On the nature of global dominance in object perception.

Abstract

One of the most enduring issues in cognition is whether object perception is more influenced by the overall configuration of the parts (global aspect) or by the identity of the component parts (local aspect). A recent controversy has centered on the proposal that the global aspect dominates during early perceptual processing (Navon, 1977). Instead, it has been proposed that during early preattentive processing both aspects have equivalent processing status, and that the priority for global information develops later during the course of attentive processing (Boer & Keuss, 1982; Miller, 1981). Experiment 1 (Paquet & Merikle, 1988) examined this issue by investigating global/local processing in both attended and nonattended objects. Of particular interest were the questions of whether an object must receive focal attention in order to observe global dominance and of whether global dominance in nonattended objects may consist of an automatic identification of global information. The paradigm involved the simultaneous presentation of two compound stimuli, one placed in a large square, the other one enclosed in a large circle. Subjects had to decide whether either the large letter (global - directed instructions) or the small letters (local - directed instructions) contained in the large square (attended object) were Hs or Es (target set), and to ignore the letters appearing in the circle (nonattended object). On some of the trials, a member of the target set (H or E) that was either consistent or conflicting with the target identity, appeared either 1) within the attended object, 2) at the local or 3) at the global aspect of the nonattended object. On other trials, only the letter 0, which was not a target set member, appeared as distractor.The results indicated that global dominance occurred for the attended object, as the global aspect was identified faster and was more difficult to ignore than the local aspect. On the other hand, the Stroop - like effects associated with the aspects of the nonattended object were dimension specific; global congruency effects were obtained only when attention was global - directed, and similarly, local congruency effects were observed only when attention was local - directed. However, despite the absence of global congruency effects under local - directed attention, global dominance was nonetheless observed for nonattended objects. Thus, the mere presence of a target set member at the global aspect of the nonattended object slowed reaction time in comparison with when only the letter 0 appeared as distractor, this under both local and global - directed instructions. Furthermore, this category effect was not obtained for the local aspect in the global - directed task. These asymmetrical category effects favouring global information suggest that obligatory global processing occurs and leads to an automatic categorization (i.e., membership in the target set) of the global features of objects.Three further experiments examined a variety of different accounts of the observed effects. Experiment 2 demonstrated that the asymmetrical category effects were not related to response conflict between attended aspects. Experiment 3 verified that these effects were not specific to the size of the information used to designate the attended object; they were still observed when the attended object was designated by a small gap, by small white lines on the surrounding frame, or by a precue presented 100 ms in advance of the attended object. …