Academic journal article Attention, Perception and Psychophysics

Evidence for Scene-Based Motion Correspondence

Academic journal article Attention, Perception and Psychophysics

Evidence for Scene-Based Motion Correspondence

Article excerpt

Published online: 23 January 2014

# Psychonomic Society, Inc. 2014

Abstract To maintain stable object representations as our eyes or the objects themselves move, the visual system must determine how newly sampled information relates to existing object representations. To solve this correspondence problem, the visual system uses not only spatiotemporal information (e.g., the spatial and temporal proximity between elements), but also feature information (e.g., the similarity in size or luminance between elements). Here we asked whether motion correspondence relies solely on image-based feature information, or whether it is influenced by scene-based information (e.g., the perceived sizes of surfaces or the perceived illumination conditions). We manipulated scene-based information separately from image-based information in the Ternus display, an ambiguous apparent-motion display, and found that scene-based information influences how motion correspondence is resolved, indicating that theories of motion correspondence that are based on "scene-blind" mechanisms are insufficient.

Keywords Apparent motion . Perceptual organization . Ternus display

Imagine that you are at a busy skating rink trying to follow your friend. Being a good skater, your friend moves rapidly over the ice, one second being visible close to the entrance of the ice rink, but disappearing the next second, as he is occlud- ed by other skaters that pass between the two of you, and then reappearing again at a completely different location. How do you know that the person reappearing again is the same person that you saw somewhere else one moment ago?

One way to understand this correspondence problem (e.g., Ullman, 1979)-that is, how the visual system is able to make connections between different images across time-is in terms of objects (e.g., Enns, Lleras, & Moore, 2010). Within this framework, the visual system represents the continuous pres- ence of an object, even when it is not always visible, by linking together isolated glimpses that correspond to the same object representation within the scene. Specifically, if corre- spondence is established between two separate instances, the existing object representation that is associated with the first stimulus will be updated to accommodate the information from the more recently sampled stimulus. In contrast, if no correspondence is established, a new representation must be created to accommodate the new information, and the old representation will be left unchanged (Fig. 1).

Another way that correspondence can be understood is in terms of the interpretation of the output of motion energy models (e.g., Adelson & Bergen, 1985; Van Santen & Sperling, 1985; Werkhoven, Sperling, & Chubb, 1993). The idea of these models is that the visual system computes motion energy on the basis of changes in the activation of low-level motion detectors, due to the change of the stimulation at a certain location over time, and determines correspondence on the basis of the direction in which the most motion energy occurs.

Motion-energy-based and object-based accounts of corre- spondence imply different levels of visual processing at which correspondence is resolved. From the viewpoint of motion energy models (e.g., Adelson & Bergen, 1985; Van Santen & Sperling, 1985; Werkhoven et al., 1993), correspondence is established between relatively ?raw? representations of stimulus information (i.e., features that are represented directly in the retinal image, or ?image features?). These include features like luminance and retinal size. In contrast, according to object-based models (e.g., Enns, Lleras, & Moore, 2010), although corre- spondence can be established between ?raw? representations if nothing else is available, it is more likely to be established between representations of stimuli that have been processed to take into account scene characteristics, such as differences in illumination conditions, leading to lightness rather than lumi- nance, and relative depth, leading to perceived size rather than retinal size. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.