Abstract
Backwards masking reduces stimulus visibility and allows researchers to investigate the dynamics of perceptual processes. However, the mechanisms underlying this attenuation remain unclear. How do the visual features of a mask interact with the perception of a target scene? Is it more difficult to categorize a scene when it is masked by an image sharing visual or conceptual features? Here, we tested six different mask types created from Image Types that were either (1)identical to the target image or (2)different images in which the mask was created from a different category scene. The Mask Types were: (1)different-scene, (2)textures from the Portilla-Simoncelli algorithm (2000),3)phase-scramble,(4)pixel-scramble, (5)10-pixel block-scramble and (6)100-pixel block-scramble. Participants were shown colour photographs of scenes for either 20 or 50ms, followed by one of six mask types for 50ms. Participants were then shown a scene category name and instructed to indicate whether it matched the target scene. A mask was deemed more effective when categorization accuracy was poorer. For the 20ms condition, Bonferroni-adjusted comparisons revealed an interaction between Image Type (identical versus different) and Mask Type such that for the different-scene and 100-pixel block scramble masks, different image masks were more effective (55%) than identical image masks (85%).Interestingly, the most effective mask for both image types was the 10-pixel scramble mask (Different:59%, Identical:58%). Of the remaining mask types, masking was moderately effective with overall higher accuracy rates(68%-78%) and differences between Image Types were small or non-existent (.04%-5%). The pattern of results suggests that masks are more effective when derived from a different scene category and when some conceptual information can be derived (Potter, Staub & O’Connor, 2004), indicating that mask efficacy is increased when masks differ in both perceptual and conceptual information from the target image.
Meeting abstract presented at VSS 2012