There is no induced motion at near-threshold velocities

Duncker's classic experiments on induced motion at near-threshold velocities were based on the assumption that movement of one stimulus relative to the other (object-relative) is perfectly detected while it remains completely impossible to identify which of the two stimuli is moving (subject-relative). In the present experiments it is shown that the threshold areas of object-relative and subject-relative movement are largely overlapping. Consequently Dunker's assumption cannot be fulfilled: whenever object-relative movement is well detectable accuracy of detecting subject-relative motion will also be over 50%. In four additional experiments it is shown that the well-established effects of fixation, stimulus size and enclosure on induced motion are to be interpreted as effects of these variables on the detection of subject-relative movement: the so-called induction phenomena occured only when accuracy was over 50%.