Figure 1.
Experimental and simulation setups.
(A) Apparatus and experimental setting of the double video system and dual-EEG recording [27]. (B) Right and Top views of the pair of virtual brains. Each weighted network represents the 90 brain regions and their average anatomical connectivity. Arrows indicate the directed coupling from the motor to visual.
Figure 2.
Real and simulated functional data.
PLV matrices for real (A) and simulated (B) data and related histogram (C). h-PLV matrix for real (D) and simulated (E) data and related histogram (F). PLV and h-PLV are computed for the gamma band and averaged across either the 9 pairs of real subjects during resting state condition or 9 pairs of simulated subjects with Cintra = 0.49 and Cinter = 0. It can be seen from this example that PLV and h-PLV exhibit different distributions. Notice that the difference of the dynamics between the partners gives an asymmetry in the h-PLV matrix.
Figure 3.
Procedure Flowchart illustrating the different steps of the simulations and their comparisons with the real EEG data.
Figure 4.
Example of simulation with variation of the Cintra control parameter over time.
(A) Timecourses of all ROIs instantaneous frequency. (B) Related timecourse of the Cintra parameter.
Figure 5.
Influence of the global anatomical connectivity strength on the model’s dynamics.
(A) Average order parameter across the 90 ROIs. Influence of the global anatomical connectivity strength at the ROIs level. Results are averaged across 18 simulations with Cinter = 0. Areas stand for the standard error. Blue: real connectivity. Green: identic shuffled connectivity for the two virtual brains in same dyads. Red: different shuffled connectivity for all virtual brains. (B) Average PLV in the gamma band between all the ROIs inside each virtual brain. (C) Example of simulated EEG signals. (D) Power spectrum for each EEG signals of C.
Figure 6.
Influence of the global anatomical connectivity strength on intra-brain synchronization.
Average PLV across all pairs of electrodes inside each simulated subject helmet for the gamma (A) and alpha (B) frequency bands. The decrease of PLV after Cintra = 0.7 for the alpha band seems caused by fluctuations of the mean low-frequency rhythm peak at strong anatomical coupling.
Figure 7.
Mahalanobis distances between simulated and real resting state data based on PLV matrices in the gamma band.
Figure 8.
Influence of the anatomical connectivity on inter-brain synchronization.
(A) Average response of the artificial sensorimotor coupling strength on h-PLV across the best fitting area (Cintra between 0.5 and 0.6). (B) Effect of the anatomical topology and similarity on the normalized h-PLV for Cinter = 0.01. Each point is computed for a normalized linear combination of the three cases: same real anatomy, same shuffled version and different shuffled versions.