This paper proposes a new model of perception called Perceptual Logic and applies it to the domain of art to understand artistic style. We describe style in terms of affordances, or ways in which an artist can interact with and contribute to an artwork. Different types of Perceptual Logic are found to influence the perceived affordances of an artwork. We present a computational collaborative art program called PERLOGICART that uses a computational model of Perceptual Logic to learn an artist's style through collaboration. The research is conducted using a practice-based method --- we are working on building an interactive tool to support the making of artworks and the understanding of the creative process at the same time. PERLOGICART is a compelling interactive artwork as well as a valuable research tool that records and categorizes the creative process in a systematic manner.