Context effects

Context can exert a large influence on perception.  We found that non-target elements in a visual search array can influence the interpretation fitted to an ambiguous target item (Rauschenberger, Peterson, Mosca, & Bruno, 2004).  We are pursuing the question of how context effects operate spatially and temporally.  Experiments exploring these issues are funded by a grant from the NSF (BCS 0418179).

 

 

The center element in each display is ambiguous: it can be perceived as either a
blue disk lying behind a green square or as a blue pacman shape abutting a green
square.  The first interpretation involves amodal completion.  When the center
element is the target in a visual search experiment, its interpretation is
affected by the non-targets surrounding it.  Disk non-targets favor the disk
interpretation; pacman non-targets favor the pacman interpretation, even when
the displays are exposed long enough for amodal completion to have occurred.
 

 

Context effects also extend to shape segmentation.  We found that convexity, one of the classic stimulus cues to shape, varies in its effectiveness in different contexts (Kim & Peterson, 2001 - 2003).  Specifically, convexity is an extremely effective figure cue in displays containing a number of alternating black and white convex and concave regions, but it is relatively ineffective when there is only one convex region.  The same does not appear to be true for symmetry.  We are attempting to understand these results by modeling the interactions between neurons with different size receptive fields at different levels of the visual system.