Striking Deficiency in Top-Down Perceptual Reorganization of Two-Tone Images in Preschool Children J.M.D. Yoon * J. Winawer N. Witthoft E.M. Markman Stanford University Stanford University Stanford University Stanford University Stanford, CA 94305 Stanford, CA 94305 Stanford, CA 94305 Stanford, CA 94305 jennifer.yoon@stanford.edu winawer@stanford.edu witthoft@stanford.edu markman@stanford.edu * Supported by an NSF Graduate Research Fellowship Abstract - Two-tone images (black and white transformations of gray-scale photographs) can be difficult for adult observers to recognize. However, following a brief presentation of the original photograph from which the two-tone image was created, adults experience rapid and long-lasting perceptual reorganization, such that after the initial presentation, the two-tone image becomes immediately and easily recognizable. Following a previously reported observation [1], we present evidence that, in contrast to the effortless recognition seen in adults, preschool- aged children are generally unable to recognize two-tone images even when the photograph is simultaneously available. When asked to draw corresponding parts of the photo and two-tone images, children often marked correct regions of the photo and nonsensical regions of the two-tone image. A control experiment showed that children are fully able to mark corresponding parts of two identical photographs. These results point to a dramatic lack of cue-driven perceptual reorganization in young children under conditions that trigger instant recognition in adults. We suggest that this robust phenomenon may provide a window into the development of top-down mechanisms of perceptual learning and consider interventions (e.g., [2-4]) that may improve young children's ability to use one image to reorganize another. Index Terms – Development, Perceptual Reorganization, Preschool Children, Two-Tone Images. INTRODUCTION Figure 1 shows a two-tone image often seen in perception textbooks that illustrates how information not derived directly from the image can be used to guide perceptual organization. In this case, naïve observers often fail to see the Dalmatian dog in the snow and instead perceive a disorganized smattering of black and white patches. However, when instructed to look for a Dalmatian, viewers find the image suddenly and vividly transformed into a coherent percept [5, 6]. Moreover, following reorganization, it is difficult if not impossible to see the image as a meaningless array of patches. This serves as a powerful demonstration of the ability of top- down information to drive perceptual reorganization. Other cues, such as a single presentation of the photo from which the two-tone image was derived, can also trigger immediate perceptual reorganization of the corresponding two-tone image in human and non-human adult primates. Long-lasting recognition for the two-tone image obtains without further need for the cue, resulting in enhanced responsivity in neurons in the inferior temporal cortex [7] as well as increased functional connectivity between temporal and frontal cortical regions [8] when viewing previously cued two-tone images in comparison to equivalent uncued and unrecognised images. That an extrinsic cue can trigger perceptual reorganization following a single presentation suggests the influence of a process high in the visual hierarchy driving a top-down reinterpretation of an otherwise unchanged stimulus. Fig. 1 Dalmatian dog two-tone image (hint: head is on the left). Kovacs and Eisenberg [1] suggested that this ability might only develop over time. They presented two-tone images and their corresponding photos to 4-5 year-old children. None of the eight children observed were able to verbally identify the two-tone images, even with simultaneous presentation of the corresponding photo. This phenomenon is striking in that children appear to not just be slower or less accurate; rather, when viewing the two-tone images and photos side-by-side, they fail to see what adults see automatically. The experiments reported below document our efforts to probe the robustness of Kovacs and Eisenberg’s observations, develop a measure that allows children to show us how well they can or cannot perceive the correspondence between a two-tone image and its cue, and to quantify the