Mu rhythm suppression reflects mother-child face-to-face interactions: a pilot study with simultaneous MEG recording

Spontaneous face-to-face interactions between mothers and their children play crucial roles in the development of social minds; however, these inter-brain dynamics are still unclear. In this pilot study, we measured MEG mu suppression during face-to-face spontaneous non-linguistic interactions between mothers and their children with autism spectrum disorder (ASD) using the MEG hyperscanning system (i.e., simultaneous recording). The results demonstrated significant correlations between the index of mu suppression (IMS) in the right precentral area and the traits (or severity) of ASD in 13 mothers and 8 children (MEG data from 5 of the children could not be obtained due to motion noise). In addition, higher IMS values (i.e., strong mu suppression) in mothers were associated with higher IMS values in their children. To evaluate the behavioral contingency between mothers and their children, we calculated cross correlations between the magnitude of the mother and child head-motion during MEG recordings. As a result, in mothers whose head motions tended to follow her child’s head motion, the magnitudes of mu suppression in the mother’s precentral area were large. Further studies with larger sample sizes, including typically developing children, are necessary to generalize this result to typical interactions between mothers and their children.


2
Legend for supplementary movie S1. 1 2 Movie S1. Head-motion-graphy (HMG). 3 The magnitudes of the head motions were quantified using motion artifacts in the MEG 4 signals. As shown in this movie, HMG (right movie) practically reflected the head 5 motions (left movie).    Table S1. Spearman's rank correlation coefficients (ρ) between latency/intensity in 2 positive peak (within 2sec) with index of mu suppression in precentral area in mothers 3 and her children (n = 8). Brain anatomical estimation for children 19 We could not obtain individual brain structural data because it is difficult for young 20 children to perform MRI recordings without sedation. To superimpose the coordinate 21 system of the MEG on the collected anatomical information, we estimated the brain 22 structures from the individual head surface shapes in young children using the following 23 methods, which are modified versions of our previous estimation algorithm 1 . Our 24 algorithm was developed to find an optimal structural image from the 98 brain examples 25 using the head surface points of a child.

26
The estimation of the brain structure consisted of the following three steps.

27
(1) We prepared a database of T1-weighted MR images from 98 children (age range: 0 -28 8 years), which were regarded as templates of head surfaces and cortical structure for 29 Japanese children. Then, for each of the above 98 template images, five fiduciary points 30 (right preauricular, left preauricular, nasion, vertex, inion) on the head surface were 31 determined.

32
(2) The root mean square error (RMSE) was calculated using the distance between the 33 corresponding surface points of the child participants and a template. The RMSE was 34 defined using the following formula: where XC are the coordinates of the child participant, XO are the coordinates of the 2 template, and N is the number of surface points.

3
(3) The template with the lowest RMSEs of all templates was selected as an optimal 4 brain template for the child participant. fT in all channels were automatically rejected to exclude data with motion noise or other 12 noise contamination. In the children, epochs with a magnetic amplitude greater than 13 4000 fT were automatically rejected. Because of this preprocessing, we excluded the 14 MEG data of 5 children due to excessive motion noise and excluded the MEG data of 15 one mother due to excessive magnetic noise resulting from the presence of dental metals.

16
In these excluded subjects, we could not obtain a sufficient period of noise-free MEG 17 data (i.e., we captured less than a 50 sec period).

19
Source estimation 20 We performed signal source estimation of the mu rhythm using the individual 21 anatomy of the mothers and using the individually estimated anatomy of their children. 22 We focused on the upper mu band (10-12 Hz frequency band) in this study. Source   The following procedure was common in mothers and children: the lead field was then 9 computed using the overlapping spheres algorithm 8 with a cortical surface tessellated 10 with 15000 vertices. The inverse solution was calculated for each individual using 11 Tikhonov-regularized minimum-norm estimates 9 . A noise covariance matrix was 12 calculated from MEG recordings from resting states (i.e., DVD condition) to estimate 13 the noise level. Then, a weighted minimum-norm estimation with source orientation 14 constraints was chosen to compute the source activity for mu rhythms. We used 15 Desikan-Killiany atlases to estimate the region of interest 10 .

17
Analysis of the behavioral contingency between mother and child 18 We analyzed the behavioral contingency between mother and child during face-to-face 19 interactions using MEG data from the whole period of the "Live" condition, including 20 motion noise. For correlation analysis with the index of mu suppression, the MEG data 21 from eight mother-child pairs were analyzed. To quantify the mutual face-to-face 22 interactions between mother and child, we calculated the correlation coefficient 23 sequence between the magnitudes in the mother and child head motions (e.g., nod, nod direction of the nod (e.g., the up and down direction). We addressed the processed time 32 series data as head-motion-graphy (HMG) in the present study. As shown in 33 supplementary movie S1, this HMG practically reflects head motions (e.g., nod, nod no,  : 8 -14). Fifth, the cross-correlation analysis between the time series of the 3 HMGs in the mothers and children during the "Live" condition was completed for each 4 15 sec segment. Sixth, the time series of cross correlation coefficients calculated using 5 these segments were averaged for each pair. Finally, we defined a peak value for the 6 correlation coefficient within a 2 sec time window before and after time 0 as a strength 7 of contingency, and we defined the latency of the peak as the direction of contingency 8 (i.e., a positive value indicates that the mother is the leader, and a negative value 9 indicates that the mother is the follower in terms of head movements during the