Common network for the processing of dynamic emotional bodies contains information to discriminate individual basic emotions

Mechanisms underlying the processing of basic emotional expressions have been proposed to reside in discrete gross anatomical locations. Recent meta-analytical studies, however, provide conflicting results with some in favor (Vytal & Hamann 2010) and others disagreeing (Lindquist et al. 2012) with this proposition. Here, we used dynamic stimuli of emotionally expressive gaits to address the question whether common or distinct networks subserve processing of different emotions displayed by human bodies. Furthermore, instead of relying on general activation differences between emotions, we investigated which regions contain information to reliably discriminate between different emotions. Stimuli were generated from a motion capture database of actors performing neutral and emotionally expressive (anger, happiness, sadness, fear) gaits. A custom-built volumetric avatar model was used for animation. This procedure has been shown to result in highly recognizable emotional displays (Roether et al. 2009). Six stimuli of each of the four emotions and the neutral gaits were selected for the event-related fMRI experiment, resulting in 30 stimuli. These were presented twice within one run (8 & 5 vis. degrees respectively) for two consecutive gait cycles in a pseudo-random order. To identify regions commonly activated by all emotions, we contrasted each emotion individually with the neutral condition and selected voxels significantly activated in all 4 contrasts. These were located in the posterior STS, the temporal poles, posterior, middle and anterior cingulate cortex, the left parahippocampal gyrus and the medial orbitofrontal cortex. Investigating the information present in these ROIs, SVM classification showed above chance performance for at least 3 out of the 4 emotions in the left posterior STS, left parahippocampal gyrus, medial orbitofrontal cortex and middle cingulate cortex. Contrasting each emotion separately with the three others did not yield any significant voxels common to all 3 contrasts. Furthermore, information about the specific emotion was not limited to the emotion network, but was also present in the posterior MTG, left inferior frontal sulcus, and the fusiform gyrus; all part of the action observation network. We did not obtain evidence for discriminable brain correlates of individual basic emotions. Nevertheless, several regions commonly activated for all emotions also contained information about the specific emotion displayed. In our ongoing work, including patients with neurodegenerative diseases, we intend to identify the contribution of the localized emotion network for the processing of emotional body expressions.status: publishe