Modeling Emotional Arousal With Turbulence Measured by EEG

Abstract

Turbulence-like dynamics in brain activity have been proposed as a signature of systems operating near criticality, and may reflect changes in neuronal function associated with emotional states. In this paper, we hypothesize that motor behavior linked to emotional expression modulates turbulence, reflecting a shift towards more streamlined brain dynamics characteristic of emotional motor control. We assessed EEG turbulence in 30 healthy participants in a motor paradigm varying in both task demand and degree of emotionality. Conditions included singing, swaying, responding to a virtual conductor of variable expressivity, having your own body movements mirrored by a virtual agent, and combinations thereof. Results showed an inverse relation of turbulence intensity in the alpha range to both degree of movement and perceived level of task emotionality, which was also true for the high gamma range, but to a lesser extent. When factoring in task demand, the effect of level of emotionality in the alpha range deteriorated. This is physiological evidence for why physical arousal is likely to increase the level of perceived emotional engagement or even be misinterpreted as such. Our findings suggest high gamma activity is a more accurate indicator of emotionality during motor tasks and can be key to differentiating EEG signatures of emotional motor control, which has been shown to be partly autonomous from voluntary motor control.

Keywords: criticality; emotional motor control; functional data; immersive VR; mind–brain–body; multisensory processing; naturalistic neuroscience.

FIGURE 1

Experimental paradigm. Participants engaged in a set of conditions combining singing, swaying, and interacting with three virtual agents conducting music (robotically, expressively and mirroring participants' movements), while wearing a mocap suit, VR headset, and EEG cap. The experiment was organized in two blocks of randomized conditions, without and with VA. A short familiarization task with the VA took place before the VA block.

FIGURE 2

Spatiotemporal functional data model of EEG dynamics based on Pfaff (2009) generalized arousal (dominant dynamics), used to characterize turbulence‐like patterns in brain activity.

FIGURE 3

Schematic of the pre‐processing pipeline. Visual inspection steps are marked with an eye icon.

FIGURE 4

Behavioral data: Percentage distributions of the participants' emotional engagement and VA interaction ratings (1 to 5) across task conditions with accompanying graphs of the rated items directed with colors towards the condition with higher mean rates with the indication of the test significance and effect sizes.

FIGURE 5

Turbulence analysis of EEG signal. (A) Boxplots of turbulence intensity across conditions in alpha (8–13 Hz) and high gamma (50–80 Hz). The centre line of each boxplot represents the data median and the bounds of the box show the interquartile range. The whiskers represent the bottom 25% and top 25% of the data range. The baseline condition is NM.NS.NA, only listening to the music. (B) 3D scatterplots showing interactions between the variable emotional engagement, turbulence intensity and movement velocity rates in a log‐scale.

FIGURE 6

Spatial patterns in alpha band during the different motor conditions. Colors represent the absolute value of the participant‐average of the turbulence scores after application of spherical spline surface Laplacian (Carvalhaes and de Barros 2015). This technique is used to mitigate artifactual volume conduction effects and localize interesting cortical areas involved in the cognitive task under examination. High turbulence levels are observed in the sensorimotor area, particularly in central and centro‐left lateral regions, during conditions involving movement and/or the visualization of movement. The occipito‐parietal region shows a reduction in turbulence as the perceived level of emotional engagement (EE) and physical task demand increases. Percentages represent the mean value of the EE ratings.