The association between working memory precision and the nonlinear dynamics of frontal and parieto-occipital EEG activity

Chang, Wen-Sheng; Liang, Wei-Kuang; Li, Dong-Han; Muggleton, Neil G.; Balachandran, Prasad; Huang, Norden E.; Juan, Chi-Hung

doi:10.1038/s41598-023-41358-0

The association between working memory precision and the nonlinear dynamics of frontal and parieto-occipital EEG activity

Article
Open access
Published: 31 August 2023

Volume 13, article number 14252, (2023)
Cite this article

Download PDF

You have full access to this open access article

Scientific Reports

The association between working memory precision and the nonlinear dynamics of frontal and parieto-occipital EEG activity

Download PDF

Wen-Sheng Chang¹,
Wei-Kuang Liang^1,2,
Dong-Han Li^1,2,
Neil G. Muggleton^1,2,3,4,
Prasad Balachandran^1,7,
Norden E. Huang^2,5 &
…
Chi-Hung Juan^1,2,6

1821 Accesses
2 Citations
1 Altmetric
Explore all metrics

Abstract

Electrophysiological working memory (WM) research shows brain areas communicate via macroscopic oscillations across frequency bands, generating nonlinear amplitude modulation (AM) in the signal. Traditionally, AM is expressed as the coupling strength between the signal and a prespecified modulator at a lower frequency. Therefore, the idea of AM and coupling cannot be studied separately. In this study, 33 participants completed a color recall task while their brain activity was recorded through EEG. The AM of the EEG data was extracted using the Holo-Hilbert spectral analysis (HHSA), an adaptive method based on the Hilbert-Huang transforms. The results showed that WM load modulated parieto-occipital alpha/beta power suppression. Furthermore, individuals with higher frontal theta power and lower parieto-occipital alpha/beta power exhibited superior WM precision. In addition, the AM of parieto-occipital alpha/beta power predicted WM precision after presenting a target-defining probe array. The phase-amplitude coupling (PAC) between the frontal theta phase and parieto-occipital alpha/beta AM increased with WM load while processing incoming stimuli, but the PAC itself did not predict the subsequent recall performance. These results suggest frontal and parieto-occipital regions communicate through theta-alpha/beta PAC. However, the overall recall precision depends on the alpha/beta AM following the onset of the retro cue.

EEG dynamics reveal a dissociation between storage and selective attention within working memory

Article Open access 18 September 2019

Effects of load and emotional state on EEG alpha-band power and inter-site synchrony during a visual working memory task

Article 24 August 2020

Gamma oscillations modulate working memory recall precision

Article Open access 05 July 2021

Introduction

WM enables us to store and utilize information without external stimuli for a brief period. Previous studies have associated WM performance with oscillatory activities and interareal connectivity between frontal and posterior brain regions^1,2,3,4. Specifically, frontal theta (4–8 Hz) and posterior alpha/beta (8–30 Hz) activities have been widely studied as indicators of WM performance^5,6,7,8,9,10. High frontal theta power is associated with increased executive control demands in memory tasks, with larger power indicating higher cognitive requirements^11,12,13. The directionality of parieto-occipital alpha/beta power, however, has shown inconsistent findings in relation to WM performance¹⁰. In general, suppression of alpha/beta power is associated with processes involved in forming and maintaining active WM representations, while an increase in alpha/beta power reflects the protection of memorized information from irrelevant stimuli^{8,14,15,16,17,18}. The direction of parieto-occipital alpha/beta power modulation depends on various factors, including the specific WM task employed, the definition of the target, and the memory stages being investigated¹⁰. Recent advances in electrophysiological WM research have focused on interactions across different frequency components and brain areas^{19,20,21,22,23,24}. Measures of cross-frequency coupling (CFC) are often used to quantify these WM-related cross-frequency interactions, e.g., the phase-amplitude coupling (PAC) between fast and slow oscillations^25,26. In PAC, fast and slow oscillatory components are filtered from the raw signal, then the phase of the slower component is compared to the amplitude of the faster component with synchronous measures. Typically, stronger PAC is associated with greater WM demands or better WM performance^{19,23,27,28,29}.

Although frontal theta and parieto-occipital alpha power modulations during WM maintenance have been frequently reported, their coupling is rarely discussed. Indirect evidence (e.g., Granger causality) has shown that frontal to parieto-occipital theta connectivity is transiently modulated by incoming executive demands, whereas parieto-occipital to frontal alpha/beta connectivity is sustained throughout the task procedure^20,22. One recent dual-task study reported that frontal-midline theta power predicted the subsequent parieto-occipital alpha power lateralization. In addition, enhanced coupling was observed between frontal theta phase and parieto-occipital alpha amplitude in the cued region²⁴. Nevertheless, the correlation between frontoparietal theta-alpha PAC and subsequent recall performance remained unexplored.

The standard PAC methods are increasingly criticized due to potential confounding factors arising from the nonstationary and nonlinear nature of biological signals^30,31,32. The most fundamental issue comes from the physical meaning of PAC, which is a multiplicative process that cannot directly be measured with additive time–frequency decomposition methods such as short-time Fourier and wavelet decompositions^33,34. The same limitation holds for bispectral measures frequently used to study the interactions between various frequency components of a signal^35,36. In these methods, the amplitude modulation (AM) of a fast wave is measured by its projection to a slow reference wave. Therefore PAC and AM are inseparable, and the rhythmic AM of a fast wave cannot be measured if the frequency-matched slow wave does not present in the observed data. Such constraint limits the scope of studying the nonlinear properties of oscillatory signals.

The study's primary purpose is to investigate how nonlinear interactions of EEG oscillations are associated with WM performance. Following previous literature, we hypothesized that frontal theta and parieto-occipital alpha oscillations communicate through PAC during WM. The communication process leads to parieto-occipital alpha AM, and the pattern of AM predicts performance. Participants took part in a color recall task with simultaneous EEG recording. On each trial, they were required to remember a set of color squares. During WM maintenance, a retro cue was presented to indicate the target item. Participants responded by matching the target color with a color wheel presented one second after the cue onset. Memory precisions under different WM loads served as an index of WM performance^37,38. The observed EEG data was decomposed with Holo-Hilbert spectral analysis (HHSA), which can directly measure the energy of AM in nonlinear signals³³. In HHSA, the AMs of signal carriers are decomposed into different amplitude-modulating frequencies (AMF). The time profile of signal carriers and their AM structure are expressed in Holo-Hilbert spectra (HHS). To verify the hypothesis, we first replicated previous findings on the effect of load manipulations in EEG signals and their correlation to individual differences in WM performance. After confirmation, we investigated whether AM could reveal more information. The load effect for frontoparietal theta-alpha PAC and its correlation to WM precision were also examined for an explicit communication mechanism between frontal and parieto-occipital regions.

Methods

Participants

Thirty-three university students were recruited to participate in the experiment (21 males and 12 females, mean age = 23.13 years and standard deviation = 5.16 years). The sample size was determined following conventional EEG studies⁷. The recruiting process stopped when the sample size reached 30. Three more participants registered for the study before the termination of the recruitment, thus their data were also collected. All participants had a normal or corrected-to-normal vision. All participants gave informed consent before participation. The study procedures were approved by the local IRB committee (National Taiwan University Research Ethics Committee, 201606ES031). All methods were carried out in accrodance with the relevant guideliines and regulations.

Experimental design

The behavioral task was modified from the color recall tasks for WM precision^37,38. The stimuli were presented on an LCD monitor on a black background at a viewing distance of 60 cm. The colors were selected from a circle in CIELAB space³⁹, centered at L = 70, a = 20, b = 38, in CIE coordinates with a radius of 60. The items to be remembered were a set of color squares with a size 2° × 2° of visual angle. The color of each item was sampled from 45 possible color values at every 8° of the circle. The locations of the items were randomly chosen but equally spaced concentrically, 4.5° away from the center of the monitor. The color wheel for color recall was constructed with 360 color values, sampled from 1° to 360° of the circle in the color space. The color wheel was centered on the screen, with an inner and outer radius of 6° and 8.2° of visual angle.

Each trial began with a concurrent presentation of a beep sound (500 Hz) and a central fixation cross (gray, 1°$\times$ 1°). The beep sound lasted for 100 ms, and the fixation cross remained on display throughout the trial. After 500 ms, a set of color squares was displayed for 100 ms, followed by a 1000 ms delay period. After the delay period, a set of outlined squares was presented at the location of each item during stimulus presentation for 1 s (the probe period). One of the squares had thicker edges, indicating the target item for the observer to recall. A color wheel was displayed as the probe squares vanished. The observer had to identify the target color by mouse-clicking the corresponding color on the color wheel. During the response period, the color in the target location kept updating to the color selected by the mouse cursor (Fig. 1a). The subsequent trial began 500 ms after a target color was picked from the color wheel. In different experimental conditions, the set size of the sample array varied between 1, 2, 4, or 6 items. The four set-size conditions were randomly interleaved across trials. All participants performed 90 trials for each condition, making 360 trials in total.

Mixture model analysis

Behavioral performance was assessed using the mixture model analysis³⁷. In each trial, the degree of error was obtained by calculating the angular deviation between the coordinates of the reported color and the correct target color in the color wheel. The model proposed three sources of WM errors: (1) the probability that the target color was not stored in the WM system, (2) the probability of misreporting a non-target color instead of the target color, and (3) information was stored with noise in WM. Participants were assumed to make a uniformly random response when they could not recall the target color. In addition, all items were assumed to be encoded with equal resolution and encoding to decay at an equal rate. The probability of the reported color value in each trial can therefore be described as follows:

$$p\left(x\right)=\left(1-pU-pNT\right){\Phi }_{\sigma }\left(x-\theta \right)+pU\frac{1}{2\pi }+pNT\frac{1}{m}\sum_{i=1}^{m}{\Phi }_{\sigma }\left(x-{\theta }_{i}^{*}\right)$$

where ${\Phi }_{\sigma }$ represented the von Mises (circular Gaussian) distribution with mean zero and standard deviation $\upsigma$, $x$ was the reported color value, $\theta$ was the target color value, ${\theta }_{i}^{*}, i\in \{1,\dots ,m\}$ were the m non-target color values, $pT$ was the probability of correctly remembering the target color, $pU$ was the probability of random guessing, and $pNT$ was the probability of misremembering the target location (Fig. 1b). The index for WM precision ($\kappa =1/{\upsigma }^{2}$) was the concentration number of the von Mises distribution, which was the reciprocal of error variance.

EEG recording

EEG was continuously recorded throughout the experiment using SynAmps2 amplifiers (Compumedics Neuroscan) using Ag/AgCl electrodes mounted in a 64-channel electrode cap (Quik-Cap, Neuromedical Supplies). The EEG cap followed the international 10–20 system for electrode placement. Impedances of all electrodes were kept below 5 kΩ. The sampling rate was set at 2000 Hz. The horizontal and vertical electrooculograms were recorded along with the EEG recordings. AFz served as the ground electrode, whereas Cz served as the online reference. The recordings were filtered with a low-pass filter of 500 Hz.

EEG data preprocessing

The raw EEG data were offline-referenced to the average of M1 and M2 electrodes. The data were epoched from – 1500 to 3000 ms relative to the onset of the sample array. Epochs containing artifacts of muscle activity, head or body movements, and electrode noise were rejected by visual inspection. Eye movements and blink artifacts were identified and removed with independent component analysis using the EEGLAB toolbox ⁴⁰. Epochs deviating more than 200 μV were also rejected. Electrodes containing more than 50% of such ‘bad’ trials were replaced by spline interpolation of their neighbors. 19 out of 33 participants required such interpolation (three channels for one participant, two channels for four participants, and one channel for 14 participants). Most removed electrodes were located in the forehead or temporal sites (FP1/2, FPz, T7/8, TP7/8, and C6). After this procedure, none of the participants was rejected for further analysis. The averaged survived epoch ratio for all participants is 82.9% (298 trials) with 9.4% standard deviation. The preprocessed EEG signal was further transformed into its scalp current sources via the surface Laplacian method⁴¹. After applying the surface Laplacian transformation, the resulting signals underwent HHSA for time–frequency decomposition and phase-based statistics (Fig. 2a).

Holo-Hilbert Spectral Analysis (HHSA)

The preprocessed EEG data were decomposed with HHSA using in-house MATLAB codes. The codes for HHSA are available in the public repository (http://osf.io/243ps). The HHSA involves two layers of nested EMD and a 3D spectral representation for the signal³³. EMD is an iterative process that decomposes data into a set of intrinsic mode functions (IMF, Huang et al. 1998). Each IMF is symmetric locally with zero mean by definition. Therefore, the instantaneous phase can be obtained directly with the Hilbert transform or later proposed methods such as normalized Hilbert transform or direct quadrature (DQ, Huang et al. 2009). Compared to conventional approaches, HHSA has the following advantages: (1) it is not limited to the uncertainty principle of time–frequency analysis (i.e., the trade-off between time and frequency resolutions) by providing instantaneous frequency^42,43, so the data can be represented in high precision in the spectral domain; (2) the structure of data itself determines the timescales being examined, so a priori assumptions for frequency bands are not needed; (3) the envelope function of each frequency component can be decomposed again into a set of second-layer IMFs, where each second-layer IMF represents modulation of the envelope function within a dyadic timescale; (4) EMD is an additive decomposition method; therefore the original signal can be reconstructed by direct summation over IMFs.

In HHSA, the first-layer IMFs are obtained by applying EMD to the input data (Fig. 2b, the left column). In practice, EMD acts as a dyadic filter bank to the data so that each IMF represents oscillatory activity in distinct log-2 timescales^44,45. For instance, in an EEG signal with a 500-Hz sampling rate, the frequency range for the first IMF would be a Gaussian-like bandpass filter centering at around 128 Hz; the second IMF peaks at 64 Hz, and so on for consecutive IMFs (Fig. 4a). Since the envelopes of the first-layer IMFs are also broadband signals, they can be decomposed again with EMD (Fig. 2b, the right column). The second-layer IMFs describe the variation of envelope functions in distinct log-2 timescales; hence we obtain a representation for each envelope in the frequency domain. To distinguish the instantaneous frequencies of the first- and second-layer IMFs, we will refer to instantaneous frequencies of the first-layer IMFs as “carrier frequency” (CF) and to second-layer IMFs as “AM frequency” (AMF). The Holo-Hilbert spectrum (HHS) is the 3D representation of a signal over time (t), CF, and AMF (Fig. 2c). Extending the time–frequency representation with AMF shows the energy fluctuation induced by multiplicative interactions. Note that the AMF domain describes the results of multiplicative interaction from all possible sources. The PAC between any two signals can be evaluated via phase synchronization indices between the first- and second-layer IMFs (Fig. 2d). For instance, the theta-alpha PAC can be evaluated by calculating the directed phase-lag index (dPLI)⁴⁶ between the theta-band carrier and the frequency-matched AM component of the alpha-band envelope. The difference between AM and PAC under HHSA is illustrated in the supplementary files (Fig. S1).

In this study, HHS is projected along the three dimensions to highlight different aspects of the full spectrum. Projecting the HHS onto the CF-time plane created the CF-t spectrum (Fig. 3a), which captured the intra-mode frequency variation of the carriers. CF-t spectrum was equivalent to the spectrogram obtained from the Hilbert-Huang Transform (HHT)⁴² that has been frequently used for analyzing the instantaneous frequency and amplitude variations of biological signals (e.g.,^47,48). When HHS was projected onto the AMF-time plane, the result highlighted the dynamic of cross-scale interaction for a given range of carrier frequencies (the AMF-t spectrum, Fig. 3b). When HHS was projected onto the AMF-CF plane, the results showed the overall structure of cross-scale interactions between different frequency components in a given period (the AMF-CF spectrum, Fig. 3c). In this study, the resolution of HHS was set as 4 ms for time and 0.5 Hz for both CF and AMF.

Transient bursts in biological signals sometimes cause mode mixing, i.e., a single IMF contains multiple frequency components with classical EMD. Here, we applied an improved version of EMD utilizing sinusoidal masking signals in conjunction with the original EMD algorithm to reduce the problem^47,49,50. A set of masking signals with initial phases 0, π/2, π, and 3π/2 were added into a pre-decomposed IMF obtained from regular EMD to enhance decomposition. The optimal amplitude of the masking signals is 1.6 times above the average amplitude of the components through experience⁵⁰. We used a factor of 2 for both layers of EMD. The frequency of masking signals was the same as the mean instantaneous frequency of the pre-decomposed IMF. Instantaneous frequencies of IMFs were directly calculated by taking time derivatives of instantaneous phases obtained from DQ.

Intra- and inter-areal PAC under the HHSA framework

IMF is a function that requires (1) symmetric locally to zero, and (2) there exists only one extremum between two successive zero-crossing points. According to the above definition, the instantaneous phases for the first- and second-layer IMFs can be obtained directly with DQ. Given the phase functions of carriers and their AMs, the degree of PAC can be measured by phase synchronization indices. Here, we used the dPLI to quantify the degree of PAC while taking the leading-lagging relationship into account:

$${\mathrm{dPLI}}_{fg}= \frac{1}{N}\sum_{n=1}^{N}{H(Im(e}^{i\left({\varphi }_{f}\left(n\right)-{\varphi }_{g}\left(n\right)\right)})),$$

where ${\varphi }_{f}$ and ${\varphi }_{g}$ denote the phase functions of two oscillations, and N denotes the number of sampling points within a time window. H is the Heaviside function (H(x) = 0 if x ≤ 0; H(x) = 1 if x > 0), and $Im$ denotes that only the imaginary part of the relative phase is considered. The dPLI is a directed synchronization measure bounded by 0 and 1. If signal f is phase-leading to signal g, then ${0.5<\mathrm{dPLI}}_{fg}\le 1$; if signal g is phase-leading to signal f, then ${0\le \mathrm{dPLI}}_{fg}<0.5$. When PAC employs the dPLI formula under the HHSA framework, one phase function is from the carrier of a slower oscillation, and the other is from the frequency-matched AM component of a faster oscillation. This approach is similar to the Holo-Hilbert cross-frequency phase clustering (HHCFPC) that has been demonstrated as efficient in quantifying intra- and inter-areal CFC in the process of WM²¹.

Combining HHSA and phase-based statistics to evaluate PAC has the following advantages over conventional methods: first, phase-based statistics, such as the phase-lag index, only consider the imaginary part of the spectrum. Therefore, they are robust to noise. Second, nonstationarity in the high-frequency power leads to instability in the Hilbert spectrum and is resolved via EMD. Third, different modulating waves may interfere when the modulation strength is determined by the distribution of high-frequency amplitude over low-frequency phase bins (e.g.,²⁶). In summary, the confounding of power is eliminated by utilizing phase-based statistics. The waveform shape is essentially the nonstationarity issue, which is resolved by EMD. The phase-reset factor is related to separating co-occurring phenomena from actual modulation.

Statistical analyses of EEG data

We used MATLAB and the FieldTrip Analysis Toolbox for the statistical analysis of EEG data⁵¹. Epochs were time-locked to the onset of the sample array for all analyses.

Evaluating the effect of WM load on EEG activity

For preliminary analysis, we calculated the mean amplitude of each IMF across all electrodes for each set size and participant, then tested the effect of load manipulation with a repeated-measures ANOVA for set sizes 2, 4, and 6 at each time point. The single-item condition served as active control for the task. ANOVA results were corrected for multiple comparisons by a nonparametric cluster-based permutation test of 5000 iterations. The method only assumes that every experimental condition is associated with a probability distribution. Unlike parametric tests, the validity of the permutation test does not depend on the probability distribution of the data or the cluster-forming method⁵². Both the cluster-forming threshold and the critical value of significance were set at p = 0.05.

Based on previous studies^6,7,24, we choose mid-frontal electrodes (AF3/4, F1/2/3/4, FC1/2/3/4, Fz, and FCz) and parieto-occipital electrodes (P1/2/3/4/5/6/7/8, PO3/4/5/6/7/8, O1/2, Pz, POz, and Oz) for area-based analysis. Averaged CF-t power spectra in frontal and parieto-occipital electrodes were calculated separately, ranging from 1 to 50 Hz. Load effects for the CF-t spectra were evaluated with the same procedure of repeated-measures ANOVA as the preliminary analysis. If any significant cluster was identified, the AMF-t spectrum of the corresponding CF range was calculated and tested with the same ANOVA procedure.

Correlation between individual WM precision and EEG activity

The same CF-t and AMF-t representations of parieto-occipital EEG power were also used for correlation analyses. EEG spectra and $\kappa$ values from the set-size one condition were subtracted from those from set sizes 2, 4, and 6, and were then transformed into z-scores for regression analysis. For each spectrum, a general-linear model (GLM) was constructed in which the z-transformed $\kappa$ values served as a regressor to predict the z-transformed EEG power in each spectral point. Since both $\kappa$ values and EEG power varied systematically with increasing set size, the regression weights were separately evaluated for set sizes 2, 4, and 6, then averaged together for statistical tests (i.e., set size served as a control variable for the GLM). The approach was similar to previous within-subject studies^53,54 but on a between-subject basis. The same cluster-based permutation procedure was applied to correct the results of GLM for load-effect analysis, but both cluster-forming threshold and critical value were set at p = 0.05 for two-tailed tests. In addition, we also tested EEG activity in frontal electrodes to see if there was any WM correlate in the cluster. Correlations between frontal CF-t spectrum and WM precision were analyzed first; if any significant correlate was identified, the AMF-t spectrum of the corresponding CF range was analyzed.

Intra- and inter-areal PAC

The oscillations of interest were beta, alpha, and theta waves for this study. In this dataset, the frequency ranges of IMFs 4, 5, and 6 represented oscillatory activities in the beta-, alpha-, and theta-band, respectively. The intra- and inter-regional PAC were calculated as follows: first, we obtained the phase of theta waves (from IMF 6) and the phase of frequency-matched alpha/beta AM components (from IMFs 4 and 5) through HHSA. Second, the cross-frequency dPLI between theta phase and frequency-matched alpha/beta AMs across all possible electrode pairs was calculated and then averaged together according to the frontal and parieto-occipital electrodes of interest for each set-size condition. Time windows for dPLI evaluation were determined according to the previous results of load effects over HHS. The first interval ranged from 0 to 0.5 s after the onset of the sample array, and the remaining part of the delay period constituted the second interval. Time windows for the probe period were also divided into the first 0.5 s and the remainder interval for analysis. The effect of load manipulation was evaluated separately by a repeated-measures ANOVA for different time windows and electrode clusters. In addition, the phase synchronization of frontal and parieto-occipital regions was tested by the same procedure. The overall results were corrected for multiple testing using the Benjamini–Hochberg procedure⁵⁵ with a false-discovery rate (FDR) of 0.05.

Results

Behavioral results

Thirty-three participants performed a color-recall task with varying WM loads. Response errors in different loads were fitted separately by a mixture model. As expected, participants performed worse as set size increased (Fig. 1c). With increasing set size, the probability for remembered response (pT) decreased (F_(3,96) = 55.52, p < 10^–5; Cohen’s f = 1.7352), whereas the probability for misreported responses (pNT; F_(3,96) = 12.05, p < 10^–5; Cohen’s f = 0.3820) and guessed responses (pU; F_(3,96) = 26.20, p < 10^–5; Cohen’s f = 0.8185) increased. On top of the decreasing recall rates, WM precision (κ) also decreased as set size increased (F_(3,96) = 48.54, p < 10^–5, Cohen’s f = 1.5170). Post hoc analyses revealed significant differences of κ between set sizes 1 and 2 (t₃₂ = − 2.728, p = 0.0103) but no significant difference for pT (t₃₂ = − 1.2182, p = 0.2321), indicating that WM precision (κ) was more sensitive in small WM loads. Parameters $\upkappa$, pT, and pU show significant differences between set sizes 2 and 4 ($\upkappa$, t₃₂ = − 4.7273, p = 4.38 × 10^–5; pT, t₃₂ = − 7.4262, p = 1.89 × 10^–8; pU, t₃₂ = 5.7989, p = 1.94 × 10^–6), as well as 4 and 6 ($\upkappa$, t₃₂ = − 3.889, p = 4.78 × 10^–4; pT, t₃₂ = -4.6453, p = 5.55 × 10^–5; pU, t₃₂ = 2.581, p = 0.0146) after FDR correction at 0.05 level. pNT differences between set sizes 2 and 4 (t₃₂ = 2.3583, p = 0.0246) as well as set sizes 4 and 6 (t₃₂ = 2.4185, p = 0.0215) were not significant after FDR correction. We performed the sensitivity analysis for the four mixture model parameters given 33 participants, 0.05 significant threshold, and 0.80 statistical power. The critical effect sizes for $\upkappa$, pT, pNT, and pU were 0.2459, 0.2345, 0.2926, and 0.2427, respectively.

WM load modulated parieto-occipital alpha/beta power

The HHSA is based on the empirical mode decomposition (EMD), which decomposes the temporal signal into a set of intrinsic-mode functions (IMF, Huang et al., 1998). The IMFs serve as a dyadic filter bank for the data⁵⁰. Each IMF represents EEG activity in a different dyadic timescale (Fig. 4a). Between-participant variations of the peak frequency and frequency range for each IMF were summarized in Table 1. IMFs 4, 5, and 6 represented the beta-, alpha-, and theta-band activities, respectively. Standard deviations of the frequency range and peak frequency were smaller than 1 Hz for the three IMFs.

Table 1 Descriptive statistics of peak and the lower-bound frequencies for each IMF.

Full size table

Our EEG analysis first verified whether the task used in the study replicated the load-dependent alpha power suppression in the previous literature^6,7,14,56. For each IMF, the grand-average powers across all electrodes were calculated first, then the effect of load manipulation was evaluated by performing a repeated-measures ANOVA at each time point and each IMF. They were corrected with a cluster-based permutation procedure of 5000 iterations. The set-size 1 condition served as the baseline for comparison. The effects were quantified as the deviation from the baseline in set sizes 2, 4, and 6. Significant clusters were observed in IMFs 4 to 7, in which IMFs 4 (12.03–23.68 Hz) and 5 (6.46–12.03 Hz) showed negative load effects in the delay and probe periods, whereas IMFs 6 (3.03–6.46 Hz) and 7 (1.46–3.03 Hz) showed transient positive load effects following the onsets of sample and probe arrays (Fig. S2). For the IMF capturing alpha activity, two negative clusters could be observed within the latency of interest (-0.5 to 2.5 s relative to the onset of the sample array, Fig. 4b). The first cluster was observed in 0.5–1.15 s (cluster-corrected p = 4 × 10^–4), and the second cluster was observed in 1.55–2.1 s (cluster-corrected p ≤ 2 × 10^–4). A marginal positive load effect was observed following the onset of the sample array (0–0.2 s; cluster-corrected p = 0.057). On average, the F-statistics from 0 to 2.1 s after the onset of the sample array concentrated in parieto-occipital electrodes (Fig. 4c).

Next, we mapped parieto-occipital EEG activity into HHS, then verified the load effect using the same ANOVA procedure as the above. The results were similar to previous studies^7,56. A sustained negative cluster was observed throughout the delay and probe periods in the alpha- and beta-bands (8–23 Hz; cluster-corrected p = 4 × 10^–4; Fig. 5a). Compared with alpha power suppression, the effect of beta power (12–23 Hz) suppression was more pronounced in the probe period (1.1 to 2.1 s from the onset of sample array). On the other hand, transient positive clusters were observed in the theta-band (2–8 Hz) frequency during the onsets of the sample (cluster-corrected p = 6.2 × 10^–3) and probe arrays (cluster-corrected p = 4 × 10^–4). The observed alpha/beta power suppression was not driven by the evoked responses, since the the same result could be replicated in the induced activity (Fig. S7). The theta-band clusters, on the other hand, was more related to evoked responses.

To further quantify the time course of inter-frequency power modulation in terms of AMF for the sustained negative cluster observed, we calculated the AMF-t spectrum within the alpha and beta time scales (i.e., IMFs 4 and 5), then performed the same analysis as for the CF-t spectrum. The results showed that the sustained negative cluster observed consisted of two AMF components (cluster-corrected p = 5 × 10^–4, Fig. 5b): a low-frequency component (0–4 Hz) extended throughout the task procedure (Fig. 5c), and a 4–12 Hz component that was only observed in the latter parts of the delay and probe periods (Fig. 5d). Note that the 4–12 Hz component was observed in the same time window in which the power of alpha CF gradually increased from the dips following the onsets of the sample and probe arrays (the blue trace in Fig. 5b).

We also tested the load effect for frontal EEG power, but no significant effect was observed (Fig. S3). For comparison between HHSA and traditional time–frequency analysis, a separate analysis with decomposition using the Morlet wavelet transforms was illustrated in Fig. S4. The patterns of the results were similar for the two methods. However, HHSA showed more precise time–frequency resolution, and only the HHSA could provide information in terms of AMF.

WM precision and HHS in the parieto-occipital region

After evaluating the load effects across different set sizes, we conducted a correlation analysis for the individual differences between WM precision (κ) and parieto-occipital HHS. The set-size 1 served as the baseline condition. The κ value and HHS were subtracted from the single-item condition and transformed into z-scores for analysis. The correlation was evaluated by a general linear model between the κ value and each spectral point of the HHS, with WM load as a covariate. The results were corrected with a cluster-based permutation test with 5000 iterations. The results for the CF-t spectrum showed a sustained negative correlation in the alpha band for both the delay (cluster-corrected p = 0.007) and probe periods (cluster-corrected p = 0.002; Fig. 6a). Beta power was negatively correlated with the κ value in the last 0.4 s time window of the probe period (cluster-corrected p = 0.013). In addition, theta power also negatively correlated with the κ value toward the end of the probe period (the cluster was connected to the alpha-band cluster in the response period; cluster-corrected p = 0.002). The same analysis with wavelet decomposition yielded similar results (Fig. S5).

For the AMF-t spectrum, three AMF correlates of WM precision could be identified (Fig. 6b). First, an alpha-AMF correlate (7–13 Hz) was observed in the latter part of the probe period (1.5–2.1 s from the onset of sample array; cluster-corrected p = 0. 0152). Second, a low-frequency AMF component (1.5–4 Hz) was observed from the onset of the probe array, and it extended to the response period. Finally, the trend component (0–0.5 Hz) showed a sustained negative correlation throughout the task procedure. The latter two correlates were connected in the response period (cluster-corrected p = 0.0044). Further GLM analyses for the CF-AMF spectra showed that the parieto-occipital alpha/beta AM predicted individual WM precision after a target-defining cue was presented to the participants (Fig. 6c).

WM precision and HHS in the frontal region

The GLM results for frontal HHS showed a positive correlation with individual WM precision (Fig. 7). For the CF-t spectrum, the theta-band power was positively correlated to $\kappa$ value during WM retention. However, the frequency range of frontal correlates was slightly different in the delay and probe periods (Fig. 7a). The frequency range was 3–5 Hz for the cluster observed in the delay period (cluster-corrected p = 0.0054), whereas the frequency range was 5–8 Hz for the cluster observed in the probe period (cluster-corrected p = 0.0092). The comparable analysis with the Morlet wavelet transforms only showed a positive correlation trend, but the statistics did not pass the significant level after multiple corrections (Fig. S5).

When the frontal theta power (IMFs 6 and 7) was mapped into the AMF-t spectrum, a 1.5–4 Hz AMF component was observed in 0.5 s before and after the onset of the probe array (cluster-corrected p = 0.0074; Fig. 7b). The result reflected the transition from the lower-theta (3–5 Hz) to the upper-theta (5–8 Hz) band as observed in Fig. 5a. The results of GLM analyses for the CF-AMF spectra in the same area showed that the AM component of the frontal theta power showed a significant correlation during the processing of incoming stimuli. A significant cluster could be observed in the first 500-ms time window of the delay (Fig. 8c, the first diagram; cluster-corrected p = 0.0168) and probe periods (Fig. 7c, the third diagram; cluster-corrected p = 0.0076).

Intra- and inter-regional PACs between theta and alpha/beta oscillations

dPLIs between theta-, alpha-, and beta-band oscillations (i.e., IMFs 6, 5, and 4, respectively) were calculated across all electrode pairs, then were averaged according to set-size conditions, frequency combination, and the locations of electrode pairs. Since the theta power increased but the alpha/beta power decreased in response to the task stimuli, a 180-degree offset should be considered when inferring the phase leading-lagging relationship from the cross-frequency dPLI. We found significant load effects in the first 0.5-s time window of the delay and probe periods (Fig. 8). The cross-frequency dPLI between frontal theta phase and parieto-occipital alpha/beta AM decreased from 0.5 to even smaller values during the onset of the probe array (F_(3,64) = 14.7782, p = 5.29 × 10^–6 and F_(3,64) = 17.5511, p = 8.38 × 10^–7; Fig. 9a,b). For cross-frequency connectivity within the parieto-occipital region, the dPLI between theta phase and alpha/beta AM increased from 0.5 to even larger values as set size increased during the onset of the sample and probe arrays (F_(3,64) = 6.4161, p = 0.0029 and F_(3,64) = 17.9176, p = 6.62 × 10^–7 for theta-beta PAC; F_(3,64) = 7.8448, p = 0.0009 for theta-alpha PAC; Fig. 9c,d). Finally, the dPLI between frontal and parieto-occipital theta phases increased from 0.5 to larger values as the set size increased (F_(3,64) = 22.7148, p = 3.51 × 10^–8; Fig. 9e).

In summary, the frontal theta phase preceded the phase of frequency-matched parieto-occipital alpha/beta AM under high WM loads, whereas the parieto-occipital theta phase followed the phase of frequency-matched alpha/beta AM within the same region. The frontal theta phase also preceded the parieto-occipital theta phase during the onset of the sample array. Note that the phase offsets between inter- and intra-regional dPLI were closed to 180 degrees under high WM load (e.g., N = 6), and the offsets between frontal and parieto-occipital theta phase were also closed (despite not equal) to 180 degrees in the same condition. Therefore, the pattern of intra-regional dPLI could arise from the field-spreading of the frontal theta wave.

We performed GLM analyses between the individual differences in WM precision and PAC, but neither intra- nor inter-regional PACs predicted individual WM precision. Finally, the degree of PAC did not predict the subsequent suppression of the parieto-occipital alpha/beta AM. These results indicated that PAC and AM are independent measurements in WM processing.

Discussion

This study investigated the electrophysiological bases of AM and PAC in WM processing. The main result showed that stronger load-dependent alpha/beta power suppression in parieto-occipital electrodes was associated with better WM precision. Furthermore, the observed sustained alpha/beta power suppression could be separated into continuous low-frequency (< 2 Hz) and 4–12 Hz AMF components. The low-frequency component extended throughout the task procedure, whereas the 4–12 Hz AMF component was observed only after processing the sample and probe stimuli. When correlating individual WM precision to alpha/beta AMs, we found negative correlations for the tonic alpha/beta power suppression (AMF < 1 Hz) throughout the task procedure. On the other hand, higher AMFs of alpha/beta power suppression only showed significant correlations after the onset of the probe array.

Connectivity analysis with cross-frequency dPLI showed a load-dependent increment of theta-alpha/beta PAC between the frontal and parieto-occipital regions and within the parieto-occipital region itself during the onsets of the sample and probe arrays. Therefore, the theta-AMF component of parieto-occipital alpha/beta AM was likely driven by frontal and parieto-occipital theta activity. The overall pattern of results suggests that WM precision, like capacity measurements, is supported by frontoparietal oscillatory activities^21,57,58.

Alpha/beta power suppression reflects the ability to access feature information in WM

Our findings in parieto-occipital alpha/beta activity are consistent with previous reports showing that load-dependent alpha/beta suppression predicts individual differences in WM capacity^6,7,56,59. In memory formation and retrieval, reducing alpha/beta power facilitates information-processing ability in task-relevant sensory areas⁶⁰. Therefore, the precision of memorized information is associated with the ability to maintain alpha/beta power suppression in areas storing stimulus-specific information ¹⁶. Since the κ value measured the quality of WM representation³⁸, our results suggest that maintaining fine-grained perceptual information in WM requires engaging the parieto-occipital region¹⁴. This argument is consistent with studies using multivariate regression methods, which have shown that non-spatial feature information of visual WM could be decoded from bilateral posterior alpha/beta activities^8,61,62. Similar results have also been reported in fMRI studies, in which the stimulus information can be decoded from the pattern of BOLD responses in visual areas^63,64,65. In addition, the same pattern of results can extend to other modalities, e.g., the load-dependent alpha/beta power suppression could either be observed within the visual or somatosensory area, depending on which modality was prompted for a response in an intermodal WM task⁵.

The alpha/beta power suppression pattern indicates two possible roles in WM processing. First, it is the neural substrate of the number of items held in memory^7,66. Second, it reflects a necessary mechanism that keeps the target information activated, i.e., keeping relevant information in the focus of attention^67,68. The first argument suggests that the alpha/beta power suppression should be reduced after presenting the retro cue since irrelevant information should be deactivated to prevent interference. The second argument does not assume the release of suppression after retro cue presentation. Our result finds alpha/beta power suppression does not reduce after presenting the retro-cue. The result is consistent with a previous report using lateralized cueing paradigm⁵³. In that study, only the alpha power representing the cued location was suppressed, but the alpha power representing irrelevant item locations did not increase. One recent study has shown that the parieto-occipital alpha power suppression not only scales with WM load but also to the spatial extent between items and the target-distractor similarity⁶⁹. Therefore, the alpha/beta power suppression observed in our study is better explained as the difficulty of maintaining the target information in the focus of attention.

By decomposing parieto-occipital alpha/beta AMs into different AMFs, we found differences between the trend (< 1 Hz) and higher AMF components (4–12 Hz) in their correlation to κ values. Specifically, the trend component showed a sustained negative correlation throughout the maintenance period, whereas the higher AMF components only showed a significant correlation after the onset of the probe array. These results might imply different cognitive factors contributing to the maintenance of WM information. The low-frequency component reflected tonic psychophysiological states such as alertness and sustained attention⁷⁰. In contrast, the higher AMF components reflected within-trial target select processes²⁴. The two levels of modulation are further evidence supporting parieto-occipital alpha/beta power as the access rather than the neural substrate of memorized visual information.

The CF-AMF structure of the higher AMF components shifted from alpha AM in the early (t = [1.1, 1.6] s) and to beta AM in the later (t = [1.6, 2.1] s) time windows of the probe period. One recent study has shown that the load-dependent increment of the parietal alpha/beta oscillation speed predicts individual WM capacity⁷¹. Therefore, the beta AM correlate observed might indicate a more efficient target selection for high-precision participants.

Relationship between the frontal theta power and WM precision

Frontal midline theta activity has been considered the oscillatory substrate for cognitive control or cognitive efforts in completing the task^72,73. Prior studies have associated the increment of frontal midline theta power with increasing memory load^5,22,74, increasing task difficulty⁷⁵, or as an index of successful WM manipulation⁷⁶. The increment of frontal theta power is also associated with better performance in WM tasks^77,78. Our results showed a positive correlation between frontal theta power and WM precision. However, frontal theta power did not scale with WM loads (Figs. S3 and 7a). The seeming contradiction of load effects could be explained as the following. First, frontal theta power increment is more sensitive in maintaining temporal order information than visual item information^79,80. The effect size of frontal theta power increment could be smaller for concurrent visuospatial WM tasks. Second, other factors such as individual WM capacity and acute stress could interact with the frontal theta load effect so that frontal theta power peaks at some critical load and then decreases at higher loads^77,78. One recent intracranial EEG study also failed to replicate the load-dependent theta power increment in the anterior cingulate cortex, which was considered the primary source of frontal midline theta activity⁸¹. The study argued that the hippocampus indeed contributed a portion of the frontal midline activity observed from the scalp.

Frontoparietal cross-frequency connectivity indicates the top-down modulation of memorized contents

Our results show the load-dependent modulation of cross-frequency dPLI between frontal theta phase and frequency-matched parieto-occipital alpha/beta AMs (Fig. 9). The degree of dPLI increases as WM load increases, and such long-range modulation is only transiently involved during the processing of probe stimuli. The frontal theta phase precedes the phase of parieto-occipital alpha/beta AM. Therefore, the inter-regional PAC observed can be associated with the selection of visual information in WM⁸². Load effects for cross-frequency dPLI within the parieto-occipital cluster are also observed. However, the parieto-occipital theta phase falls behind the phase of alpha/beta AM. The result suggests that the state of alpha/beta AM may guide the upcoming sensory processing, thus forming a reciprocal interaction between the goal at the moment and perceptual inputs.

Previous studies have suggested that the prefrontal cortex serves as the central executive module during WM processing^1,83. In particular, the prefrontal cortex exerts control through low-frequency oscillations around the theta-band, coordinating posterior areas that store WM representations^{12,13,84,85,86}. Studies with spatial retro-cueing consistently reported stronger alpha power suppression in the contralateral sites^7,53,54. Such alpha lateralization was transient following the retro-cue⁸⁷, and the degree of lateralization scaled with task difficulty⁸⁸. The transient lateralization of posterior alpha power was associated with the power of frontal low-frequency oscillations^62,87. Our result is consistent with these findings.

Recent studies have proposed two mechanisms for frontal theta activity exerting control over posterior regions during memory. For the direct mechanism, the frontal theta activity determines the timing of long-range communication between the central executive and sensory areas⁸⁹. The direct mechanism does not require long-range theta coherence as well as local PAC. On the other hand, the indirect mechanism assumes that the frontal region exerts control through long-range theta coherence. The activated representations are modulated by local PAC in the posterior region^19,90,91,92. The direct modulation better explains our result since the long-range theta coherence is not necessary for frontoparietal theta to alpha/beta PAC.

HHSA reveals the relationship between oscillatory power variations and PAC

While conventional PAC methods calculate the distribution of the power of a high-frequency oscillation to the phase of a low-frequency oscillation, the current study isolates the estimation of cross-frequency coupling in two steps: first, the amplitude of an oscillation was decomposed into different AMs with HHSA; second, PAC was evaluated by cross-frequency dPLI between the slow oscillation and the scale-matched AM of the fast oscillation^21,34. We have shown that WM load manipulation modulates theta-alpha and theta-beta PACs during the processing of probe stimuli, followed by a corresponding change in alpha/beta power suppression. However, only alpha/beta AM predicts individual WM precision. The results imply PAC and AM have distinct meanings in WM processing. Theta-alpha/beta PAC reflects the mechanism of top-down selection over WM representations¹², whereas alpha/beta AM reflects the fidelity of target information after selection.

The possible roles of alpha/beta power suppression in the WM subprocesses

The 100 ms stimulus presentation time in this study falls short of the duration required for full encoding of visual stimuli into WM⁹³. Therefore, the identified load effects on parieto-occipital alpha/beta power suppression may be resulted from unfinished encoding process, rather than the active maintenance of WM representation.

Parieto-occipital alpha/beta power suppression has been frequently documented in memory studies involving simple features such as color and orientation^6,7,56,94. This sustained suppression is observed during more extended encoding and maintenance periods. For instance, Proskovec et al. (2019) utilized 1.5 s for encoding and 2.5 s for maintenance in their research, and alpha power suppression was observed across the trial sequence. Their findings suggest a maintenance role for the sustained parieto-occipital alpha/beta power suppression. On the other hand, when dealing with information that can be transformed into other modalities, such as shapes or letters, parieto-occipital alpha/beta power decreases during encoding and increases during maintenance^18,95. These studies suggest that task-relevant information can be transferred to other brain areas for sustained maintenance. In this context, the observed parieto-occipital alpha/beta power suppression indicates essential visual information processing for memory encoding.

Taken together, it is evident that the occurrence of parieto-occipital alpha/beta power suppression during maintenance is contingent upon the nature of the tested features. If the alpha/beta suppression is solely a consequence of visual information encoding, it would not persist during the maintenance period for tasks involving simple features with sufficient encoding duration. Therefore, we conclude that the alpha/beta power suppression observed in this study is related to the active maintenance of memorized information. Nevertheless, it is important to note that this does not preclude its potential involving in the encoding process. Further investigation is required to elucidate the distinct contributions of encoding and maintenance to alpha power suppression in WM tasks of similar design.

Conclusion

Our results demonstrate that individual differences in parieto-occipital alpha/beta power suppression during WM maintenance predict their recall precision. The alpha/beta power suppression can be decomposed into different AMF components. The faster (4–12 Hz) and the slower (< 2 Hz) AMF components represent within- and across-trial attentional mechanisms that are crucial for maintaining the activated WM information. Enhanced theta-alpha/beta frontoparietal PAC under high WM load suggests an explicit mechanism for inter-regional communication. When a target-defining is presented, the frontal theta wave modulates parieto-occipital alpha/beta AM with PAC, and the subsequent alpha/beta AM variation determines the recall precision.

Limitations

This study finds that frontoparietal theta-alpha/beta PAC increases with increasing WM load, and individual differences in WM precision are associated with parieto-occipital alpha/beta AM around the theta-band (4–12 Hz) AMF. Nevertheless, since AM summarizes multiplicative interactions from all possible sources, the reported results could also arise from the task design itself. More experiments are necessary to verify whether the observed alpha/beta AM correlates depending on the task design. For instance, if the same periodic EEG activities modulate WM maintenance, the same AM correlates are expected under different delay periods. From the pattern of results, we inferred that frontal theta oscillation served as the executive control, whereas parieto-occipital alpha/beta AM reflected attentional selections that are required for maintaining activated WM representations. Both arguments need to be examined with direct testing, such as manipulating different levels of cognitive control under the same stimulus display. Finally, the paradigm was modified from the standard WM precision task³⁷, in which a retro cue was presented the probe participants’ memory one seconds before the response stage. The comparable results in the CF-t spectrum might indicate the generalizability of HHSA results in similar tasks.

Data availability

The data and necessary codes supporting the findings of this study are available in the OSF repository (http://osf.io/243ps).

Abbreviations

AM:: Amplitude modulation
AMF:: Amplitude-modulation frequency
CF:: Carrier frequency
CFC:: Cross-frequency coupling
dPLI:: Directed phase-lag index
DQ:: Direct quadrature
EMD:: Empirical mode decomposition
FDR:: False-discovery rate
FO:: Fast oscillation
GLM:: General linear model
HHCFPC:: Holo-Hilbert cross-frequency phase clustering
HHS:: Holo-Hilbert spectrum
HHSA:: Holo-Hilbert spectral analysis
IMF:: Intrinsic-mode function
PAC:: Phase-amplitude coupling
SO:: Slow oscillation
WM:: Working memory

References

D’Esposito, M. & Postle, B. R. The cognitive neuroscience of working memory. Annu. Rev. Psychol. 66, 115–142 (2015).
PubMed Google Scholar
Dipoppa, M., Szwed, M. & Gutkin, B. S. Controlling working memory operations by selective gating: The roles of oscillations and synchrony. Adv. Cogn. Psychol. 12, 209–232 (2016).
PubMed PubMed Central Google Scholar
Eriksson, J., Vogel, E. K., Lansner, A., Bergström, F. & Nyberg, L. Neurocognitive architecture of working memory. Neuron 88, 33–46 (2015).
CAS PubMed PubMed Central Google Scholar
Gazzaley, A. & Nobre, A. C. Top-down modulation: Bridging selective attention and working memory. Trends Cogn. Sci. 16, 129–135 (2012).
PubMed Google Scholar
van Ede, F., Jensen, O. & Maris, E. Supramodal theta, gamma, and sustained fields predict modality-specific modulations of alpha and beta oscillations during visual and tactile working memory. J. Cogn. Neurosci. 29, 1455–1472 (2017).
PubMed Google Scholar
Erickson, M. A., Smith, D., Albrecht, M. A. & Silverstein, S. Alpha-band desynchronization reflects memory-specific processes during visual change detection. Psychophysiology 56, 1–9 (2019).
Google Scholar
Fukuda, K., Mance, I. & Vogel, E. K. Α power modulation and event-related slow wave provide dissociable correlates of visual working memory. J. Neurosci. 35, 14009–14016 (2015).
CAS PubMed PubMed Central Google Scholar
Fukuda, K., Kang, M. S. & Woodman, G. F. Distinct neural mechanisms for spatially lateralized and spatially global visual working memory representations. J. Neurophysiol. 116, 1715–1727 (2016).
PubMed PubMed Central Google Scholar
Hsieh, L. T. & Ranganath, C. Frontal midline theta oscillations during working memory maintenance and episodic encoding and retrieval. Neuroimage 85, 721–729 (2014).
PubMed Google Scholar
Pavlov, Y. G. & Kotchoubey, B. Oscillatory brain activity and maintenance of verbal and visual working memory: A systematic review. Psychophysiology 59, e13735 (2022).
PubMed Google Scholar
Hanslmayr, S. & Staudigl, T. How brain oscillations form memories—a processing based perspective on oscillatory subsequent memory effects. Neuroimage 85, 648–655 (2014).
PubMed Google Scholar
de Vries, I. E. J., Slagter, H. A. & Olivers, C. N. L. Oscillatory control over representational states in working memory. Trends. Cogn. Sci. 24, 150–162 (2020).
PubMed Google Scholar
Sauseng, P., Griesmayr, B., Freunberger, R. & Klimesch, W. Control mechanisms in working memory: A possible function of EEG theta oscillations. Neurosci. Biobehav. Rev. 34, 1015–1022 (2010).
PubMed Google Scholar
vanEde, F. Mnemonic and attentional roles for states of attenuated alpha oscillations in perceptual working memory: A review. Eur. J. Neurosci. 48, 2509–2515 (2018).
Google Scholar
Foster, J. J., Sutterer, D. W., Serences, J. T., Vogel, E. K. & Awh, E. The topography of alpha-band activity tracks the content of spatial working memory. J. Neurophysiol. 115, 168–177 (2016).
PubMed Google Scholar
Griffiths, B. J. et al. Alpha/beta power decreases track the fidelity of stimulus specific information. Elife 8, 1–22 (2019).
Google Scholar
Schneider, D., Mertes, C. & Wascher, E. The time course of visuo-spatial working memory updating revealed by a retro-cuing paradigm. Sci. Rep. 6, 21442 (2016).
CAS PubMed PubMed Central ADS Google Scholar
Heinrichs-Graham, E. & Wilson, T. W. Spatiotemporal oscillatory dynamics during the encoding and maintenance phases of a visual working memory task. Cortex 69, 121–130 (2015).
PubMed PubMed Central Google Scholar
Daume, J., Gruber, T., Engel, A. K. & Friese, U. Phase-amplitude coupling and long-range phase synchronization reveal frontotemporal interactions during visual working memory. J. Neurosci. 37, 313–322 (2017).
CAS PubMed PubMed Central Google Scholar
Johnson, E. L. et al. Bidirectional frontoparietal oscillatory systems support working memory. Curr. Biol. 27, 1829-1835.e4 (2017).
MathSciNet CAS PubMed PubMed Central Google Scholar
Liang, W.-K., Tseng, P., Yeh, J.-R., Huang, N. E. & Juan, C.-H. Frontoparietal beta amplitude modulation and its interareal cross-frequency coupling in visual working memory. Neuroscience 460, 69–87 (2021).
CAS PubMed Google Scholar
Popov, T. et al. Cross-frequency interactions between frontal theta and posterior alpha control mechanisms foster working memory. Neuroimage https://doi.org/10.1016/j.neuroimage.2018.07.067 (2018).
Article PubMed Google Scholar
Roux, F. & Uhlhaas, P. J. Working memory and neural oscillations: Alpha-gamma versus theta-gamma codes for distinct WM information?. Trends Cogn. Sci. 18, 16–25 (2014).
PubMed Google Scholar
de Vries, I. E. J., Van Driel, J., Karacaoglu, M. & Olivers, C. N. L. Priority switches in visual working memory are supported by frontal delta and posterior alpha interactions. Cereb. Cortex 28, 4090–4104 (2018).
PubMed PubMed Central Google Scholar
Canolty, R. T. & Knight, R. T. The functional role of cross-frequency coupling. Trends Cogn. Sci. 14, 506–515 (2010).
PubMed PubMed Central Google Scholar
Tort, A. B. L., Komorowski, R., Eichenbaum, H. & Kopell, N. Measuring phase-amplitude coupling between neuronal oscillations of different frequencies. J. Neurophysiol. 104, 1195–1210 (2010).
PubMed PubMed Central Google Scholar
Axmacher, N. et al. Cross-frequency coupling supports multi-item working memory in the human hippocampus. Proc. Natl. Acad. Sci. USA 107, 3228–3233 (2010).
CAS PubMed PubMed Central ADS Google Scholar
Siegel, M., Warden, M. R. & Miller, E. K. Phase-dependent neuronal coding of objects in short-term memory. Proc. Natl. Acad. Sci. USA 106, 21341–21346 (2009).
CAS PubMed PubMed Central ADS Google Scholar
Friese, U. et al. Successful memory encoding is associated with increased cross-frequency coupling between frontal theta and posterior gamma oscillations in human scalp-recorded EEG. Neuroimage 66, 642–647 (2013).
PubMed Google Scholar
Aru, J. et al. Untangling cross-frequency coupling in neuroscience. Curr. Opin. Neurobiol. 31, 51–61 (2015).
CAS PubMed Google Scholar
Cole, S. R. & Voytek, B. Brain oscillations and the importance of waveform shape. Trends Cogn. Sci. 21, 137–149 (2017).
PubMed Google Scholar
Hyafil, A. Misidentifications of specific forms of cross-frequency coupling: Three warnings. Front. Neurosci. 9, 1–6 (2015).
Google Scholar
Huang, N. E. et al. On Holo-Hilbert spectral analysis: A full informational spectral representation for nonlinear and non-stationary data. Philos. Trans. R. Soc. A. 374, 20150206 (2016).
ADS Google Scholar
Juan, C. H. et al. Revealing the dynamic nature of amplitude modulated neural entrainment with Holo-Hilbert spectral analysis. Front. Neurosci 15, 1–18 (2021).
Google Scholar
Kovach, C. K., Oya, H. & Kawasaki, H. The bispectrum and its relationship to phase-amplitude coupling. Neuroimage 173, 518–539 (2018).
PubMed Google Scholar
Shahbazi Avarvand, F. et al. Localizing bicoherence from EEG and MEG. Neuroimage 174, 352–363 (2018).
PubMed Google Scholar
Bays, P. M., Catalao, R. F. G. & Husain, M. The precision of visual working memory is set by allocation of a shared resource. J. Vis. 9, 7–7 (2009).
Google Scholar
Zhang, W. & Luck, S. J. Discrete fixed-resolution representations in visual working memory. Nature 453, 233-U13 (2008).
CAS PubMed PubMed Central ADS Google Scholar
Robertson, A. R. The CIE 1976 color-difference formulae. Color Res. Appl. 2, 7–11 (1977).
ADS Google Scholar
Delorme, A. & Makeig, S. EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J. Neurosci. Methods 134, 9–21 (2004).
PubMed Google Scholar
Perrin, F., Pernier, J. & Bertrand, O. Spherical splines for scalp potential and current density mapping. Electroencephalogr. Clin. Neurophysiol. 72, 184–187. https://doi.org/10.1016/0013-4694(89)90180-6 (1989).
Article CAS PubMed Google Scholar
Huang, N. E. et al. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Philos. Trans. R. Soc. A. 454, 903–995 (1998).
MathSciNet MATH Google Scholar
Huang, N. E. et al. On instantaneous frequency. Adv. Adapt. Data. Anal. 01, 177–229 (2009).
MathSciNet Google Scholar
Flandrin, P., Rilling, G. & Gonçalvés, P. Empirical mode decomposition as a filter bank. IEEE Signal Process. Lett. 11, 112–114 (2004).
ADS Google Scholar
Wu, Z. & Huang, N. E. A study of the characteristics of white noise using the empirical mode decomposition method. Philos. Trans. R. Soc. A. 460, 1597–1611 (2004).
MATH Google Scholar
Stam, C. J. & van Straaten, E. C. W. Go with the flow: Use of a directed phase lag index (dPLI) to characterize patterns of phase relations in a large-scale model of brain dynamics. Neuroimage 62, 1415–1428 (2012).
CAS PubMed Google Scholar
Quinn, A. J. et al. Within-cycle instantaneous frequency profiles report oscillatory waveform dynamics. J. Neurophysiol. 126, 1190–1208 (2021).
PubMed PubMed Central Google Scholar
Lopes-dos-Santos, V. et al. Parsing hippocampal theta oscillations by nested spectral components during spatial exploration and memory-guided behavior. Neuron https://doi.org/10.1016/j.neuron.2018.09.031 (2018).
Article PubMed PubMed Central Google Scholar
Nguyen, K. T. et al. Unraveling nonlinear electrophysiologic processes in the human visual system with full dimension spectral analysis. Sci. Rep. 9, 1–13 (2019).
ADS Google Scholar
Deering, R. & Kaiser, J. F. The use of a masking signal to improve empirical mode decomposition. In Proc. ICASSP’05, Vol. 4 (2005).
Oostenveld, R., Fries, P., Maris, E. & Schoffelen, J. M. FieldTrip: Open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data. Comput. Intell. Neurosci. https://doi.org/10.1155/2011/156869 (2011).
Article PubMed PubMed Central Google Scholar
Maris, E. & Oostenveld, R. Nonparametric statistical testing of EEG- and MEG-data. J. Neurosci. Methods 164, 177–190 (2007).
PubMed Google Scholar
Myers, N. E., Stokes, M. G., Walther, L. & Nobre, A. C. Oscillatory brain state predicts variability in working memory. J. Neurosci. 34, 7735–7743 (2014).
CAS PubMed PubMed Central Google Scholar
Poliakov, E., Stokes, M. G., Woolrich, M. W., Mantini, D. & Astle, D. E. Modulation of alpha power at encoding and retrieval tracks the precision of visual short-term memory. J. Neurophysiol. 112, 2939–2945 (2014).
CAS PubMed PubMed Central Google Scholar
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: A practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B Stat. Methodol. 57, 289–300 (1995).
MathSciNet MATH Google Scholar
Proskovec, A. L., Wiesman, A. I., Heinrichs-Graham, E. & Wilson, T. W. Load effects on spatial working memory performance are linked to distributed alpha and beta oscillations. Hum. Brain Mapp. 40, 3682–3689 (2019).
PubMed PubMed Central Google Scholar
Palva, S., Monto, S. & Palva, J. M. Graph properties of synchronized cortical networks during visual working memory maintenance. Neuroimage 49, 3257–3268 (2010).
PubMed Google Scholar
Siebenhühner, F., Wang, S. H., Palva, J. M. & Palva, S. Cross-frequency synchronization connects networks of fast and slow oscillations during visual working memory maintenance. Elife 5, 15–30 (2016).
Google Scholar
Zammit, N., Falzon, O., Camilleri, K. & Muscat, R. Working memory alpha–beta band oscillatory signatures in adolescents and young adults. Eur. J. Neurosci. 48, 2527–2536 (2018).
PubMed Google Scholar
Hanslmayr, S., Staudigl, T. & Fellner, M. C. Oscillatory power decreases and long-term memory: The information via desynchronization hypothesis. Front. Hum. Neurosci. 6, 1–20 (2012).
Google Scholar
Bae, G. Y. & Luck, S. J. Dissociable decoding of spatial attention and working memory from EEG oscillations and sustained potentials. J. Neurosci. 38, 409–422 (2018).
CAS PubMed PubMed Central Google Scholar
de Vries, I. E. J., van Driel, J. & Olivers, C. N. L. Decoding the status of working memory representations in preparation of visual selection. Neuroimage 191, 549–559 (2019).
PubMed Google Scholar
Ester, E. F., Serences, J. T. & Awh, E. Spatially global representations in human primary visual cortex during working memory maintenance. J. Neurosci. 29, 15258–15265 (2009).
CAS PubMed PubMed Central Google Scholar
Harrison, S. A. & Tong, F. Decoding reveals the contents of visual working memory in early visual areas. Nature 458, 632–635 (2009).
CAS PubMed PubMed Central ADS Google Scholar
Pratte, M. S. & Tong, F. Spatial specificity of working memory representations in the early visual cortex. J. Vis. 14, 1–12 (2014).
Google Scholar
Adam, K. C. S., Robison, M. K. & Vogel, E. K. Contralateral delay activity tracks fluctuations in working memory performance. J. Cogn. Neurosci. 30, 1229–1240 (2018).
PubMed PubMed Central Google Scholar
Woodman, G. F., Wang, S., Sutterer, D. W., Reinhart, R. M. G. & Fukuda, K. Alpha suppression indexes a spotlight of visual-spatial attention that can shine on both perceptual and memory representations. Psychon. Bull. Rev. https://doi.org/10.3758/s13423-021-02034-4 (2021).
Article PubMed PubMed Central Google Scholar
Zickerick, B., Rösner, M., Sabo, M. & Schneider, D. How to refocus attention on working memory representations following interruptions—evidence from frontal theta and posterior alpha oscillations. Eur. J. Neurosci. 54, 7820–7838 (2021).
CAS PubMed Google Scholar
Wang, S., Megla, E. E. & Woodman, G. F. Stimulus-induced alpha suppression tracks the difficulty of attentional selection, not visual working memory storage. J. Cogn. Neurosci. 33, 536–562 (2021).
PubMed Google Scholar
Oken, B. S., Salinsky, M. C. & Elsas, S. M. Vigilance, alertness, or sustained attention: physiological basis and measurement. Clin. Neurophysiol. 117, 1885–1901 (2006).
CAS PubMed PubMed Central Google Scholar
Noguchi, Y. & Kakigi, R. Temporal codes of visual working memory in the human cerebral cortex: Brain rhythms associated with high memory capacity. Neuroimage 222, 117294 (2020).
PubMed Google Scholar
Cavanagh, J. F. & Frank, M. J. Frontal theta as a mechanism for cognitive control. Trends Cogn. Sci. 18, 414–421 (2014).
PubMed PubMed Central Google Scholar
McFerren, A., Riddle, J., Walker, C., Buse, J. B. & Frohlich, F. Causal role of frontal-midline theta in cognitive effort: A pilot study. J. Neurophysiol. 126, 1221–1233 (2021).
CAS PubMed PubMed Central Google Scholar
Jensen, O. & Tesche, C. D. Frontal theta activity in humans increases with memory load in a working memory task. Eur. J. Neurosci. 15, 1395–1399 (2002).
PubMed Google Scholar
Wisniewski, M. G., Thompson, E. R. & Iyer, N. Theta- and alpha-power enhancements in the electroencephalogram as an auditory delayed match-to-sample task becomes impossibly difficult. Psychophysiology 54, 1916–1928 (2017).
PubMed Google Scholar
Itthipuripat, S., Wessel, J. R. & Aron, A. R. Frontal theta is a signature of successful working memory manipulation. Exp. Brain. Res. 224, 255–262 (2013).
PubMed Google Scholar
Gärtner, M., Rohde-Liebenau, L., Grimm, S. & Bajbouj, M. Working memory-related frontal theta activity is decreased under acute stress. Psychoneuroendocrinology 43, 105–113 (2014).
PubMed Google Scholar
Zakrzewska, M. Z. & Brzezicka, A. Working memory capacity as a moderator of load-related frontal midline theta variability in sternberg task. Front. Hum. Neurosci. 8, 1–7 (2014).
Google Scholar
Hsieh, L.-T., Ekstrom, A. D. & Ranganath, C. Neural oscillations associated with item and temporal order maintenance in working memory. J. Neurosci. 31, 10803–10810 (2011).
CAS PubMed PubMed Central Google Scholar
Liu, X. L. et al. Task-specific disruptions in theta oscillations during working memory for temporal order in people with schizophrenia. J. Cogn. Neurosci. 32, 2117–2130 (2020).
PubMed Google Scholar
Brzezicka, A. et al. Working memory load-related theta power decreases in dorsolateral prefrontal cortex predict individual differences in performance. J. Cogn. Neurosci. 31, 1290–1307 (2019).
PubMed PubMed Central Google Scholar
Myers, N. E., Stokes, M. G. & Nobre, A. C. Prioritizing information during working memory: Beyond sustained internal attention. Trends Cogn. Sci. 21, 449–461 (2017).
PubMed PubMed Central Google Scholar
Sreenivasan, K. K., Curtis, C. E. & D’Esposito, M. Revisiting the role of persistent neural activity during working memory. Trends Cogn. Sci. 18, 82–89 (2014).
PubMed PubMed Central Google Scholar
Fell, J. & Axmacher, N. The role of phase synchronization in memory processes. Nat. Rev. Neurosci. 12, 105–118 (2011).
CAS PubMed Google Scholar
Herweg, N. A., Solomon, E. A. & Kahana, M. J. Theta oscillations in human memory. Trends Cogn. Sci. 24, 208–227 (2020).
PubMed PubMed Central Google Scholar
Staresina, B. P. & Wimber, M. A neural chronometry of memory recall. Trends Cogn. Sci. 23, 1071–1085 (2019).
PubMed Google Scholar
Wallis, G., Stokes, M., Cousijn, H., Woolrich, M. & Nobre, A. C. Frontoparietal and cingulo-opercular networks play dissociable roles in control of working memory. J. Cogn. Neurosci. https://doi.org/10.1162/jocn_a_00838 (2015).
Article PubMed Google Scholar
van Driel, J., Ort, E., Fahrenfort, J. & Olivers, C. Beta and theta oscillations differentially support free versus forced control over multiple-target search. J. Neurosci. 39, 2518–2547 (2018).
Google Scholar
Berger, B. et al. Dynamic regulation of interregional cortical communication by slow brain oscillations during working memory. Nat. Commun. 10, 1–11 (2019).
CAS ADS Google Scholar
Daume, J., Graetz, S., Gruber, T., Engel, A. K. & Friese, U. Cognitive control during audiovisual working memory engages frontotemporal theta-band interactions. Sci. Rep. 7, 1–13 (2017).
CAS Google Scholar
Reinhart, R. M. G. & Nguyen, J. A. Working memory revived in older adults by synchronizing rhythmic brain circuits. Nat. Neurosci. 22, 820–827 (2019).
CAS PubMed PubMed Central Google Scholar
von Nicolai, C. et al. Corticostriatal coordination through coherent phase-amplitude coupling. J. Neurosci. 34, 5938–5948 (2014).
Google Scholar
Bays, P. M., Gorgoraptis, N., Wee, N., Marshall, L. & Husain, M. Temporal dynamics of encoding, storage, and reallocation of visual working memory. J. Vis. 11, 1–15 (2011).
Google Scholar
Pahor, A. & Jaušovec, N. The effects of theta and gamma tacs on working memory and electrophysiology. Front. Hum. Neurosci. 11, 1–16 (2018).
Google Scholar
Heinz, A. J. & Johnson, J. S. Load-dependent increases in delay-period alpha-band power track the gating of task-irrelevant inputs to working memory. Front. Hum. Neurosci. 11, 1–13 (2017).
Google Scholar

Download references

Acknowledgements

We are grateful to Dr. Shih-Kuen Cheng for sharing his expertise on EEG and the recording equipment.

Funding

Ministry of Science and Technology, Taiwan, MOST110-2410-H-008-039, MOST110-2410-H-008-040-MY3, MOST110-2639-H008-001-ASP.

Author information

Authors and Affiliations

Institute of Cognitive Neuroscience, College of Health Sciences and Technology, National Central University, Taoyuan City, Taiwan
Wen-Sheng Chang, Wei-Kuang Liang, Dong-Han Li, Neil G. Muggleton, Prasad Balachandran & Chi-Hung Juan
Cognitive Intelligence and Precision Healthcare Center, National Central University, Taoyuan City, Taiwan
Wei-Kuang Liang, Dong-Han Li, Neil G. Muggleton, Norden E. Huang & Chi-Hung Juan
Institute of Cognitive Neuroscience, University College London, London, UK
Neil G. Muggleton
Department of Psychology, Goldsmiths, University of London, London, UK
Neil G. Muggleton
Data Analysis and Application Laboratory, The First Institute of Oceanography, Qingdao, China
Norden E. Huang
Department of Psychology, Kaohsiung Medical University, Kaohsiung, Taiwan
Chi-Hung Juan
Taiwan International Graduate Program in Interdisciplinary Neuroscience, National Cheng Kung University and Academia Sinica, Taipei, Taiwan
Prasad Balachandran

Authors

Wen-Sheng Chang
View author publications
You can also search for this author in PubMed Google Scholar
Wei-Kuang Liang
View author publications
You can also search for this author in PubMed Google Scholar
Dong-Han Li
View author publications
You can also search for this author in PubMed Google Scholar
Neil G. Muggleton
View author publications
You can also search for this author in PubMed Google Scholar
Prasad Balachandran
View author publications
You can also search for this author in PubMed Google Scholar
Norden E. Huang
View author publications
You can also search for this author in PubMed Google Scholar
Chi-Hung Juan
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

Conceptualization, W.-S.C. and C.-H.J.; formal analysis, W.-S.C., D.-H.Li, N.G.M., W.-K.L. and C.-H.J.; funding acquisition, C.-H.J., N. G. M., & W.-K. L; methodology, W.-K.L., N. E. H. and C.-H.J.; resources, C.-H.J.; software, N. E. H. and W.-K.L.; supervision, C.-H.J.; validation, W.-S.C., P.B., and C.-H.J.; visualization, W.-S.C.., P.B., N.G.M.,W.-K.L. and C.-H.J.; writing—original draft, W.-S.C.., P.B., N.G.M. and C.-H.J.; writing, review and editing, W.-S.C.., P.B., N.G.M., W.-K.L. and C.-H.J. All authors have read and agreed to the published version of the manuscript.

Corresponding author

Correspondence to Chi-Hung Juan.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary Information.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Chang, WS., Liang, WK., Li, DH. et al. The association between working memory precision and the nonlinear dynamics of frontal and parieto-occipital EEG activity. Sci Rep 13, 14252 (2023). https://doi.org/10.1038/s41598-023-41358-0

Download citation

Received: 24 April 2023
Accepted: 24 August 2023
Published: 31 August 2023
DOI: https://doi.org/10.1038/s41598-023-41358-0
Springer Nature Limited

This article is cited by

Discriminating orientation information with phase consistency in alpha and low-gamma frequency bands: an EEG study
- Alireza Khadir
- Shamim Sasani Ghamsari
- Borhan Beigzadeh
Scientific Reports (2024)

The association between working memory precision and the nonlinear dynamics of frontal and parieto-occipital EEG activity

Abstract

Similar content being viewed by others

EEG dynamics reveal a dissociation between storage and selective attention within working memory

Effects of load and emotional state on EEG alpha-band power and inter-site synchrony during a visual working memory task

Gamma oscillations modulate working memory recall precision

Introduction

Methods

Participants

Experimental design

Mixture model analysis

EEG recording

EEG data preprocessing

Holo-Hilbert Spectral Analysis (HHSA)

Intra- and inter-areal PAC under the HHSA framework

Statistical analyses of EEG data

Evaluating the effect of WM load on EEG activity

Correlation between individual WM precision and EEG activity

Intra- and inter-areal PAC

Results

Behavioral results

WM load modulated parieto-occipital alpha/beta power

WM precision and HHS in the parieto-occipital region

WM precision and HHS in the frontal region

Intra- and inter-regional PACs between theta and alpha/beta oscillations

Discussion

Alpha/beta power suppression reflects the ability to access feature information in WM

Relationship between the frontal theta power and WM precision

Frontoparietal cross-frequency connectivity indicates the top-down modulation of memorized contents

HHSA reveals the relationship between oscillatory power variations and PAC

The possible roles of alpha/beta power suppression in the WM subprocesses

Conclusion

Limitations

Data availability

Abbreviations

References

Acknowledgements

Funding

Author information

Authors and Affiliations

Contributions

Corresponding author

Ethics declarations

Competing interests

Additional information

Publisher's note

Supplementary Information

Supplementary Information.

Rights and permissions

About this article

Cite this article

Share this article

This article is cited by

Discriminating orientation information with phase consistency in alpha and low-gamma frequency bands: an EEG study

Search

Navigation