Visual Arts Creation Assisted by BICASSO: Brain-Inspired Computationally Aesthetic Selective Savant and Observer

Visual Arts Creation Assisted by BICASSO: Brain-Inspired Computationally Aesthetic Selective Savant and Observer

(Short paper)

Keywords: neuroaesthetics, collaboration, reinforcement learning, beauty evaluation, combinatorial creativity.

This paper describes the design of a computer program which will assist artists in producing aesthetically interesting pieces of visual art. In contrast of existing creative drawing computer programs, the proposed software will attempt to simulate the creation and perception of visual art. The program, called BICASSO, will be based on previous neuroaesthetics findings which offer an understanding of what the human brain considers beautiful. The program will include features representing the roles of the brain regions enrolled in visual perception, memory and decision-making. BICASSO will modify visual elements of the visual art being created, depending on aesthetic rules to render visual components easy to process by the human visual system. The aesthetic rules rely on the hypothesis that processing fluency and aesthetic pleasure are linked. The artist will give feedback so the program can learn and modify its behavior when its collaboration is not considered beneficial. As the program will only suggest aesthetic improvements (based on its knowledge, and ratings of the collaboration and the final products), we will be able to observe if this asynchronous collaboration can generate creative products which are considered as more valuable to the human eye.