Abstract
Visual memory for close-up views of scenes consistently exhibits a distortion known as boundary extension (BE), characterized by the report of seeing a wider-angle view than what was actually shown (Intraub & Richardson, 1989). Last year, we presented a series of experiments showing the first reported case (in a normal population) of memory for intact scenes not showing BE (McDunn et al., VSS 2013). Instead, a memory normalization effect was found in which participants seemed to be comparing test images to an averaged view of the studied stimulus set. This unusual finding was verified by several replications. The current study explores potential causes of this result using the standard paradigm used in the BE literature, a free-viewing study phase followed by a recognition test. Four conditions were created by independently manipulating two attributes of the scenes: whether the scenes appear in black and white versus full color, and whether the scenes depict the same location versus independent locations. The results indicate that strict memory normalization only occurs when both variables increase perceptual similarity between images in the set (black/white and same location). A clear effect of BE can be detected when color or different locations are applied to the images, with a slightly greater effect being found with the location variable. These results are explained in terms of the Fuzzy Trace Theory of memory (Reyna and Brainerd, 1995). Increasing the perceptual similarity within the image set likely creates a more gist-like memory trace that include all scenes viewed during the study phase of the experiment. An averaged episodic memory of the viewing experience increases the likelihood of gist recall during the recognition test, leading to normalization. When the images appear more perceptually dissimilar there is a greater likelihood of verbatim recall based on specific visual features, leading to the BE.
Meeting abstract presented at VSS 2014