Cross-modal decoding of emotional expressions in fMRI-Cross-session and cross-sample replication

Area of Science:

• Neuroscience
• Cognitive Science
• Social Cognition

Background:

• The embodied simulation theory proposes mirror neurons (MN) create shared representations for action and perception, enabling understanding of others' mental states.
• Multivariate pattern analysis (MVPA) of fMRI data offers high spatial sensitivity to investigate MN properties like cross-modality and action specificity.

Purpose of the Study:

• To investigate the cross-modal and action-specific properties of mirror neurons (MN) in representing emotions.
• To test the embodied simulation theory by examining neural representations of observed versus executed/imitated emotional facial expressions.

Main Methods:

• BOLD-fMRI data were collected during imitation, execution, and observation of fearful or angry facial expressions.
• Support vector machines and searchlight analysis were used to classify emotional content within and across modalities.
• Classification accuracies were evaluated using a permutation scheme, with cross-session and cross-sample replicability assessed.

Main Results:

• Emotional content was classifiable within modalities for execution and imitation, with cross-session and cross-sample replication.
• Cross-modal classification was successful when training on execution and testing on imitation, showing cross-session replication.
• Searchlight analysis identified prefrontal cortex regions with action-specific and cross-modal representations; however, observation modality lacked shared neural representation.

Conclusions:

• Brain regions demonstrate replicable action-specific and cross-modal representations of fear and anger for execution and imitation tasks.
• The absence of a shared neural representation within the observation modality only partially supports the embodied simulation theory.
• Mirror neuron (MN) region activation is less robust and distinguishable during observation compared to motor tasks.