Are the original Roelofs effect and the induced Roelofs effect confounded by the same expansion of remembered space?

Are the original Roelofs effect and the induced Roelofs effect confounded by the same expansion of remembered space? Roelofs (1935) first described a perceptual phenomenon in which an observer, when viewing a large rectangular frame whose center is positioned to the left or right of the objective midline, will underestimate the degree to which the frame is offset. More recently, Bridgeman, Peery, and Anand (1997) described a related phenomenon (the induced Roelofs effect) in which a target enclosed by a frame will appear to be displaced in a direction opposite the frame offset; for example, a leftward-shifted frame will induce a rightward error in the perception of the target's location. In theory, both the original and induced Roelofs effects could be explained by a frame-induced distortion of the observer's A recent article by de Grave, Brenner, and Smeets (2002) sought to directly test the idea that both versions of the Roelofs effect had the same underlying cause. In their initial experiment, de Grave et al. replicated the finding of an induced Roelofs effect (Bridgeman et al., 1997) by asking observers to report the location of a target with respect to straight-ahead. In a subsequent key experiment (experiment 2), observers were asked to report the locations of both the target and the surrounding frame, with the prediction that if the original and induced Roelofs effects were caused by the same distortion of the apparent midline, the perceptions of both the target and frame should be affected equally. To the contrary, de Grave et al. found that while the frame was mislocalized as expected, the perception of target location was unaffected by the frame in this condition (i.e., no induced Roelofs effect was seen). Based on these findings, de Grave et al. concluded that the original and induced Roelofs effects must have separate and disso-ciable underlying causes. However, we believe that the conclusions of de Grave et al. were incorrectly based on two factors that subtly confounded their findings. Most importantly, de Grave et al. provided a training period before each task, in which observers first attempted to report the location of either a target or a frame (with trial types presented in random order), and then were provided accurate feedback of performance so that accuracy could be improved. However, this feedback would have had an additional unintended consequence in the second experiment. Specifically, on those practice trials in which observers were …