This study investigated the interaction between hemispheric specialization and pictorial recognition memory for pictures presented in three different color modes: realistic color, non-realistic color, and monochrome (back and white). The study was also designed to confirm the efficacy of applying signal detection analysis to color recognition memory data to obtain a more accurate assessment of the role of color in visual information processing. The stimuli consisted of 120 stimulus slides and 60 distractor slides, all photographs of North American and Canadian geography scenery. Twenty-six graduate students were first shown the set of 120 stimulus slides. They were then instructed to perform a complex verbal masking task designed to localize processing to the right hemisphere while viewing either the first or second 60 slides; the alternate slides were viewed without the masking task. The subjects were then presented with a random distribution of the stimulus and distractor slides for five seconds each, and asked to indicate whether the slide was one they had seen before or a "new" one. It was found that under the right localized treatment, both the realistic color and black and white treatments scored significantly lower than the non-realistic color treatment, although the non-realistic group did not differ significantly from any of the treatment groups under the integrated treatment. It is concluded that color processing is bi-locational, with realistic/verbal image processing being done in the left hemisphere and pure color processing in the right hemisphere. It is further concluded that visual realism probably constitutes a combination of imaginal and verbal information. Three figures are appended. (49 references) (BBM)