Change Detection as a Tool for Assessing Attentional Deployment in Atypical Populations: The Case of Williams Syndrome

ABSTRACT When unexpected changes occur in a visual scene, people often fail to notice them. Because change detection depends on attentional mechanisms, people tend to notice changes that are of special significance. People with Williams syndrome (WMS) have an unusually strong interest in other people that is manifest in relatively spared face recognition skills, heightened social attention and hypersociability. We hypothesized that in a change blindness paradigm participants with WMS would be more sensitive to changes in people in social scenes compared to age, IQ and language matched participants with learning or intellectual disabilities. Two videos were presented, one showing an unexpected change to the identity of an actor and one with numerous unexpected changes during a conversation scene. Subjects in both the WMS and the learning disabilities groups noticed fewer overall changes than age-matched normal controls, suggesting that change detection is especially challenging to people with intellectual disabilities. Consistent with our hypothesis, WMS subjects noticed more person-related changes in the complex scene than did subjects with other intellectual/learning disabilities. WMS subjects attend to social elements of dynamic scenes, decreasing change blindness for changes associated with people. KEYWORDS: Williams syndrome, change detection, change blindness, attentional biases. INTRODUCTION How people encode their visual environment depends to a large extent on how attention is deployed in real-time. Although our perceptual experience seems rich in detail, people are surprisingly poor at noticing large changes to visual scenes if the changes occur during a visual disruption, and this "change blindness" is accentuated when the changes are unexpected (see Rensink, 2002; Simons & Ambinder, 2005 for recent reviews). For example, nearly two-thirds of observers failed to notice when the only actor in a brief motion picture was unexpectedly replaced by a different actor during a cut from one shot to the next (Levin & Simons, 1997). Change detection is enhanced when the change occurs instantly, producing a visible transient signal (Rensink, O'Regan, & Clark, 1997). However, change blindness ensues when the perceptibility of the transient signal is disrupted by a blank screen (e.g., Rensink et al., 1997), an eye movement (e.g., Grimes, 1996; Henderson & Hollingworth, 1999; McConkie & Currie, 1996), a blink (O'Regan, Deubel, Clark, & Rensink, 2000), or a cut or pan in a motion picture (e.g., Simons & Levin, 1998; Levin & Simons, 1997). This pervasive change blindness occurs both when observers intentionally search for change and when changes occur unexpectedly. Evidence from a variety of paradigms suggests that attention to the change is necessary for change detection. Observers must encode the pre-change scene and compare it to the post-change scene, a seemingly attention-demanding process. To the extent that attention is needed for change detection, successful change detection implies that the changing element was attended and encoded (Tse, 2004). In support of this assumption, changes to objects rated as more important to the scene are noticed more readily than less important objects (Rensink et al., 1997). Several studies have relied on this assumption, using change detection tasks to measure the capacity of attention (Rensink, 2000) and to map the spatial locus of attention (Tse, Sheinberg, & Logothetis, 2003; Tse, 2004). Individual and group differences in expectations, interest, and expertise also influence the focus of attention in scenes, leading to enhanced or impoverished change detection performance. For example, recreational drug users and problem drinkers are more likely than non-users to notice changes to drug paraphernalia and alcohol-related items, respectively (Jones, Jones, Smith, & Copley, 2003; Jones, Bruce, Livingstone, & Reed, 2006). …