Automatic recognition of the unconscious reactions from physiological signals

While the research in affective computing has been exclusively dealing with the recognition of explicit affective and cognitive states, carefully designed psychological and neuroimaging studies indicated that a considerable part of human experiences is tied to a deeper level of a psyche and not available for conscious awareness. Nevertheless, the unconscious processes of the mind greatly influence individuals' feelings and shape their behaviors. This paper presents an approach for automatic recognition of the unconscious experiences from physiological data. In our study we focused on primary or archetypal unconscious experiences. The subjects were stimulated with the film clips corresponding to 8 archetypal experiences. Their physiological signals including cardiovascular, electrodermal, respiratory activities, and skin temperature were monitored. The statistical analysis indicated that the induced experiences could be differentiated based on the physiological activations. Finally, a prediction model, which recognized the induced states with an accuracy of 79.5%, was constructed.