Frames of reference and motion aftereffects.

Evidence concerning the origin of the motion aftereffect (MAE) is assessed in terms of a model of levels of representation in visual motion perception proposed by Wade and Swanston. Very few experiments have been designed so as to permit unambiguous conclusions to be drawn. The requirements for such experiments are identified. Whereas retinocentric motion could in principle give rise to the MAE, data are not available which would enable a conclusion to be drawn. There is good evidence for a patterncentric origin, indicating that the MAE is primarily the result of adaptation in the systems responsible for detecting relative visual motion. There is evidence for a further contribution from the process that compensates retinocentric motion for eye movements, in the form of nonveridical information for eye movements. There may also be an effect at the level at which perceived distance and self-movement information are combined with egocentric motion to give a geocentric representation which provides the basis for reports of phenomenal experience. It is concluded that the MAE can be caused by changes in activity at more than one level of representation, and cannot be ascribed to a single underlying process.