Absence of perceptual segmentation in image generation by normals

We used kanji characters (Chinese ideographic characters) and hiragana characters (Japanese phonographic characters) in the transfer‐appropriate processing paradigm to examine whether or not people without brain damage conduct perceptual segmentation while generating a visual mental image. Ninety‐six participants were divided into three study conditions: (a) seeing hiragana characters and generating a mental image of the corresponding kanji characters; (b) seeing the kanji characters; and (c) seeing the hiragana characters without generating the image of the kanji characters. Generating an image of a kanji character did not transfer at test to the decision as to whether the visually presented kanji character was vertically segmented or not, whereas it did transfer to a semantic decision as to whether the kanji character had a concrete or abstract meaning. Seeing a kanji character transferred to both decisions. Seeing hiragana characters without generating an image of the kanji characters transferred to neither decision. These results suggest that perceptual segmentation is not routinely conducted by normals in the process of image generation.