This study aimed to find out whether image segmentation is affected by the object's meaning. In Experiment 1, 20 subjects judged whether or not two probe circles embedded in a meaningless two-part figure have the same color. The probes shared the same part (same object) or different parts (different objects) of the figure. The metric distance from fixation point to probes was identical across conditions. Responses were faster and more accurate in the same object than in the different object condition (same-object-advantage). In experiment 2, meaningful Japanese kanji and meaningless jumbled-kanji were used as stimuli. The kanji measured the effect of meaning on segmentation while the jumbled-kanji measured the applicable range of this effect. Data from 20 subjects reveal that reaction time for same-object was longer than the different-object condition (same abject-disadvantage) irrespective of whether they were kanji or jumbled-kanji. These findings suggest that meaning and it's influence affect image segmentation.

Key words: image segmentation, object-based effect, bottom-up and top-down effects