With the proliferation of consumer virtual reality (VR) headsets and creative tools, content creators are experimenting with new forms of interactive audience experience using immersive media. Understanding user attention and behaviours in virtual environment can greatly inform the creative processes in VR. We developed an abstract VR painting and an experimentation system to study audience art encounters through eye gaze and movement tracking. The data from a user experiment with 35 participants reveal a range of user activity patterns in art exploration. Deep learning models are used to study the connections between the behavioural data and the audience’s background. The work also introduced new integrated methods to visualise user attention for content creators.