Change Blindness Theory and Consequences

People often fail to notice large changes to visual scenes, a phenomenon now known as change blindness. The extent of change blindness in visual perception suggests limits on our capacity to encode, retain, and compare visual information from one glance to the next; our awareness of our visual surroundings is far more sparse than most people intuitively believe. These failures of awareness and the erroneous intuitions that often accompany them have both theoretical and practical ramifications. This article briefly summarizes the current state of research on change blindness and suggests future directions that promise to improve our understanding of scene perception and visual memory. KEYWORDS—change blindness; change detection; visual representation; attention; perception; consciousness; awareness; memory Would you notice if a person you were talking to were surreptitiously replaced by a different person during a brief interruption? Do you think you would readily notice if two people in a photograph exchanged heads while you shifted your eyes from one part of the photo to another? If you are like most people, and you have not heard of change blindness before, you might confidently answer ‘‘yes’’ to both questions. Yet in studies involving just such scenarios, 50% of observers missed these changes (Grimes, 1996; Simons & Levin, 1998)! In fact, even when actively searching for changes, observers often struggle to find them. For example, when an original and changed photograph alternate repeatedly, separated by a brief blank screen, observers often require dozens of alternations to spot large changes (e.g., the disappearance of an airplane engine; Rensink, O’Regan, & Clark, 1997). Although change detection is quite good when the change signal is clearly visible (i.e., when the shift from the original version to the changed version is instantaneous and visible as it happens) and when no other distractions draw attention away from it, people are surprisingly inept at change detection whenever the change signal is masked or hidden from view. This failure to detect changes, or change blindness, has developed from a laboratory curiosity into a central phenomenon in the field of visual cognition, and has both theoretical and practical implications. It has been used to motivate conclusions about the nature of visual memory, the role of attention in scene perception, and even the mechanisms underlying conscious awareness of our visual world. A BRIEF OVERVIEW OF CHANGE-BLINDNESS RESEARCH Interest in change blindness has surged since the mid-1990s, but the phenomenon itself has much older roots. The use of change-detection tasks and experimental evidence for change blindness appeared sporadically in the literature beginning as early as the 1950s, with most studies revealing failures to detect changes to simple, sequentially presented arrays of dots or letters when they were separated by a brief blank interval (e.g., Di Lollo, 1980). Change blindness also surfaced in studies exploring the integration of information across eye movements (saccades), with subjects failing to detect changes that occurred while the eyes were moving. Although these early studies documented the existence of change blindness, more recent studies have illustrated the extent of our blindness to large changes in contexts more closely approximating real-world perception. The most prominent task used to study change detection was developed by Ron Rensink and his colleagues in the 1990s (Rensink et al., 1997). The task was inspired by the finding that changes introduced during eye movements (known as saccadecontingent changes) often go undetected. One explanation for such change blindness is that the mechanisms that generate an eye movement actively suppress perception during the eye movement, a phenomenon known as saccadic suppression. Alternatively, rapid movement of the eye produces blur on the retina, and it might be this blurring itself that masks the change signal. If it is motion blur that masks changes, then other The authors contributed equally to this manuscript. Address correspondence to Daniel J. Simons, Department of Psychology & Beckman Institute, University of Illinois, 603 E. Daniel St., #807, Champaign, IL 61820; e-mail: [email protected]. CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 44 Volume 14—Number 1 Copyright r 2005 American Psychological Society disruptions should mask them as well. In Rensink’s flicker task, an original and changed image alternate repeatedly, separated by a blank screen, until observers detect the change (see Fig. 1). The blank screen produces a luminance change everywhere in the image, which serves to mask the signal produced by the change—that is, observers cannot see the change to the object while it happens. While performing this task, observers know something is changing but simply cannot find it. In most saccade-contingent change-detection tasks, changes typically occur only once; observers do not experience the ‘‘struggle’’ of change detection. This flicker task contributed to interest in change blindness by allowing audience members at talks or in classes to experience the phenomenon for themselves. Change blindness has been observed in many other tasks, all of which disrupt or hide the localizable signal that would otherwise accompany an immediate change (see Rensink, 2002, for a recent review). For example, observers fail to notice changes introduced during a blank screen, a blink, an eye movement, or a motion-picture cut or pan. Similarly, they fail to notice a change that is accompanied by other visual signals that distract attention from the change location or a change that occurs gradually over a period of several seconds so that the change signal is not sufficiently strong to draw attention. The extent of change blindness is particularly striking, with remarkably large changes going unnoticed when the change is unexpected and incidental to the observer’s task. WHAT WE KNOWAND WHAT REMAINS TO BE DETERMINED The change-blindness literature has converged on a core set of findings: First, change blindness occurs whenever attention is diverted from the change signal. Second, changes to objects that are central to the meaning of the scene or changes to visually distinctive objects are detected more readily than other changes, presumably because observers focus attention on important objects (Rensink et al., 1997). Third, attention may be necessary for change detection, with changes to unattended objects going unnoticed. Fourth, attention to a changing object may not be sufficient for change detection; observers frequently fail to detect changes to the central actors in motion pictures and to real-world conversation partners even though these people clearly are attended (Levin & Simons, 1997; Simons & Levin, 1998), suggesting that change detection requires observers to encode the changing features before and after the change and compare them (see Fig. 2). Although these core conclusions are fairly well established in the change-detection literature, a number of open issues remain. For example, more research is needed to establish what draws attention to some scene elements and not others in a change-detection task. Image features might attract attention by virtue of their distinctiveness, or expectations about a scene might drive attention to an object. Few studies have examined Fig. 1. Schematic illustration of the flicker task. In this task, an original and modified image alternate repeatedly, separated by brief blank displays, until observers find the change. Even large changes can go unnoticed for many seconds. In this image, the change is the appearance and disappearance of a building in the background. Volume 14—Number 1 45 Daniel J. Simons and Michael S. Ambinder