ISSN : 1226-9654
Varying the size and width of circles in Treisman and Paterson's (1984) experimental displays, three experiments were designed to test predictions derived from Treisman and her associates' feature integration theory of attention as against those from a visual routines hypothesis (Ullman, 1986) about illusory conjunctions. The former theory emphasizes the importance of spotlight of attention in the perception of objects, whereas the latter does non-attentinonal processes such as coloring and boundary tracing. Three experiments showed that the extraction of emergent features such as closure is heavily dependent on the size of spotlight of attention, thus rejecting the visual routines hypothesis in encoding of spatial relations. However, our results also indicate the need for further refinements in the feature rote-gration theory of attention, especially its explanations of illusory conjunctions.