Graph-theoretical exploration of MEG hyperscanning data in the mirror game

Loading...
Thumbnail Image
Journal Title
Journal ISSN
Volume Title
Perustieteiden korkeakoulu | Master's thesis
Date
2024-05-20
Department
Major/Subject
SCI3601
Mcode
SCI3601
Degree programme
Master’s Programme in Life Science Technologies
Language
en
Pages
71+8
Series
Abstract
Social interactions are a core element of our everyday lives. The brain has specialized circuits for social interaction that can be studied with brain connectivity. We investigated the neural signatures underlying mirror game interactions using magnetoencephalography (MEG) hyperscanning recordings. The mirror game is a movement improvisation task, where individuals mirror each other, with or without a designated leader. We used data from 10 pairs of subjects performing a 1D finger-movement mirror game while their MEG was simultaneously recorded. They performed the task in interactive (leader, follower, joint improvisation) and non-interactive (moving in isolation) conditions. We used source-level weighted-phase-lag-index functional connectivity within and across interaction brains. We created binary and weighted graphs for the theta [4--7 Hz], alpha [8--13 Hz], and beta [14--25 Hz] frequency bands, with densities of 5, 10, and 20 percent, for both intra- and inter-brain networks. We computed network centrality metrics including degree and strength. In addition, we used the network-based statistic (NBS) to identify interaction and task-relevant subnetworks within a range of component-forming t-thresholds. We observed, in the 5 percent density alpha-band intra-brain graphs, a higher degree in the right supramarginal regions for interactive, compared to non-interactive conditions. We also found a significant increase in strength for the non-interactive condition in a set of distributed brain regions, potentially influenced by signal-to-noise differences between conditions. The NBS-based contrast between interactive and non-interactive conditions revealed significantly different subnetworks, including areas associated with mentalizing and the mirror neuron systems. Importantly, our results showed a threshold-dependent variability. Further work is needed to determine the optimal threshold selection approach when applying network centrality metrics and NBS. Overall, the graph-theoretical approach is a promising tool for summarizing the complexity of MEG hyperscanning data.
Description
Supervisor
Parkkonen, Lauri
Thesis advisor
Avendano, Juan
Korhonen, Onerva
Keywords
hyperscanning, graph theory, network-based statistic, MEG, centrality, functional connectivity
Other note
Citation