[Paper Review] Somatosensory and motor contributions to emotion representation

Marianne C. Reddan, Luke J. Chang+2 more|arXiv (Cornell University)|Nov 13, 2024

Action Observation and SynchronizationPsychology3 citations

TL;DR

This fMRI study investigates whether emotional experiences are grounded in somatosensory and motor representations by exposing 21 participants to emotionally arousing images while measuring neural activity and body-sensation maps. Using representational similarity analysis (RSA), the study identifies a distributed brain network—including primary somatosensory and motor cortices, insula, precuneus, and medial prefrontal cortex—that encodes the spatial and emotional geometry of bodily sensations, supporting the embodied emotion theory over purely cognitive or visual explanations.

ABSTRACT

Emotion is often described as something people 'feel' in their bodies. Embodied emotion theorists propose that this connection is not purely linguistic; perceiving an emotion may require somatosensory and motor re-experiencing. However, it remains unclear whether self-reports of emotion-related bodily sensations (i.e., 'lump in my throat') are related to neural simulations of bodily action and sensation or whether they can be explained by cognitive appraisals or the visual features of socioemotional signals. To investigate this, participants (N = 21) were shown arousing emotional images that varied in valence, complexity, and content while undergoing fMRI scans. Participants then rated the images on a set of emotion appraisal scales and indicated where, on a body map, they experienced sensation in response to the image. To derive normative models of responses on these scales, a separate larger online sample online (N = 56 - 128) also rated these images. Representational similarity analysis (RSA) was used to compare the emotional content in the body maps with appraisals and visual features. A pairwise distance matrix between the body maps generated for each stimulus was then used in a whole brain voxel-wise searchlight analysis to identify brain regions which reflect the representational geometry of embodied emotion. This analysis revealed a network including bilateral primary somatosensory and motor cortices, precuneus, insula, and medial prefrontal cortex. The results of this study suggest that the relationship between emotion and the body is not purely conceptual: It is supported by sensorimotor cortical activations.

Motivation & Objective

• To test whether self-reported bodily sensations during emotion are linked to neural simulations of sensorimotor processes rather than cognitive appraisals or visual features.
• To determine if the spatial distribution of bodily sensations in response to emotional stimuli reflects a neural representation in the brain.
• To isolate the contribution of sensorimotor cortical activity to emotion representation, independent of higher-level cognitive or perceptual factors.
• To develop normative models of emotion appraisal and bodily sensation using a large online sample (N = 56–128) to inform fMRI analysis.
• To identify brain regions whose representational geometry matches the structure of body maps of emotional sensation.

Proposed method

• Participants underwent fMRI while viewing emotionally arousing images varying in valence, complexity, and content.
• Participants rated each image on emotion appraisal scales and marked body maps indicating where they felt sensations.
• A separate online sample (N = 56–128) provided normative ratings for appraisal scales and body maps to standardize responses.
• Representational similarity analysis (RSA) compared the similarity structure of body maps with those of visual features and appraisals.
• Whole-brain voxel-wise searchlight analysis identified brain regions whose activity patterns reflected the representational geometry of body maps.
• Pairwise distance matrices of body maps were used as a neural representational model in RSA to test for correspondence with fMRI activity patterns.

Experimental results

Research questions

• RQ1 Do self-reported bodily sensations during emotion correspond to neural representations in sensorimotor cortices?
• RQ2 Is the spatial pattern of bodily sensation during emotion uniquely represented in the brain beyond visual or cognitive appraisals?
• RQ3 Which brain regions show representational geometry that matches the structure of emotion-related body maps?
• RQ4 To what extent can emotional body sensations be explained by visual features of stimuli or cognitive appraisals?
• RQ5 Is there a distributed neural network that encodes the spatial and emotional structure of bodily sensations in emotion?

Key findings

• A distributed brain network, including bilateral primary somatosensory and motor cortices, insula, precuneus, and medial prefrontal cortex, showed representational geometry matching the structure of body maps of emotional sensation.
• The neural representation of bodily sensations during emotion was not explained by visual features of the images or by cognitive appraisal scales.
• The searchlight analysis revealed significant representational similarity between body map patterns and fMRI activity in sensorimotor and parietal regions.
• The study’s normative models from the online sample (N = 56–128) provided reliable benchmarks for emotion appraisal and body sensation ratings.
• The results support the embodied emotion theory by demonstrating that sensorimotor cortical areas are involved in representing emotion through bodily states.
• The findings indicate that emotion is not merely conceptual but is grounded in neural simulations of bodily sensation and action.