Abstract
Our visual worlds are composed of many objects. Despite the fruitful accumulation of object recognition studies, it is still unclear how we see entire visual scenes. We conducted three experiments to verify the application of Edelman's (1999) object recognition model to scene recognition. Experiment 1 showed participants' superior performance recognizing target scenes in novel views nearly located in experienced views. This was due to the combination of multiple views, but not due to comparing the novel view to the one most similar experienced view (the view combination effect). Experiment 2 indicated that these processes are elicited by the “summation” of stored multiple previous experiences, and not by comparison to a prototype. Experiment 3 was conducted to test whether these processes of scene recognition would apply to another context, by a spatial bisection task. Performance did not differ due to prior view experience. These results emphasize the difference between recognition and perceptual processes involved in bisection tasks. Thus, our results suggest that the psychological processing of visual scenes includes various phases, and may be subject to task-dependent effects.
