Data-driven analysis of facial thermal responses to an emotional movie reveals consistent stimulus-locked physiological changes

Facial infra-red imaging (IRI) is a contact-free technique complimenting the traditional psychophysiological measures to characterize physiological profile. However, its full potential in affective research is arguably unmet due to the analytical challenges it poses. Here we acquired facial IRI data, facial expressions and traditional physiological recordings (heart rate and skin conductance) from healthy human subjects whilst they viewed a 20-minute-long unedited emotional movie. We present a novel application of motion correction and the results of spatial independent component analysis of the thermal data. Three distinct spatial components are recovered associated with the nose, the cheeks and a respiratory component. We first benchmark this methodology against a traditional region-of-interest based technique. We then show significant correlation of all the physiological responses across subjects, including the thermal signals, suggesting common dynamic shifts in emotional state induced by the movie. Finally, we show that thermal responses were significantly anti-correlated with the positive emotional content of the movie thus an index of emotionally-driven physiological response. In sum, this study introduces an innovative approach to analyse facial IRI data and highlights the potential of thermal imaging to robustly capture emotion-related changes in ecological contexts.