Abstract
When working with generative systems, designers enter into a loop of discrete steps; external evaluations of the output feedback into the system, and new outputs are subsequently reevaluated. In such systems, interacting low level elements can engender a difficult to predict emergence of macro-level characteristics. Furthermore, the state space of some systems can be vast. Consequently, designers generally rely on trial-and-error, experience or intuition in selecting parameter values to develop the aesthetic aspects of their designs. We investigate an alternative means of exploring the state spaces of generative visual systems by using a gaze- contingent display. A user's gaze continuously controls and directs an evolution of visual forms and patterns on screen. As time progresses and the viewer and system remain coupled in this evolution, a population of generative artefacts tends towards an area of their state space that is 'of interest', as defined by the eye tracking data. The evaluation-feedback loop is continuous and uninterrupted, gaze the guiding feedback mechanism in the exploration of state space.