindle0.00Frontal × Fast SP0.15 (0.01, 0.28)
Central × Fast SP0.02 (−0.09, 0.14)
Posterior × Fast SP0.08 (−0.12, 0.28)
Frontal × Slow SP0.04 (−0.11, 0.18)
Central × Slow SP0.05 (0.10, 0.20)
Posterior × Slow SP0.15 (−0.05, 0.35)
Time-lag0.43Time-lag bias0.024 (−0.015, 0.060)
Prior sensitivity0.26N(0, 2.5), InvGamma(2, 0.5)0.08 (0.00, 0.16)
1.00Non-informative0.08 (0.02, 0.15)
Main model1 − WeightNone0.08 (0.02, 0.15)
Appendix 3—table 4
Summary of interaction and sensitivity models for the coupling percentage–memory association.
ModelsWeightFactorsEstimate (95% CrI)Age/Year slope (95% CrI)
Age× Channel0.40Age × Frontal−0.04 (−0.76, 0.72)0.005 (−0.038, 0.045)
Age × Central−0.00 (−0.59, 0.63)-0.001 (-0.036, 0.031)
Age × Posterior0.18 (−0.61, 0.98)−0.014 (−0.062, 0.030)
Age× Task0.65Age × Verbal0.42 (−0.24, 1.16)−0.035 (−0.085, 0.008)
Age × Spatial−0.04 (−1.00, 1.07)0.003 (−0.072, 0.072)
Age × Emotional0.21 (-0.54, 0.84)−0.017 (−0.073, 0.034)
Age × Motor0.39 (−0.91, 1.49)−0.015 (−0.085, 0.055)
Channel× Spindle0.00Frontal × Fast SP0.02 (−0.16, 0.19)
Central × Fast SP−0.04 (−0.20, 0.13)
Posterior × Fast SP0.00 (−0.29, 0.27)
Frontal × Slow SP0.10 (−0.10, 0.29)
Central × Slow SP0.01 (−0.18, 0.19)
Posterior × Slow SP−0.07 (−0.36, 0.95)
Time-lagNaNTime-lag biasDid not perform
Prior sensitivity0.09N(0, 2.5), InvGamma(2, 0.5)−0.04 (−0.17, 0.09)
1.00Non-informative−0.03 (−0.16, 0.07)
Main model1 − WeightNone−0.03 (−0.15, 0.07)

Additional files

MDAR checklist
https://cdn.elifesciences.org/articles/101992/elife-101992-mdarchecklist1-v1.docx
Supplementary file 1

PRISMA statements checklist.

https://cdn.elifesciences.org/articles/101992/elife-101992-supp1-v1.docx
Supplementary file 2

Model diagnostics.

Example plots demonstrating the diagnostic process for each fitted Bayesian model, including posterior predictive checks, trace plots, and autocorrelation plots.

https://cdn.elifesciences.org/articles/101992/elife-101992-supp2-v1.pdf
Supplementary file 3

Effect size-level forest plot.

https://cdn.elifesciences.org/articles/101992/elife-101992-supp3-v1.pdf
Supplementary file 4

Pareto k diagnostic statistics.

https://cdn.elifesciences.org/articles/101992/elife-101992-supp4-v1.pdf
Supplementary file 5

Frequentist analysis.

Forest plots generated using frequentist analysis for each fitted overall Bayesian model.

https://cdn.elifesciences.org/articles/101992/elife-101992-supp5-v1.pdf
Supplementary file 6

Posterior distributions of moderators.

https://cdn.elifesciences.org/articles/101992/elife-101992-supp6-v1.pdf
Source data 1

Main table of study characteristics and participant demographics.

https://cdn.elifesciences.org/articles/101992/elife-101992-data1-v1.zip

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Thea Ng
  2. Eunsol Noh
  3. Rebecca MC Spencer
(2025)
Bayesian meta-analysis reveals the mechanistic role of slow oscillation-spindle coupling in sleep-dependent memory consolidation
eLife 13:RP101992.
https://doi.org/10.7554/eLife.101992.3