Conversation Disrupts Change Detection in Complex Traffic Scenes

Article excerpt

INTRODUCTION

The ability to notice behaviorally meaningful objects and events in the visual surroundings is fundamental to an operator's capacity to maintain performance in a complex environment. Indeed, Endsley's (1995) widely cited model of situation awareness recognizes the perception of task-relevant information in the environment as the foundational stage of knowing and understanding "what is going on around you" (Endsley, 2000, p. 5). It is less obvious, however, how easily perception and attention may fail. Despite people's impressions of a detailed and continuous visual world, human performance data indicate that lapses of perception and attention are frequent and often consequential. Jones and Endsley (1996), for example, found that 76% of pilot errors were attributable to failures of perception and attention. Similarly, attentional lapses have been implicated as an important cause of various forms of traffic error (Langham, Hole, Edwards, & O'Neil, 2002; Larsen & Kines, 2002).

Recent findings in the study of visual performance have corroborated the suggestion that perception is less comprehensive than introspection suggests. Evidence demonstrates that attention is generally necessary, for the conscious perception of objects within a static scene (Mack & Rock, 1998) as well as for the detection of events within a scene (Pringle, Irwin, Kramer; & Atchley, 2001; Pringle, Kramer, & Irwin, 2004; Rensink, O'Regan, & Clark, 1997; Simons & Levin, 1997). Under common circumstances, visual events generate localized transient signals--motion or flicker--which capture attention and ensure that changes within an operator's surroundings are noticed. When the transient signal produced by an event is somehow masked, however, the event itself may go unattended and therefore undetected.

Even seemingly obvious changes can thus fail to reach awareness, a phenomenon known as change blindness (Simons & Levin, 1997). To avoid such perceptual failure, an observer must rely on effortful, attentive scanning (Hollingworth, Schrock, & Henderson, 2001; Rensink et al., 1997) to actively encode objects and note changes to them. Bottom-up/stimulus-driven and top-down/knowledge-driven processes guide such scanning, helping to ensure that changes are detected more easily when made to objects that are physically salient or meaningful within the context of a scene (Pringle et al., 2001, 2004). Nonetheless, changes to important and physically conspicuous objects often go unnoticed.

The study of change blindness has provided basic researchers with insight into the cognitive and neural bases of conscious perception. The implications of the phenomenon, however, extend into applied domains. Data indicate that change blindness can result from a variety naturalistic visual events, including saccades (Grimes, 1996), blinks (O'Regan, Deubel, Clark, & Rensink, 2000), egomotion (Wallis & Bulthoff, 2000), occlusion of a changing item (Simons & Levin, 1998), and the presence of irrelevant transient signals (O'Regan, Rensink, & Clark, 1999). As such, change blindness can occur outside of contrived laboratory settings (Simons & Levin, 1998) and is likely to mediate visual performance in real-world tasks and circumstances. An understanding of the mechanisms and processes underlying change detection might therefore provide insight into the limits of human perception and cognition outside the lab. By the same token, change detection may serve as a gauge of perceptual-cognitive performance under varying circumstances.

One applied use of the change detection paradigm has been in the study of the effects of cognitive distraction on perceptual performance. An experiment by Richard et al. (2002) examined the effects of a secondary task on reaction times (RTs) for the detection of changes in traffic scenes. Meant to simulate a hands-flee cellular phone conversation, the loading task required participants to listen to and remember a short declarative sentence presented through a speaker. …