Elicited emotion: effects of inoculation of an art style on emotionally strong images.

Abstract

The objective of this research is to study how the application of the Convolutional Neural Network (CNN) artistic filter can be an alternative to mitigate the emotional response to photographs with strong emotional content published in Internet news. Van Gogh's artistic style was extracted through a CNN and inoculated with 64 IAPS images chosen to cover the entire emotional space. 140 university students of both sexes (70 men and 70 women) with an average age of 22 years, evaluated 128 stimuli, 64 original and 64 digitally inoculated, giving the appearance that they were painted with the artistic style of Van Gogh. For the evaluation of the stimuli, four groups were established under the conditions: 1 high arousal-positive valence, 2 negative valence-low arousal, 3 high arousal-negative valence and 4, low arousal, positive valence. The original images (OI) tended to produce less pleasant effects, while the images inoculated with filters made with CNN provoked reactions with a tendency to calm. The most significant changes in the emotional states are observed in the valence, the stimuli with the inoculation of the artistic style produces alterations with a tendency to pleasant effects. The averages of the comparisons of the dimensions valence and arousal of the OI and the CNN allow to show that there are differences in the emotional states, the results can permit the development of a methodology that, based on the inoculation of the artistic style of original paintings through CNN in emotionally strong images, a new image is created that replaces the strong images published in the Internet news.