Decoding brain basis of laughter and crying in natural scenes

Abstract

AbstractLaughter and crying are universal signals of prosociality and distress, respectively. Here we investigated the functional brain basis of perceiving laughter and crying using naturalistic functional magnetic resonance imaging (fMRI) approach. We measured haemodynamic brain activity evoked by laughter and crying in three experiments with 100 subjects in each. The subjects i) viewed a 20-minute medley of short video clips, and ii) 30 minutes of a full-length feature film, and iii) listened to 15 minutes of a radio play that all contained bursts of laughter and crying. Intensity of laughing and crying in the videos and radio play was annotated by independent observes, and the resulting time series were used to predict hemodynamic activity to laughter and crying episodes. Multivariate pattern analysis (MVPA) was used to test for regional selectivity in laughter and crying evoked activations. Laughter induced widespread activity in ventral visual cortex and superior and middle temporal and motor cortices. Crying activated thalamus, cingulate cortex along the anterior-posterior axis, insula and orbitofrontal cortex. Both laughter and crying could be decoded accurately (66-77% depending on the experiment) from the BOLD signal, and the voxels contributing most significantly to classification were in superior temporal cortex. These results suggest that perceiving laughter and crying engage distinct neural networks, whose activity suppresses each other to manage appropriate behavioral responses to others’ bonding and distress signals.Significance statementLaughter and crying are universal signals of prosociality and distress, respectively. They occur in complex, dynamic social settings with variable and dynamically evolving time courses. Here we used functional magnetic resonance imaging experiments and statistical pattern recognition for disentangling the neural systems that encode laughter and crying signals from dynamic and highly naturalistic scenes. These results show that separable neural circuits are engaged in processing distinct types of social attachment cues, and that pattern recognition during dynamic scene perception allows reliable separation of laughter and crying evoked activation patterns. Coordinated activity of these networks allows managing appropriate behavioral responses to others’ bonding and distress signals.