Shifts in broadband power and alpha peak frequency observed during long-term isolation

Prolonged periods of social isolation and spatial confinement do not only represent an issue that needs to be faced by a few astronauts during space missions, but can affect all of us as recently shown during pandemic situations. The fundamental question, how the brain adapts to periods of sensory deprivation and re-adapts to normality, has only received little attention. Here, we use eyes closed and eyes open resting-state electroencephalographic (EEG) recordings to investigate how neural activity is altered during 120 days of isolation in a spatially confined, space-analogue environment. After disentangling oscillatory patterns from 1/f activity, we show that isolation leads to a reduction in broadband power and a flattening of the 1/f spectral slope. Beyond that, we observed a reduction in alpha peak frequency during isolation, but did not find strong evidence for isolation-induced changes that are of oscillatory nature. Critically, all effects reversed upon release from isolation. These findings suggest that isolation and concomitant sensory deprivation lead to an enhanced cortical deactivation which might be explained by a reduction in the mean neuronal population firing rate.

: Complete post-hoc comparisons for power offset eyes closed condition (clusterbased permutation tests, Monte-Carlo method, 1000 iterations)

Spectral slope eyes open
= Decrease in spectral slope = Increase in spectral slope Pre Iso. = Pre-Isolation (-13 days before start of isolation period), T2 = +15 days, T3 = +54 days, T4 = +79 days, T5 = + 110 days, Post Iso. = Post-Isolation (+ 7 days after end of isolation period) ------    Single-subject change in spectral offset of the aperiodic signal from isolation to post-isolation during eyes closed condition. Histograms show the probability distribution of the spectral offset on a trial basis. Fitting of the aperiodic signal was performed on a trial basis using IRASA. The red histogram reflects the probability distribution of the spectral offset during isolation. The green histogram reflects the probability distribution of the spectral offset after exposure to isolation (post-isolation). The topographical t-value distribution (obtained through random trial shuffling between the two conditions, see methods for details on single-subject cluster permutation analysis) obtained via cluster permutation is plotted for each condition difference (pairwise t-tests). Black dots represent significant sensors that are part of a negative cluster (here: spectral offset higher during isolation as compared to post-isolation). Gray dots represent significant sensors that are part of a positive cluster (here: spectral offset lower during isolation as compared to post-isolation). Individual subjects are denoted as Subject A, B, C, D, E, F. Note the consistent global reduction (5/6 subjects show a significant decrease; subject D shows both increases and decreases in spectral offset which are clustered separately approximately along the two hemispheres) in spectral power offset from isolation to post-isolation. Legend: *** p < 0.001, ** p < 0.01 Single-subject change in spectral slope of the aperiodic signal from pre-isolation to isolation during eyes closed condition. Histograms show the probability distribution of the spectral slope on a trial basis. Fitting of the aperiodic signal was performed on a trial basis using IRASA. The gray histogram reflects the probability distribution of the spectral slope prior to isolation. The red histogram reflects the probability distribution of the spectral slope during isolation. The topographical t-value distribution (obtained through random trial shuffling between the two conditions, see methods for details on single-subject cluster permutation analysis) obtained via cluster permutation is plotted for each condition difference (pairwise t-tests). Black dots represent significant sensors that are part of a negative cluster (here: spectral slope steeper during isolation as compared to pre-isolation). Gray dots represent significant sensors that are part of a positive cluster (here: spectral slope flatter during isolation as compared to pre-isolation). Individual subjects are denoted as Subject A, B, C, D, E, F. Note that 4/6 subjects show an increase in spectral slope from pre-isolation to isolation whereas 2/6 subjects show the opposite pattern. Legend: *** p < 0.001 Single-subject change in spectral slope of the aperiodic signal from isolation to post-isolation during eyes closed condition. Histograms show the probability distribution of the spectral slope on a trial basis. Fitting of the aperiodic signal was performed on a trial basis using IRASA. The red histogram reflects the probability distribution of the spectral slope during isolation. The green histogram reflects the probability distribution of the spectral slope post-isolation. The topographical t-value distribution (obtained through random trial shuffling between the two conditions, see methods for details on single-subject cluster permutation analysis) obtained via cluster permutation is plotted for each condition difference (pairwise t-tests). Black dots represent significant sensors that are part of a negative cluster (here: spectral slope flatter during isolation as compared to post-isolation). Gray dots represent significant sensors that are part of a positive cluster (here: spectral slope steeper during isolation as compared to post-isolation). Individual subjects are denoted as Subject A, B, C, D, E, F. Note that 5/6 subjects show a decrease in spectral slope from isolation to post-isolation whereas 1 subject shows the opposite pattern. Legend: *** p < 0.001 Single-subject change in spectral offset of the aperiodic signal from pre-isolation to isolation during eyes open condition. Histograms show the probability distribution of the spectral offset on a trial basis. Fitting of the aperiodic signal was performed on a trial basis using IRASA. The gray histogram reflects the probability distribution of the spectral offset prior to isolation. The red histogram reflects the probability distribution of the spectral offset during exposure to isolation. The topographical t-value distribution (obtained through random trial shuffling between the two conditions, see methods for details on single-subject cluster permutation analysis) obtained via cluster permutation is plotted for each condition difference (pairwise ttests). Black dots represent significant sensors that are part of a negative cluster (here: spectral offset higher prior to isolation as compared to within isolation). Gray dots represent significant sensors that are part of a positive cluster (here: spectral offset lower prior to isolation as compared to within isolation). Individual subjects are denoted as Subject A, B, C, D, E, F. Note the consistent global reduction (6/6 subjects show a significant decrease) in spectral power offset from pre-isolation to isolation. Legend: *** p < 0.001, ** p < 0.01 Single-subject change in spectral offset of the aperiodic signal from isolation to post-isolation during eyes open condition. Histograms show the probability distribution of the spectral offset on a trial basis. Fitting of the aperiodic signal was performed on a trial basis using IRASA. The red histogram reflects the probability distribution of the spectral offset during isolation. The green histogram reflects the probability distribution of the spectral offset post-isolation. The topographical t-value distribution (obtained through random trial shuffling between the two conditions, see methods for details on single-subject cluster permutation analysis) obtained via cluster permutation is plotted for each condition difference (pairwise t-tests). Black dots represent significant sensors that are part of a negative cluster (here: spectral offset higher during isolation as compared to post-isolation). Gray dots represent significant sensors that are part of a positive cluster (here: spectral offset lower during isolation as compared to postisolation). Individual subjects are denoted as Subject A, B, C, D, E, F. Note that 5/6 subjects show an increase in spectral offset from isolation to post-isolation. Legend: *** p < 0.001 Single-subject change in spectral slope of the aperiodic signal from pre-isolation to isolation during eyes open condition. Histograms show the probability distribution of the spectral slope on a trial basis. Fitting of the aperiodic signal was performed on a trial basis using IRASA. The gray histogram reflects the probability distribution of the spectral slope prior to isolation. The red histogram reflects the probability distribution of the spectral slope during isolation. The topographical t-value distribution (obtained through random trial shuffling between the two conditions, see methods for details on single-subject cluster permutation analysis) obtained via cluster permutation is plotted for each condition difference (pairwise t-tests). Black dots represent significant sensors that are part of a negative cluster (here: spectral slope steeper during isolation as compared to pre-isolation). Gray dots represent significant sensors that are part of a positive cluster (here: spectral slope flatter during isolation as compared to preisolation). Individual subjects are denoted as Subject A, B, C, D, E, F. Note that 4/6 subjects show an increase in spectral slope from pre-isolation to isolation whereas 2/6 subjects show the opposite pattern. Legend: *** p < 0.001, ** p < 0.01 Single-subject change in spectral slope of the aperiodic signal from isolation to post-isolation during eyes open condition. Histograms show the probability distribution of the spectral slope on a trial basis. Fitting of the aperiodic signal was performed on a trial basis using IRASA. The red histogram reflects the probability distribution of the spectral slope during isolation. The green histogram reflects the probability distribution of the spectral slope post-isolation. The topographical t-value distribution (obtained through random trial shuffling between the two conditions, see methods for details on single-subject cluster permutation analysis) obtained via cluster permutation is plotted for each condition difference (pairwise t-tests). Black dots represent significant sensors that are part of a negative cluster (here: spectral slope flatter during isolation as compared to post-isolation). Individual subjects are denoted as Subject A, B, C, D, E, F. Note that 6/6 subjects show a decrease in spectral slope from isolation to postisolation. Legend: *** p < 0.001 Single-subject change in Alpha Peak Frequency (APF) after removal of the aperiodic signal using IRASA. Bar plots show the mean APF during isolation (red) and post-isolation (green). Error bars represent the standard deviation over trials. The topographical t-value distribution (given by random trial shuffling between the two conditions, see methods for details on singlesubject cluster permutation analysis) obtained via cluster permutation is plotted for each condition difference. Black dots represent significant sensors that are part of a negative cluster (here: APF reduced during isolation as compared to pre-isolation). Gray dots represent significant sensors that are part of a positive cluster (here: APF higher during isolation as compared to pre-isolation). Individual subjects are denoted as Subject A, B, C, D, E, F. Note that APF decreased in 3/6 subjects during isolation whereas 1 subject showed the opposite pattern and in 2 subject APF was not altered. Legend: *** p < 0.001, ns = not significant ns.

*** ***
Single-subject change in peak alpha frequency eyes closed (Isolation -Pre-Isolation) Due to complexity in the space-analogue environments and limited time capacity, it is not always possible to measure at the same time of the day. To exclude possible confounds induced by the daytime at which the measurement was taken, we performed several control analyses.

S14.1)
We binned the daytime into 5 bins between 8am to 6pm (8 -10 am, 10 -12noon, 12 -2 pm, 2 -4 pm, 4 -6 pm) and adjusted the data accordingly to perform spearman correlations to examine whether there is a relationship between daytime and our dependent variables. We did not observe any significant correlation in any of our parameters.

S14.2)
(A) Displays how the different parameters are modulated by daytime. Note that, except for the higher spectral offsets for measurements at 12pm during eyes closed and at 10am and 4pm during eyes open, no modulation by daytime is visible. (B) The colormap displays the amount of measurements that have been taken at a respective daytime and isolation timepoints. Note that most of the measurements have been taken between 10 -12 noon. (C) Displays the daytime at which EEG has been recorded for each participant. Note that there is no systematic pattern which is likely to explain the consistent (across the majority of subjects as denoted in the single subject analyses) shift over time that we observed. (D) Displays the number of EEG recordings that have been taken at a certain timepoint.
Note that the majority of measurements has been taken between 10 -12 noon.