Communication of emotion via drumming: dual-brain imaging with functional near-infrared spectroscopy

Abstract

Nonverbal communication of emotion is essential to human interaction and relevant to many clinical applications, yet it is an understudied topic in social neuroscience. Drumming is an ancient nonverbal communication modality for expression of emotion that has not been previously investigated in this context. We investigate the neural response to live, natural communication of emotion via drumming using a novel dual-brain neuroimaging paradigm. Hemodynamic signals were acquired using whole-head functional near-infrared spectroscopy (fNIRS). Dyads of 36 subjects participated in two conditions, drumming and talking, alternating between 'sending' (drumming or talking to partner) and 'receiving' (listening to partner) in response to emotionally salient images from the International Affective Picture System. Increased frequency and amplitude of drum strikes was behaviorally correlated with higher arousal and lower valence measures and neurally correlated with temporoparietal junction (TPJ) activation in the listener. Contrast comparisons of drumming greater than talking also revealed neural activity in right TPJ. Together, findings suggest that emotional content communicated by drumming engages right TPJ mechanisms in an emotionally and behaviorally sensitive fashion. Drumming may provide novel, effective clinical approaches for treating social-emotional psychopathology.

Fig. 1

Experimental paradigm. Each run is 3 min, 12 epochs (8 epochs shown here). Subjects alternate ‘sending’ (speaking or drumming) and ‘receiving’ when triggered by image change. Each image was selected from the IAPS library with established arousal and valence ratings.

Fig. 2

Experimental set-up for two interacting partners in the drumming condition. The talking communication condition was identical, but without the drum apparatus.

Fig. 3

Examples of IAPS images with low/high arousal (A) and negative/positive valence (V). The figure illustrates the arousal/valence index system for emotional qualities of each image.

Fig. 4

Right and left hemispheres of a single-rendered brain illustrate average locations (red circles) for channel centroids. See Table S1 (Supplementary Material) for average MNI coordinates and anatomical locations.

Fig. 5

A, A positive correlation (r = 0.37) was observed between the quantified drumming response (number of drum strikes multiplied by average drum strike force) and the arousal ratings of IAPS image stimuli. The bars represent two brackets equally dividing our range of IAPS image stimuli arousal ratings (lowest arousal 2.63 to highest arousal 7.35), P < 0.001. B, A negative correlation (r = −0.22) was observed between the quantified drumming response (number of drum strikes multiplied by average drum strike force) and the valence ratings of IAPS image stimuli. The bars represent two brackets equally dividing our range of IAPS image stimuli valence ratings (lowest valence 2.16 to highest valence 8.34), P < 0.001.

Fig. 6

Convolving strike by strike drumming intensities with the hemodynamic response function for the drumming (‘sending’) block, the listening condition shows greater activity than the drumming condition in two loci (blue), both in the right hemisphere. The first peak voxel was located at 64, −52, 24 (T = −3.49, P < 0.00078, P < 0.05 FDR corrected), and it included SMG (BA40) 49%, STG (BA22) 35% and angular gyrus (BA39) 16%. The second peak voxel was located at 64, −46, 6 (T = −3.84, P < 0.00031, P < 0.05 FDR corrected), and it included STG (BA22) 56% and middle temporal gyrus (BA21) 40%. In contrast, the drumming (‘sending’) condition shows greater activity than the listening (‘receiving’) condition in two loci (red), one in each hemisphere. The right hemisphere peak voxel was located at 60, −16, 42 (T = 3.58, P < 0.00061, P < 0.05 FDR corrected), and it included pre-motor and supplementary motor cortex (BA6) 43% and primary somatosensory cortex (BA 1, 2, 3) 18%, 12%, 17%. The left hemisphere peak voxel was located at −50, −6, 36 (T = 3.11, P < 0.00211), and it included pre-motor and supplementary motor cortex (BA6) 100%.

Fig. 7

Collapsing across qualities of valence and arousal, the drumming condition shows greater activity than the talking condition in two loci, one in each hemisphere, mapped in accordance with the NIRS-SPM atlas (Mazziotta et al., ; Tak et al., 2016). The right hemisphere peak voxel was located at 62, −36, 28 (T = 5.32, P < 0.00001, P < 0.05 FDR corrected), and it included SMG (BA 40) 41%, STG (BA 22) 25% and primary somatosensory cortex (BA 2) 21%. The cluster in the left hemisphere had a peak voxel at −66, −30, 30 (T = 3.78, P = 0.00030, P < 0.05 FDR corrected), spatial distribution including primary somatosensory cortex (BA 2) 42% (BA 2) and SMG 20% (BA 40).