Dreaming as fascinated predictions: bridging Sartre’s phenomenology and predictive processing

Abstract Jean-Paul Sartre develops a phenomenological account of dreaming in his early work on the imaginary. Providing a brief but rich analysis, he emphasizes the active and imaginative nature of consciousness during dreaming and distinguished it sharply from perception. This paper revisits Sartre’s dream theory in light of the neuroscientific and philosophical framework of Predictive Processing (PP). PP posits that the brain engages in continuous hypothesis testing, predicting external and internal states over hierarchically organized neural levels. Under PP, dreaming is conceptualized as resulting from an active process of the predictive brain, constrained by the physiological conditions of REM sleep and largely disconnected from external inputs. While PP generally suggests a strong overlap between imagination and perception, Sartre treats them as distinct, sui generis mental phenomena. This implies a conflict seemingly rendering the two accounts of dreaming inconsistent. We argue, however, that after resolving this conflict across different levels of analysis—sub-personal mechanisms and personal-level phenomenology—PP has the potential to be an interesting theoretical candidate for illuminating aspects of Sartre’s imagination view on dreaming. These include a “perfectly closed imaginative consciousness” and a “worldly atmosphere” in dreams. We suggest that relentless prediction generation without sensory input constraining the experience can be coherently paralleled to what Sartre conceptualizes as the dreaming consciousness being “fascinated” by its own images. This integration helps refine a PP account by critically addressing assumptions about the imagination–perception relationship and enables us to situate Sartre’s phenomenology as a valuable resource for contemporary discussions of dreaming.