Absorbed listening to music often connects visual imagery with other experiential attributes such as emotions, memories, and personal thoughts, and is therefore closely related to questions surrounding the organizational structure of altered consciousness. However, it is still unclear what such a model would look like, nor what the functional role is of visual imagery. The present study used network analysis to determine and visualize the conditional (in)dependencies among key phenomenal dimensions of consciousness when listening to a piece of music, with particular focus on visual imagery during absorption. Based on self-report data from an online study (N = 622), a sparse regularized partial correlation network was computed using the graphical lasso. We then implemented a Bayesian approach to obtain a directed acyclic graph as a model for connecting the dimensions of consciousness, and to examine the scenario of absorption. Findings suggested that dissociation, positive emotions, and internaldirected attention predicted visual imagery, which in turn determined mixed emotions and, counterintuitively, short-and long-term aspects of memory. Being moderately central in the regularized partial correlation network and acting as one of the driving forces in the Bayesian network, visual imagery was found to be an important intermediary attribute of consciousness. This study highlights the utility of a network approach in rethinking the conceptualization of subjective experiential phenomena like music absorption and aesthetic experiences, and the functional role of visual imagery therein.