Perceptual Grouping and Change Detection Perceptual Grouping in Change Detection Perceptual Grouping and Change Detection

Detection of an item changing its location from one instance to another is typically unaffected by changes in the shape or color of contextual items. However, we demonstrate here that such location change detection is severely impaired if the elongated axes of contextual items change orientation, even though individual locations remain constant and even though the orientation was irrelevant to the task. Changing the orientations of the elongated stimuli altered the perceptual organization of the display, which had an important influence on change detection. In detecting location changes, subjects were unable to ignore changes in orientation unless additional, invariant grouping cues were provided or unless the items changing orientation could be actively ignored using feature-based attention (color cues). Our results suggest that some relational grouping cues are represented in change detection even when they are task irrelevant. (137 words) Perceptual grouping and change detection Jiang, Chun, & Olson 3 Change detection has been a useful tool to study visual attention and short-term memory in visual perception (Rensink, 2002; for a nice collection of studies, see Visual Cognition, 7, 2000). It creates surprising but convincing demonstrations of how visual representations are impoverished (Simons & Levin, 1997) and also how such representations can be enhanced by focal attention (Rensink, O'Regan, & Clark, 1997). It has become a standard paradigm for studying properties of visual short-term memory (VSTM; Luck & Vogel, 1997; Pashler, 1988; Philips, 1974). Most studies have focused on change detection of a collection of individual items, with an interest in how each item is represented and how its change is detected. Change detection can also be used to probe the nature of relational encoding among items: how items are perceptually grouped and how such grouping is carried from perception to short-term memory. One approach is to cue attention to an element within a perceptual group, defined by proximity or connectedness, and test the success in detecting changes of items in the same or a different group (Woodman, Vecera, & Luck, 2003). Same-group items are better detected than are different-group items, suggesting that perceptual grouping is preserved in transferring information to short-term memory. Another approach relies on the principle of "encoding specificity" (Tulving, 1974) and varies the retrieval context such that across frames of changes, the context surrounding the critical change either matches or mismatches the context during encoding (Jiang, Olson, & Chun, 2000). Using the latter approach, we found that detecting location changes is unaffected by changes in the color or shape of the contextual items, but is disrupted by changes in the locations of the surrounding items. This suggests that the spatial relation formed by adjacent items is encoded and represented in VSTM, but surface features are not. Perceptual grouping and change detection Jiang, Chun, & Olson 4 In this study we further pursue the representation of perceptual grouping in change detection tasks. What role does perceptual grouping play in detection of changes in item locations? Our own research in the past had suggested that perceptual grouping plays a significant role if it is relevant to the task, but can be largely ignored if it is irrelevant to the task. If, for example, subjects are asked to remember the locations of chromatic items and ignore the locations of achromatic items, then change detection is completely determined by changes of the chromatic items and is unaffected by changes of the achromatic items (Jiang et al., 2000). In this case, color grouping is relevant, and hence the target group (chromatic items) is attended and transferred into VSTM. If, however, color grouping is irrelevant, it is largely ignored. For instance, consider a set of red and green items forming two color groups red items and green items. The perceptual organization (grouping) is preserved when these items maintained their colors across frames, but is changed if some of the red items swapped colors with some of the green items. Nonetheless, if subjects are asked to remember the locations of all the items, then location change detection is unaffected by such color swapping (Jiang et al., 2000). Thus, it appears that if color grouping is relevant to the task, then the target group is effectively selected and stored in VSTM; but if it is irrelevant to the task, then change detection of locations can proceed independently of changes in the perceptual organization of the display. The purpose of our current study is to demonstrate that some types of perceptual grouping cues that are irrelevant to the task cannot be effectively ignored in change detection. We presented subjects with two dot arrays separated by a brief inter-stimulus-interval (ISI) of 1 s, and subjects were required to remember the locations of these dots and judge whether the two arrays occupied identical locations, or whether there was a change in one location. Using these displays, we had previously shown that change detection of dot locations was unaffected by the Perceptual grouping and change detection Jiang, Chun, & Olson 5 color change of these items (Jiang et al., 2000). In the present study, we added a short line segment interposing each dot. From the memory to the probe display, these line segments either maintained their orientation, or changed orientation. A schematic sample of an orientation change is illustrated in Figure 1. ----------Insert Figure 1 Here-------Note that in this design, the locations of the dots were relevant while the orientations of the line segments were not. Nonetheless, elongated lines tend to group by spatial proximity. The change in their orientations produced a change in the proximity relations, and hence induced changes in perceptual grouping. Can irrelevant perceptual grouping always be ignored in change detection of item locations? Experiment 1