Abstract
Everyday vision is often carried out over an extended period of time. Successful visual cognition relies on an ability to retain relevant visual information not only for a single percept, but also from several sequential percepts. In this study we test how normal adults manage to relate the information across two sequential visual displays. We ask: are sequential visual arrays represented as separate images or as a combined image in visual short-term memory (VSTM)? Proponents of the integration account suggest that an image of the first array is gradually formed and integrated with an image of the second to produce a combined representation. This view is evidenced by successful performance in an empty-cell detection task. In this task, on a 4 × 4 square matrix, 7 locations are occupied on a first array, followed by a variable interval, and then by 8 other occupied locations on a second array. Subjects' success in identifying the remaining empty cell has been taken as proof for integration. We propose, instead, that the visual system only possesses a limited ability to integrate arbitrarily related images in STM. Experiment 1 shows that when subjects have to remember the spatial locations of two sequential arrays, they retain each array separately and do not combine the two arrays to form an integrated image. Experiment s2–4 show that success in the empty-cell detection task is achieved by a convert-and-compare strategy, rather than an integration strategy. We suggest that people's ability to relate multiple visual percepts is achieved largely by representing events that have spatio-temporal continuities. The presentation of two arbitrary arrays does not support any spatio-temporal continuity. Under such conditions, perhaps people do not, and cannot, integrate sequential arrays.