Abstract
Afterimages have long been understood as visual experiences of past visual stimulation. Importantly, a large set of phenomenological evidence indicates that afterimages are not only based on past visual stimulation but also interact with present visual representations. Recent theoretical and empirical findings suggest that perceptual properties of afterimages strongly rely on the interactions of surface and boundary structures that derive from both past inducing stimuli and current stimuli. In some cases, observers experience afterimages that match with neither of the stimuli. The mixing of present and past stimulation suggests that afterimages are of great importance for investigating the human visual system, because they provide a means of revealing visual processes that cannot be invoked by a single stimulus. However, despite their potential importance, over a century of afterimage investigations have not come to a theoretical consensus on how afterimages are produced. Earlier hypotheses assumed that the adaptation of the retinal visual system generates afterimages, while later studies argued that cortical processing plays an important role in afterimage perception. We argue that, to use afterimages as a research tool, researchers need a better understanding of how afterimage percepts form. We start this process by reviewing key previous research that impacts afterimage theories, and we pay special attention to some neglected theoretic connections between the retinal and the cortical hypotheses. The main finding of our work indicates that the classic photopigment bleaching hypothesis has been significantly modified with a better understanding of functional retinal structures. However, this modification is often overlooked in recent literature that emphasizes cortical visual processing. Scrutiny of the afterimage literature implies that both retinal and cortical processes are likely to jointly produce afterimages and that afterimage formation cannot be completely understood based solely on either one of these processes.
Meeting abstract presented at VSS 2012