Abstract

Recent empirical work suggests that, during healthy aging, the variability of network dynamics changes during task performance. Such variability appears to reflect the spontaneous formation and dissolution of different functional networks. We sought to extend these observations into resting-state dynamics. We recorded EEG in young, middle-aged, and older adults during a “rest–task–rest” design and investigated if aging modifies the interaction between resting-state activity and external stimulus-induced activity. Using multiscale entropy as our measure of variability, we found that, with increasing age, resting-state dynamics shifts from distributed to more local neural processing, especially at posterior sources. In the young group, resting-state dynamics also changed from pre- to post-task, where fine-scale entropy increased in task-positive regions and coarse-scale entropy increased in the posterior cingulate, a key region associated with the default mode network. Lastly, pre- and post-task resting-state dynamics were linked to performance on the intervening task for all age groups, but this relationship became weaker with increasing age. Our results suggest that age-related changes in resting-state dynamics occur across different spatial and temporal scales and have consequences for information processing capacity.

INTRODUCTION

The brain's intrinsic activity during resting-state has been of great interest to neuroscientists since imaging studies described coherent spatial patterns in the absence of overt external stimuli (Raichle et al., 2001; Biswal, Zerrin Yetkin, Haughton, & Hyde, 1995). It is well accepted now that resting-state activity plays an important role in cortical function and cannot be ignored in assessing brain–behavior relationships (Raichle & Snyder, 2007; Arieli, Sterkin, Grinvald, & Aertsen, 1996). A recent review (Deco, Jirsa, & McIntosh, 2011) further emphasized the importance of resting-state activity in the organization of a series of highly coherent functional networks and proposed that the dissolution and formation of resting-state patterns reflects the exploration of possible functional network configurations around a more deterministic anatomical skeleton.

Intuitively, the dissolution and formation of brain resting-state can be closely related to the interaction between resting-state activity and external stimulus-induced activity. In recent years, this interaction has been explored by several studies from two directions to better understand the functional role of resting-state activity (Northoff, Qin, & Nakao, 2010). On one hand, resting-state activity interacts with external input to impact stimulus-induced changes in the brain. This is related to predicting subsequent neural activity, behavioral and mental states. For example, using fMRI, Greicius and Menon (2004) observed an association between decreased activity during stimulation in the default mode network (DMN; one of the most studied and easily visualized networks during resting state) and increased stimulus-induced activity in visual and auditory cortices. Also using fMRI, Fox, Snyder, Vincent, and Raichle (2007) showed that intrinsic brain activity not only persists during task performance and contributes to variability in evoked brain responses but also predicts a high percentage of the trial-to-trial variability in RT in a subsequent button press task. Sala-Llonch et al. (2012) found that, during pre-task resting state, functional connectivity in the posteromedial parts of the DMN predicts behavioral performance on a subsequent working memory task. On the other hand, external stimulus-induced brain activity can also modulate subsequent resting-state activity. Barnes, Bullmore, and Suckling (2009) showed that endogenous dynamics parameters (fractal scaling properties of fMRI time series) recover slowly to their pre-task values after a working memory task, and the rate of recovery was related to the difficulty of the task. Similarly, Pyka et al. (2009) demonstrated increased activation in DMN at rest when the cognitive load of a preceding working memory task was high. Finally, Grigg and Grady (2010) examined resting-state functional connectivity of the DMN in two runs that were separated by a cognitive task in young participants and revealed more variable connectivity during the resting-state after the cognitive task. Taken together, these findings support the idea that endogenous oscillatory dynamics are relevant to the brain's response to external stimulation.

Importantly, resting-state activity is influenced not only by external stimulation but also by age. For example, compared with younger adults, older adults show weaker functional connectivity among DMN regions (Huang et al., 2015; Sambataro et al., 2010; Esposito et al., 2008; Andrews-Hanna et al., 2007) and a diminished reduction of DMN activity during cognitive tasks (Damoiseaux et al., 2008; Miller et al., 2008; Persson, Lustig, Nelson, & Reuter-Lorenz, 2007; Grady, Springer, Hongwanishkul, McIntosh, & Winocur, 2006; Lustig et al., 2003). A recent study examined the effect of age on both the DMN and a task-positive network and found an age-related decrease in the extent of the DMN and an expansion of the task-positive network (Grady et al., 2010).

In the current study, we were interested in the effects of age on how an intervening task changes pre- to post-task resting state. Most studies examining the interaction between resting-state activity and external stimulus-induced activity focus on signal amplitude. However, recent computational modeling suggests that brain signal variability (i.e., transient temporal fluctuations in brain signal) conveys important information about network dynamics (Deco et al., 2011). In a complex nonlinear system such as the brain, variability facilitates the transition between possible functional network configurations, in the presence or absence of external stimulation (McIntosh et al., 2010). Empirically, brain signal variability has been shown to track maturation and disease and reflect cognitive capacity (as examples, see McIntosh et al., 2014; Protzner, Kovacevic, Cohn, & McAndrews, 2013; Catarino, Churches, Baron-Cohen, Andrade, & Ring, 2011; Garrett, Kovacevic, McIntosh, & Grady, 2010; Mizuno et al., 2010; Protzner, Valiante, Kovacevic, McCormick, & McAndrews, 2010; Takahashi et al., 2010; Lippe, Kovacevic, & McIntosh, 2009; McIntosh, Kovacevic, & Itier, 2008). For example, our recent work with EEG and MEG suggests that, during healthy aging, brain signal variability increases in the context of local communication between neural populations and decreases for distal communication between neural populations (McIntosh et al., 2014). Thus, with increasing age, there is a shift from long-range connections to more local processing. Sleimen-Markoun and colleagues (2015) showed that the pattern of variability differentiating resting-state from an oddball counting task resembled the one differentiating age groups. EEG was more variable at rest than during an attentionally demanding condition for more local communication and less variable for more distal communication. This task-related differentiation was reduced in older participants. Finally, a study by Garrett and colleagues demonstrated that BOLD signal variability provided more than five times the age-predictive power as compared with BOLD mean activation (Garrett et al., 2010). These findings suggest that the relation between aging, cognition, and brain function are underappreciated by using the traditional mean-based brain measures exclusively (Garrett, Kovacevic, McIntosh, & Grady, 2011).

To our knowledge, only one study has examined age-related changes in the interplay between brain activity at rest as measured before and after an intervening task using a variability measure. Takahashi and colleagues (2009) recorded EEG during a “rest–task–rest” design, using photic stimulation as their task. To examine variability in EEG signal, they used multiscale entropy (MSE), a measure that is sensitive to linear and nonlinear variability and can differentiate variability of a complex system (e.g., the brain) from a purely random system (Costa, Goldberger, & Peng, 2002, 2005). They found a significant increase in complexity from pre- to post-photic stimulation in young participants but not in old participants and suggested that there is a diminished functional response to stimuli with aging.

In our study, like Takahasi et al., we employed a “rest–task–rest” design to investigate the interaction between resting-state activity, external stimulus-induced activity, and aging. Here, however, we acquired true resting-state data pre- and post-task (Takahashi et al., instead, used only a 20-sec segment of resting EEG identified pre- and postphotic stimulation). Additionally, we used cognitive tasks as our intervening “external stimulation” because we wanted to link potential task-induced changes in resting-state dynamics to behavioral performance. We recorded EEG in young, middle-aged, and older adults during rest and during two cognitive tasks (visual perceptual matching and delayed match-to-sample[DMS]) and examined age-related changes in MSE and spectral power density (SPD) in the resting-state data. We first established how resting-state dynamics change with age, using resting-state data collected prior to task performance. We then examined the alteration from pre- to post-task resting state to tackle the second question of this study: whether aging would modify the interaction between resting-state activity and external stimulus-induced activity.

METHODS

Participants

Sixteen young adults (six men, mean age = 22 ± 3 years, 21.75), 16 middle-aged adults (seven men, mean age = 45 ± 6 years, 45.06), and 16 older adults (five men, mean age = 66 ± 6 years, 66.06) participated in the study. All participants were right-handed with no adverse neurological histories and had normal or corrected-to-normal vision. All participants signed informed written consent before the experiment and received monetary compensation.

Apparatus and Task

EEG recordings from 76 electrodes were collected using BioSemi ActiveTwo system (Amsterdam, The Netherlands) with a bandwidth of 99.84 (0.16–100) Hz and a sampling rate of 512 Hz. Data were recorded reference-free. All channels were re-referenced offline to the vertex electrode (Cz), then converted to a common average reference during the preprocessing.

For the purposes of the current analyses, we used two recordings (5 min each) of EEG resting-state data that were collected with eyes opened. These resting-state recordings occurred before and after 60 min of cognitive task performance. The two cognitive tasks were perceptual matching (PM) and DMS, which have been described in another paper (McIntosh et al., 2014). Briefly, stimuli were one-dimensional Gaussian white noise fields with a two-octave frequency filter and were presented simultaneously in a triangular array. In the PM task, participants indicated which of the three reference stimuli (presented at the bottom of the screen) matched the test stimulus (presented at the top of the screen) by pressing one of three buttons. The task instructions for DMS were the same as for PM, except that in DMS there was a delay between the presentations of the test stimulus and the reference stimuli. We used a psychophysical threshold procedure to ensure that participants were matched in terms of accuracy by adjusting stimulus discriminability (McIntosh et al., 2014; Protzner & McIntosh, 2007). Each participant performed two blocks of PM and two blocks of DMS in random order.

Data Preprocessing

Continuous EEG recordings were bandpass filtered from 0.5 to 55 Hz. Resting-state data were split into 2.5-sec intervals that we considered as resting-state trials. Trials with excessive signal amplitude were rejected first. This was done by randomly selecting 30% of the trials with no artifacts (by visual inspection). From these selected trials, we calculated the mean global field power and used this value plus 5 standard deviations from the mean as the threshold to reject trials. To remove ocular and muscle artifacts, we used the Infomax independent component analysis algorithm implemented in the EEGLAB (Delorme & Makeig, 2004) toolbox for MatLab (The MathWorks, Natick, MA). After performing independent component analysis for each participant, we explored the signal and the topographical distribution of the components. The alignment of the vertical electrooculogram signal from the original data with the component traces reveals the artifactual nature of some components that show blink-like events time-aligned with blinks. Examining the frontal distribution of these components in the topographic representation corroborated this. Components that were spatially localized to areas of the scalp that are typical sites of muscle movement (e.g., above the ears, right and left temples) and show high-frequency traces (20–50 Hz and above) were identified as muscle artifact components.

To further localize the dynamics of source activity at specific locations, we identified 72 ROIs in Talairach space (Diaconescu, Alain, & McIntosh, 2011) and performed source estimation at these locations using Brainstorm (Tadel, Baillet, Mosher, Pantazis, & Leahy, 2011; neuroimage.usc.edu/brainstorm). Source modeling of brain activity was based on sLORETA method. Source reconstruction was constrained to the cortical mantle of the MNI/Colin27 brain template defined by the Montreal Neurological Institute. Current density along three source orientations (x, y, and z components) was estimated for 72 source locations adapted from the regional map coarse parcellation scheme, as developed in Kötter and Wanke (2005).

MSE Estimation of Temporal Signal Complexity

Full details of MSE and its relevance for the analysis of signal complexity are given in Costa et al. (2002, 2005). The MSE method calculates sample entropy as a measure of regularity (predictability) of the signal at different scales. It consists of two procedures: (1) coarse-graining of the time series and (2) calculating sample entropy for each coarse-grained time series. For scale t, the coarse-grained time series is constructed by averaging the data points within nonoverlapping windows of length t. The original time series corresponds to Scale 1. This procedure can be viewed as a smooth version of decimation. Sample entropy of each coarse-grained time series measures its regularity by evaluating the probability of repetitive patterns based on two parameters: the pattern length, m, and the tolerance level, r. For each participant, we first calculated MSE for single trials with m = 5, r = 1, respectively (www.physionet.org/physiotools/mse/). The length of single trial time series was 1280 time points corresponding to 2.5-sec epoch at 512-Hz sampling rate. MSE measures of source waveforms were calculated for each eyes open condition separately. Estimates were calculated on a single trial basis for each timescale and, subsequently, averaged across all trials within a given condition.

Spectral Power Density

In addition to our MSE analyses, we examined the SPD of the EEG signal. SPD of the signal was calculated using fast Fourier transform on single trial data. The signal was normalized (mean = 0, SD = 1) first to deal with age-related global signal power change. Relative contributions of different frequency bands to the total spectral power were calculated based on normalized data. Given a sampling rate of 512 Hz and 1280 time points in 2.5 sec per trial, the frequency resolution was 0.4 Hz. Single trial estimates were averaged across trials to obtain mean SPD for each condition.

Partial Least Squares Analysis

Partial least squares (PLS) analysis (Krishnan, Williams, McIntosh, & Abdi, 2011; McIntosh & Lobaugh, 2004; Lobaugh, West, & McIntosh, 2001; McIntosh, Bookstein, Haxby, & Grady, 1996) was used to assess age-related and condition-dependent changes in spatiotemporal distributions of MSE and SPD measures. PLS is similar to other multivariate techniques, such as canonical correlation, in that it operates on the entire data structure at once, extracting the patterns of maximal covariance between brain signals and group/condition designs. In the current study, PLS applied singular value decomposition on (1) the covariance matrix between MSE and the experimental design (for task PLS) or behavior data (for behavioral PLS) or (2) the covariance matrix between SPD and the experimental design (for task PLS) or behavior data (for behavioral PLS) across participants to identify latent variables (LVs). LVs identify distributed patterns in brain signal that show similarities or differences between participant groups and experimental conditions. Each LV contains three vectors: (1) task saliences, (2) brain saliences, and (3) singular values. Task saliences indicate the degree to which each task within each group is related to the brain signal pattern identified in the LV. These task saliences can be interpreted as the optimal contrast that codes the effect depicted in the LV. Brain saliences are numerical voxel weights that identify the collection of source locations that, as a whole, are most related to the effects expressed in the LV. The singular values represent the covariance between the contrast and the MSE or SPD and indicate the strength of the effect expressed by the LV. The conventional usage of PLS is similar to other multivariate techniques, such as PCA, in that the algorithm extracts LVs explaining the covariance between conditions and brain activity in order of the amount of covariance explained, with the LV accounting for the most covariance extracted first. However, in examining the age effect on MSE and for all our SPD analyses, we used a nonrotated version of task PLS, in which a priori contrasts restricted the spatiotemporal patterns derived from PLS. This version of PLS has the advantage of allowing direct assessment of hypothesized experimental effects.

Statistical assessment in PLS is done across two levels. First, the overall significance of each LV is assessed with permutation testing (Good, 2013). An LV was considered significant if the observed singular value exceeded the permuted singular value in more than 95% of the permutations (corresponding to p < .05). Second, bootstrap resampling is used to estimate confidence intervals around source location weights in each LV, allowing for an assessment of the relative contribution of particular sources and timescales, and the stability of the relation with age group or behavior (Efron & Tibshirani, 1986, 1994). No corrections for multiple comparisons are necessary because the source location weights are calculated in a single mathematical step on the whole brain. For the brain data, we plot bootstrap ratios (ratio of the individual weights over the estimated standard error) as a proxy for z scores. Confidence intervals are plotted for group effects. A minimum threshold of a stable 95% confidence interval was used for all analyses.

RESULTS

Behavior

Accuracy and RT data are summarized in Table 1. For all three age groups, accuracy was at or greater than 80% for PM and DMS, and this level did not differ between groups. Mean RT was significantly different between groups and increased with increasing age in both PM and DMS (PM: F(2, 45) = 4.14, p < .05; DMS: F(2, 45) = 8.92, p < .01).

Table 1. 

Behavioral Performance Measures

TaskYoungMiddleOlder
a. Percent Correct 
Perceptual matching 92.4 (5.0) 91.63 (4.7) 91.6 (2.5) 
Delayed match to sample 87.7 (3.7) 84.1 (9) 87.2 (9.1) 
 
b. RT (msec) 
Perceptual matching 1271 (303) 1351 (173) 1567 (387) 
Delayed match to sample 1015 (187) 1157 (136) 1268 (180) 
TaskYoungMiddleOlder
a. Percent Correct 
Perceptual matching 92.4 (5.0) 91.63 (4.7) 91.6 (2.5) 
Delayed match to sample 87.7 (3.7) 84.1 (9) 87.2 (9.1) 
 
b. RT (msec) 
Perceptual matching 1271 (303) 1351 (173) 1567 (387) 
Delayed match to sample 1015 (187) 1157 (136) 1268 (180) 

Standard deviations are in parentheses.

MSE

Age Effects in Pre-task Resting State

Our first task PLS examined the effect of healthy aging on resting state in terms of brain dynamics, as measured by MSE, using only the pre-task resting condition, as this is how resting-state data typically are acquired. We tested whether or not age-related changes in MSE followed a linear trend from young to old participants and found that the linear trend was significant (p = .026; Figure 1A). The same age effect persisted when we treated age as continuous measure. The effect of age was dependent on temporal scale. At fine temporal scales (scales: 2–16 msec), the young group showed reduced MSE as compared with the middle-aged and older groups. At coarse temporal scales (scales: 40–50 msec), the young group showed greater MSE than middle-aged and older groups. This age effect was not the same across all sources; it was the strongest bilaterally in primary motor cortex, precuneus, pulvinar, thalamus, superior and inferior parietal cortex, posterior cingulate, and left premotor cortex.

Figure 1. 

Task PLS results for the comparison of age-related changes in MSE (A) and SPD (B) in the pre-task resting state. Average MSE (±SE) (A) and average SPD (B) curves are plotted for each group for the regional source in left precuneus in pre-task resting state. Circles above the curves indicate the time points/frequencies with reliable confidence intervals (|bootstrap ratio| > 3). The statistical image plots (bootstrap ratio maps) represent the regional sources and timescales/frequencies at which the linear trend of age-related changes were most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales and sources showing increased MSE/SPD with age and negative values indicate age-related decreases.

Figure 1. 

Task PLS results for the comparison of age-related changes in MSE (A) and SPD (B) in the pre-task resting state. Average MSE (±SE) (A) and average SPD (B) curves are plotted for each group for the regional source in left precuneus in pre-task resting state. Circles above the curves indicate the time points/frequencies with reliable confidence intervals (|bootstrap ratio| > 3). The statistical image plots (bootstrap ratio maps) represent the regional sources and timescales/frequencies at which the linear trend of age-related changes were most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales and sources showing increased MSE/SPD with age and negative values indicate age-related decreases.

Age Effects in Pre-task versus Post-task Resting State

Our second task PLS examined potential age differences in the change in MSE from pre- to post-task resting state. We identified one significant LV, differentiating pre- to post-task resting state, specifically in the young group and not in the middle-aged and older groups (p = .048; Figure 2A). MSE increased from pre- to post-task resting state at fine temporal scales (scales: 2–14 msec) in bilateral pFC, anterior cingulate, and subgenual cingulate cortex. MSE decreased from pre- to post-task at coarse temporal scales (scales: 20–50 msec) in posterior and retrosplenial cingulate, bilateral primary visual area, and fusiform gyrus.

Figure 2. 

Task PLS results for the group by condition effect (pre- vs. post-task resting state) on (A) MSE and (B) SPD. The bar graphs depict the contrast between age groups across conditions that was significantly expressed across the entire data set ((A) Source × Timescales or (B) Source × Frequencies) as determined by permutation tests. The statistical image plots (bootstrap ratio maps) in each section represent the regional sources and timescales/frequencies at which the contrast displayed in the bar graphs was most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales and sources showing increases in MSE/SPD from pre- to post-task resting state, and negative values denote decreases in MSE/SPD from pre- to post-task resting state.

Figure 2. 

Task PLS results for the group by condition effect (pre- vs. post-task resting state) on (A) MSE and (B) SPD. The bar graphs depict the contrast between age groups across conditions that was significantly expressed across the entire data set ((A) Source × Timescales or (B) Source × Frequencies) as determined by permutation tests. The statistical image plots (bootstrap ratio maps) in each section represent the regional sources and timescales/frequencies at which the contrast displayed in the bar graphs was most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales and sources showing increases in MSE/SPD from pre- to post-task resting state, and negative values denote decreases in MSE/SPD from pre- to post-task resting state.

The Relationship between Task Performance (RT and Accuracy) and Age in Pre-task and Post-task Resting State

We investigated whether MSE measured during pre- and post-task resting state would show a relationship with cognitive task performance. For all analyses described below, our results were similar regardless of the task (PM or DMS) from which accuracy and RT measures were obtained. Thus, for the sake of parsimony, we report only the results using performance measures from the DMS task.

Our behavior PLS examining the relationship between performance and age in all age groups identified two significant LVs, both of which indicated that the relationship with performance varies with age (LV1, p = .000; LV2, p = .008). We therefore conducted separate behavior PLSs for each age group and discuss these results in detail below.

For the young group, the behavior PLS examining condition-dependent changes in MSE associated with individual differences in RT and accuracy identified one significant LV (p = .002; Figure 3A). For both pre- and post-task conditions, increased MSE at fine temporal scales (scales: 2–14 msec) was associated with faster RT and increased accuracy. These effects were not homogeneous across sources and were strongest at ACC, left V2, right fusiform gyrus, bilateral premotor cortex, right superior temporal cortex, bilateral OFC, and right anterior insula. For both pre- and post-task conditions, increased MSE at coarse temporal scales (scales: 24–50 msec) was associated with slower RTs and decreased accuracy. These effects were strongest at ACC, left medial premotor cortex, bilateral OFC, right anterior insula, right premotor cortex, and right temporal pole. A more stable correlation pattern was observed in the right hemisphere as compared with the left hemisphere (i.e., higher bootstrap ratio values were observed in the right hemisphere).

Figure 3. 

Behavioral PLS results for the correlation of (A) MSE and behavior data and (B) SPD and behavior data from the DMS task (accuracy and RT) in the young group. The bar graph depicts the contrast between conditions and behavior measures that was significantly expressed across the entire data set ((A) Source × Timescales or (B) Source × Frequencies) as determined by permutation tests. The statistical image plots (bootstrap ratio maps) in each section represent the regional sources and timescales/frequencies at which the correlation between MSE/SPD and behavior measures was most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales/frequencies and sources showing increased MSE/SPD with better performance (higher accuracy and lower RT), and negative values denote decreased MSE/SPD with better performance.

Figure 3. 

Behavioral PLS results for the correlation of (A) MSE and behavior data and (B) SPD and behavior data from the DMS task (accuracy and RT) in the young group. The bar graph depicts the contrast between conditions and behavior measures that was significantly expressed across the entire data set ((A) Source × Timescales or (B) Source × Frequencies) as determined by permutation tests. The statistical image plots (bootstrap ratio maps) in each section represent the regional sources and timescales/frequencies at which the correlation between MSE/SPD and behavior measures was most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales/frequencies and sources showing increased MSE/SPD with better performance (higher accuracy and lower RT), and negative values denote decreased MSE/SPD with better performance.

For the middle-aged group, the behavior PLS examining condition-dependent changes in MSE associated with individual differences in RT and accuracy identified one significant LV (p = .006; Figure 4A). For both pre- and post-task conditions, increased MSE at middle temporal scales (scales: 10–22 msec) was associated with greater accuracy and had no stable relationship with RT. These effects were strongest at posterior cingulate cortex, bilateral pulvinar, right A1, right precuneus, and right ventral temporal cortex. For both pre- and post-task conditions, increased MSE at coarse temporal scales (scales: 40–50 msec) was associated with lower accuracy. These effects were strongest at posterior and retrosplenial cingulate cortex, left pulvinar, bilateral V1, right superior parietal cortex, and right ventral temporal cortex.

Figure 4. 

Behavioral PLS results for the correlation of (A) MSE and behavior data and (B) SPD and behavior data from the DMS task (accuracy and RT) in the middle-aged group. The bar graph depicts the contrast between conditions and behavior measures that was significantly expressed across the entire data set ((A) Source × Timescales or (B) Source × Frequencies) as determined by permutation tests. The statistical image plots (bootstrap ratio maps) in each section represent the regional sources and timescales/frequencies at which the correlation between MSE/SPD and behavior measures was most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales/frequencies and sources showing increased MSE/SPD with better performance (higher accuracy), and negative values denote decreased MSE/SPD with better performance.

Figure 4. 

Behavioral PLS results for the correlation of (A) MSE and behavior data and (B) SPD and behavior data from the DMS task (accuracy and RT) in the middle-aged group. The bar graph depicts the contrast between conditions and behavior measures that was significantly expressed across the entire data set ((A) Source × Timescales or (B) Source × Frequencies) as determined by permutation tests. The statistical image plots (bootstrap ratio maps) in each section represent the regional sources and timescales/frequencies at which the correlation between MSE/SPD and behavior measures was most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales/frequencies and sources showing increased MSE/SPD with better performance (higher accuracy), and negative values denote decreased MSE/SPD with better performance.

For the older group, the behavior PLS examining condition-dependent changes in MSE associated with individual differences in RT and accuracy identified two significant LVs. The first LV (p = .002; Figure 5) revealed that, for both pre- and post-task conditions, increased MSE at middle to coarse temporal scales (14–49 msec) was associated with faster RT and greater accuracy. These effects were strongest bilaterally in the pFC. The second LV (p = .038; Figure 6) revealed that, for both pre- and post-task conditions, increased MSE at fine to middle temporal scales (scales: 2–18 msec) was associated with greater accuracy (but not RT). The strongest effects were located at posterior and retrosplenial cingulate cortex, bilateral precuneus, bilateral superior parietal cortex, bilateral parahippocampal cortex, bilateral pulvinar, bilateral thalamus, and right ventral temporal cortex.

Figure 5. 

Behavioral PLS LV1 for the correlation of MSE and behavior data from the DMS task (accuracy and RT) in the older group. The bar graph depicts the contrast between conditions and behavior measures that was significantly expressed across the entire data set (Source × Timescales) as determined by permutation tests. The statistical image plots (bootstrap ratio maps) in each section represent the regional sources and timescales at which the correlation between MSE and behavior measures was most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales and sources showing increased MSE with better performance (higher accuracy and faster RT), and negative values denote decreased MSE/SPD with better performance.

Figure 5. 

Behavioral PLS LV1 for the correlation of MSE and behavior data from the DMS task (accuracy and RT) in the older group. The bar graph depicts the contrast between conditions and behavior measures that was significantly expressed across the entire data set (Source × Timescales) as determined by permutation tests. The statistical image plots (bootstrap ratio maps) in each section represent the regional sources and timescales at which the correlation between MSE and behavior measures was most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales and sources showing increased MSE with better performance (higher accuracy and faster RT), and negative values denote decreased MSE/SPD with better performance.

Figure 6. 

Behavioral PLS LV2 for the correlation of MSE and behavior data from the DMS task (accuracy and RT) in the older group. The bar graph depicts the contrast between conditions and behavior measures that was significantly expressed across the entire data set (Source × Timescales) as determined by permutation tests. The statistical image plots (bootstrap ratio maps) in each section represent the regional sources and timescales at which the correlation between MSE and behavior measures was most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales and sources showing increased MSE with better performance (higher accuracy), and negative values denote decreased MSE/SPD with better performance.

Figure 6. 

Behavioral PLS LV2 for the correlation of MSE and behavior data from the DMS task (accuracy and RT) in the older group. The bar graph depicts the contrast between conditions and behavior measures that was significantly expressed across the entire data set (Source × Timescales) as determined by permutation tests. The statistical image plots (bootstrap ratio maps) in each section represent the regional sources and timescales at which the correlation between MSE and behavior measures was most stable as determined by bootstrapping. Values represent the ratio of the parameter estimate for the source divided by the bootstrap-derived standard error (roughly z scores). Positive values indicate timescales and sources showing increased MSE with better performance (higher accuracy), and negative values denote decreased MSE/SPD with better performance.

Spectral Power Density

To allow for easy comparison between MSE and SPD results, we used the nonrotated version of PLS to examine in SPD specifically the effects that we identified in our MSE analyses. Additionally, for the nonrotated behavioral PLSs, we report only the results using performance measures from the DMS task (as with MSE, our results were similar regardless of the task (PM or DMS) from which accuracy and RT measures were obtained).

Age Effects in Pre-task Resting State

We first used nonrotated task PLS to test whether or not age-related changes in SPD followed a linear trend from young to old participants and found that the linear trend was significant (p = .01; Figure 1B). For pre-task resting-state, delta (0.5–3 Hz) and low theta (4–5 Hz) band spectral power was highest for the young group, and beta band (16–23 Hz) spectral power was highest for the oldest group at several source locations, including posterior and retrosplenial cingulate, bilateral primary motor and premotor cortex, and several posterior sources bilaterally, including parietal cortex and precuneus.

Age Effects in Pre-task versus Post-task Resting State

We next used nonrotated task PLS to examine how resting-state SPD changed from pre- to post-task across our age groups and found significant pre- to post-task differences common to all three groups (p = .002; Figure 2B). Spectral power increased from pre- to post-task in high theta (6–8 Hz) and low alpha (8–10 Hz) bands throughout the brain. Spectral power also increased in low beta band (16–24 Hz) bilaterally in the pFC. Spectral power decreased in the delta band (0.5–3 Hz) throughout the cortex and low beta band (16–20 Hz) at posterior and retrosplenial cingulate cortex and left temporal regions.

The Relationship between Task Performance (RT and Accuracy) and Age in Pre-task and Post-task Resting State

For the young group, the nonrotated behavior PLS examining condition-dependent changes in SPD associated with individual differences in RT and accuracy was significant (p < .001; Figure 3B). For both pre- and post-task conditions, decreased spectral power at high theta (5–7 Hz) and low alpha (8–9 Hz) bands was associated with faster RT and greater accuracy. These effects were strongest effects in right temporal and prefrontal sources. Increased spectral power at beta (16–40 Hz) and gamma bands (40–50 Hz) was associated with faster RT and greater accuracy. These effects were strongest for beta at ACC and several sources within the right pFC and premotor cortex. For gamma, strongest effects were observed at right anterior insula, right OFC, and right fusiform gyrus.

For the middle-aged group, the nonrotated behavior PLS examining condition-dependent changes in SPD associated with individual differences in performance identified one significant LV (p = .014; Figure 4B). For both pre- and post-task conditions, decreased spectral power in the theta band (4–7 Hz) was associated with greater accuracy (and not RT), with strongest effects at bilateral V1, left fusiform gyrus, and right ventral temporal cortex. Meanwhile, for both pre- and post-task conditions, increased spectral power in the beta band (16–30 Hz) was associated with greater accuracy at posterior cingulate cortex, bilateral pulvinar, bilateral thalamus, and bilateral A1.

For the older group, the nonrotated behavior PLS examining condition-dependent changes in SPD associated with individual differences in RT and accuracy did not identify any significant LVs.

DISCUSSION

Using a “rest–task–rest” EEG experimental design, we examined the effect of aging on SPD and MSE of resting-state brain activity. We demonstrated that healthy aging not only affects intrinsic activity during resting state but also modulates the interaction between intrinsic activity and external stimulus-induced activity.

The Effects of Age on Pre-task Resting State

We first examined the effects of age on resting-state acquired prior to task performance, as resting state most commonly is acquired this way. Our pre-task resting state MSE analysis revealed a significant age effect, where middle-aged and older groups showed higher MSE at fine temporal scales bilaterally in parietal cortex and posterior cingulate, whereas the young group showed higher MSE at coarse temporal scales in the same regions. This finding supports the idea that healthy aging is characterized by increased local information processing (higher MSE at fine temporal scales) and decreased long-range interactions (lower MSE at coarse temporal scales) with other neural populations (McIntosh et al., 2014; Vakorin, Lippé, & McIntosh, 2011). It particularly suggests that aging tends to shift the integration between distributed neural populations (MSE at coarse temporal scales) toward more local neural processing (MSE at fine temporal scales) in posterior regions of the brain. However, the age effect we observed in pre-task resting state did not replicate recent work (Takahashi et al., 2009), showing no significant differences in MSE between young and older participants as measured during rest prior to photic stimulation. One likely reason that Takahashi et al. (2009) did not find an age effect in their prephotic stimulation condition is that they used 20 sec of pre-task resting data per participant in their experiment, whereas we used 5 min.

Our pre-task resting-state SPD analysis also revealed a significant age effect, indicating that middle-aged and older participants showed more relative spectral power in the beta band in posterior brain regions as compared with younger participants. These posterior effects overlap with regions identified in our MSE analysis, where increased age is associated with increased fine temporal-scale MSE. The age effect for SPD is consistent with the finding that normal aging is associated with an EEG power shift (Polich, 1997). The finding that both power in higher frequencies and MSE at fine temporal scales increased with age in overlapping posterior brain regions suggests that the two measures are associated. Mizuno et al. (2010) also found that the MSE results from smaller and larger scales correlated with power in faster and slower frequencies, respectively.

Age-related Changes Associated with the Impact of an Intervening Task on Resting-state Activity

Our results suggest that an intervening cognitive task impacts on subsequent resting-state activity differently across age groups. Specifically, we found alterations from pre- to post-task resting-state MSE in the young group only. MSE increased at fine temporal scales from pre- to post-task resting state mainly in the anterior cingulate and bilateral prefrontal regions. Interestingly, these are the same regions in which increased fine-scale entropy is linked with better intervening cognitive task performance in the young group. We additionally found that, for the young group, pre- to post-task resting-state MSE decreased at coarse temporal scales in the posterior cingulate, a key region associated with the DMN. We interpret these results as a task-related after-effect, as they are consistent with the general concept that activity in DMN decreases during cognitive tasks to facilitate the goal-directed neural activity. Several studies suggest that the reduction of DMN activity during cognitive task is more evident in younger adults, compared with older adults (Damoiseaux et al., 2008; Miller et al., 2008; Persson et al., 2007; Grady et al., 2006; Lustig et al., 2003). Our findings suggest that, even after cognitive task performance, the decreased long-range interaction (lower MSE at coarse temporal scales) or the reduced brain dynamics in parts of the DMN is more conspicuous in young adults.

Similar MSE changes in post-task resting state were observed by Takahashi et al. (2009), as they also found a significant change in signal complexity in young participants but not older participants after photic stimulation. In their study, the young group showed significant MSE increase at coarse scales (scale > 10, around 50 msec given a sampling rate at 200 Hz). However, in our study, the significant increase in MSE was observed around finer scales (scale < 10, 2–15 msec). The different findings in the two studies may demonstrate how cognitive task and simple external stimuli affect post-task resting state differently.

Unlike our MSE results, our SPD analysis examining age effects on the impact of an intervening cognitive task on subsequent resting state identified similar changes across all age groups, where spectral power decreased from pre- to post-task in the beta band in cingulate and temporal cortex. Taken together, the lack of change in MSE and significant change in SPD for older participants may indicate that healthy aging is associated with changes in the nonlinear dependencies that are not evident in SPD. Our finding fits nicely with other work, suggesting that the association between aging and brain function is underappreciated unless higher-order aspects of brain signals are considered (Garrett et al., 2010, 2011).

Age-related Changes in the Relationship between Resting-state Activity and Intervening Task Performance

Lastly, our study demonstrates that the dynamics of resting-state activity in both pre- and post-task are associated with performance on the intervening task throughout the age range, but the relationship varies with age. For the young group, increased MSE at fine temporal scales; decreased power in the delta, theta, and low alpha bands; and increased SPD in the beta and gamma bands were associated with faster RT and greater accuracy. These correlation patterns were particularly significant in regions where we found increased post-task MSE (e.g., bilateral pFC and anterior cingulate).

MSE relation to behavior for the middle-aged group was similar to those of the young group, but weaker overall, more posterior, and stable only for accuracy and not RT. SPD effects were also similar, showing an association between decreased theta power, increased beta power, and increased accuracy. As with MSE, these effects were strongest in more posterior regions of the brain than in the young.

For the older group, increased MSE at middle temporal scales bilaterally in the pFC was associated with faster RT and greater accuracy. Increased MSE at fine temporal scales in posterior regions was associated with greater accuracy (but not RT). We did not identify a significant relationship between SPD and performance on the intervening tasks for this group.

These findings suggested that pre-task MSE at fine temporal scales and SPD in beta frequencies could be good predictors of cognitive task performance across the age range. However, the regions showing this effect vary with age, as task-positive regions (e.g., bilateral prefrontal and anterior cingulate) are the most predictive for young individuals and more posterior regions are involved for older individuals. However, the relationship between SPD and performance weakens with increasing age, as we did not identify significant effects in the oldest group. This may suggest a decoupling of MSE and SPD in normal aging, where MSE reflects information processing capacity insofar as the capacity is captured in higher-order nonlinearities that are not measurable with SPD.

Conclusions

In summary, our findings suggest that aging changes the interplay between resting-state activity and an intervening cognitive task. Aging appears to shift resting-state dynamics from the integration of distributed neural populations toward more local neural processing, especially in posterior regions of the brain. The effect of an intervening cognitive task on subsequent resting state appears to be strongest for young individuals. Although SPD changes from pre- to post-resting state in middle-aged and older individuals, MSE does not. Finally, resting-state dynamics are associated with performance on the intervening task for all age groups, but this relationship weakens with increasing age. Our results reinforce the idea that age-related brain changes manifest across different spatial and temporal scales affecting the information processing capacity or complexity of the brain.

Acknowledgments

This research was supported by grants from J.S. McDonnell Foundation to A. R. M. and grants from the Natural Sciences and Engineering Research Council of Canada, Canadian Foundation for Innovation Leaders Opportunity Fund, and Alberta Enterprise and Advanced Education Research Capacity Program, Alberta Alignment grants to A. B. P.

Reprint requests should be sent to Andrea B. Protzner, Department of Psychology, University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada, T2N 1N4, or via e-mail: protzner@ucalgary.ca.

REFERENCES

Andrews-Hanna
,
J. R.
,
Snyder
,
A. Z.
,
Vincent
,
J. L.
,
Lustig
,
C.
,
Head
,
D.
,
Raichle
,
M. E.
, et al
(
2007
).
Disruption of large-scale brain systems in advanced aging
.
Neuron
,
56
,
924
935
.
Arieli
,
A.
,
Sterkin
,
A.
,
Grinvald
,
A.
, &
Aertsen
,
A. D.
(
1996
).
Dynamics of ongoing activity: Explanation of the large variability in evoked cortical responses
.
Science
,
273
,
1868
1871
.
Barnes
,
A.
,
Bullmore
,
E. T.
, &
Suckling
,
J.
(
2009
).
Endogenous human brain dynamics recover slowly following cognitive effort
.
PloS One
,
4
,
e6626
.
Biswal
,
B.
,
Zerrin Yetkin
,
F.
,
Haughton
,
V. M.
, &
Hyde
,
J. S.
(
1995
).
Functional connectivity in the motor cortex of resting human brain using echo-planar MRI
.
Magnetic Resonance in Medicine
,
34
,
537
541
.
Catarino
,
A.
,
Churches
,
O.
,
Baron-Cohen
,
S.
,
Andrade
,
A.
, &
Ring
,
H.
(
2011
).
Atypical EEG complexity in autism spectrum conditions: A multiscale entropy analysis
.
Clinical Neurophysiology
,
122
,
2375
2383
.
Costa
,
M.
,
Goldberger
,
A. L.
, &
Peng
,
C. K.
(
2002
).
Multiscale entropy to distinguish physiologic and synthetic RR time series
.
Computers in Cardiology
,
29
,
137
140
).
Costa
,
M.
,
Goldberger
,
A. L.
, &
Peng
,
C. K.
(
2005
).
Multiscale entropy analysis of biological signals
.
Physical Review E
,
71
,
021906
.
Damoiseaux
,
J. S.
,
Beckmann
,
C. F.
,
Arigita
,
E. S.
,
Barkhof
,
F.
,
Scheltens
,
P.
,
Stam
,
C. J.
, et al
(
2008
).
Reduced resting-state brain activity in the “default network” in normal aging
.
Cerebral Cortex
,
18
,
1856
1864
.
Deco
,
G.
,
Jirsa
,
V. K.
, &
McIntosh
,
A. R.
(
2011
).
Emerging concepts for the dynamical organization of resting-state activity in the brain
.
Nature Reviews Neuroscience
,
12
,
43
56
.
Delorme
,
A.
, &
Makeig
,
S.
(
2004
).
EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis
.
Journal of Neuroscience Methods
,
134
,
9
21
.
Diaconescu
,
A. O.
,
Alain
,
C.
, &
McIntosh
,
A. R.
(
2011
).
The co-occurrence of multisensory facilitation and cross-modal conflict in the human brain
.
Journal of Neurophysiology
,
106
,
2896
2909
.
Efron
,
B.
, &
Tibshirani
,
R.
(
1986
).
Bootstrap methods for standard errors, confidence intervals, and other measures of statistical accuracy
.
Statistical Science
,
1
,
54
75
.
Efron
,
B.
, &
Tibshirani
,
R. J.
(
1994
).
An introduction to the bootstrap
.
New York
:
CRC Press
.
Esposito
,
F.
,
Aragri
,
A.
,
Pesaresi
,
I.
,
Cirillo
,
S.
,
Tedeschi
,
G.
,
Marciano
,
E.
, et al
(
2008
).
Independent component model of the default-mode brain function: Combining individual-level and population-level analyses in resting-state fMRI
.
Magnetic Resonance Imaging
,
26
,
905
913
.
Fox
,
M. D.
,
Snyder
,
A. Z.
,
Vincent
,
J. L.
, &
Raichle
,
M. E.
(
2007
).
Intrinsic fluctuations within cortical systems account for intertrial variability in human behavior
.
Neuron
,
56
,
171
184
.
Garrett
,
D. D.
,
Kovacevic
,
N.
,
McIntosh
,
A. R.
, &
Grady
,
C. L.
(
2010
).
Blood oxygen level-dependent signal variability is more than just noise
.
Journal of Neuroscience
,
30
,
4914
4921
.
Garrett
,
D. D.
,
Kovacevic
,
N.
,
McIntosh
,
A. R.
, &
Grady
,
C. L.
(
2011
).
The importance of being variable
.
Journal of Neuroscience
,
31
,
4496
4503
.
Good
,
P.
(
2013
).
Permutation tests: A practical guide to resampling methods for testing hypotheses
.
New York
:
Springer Science & Business Media
.
Grady
,
C. L.
,
Protzner
,
A. B.
,
Kovacevic
,
N.
,
Strother
,
S. C.
,
Afshin-Pour
,
B.
,
Wojtowicz
,
M.
, et al
(
2010
).
A multivariate analysis of age-related differences in default mode and task-positive networks across multiple cognitive domains
.
Cerebral Cortex
,
20
,
1432
1447
.
Grady
,
C. L.
,
Springer
,
M. V.
,
Hongwanishkul
,
D.
,
McIntosh
,
A. R.
, &
Winocur
,
G.
(
2006
).
Age-related changes in brain activity across the adult lifespan
.
Journal of Cognitive Neuroscience
,
18
,
227
241
.
Greicius
,
M. D.
, &
Menon
,
V.
(
2004
).
Default-mode activity during a passive sensory task: Uncoupled from deactivation but impacting activation
.
Journal of Cognitive Neuroscience
,
16
,
1484
1492
.
Grigg
,
O.
, &
Grady
,
C. L.
(
2010
).
Task-related effects on the temporal and spatial dynamics of resting-state functional connectivity in the default network
.
PLoS One
,
5
,
e13311
.
Huang
,
C. C.
,
Hsieh
,
W. J.
,
Lee
,
P. L.
,
Peng
,
L. N.
,
Liu
,
L. K.
,
Lee
,
W. J.
, et al
(
2015
).
Age-related changes in resting-state networks of a large sample size of healthy elderly
.
CNS Neuroscience & Therapeutics
,
21
,
817
825
.
Kötter
,
R.
, &
Wanke
,
E.
(
2005
).
Mapping brains without coordinates
.
Philosophical Transactions of the Royal Society of London, Series B, Biological Sciences
,
360
,
751
766
.
Krishnan
,
A.
,
Williams
,
L. J.
,
McIntosh
,
A. R.
, &
Abdi
,
H.
(
2011
).
Partial least squares (PLS) methods for neuroimaging: A tutorial and review
.
Neuroimage
,
56
,
455
475
.
Lippe
,
S.
,
Kovacevic
,
N.
, &
McIntosh
,
A. R.
(
2009
).
Differential maturation of brain signal complexity in the human auditory and visual system
.
Frontiers in Human Neuroscience
,
3
,
48
.
Lobaugh
,
N. J.
,
West
,
R.
, &
McIntosh
,
A. R.
(
2001
).
Spatiotemporal analysis of experimental differences in event-related potential data with partial least squares
.
Psychophysiology
,
38
,
517
530
.
Lustig
,
C.
,
Snyder
,
A. Z.
,
Bhakta
,
M.
,
O'Brien
,
K. C.
,
McAvoy
,
M.
,
Raichle
,
M. E.
, et al
(
2003
).
Functional deactivations: Change with age and dementia of the Alzheimer type
.
Proceedings of the National Academy of Sciences, U.S.A.
,
100
,
14504
14509
.
McIntosh
,
A. R.
,
Bookstein
,
F. L.
,
Haxby
,
J. V.
, &
Grady
,
C. L.
(
1996
).
Spatial pattern analysis of functional brain images using partial least squares
.
Neuroimage
,
3
,
143
157
.
McIntosh
,
A. R.
,
Kovacevic
,
N.
, &
Itier
,
R. J.
(
2008
).
Increased brain signal variability accompanies lower behavioral variability in development
.
PLoS Computational Biology
,
4
,
e1000106
.
McIntosh
,
A. R.
,
Kovacevic
,
N.
,
Lippe
,
S.
,
Garrett
,
D.
,
Grady
,
C.
, &
Jirsa
,
V.
(
2010
).
The development of a noisy brain
.
Archives Italiennes de Biologie
,
148
,
323
337
.
McIntosh
,
A. R.
, &
Lobaugh
,
N. J.
(
2004
).
Partial least squares analysis of neuroimaging data: Applications and advances
.
Neuroimage
,
23
,
S250
S263
.
McIntosh
,
A. R.
,
Vakorin
,
V.
,
Kovacevic
,
N.
,
Wang
,
H.
,
Diaconescu
,
A.
, &
Protzner
,
A. B.
(
2014
).
Spatiotemporal dependency of age-related changes in brain signal variability
.
Cerebral Cortex
,
24
,
1806
1817
.
Miller
,
S. L.
,
Celone
,
K.
,
DePeau
,
K.
,
Diamond
,
E.
,
Dickerson
,
B. C.
,
Rentz
,
D.
, et al
(
2008
).
Age-related memory impairment associated with loss of parietal deactivation but preserved hippocampal activation
.
Proceedings of the National Academy of Sciences, U.S.A.
,
105
,
2181
2186
.
Mizuno
,
T.
,
Takahashi
,
T.
,
Cho
,
R. Y.
,
Kikuchi
,
M.
,
Murata
,
T.
,
Takahashi
,
K.
, et al
(
2010
).
Assessment of EEG dynamical complexity in Alzheimer's disease using multiscale entropy
.
Clinical Neurophysiology
,
121
,
1438
1446
.
Northoff
,
G.
,
Qin
,
P.
, &
Nakao
,
T.
(
2010
).
Rest-stimulus interaction in the brain: A review
.
Trends in Neurosciences
,
33
,
277
284
.
Persson
,
J.
,
Lustig
,
C.
,
Nelson
,
J.
, &
Reuter-Lorenz
,
P. A.
(
2007
).
Age differences in deactivation: A link to cognitive control?
.
Journal of Cognitive Neuroscience
,
19
,
1021
1032
.
Polich
,
J.
(
1997
).
EEG and ERP assessment of normal aging
.
Electroencephalography and Clinical Neurophysiology
,
104
,
244
256
.
Protzner
,
A. B.
,
Kovacevic
,
N.
,
Cohn
,
M.
, &
McAndrews
,
M. P.
(
2013
).
Characterizing functional integrity: Intraindividual brain signal variability predicts memory performance in patients with medial temporal lobe epilepsy
.
Journal of Neuroscience
,
33
,
9855
9865
.
Protzner
,
A. B.
, &
McIntosh
,
A. R.
(
2007
).
The interplay of stimulus modality and response latency in neural network organization for simple working memory tasks
.
Journal of Neuroscience
,
27
,
3187
3197
.
Protzner
,
A. B.
,
Valiante
,
T. A.
,
Kovacevic
,
N.
,
McCormick
,
C.
, &
McAndrews
,
M. P.
(
2010
).
Hippocampal signal complexity in mesial temporal lobe epilepsy: A noisy brain is a healthy brain
.
Archives Italiennes de Biologie
,
148
,
289
297
.
Pyka
,
M.
,
Beckmann
,
C. F.
,
Schöning
,
S.
,
Hauke
,
S.
,
Heider
,
D.
,
Kugel
,
H.
, et al
(
2009
).
Impact of working memory load on fMRI resting state pattern in subsequent resting phases
.
PloS One
,
4
,
e7198
.
Raichle
,
M. E.
,
MacLeod
,
A. M.
,
Snyder
,
A. Z.
,
Powers
,
W. J.
,
Gusnard
,
D. A.
, &
Shulman
,
G. L.
(
2001
).
A default mode of brain function
.
Proceedings of the National Academy of Sciences, U.S.A.
,
98
,
676
682
.
Raichle
,
M. E.
, &
Snyder
,
A. Z.
(
2007
).
A default mode of brain function: A brief history of an evolving idea
.
Neuroimage
,
37
,
1083
1090
.
Sala-Llonch
,
R.
,
Pena-Gomez
,
C.
,
Arenaza-Urquijo
,
E. M.
,
Vidal-Piñeiro
,
D.
,
Bargallo
,
N.
,
Junque
,
C.
, et al
(
2012
).
Brain connectivity during resting state and subsequent working memory task predicts behavioural performance
.
Cortex
,
48
,
1187
1196
.
Sambataro
,
F.
,
Murty
,
V. P.
,
Callicott
,
J. H.
,
Tan
,
H. Y.
,
Das
,
S.
,
Weinberger
,
D. R.
, et al
(
2010
).
Age-related alterations in default mode network: Impact on working memory performance
.
Neurobiology of Aging
,
31
,
839
852
.
Sleimen-Malkoun
,
R.
,
Perdikis
,
D.
,
Müller
,
V.
,
Blanc
,
J. L.
,
Huys
,
R.
,
Temprado
,
J. J.
, et al
(
2015
).
Brain dynamics of aging: Multiscale variability of EEG signals at rest and during an auditory oddball task
.
eNeuro
,
2
,
ENEURO-0067
.
Tadel
,
F.
,
Baillet
,
S.
,
Mosher
,
J. C.
,
Pantazis
,
D.
, &
Leahy
,
R. M.
(
2011
).
Brainstorm: A user-friendly application for MEG/EEG analysis
.
Computational Intelligence and Neuroscience
,
2011
,
8
.
Takahashi
,
T.
,
Cho
,
R. Y.
,
Mizuno
,
T.
,
Kikuchi
,
M.
,
Murata
,
T.
,
Takahashi
,
K.
, et al
(
2010
).
Antipsychotics reverse abnormal EEG complexity in drug-naive schizophrenia: A multiscale entropy analysis
.
Neuroimage
,
51
,
173
182
.
Takahashi
,
T.
,
Cho
,
R. Y.
,
Murata
,
T.
,
Mizuno
,
T.
,
Kikuchi
,
M.
,
Mizukami
,
K.
, et al
(
2009
).
Age-related variation in EEG complexity to photic stimulation: A multiscale entropy analysis
.
Clinical Neurophysiology
,
120
,
476
483
.
Vakorin
,
V. A.
,
Lippé
,
S.
, &
McIntosh
,
A. R.
(
2011
).
Variability of brain signals processed locally transforms into higher connectivity with brain development
.
Journal of Neuroscience
,
31
,
6405
6413
.