Looking for Emotions in Evolutionary Art

Abstract

Although evolutionary art has demonstrated its potential over the past two decades, based primarily on interactive versions of the evolutionary algorithm (IEAs), new implementations of the algorithm continue to be developed and applied both to produce works of art and to study computer-added creative processes. Recently, human artists have been asked to participate more directly in every step included in the algorithm so that we learn from them. Nevertheless, the behavior shown is so complex that retrieving useful information that allows to understand better the creative process they develop is not an easy task. Moreover, emotions have not been analysed previously in this context. Although useful information has already been obtained and conclusions reached regarding the future improvement of EAs devoted to art, the role of emotions as conveyed by artists and understood by the audience is still unexplored. This paper paves the way towards that specific goal, that we summarize as follows: (i) applying harsh constraints on the creative processes so that the number of formal elements to be analyzed is notably reduced, thus allowing an easier analysis and at the same time to draw useful conclusions when evolutionary algorithms are applied to art and creativity; (ii) being able to analyze both, the human emotions shown by artists and those felt by the audience; together with their connections with formal elements present in the artwork. The main goal is to derive new measures that may allow in the future to introduce changes to the fitness functions that allow the artificial evolution of artwork conveying emotions. Finally, the quality of the artwork produced is formally tested by submitting it to an international art competition.