Change Blindness and Its Determinants
Michaela Porubanová · 2013
Every moment, we perceive visual information in the environment, and regardless of the complexity of the information, we have a very powerful impression that the world is there, at our disposal, and the way we visually represent it is veridical to its actual image. However, the discovery of the phenomenon of change blindness, or an extreme difficulty in detecting changes in a visual field when a disruption in visual continuity occurs, has undermined the notion that visual representations are intricate, and suggests that memory for visual scenes might be very poor (O’Regan, 1992; O’Regan & Noe, 2001; Rensink, 2000). Change blindness has since become one of the most well-known and well-studied phenomenon in the study of visual attention and visual working memory, and has provided powerful methods to study visual attention in general. This dissertation can be split into two different parts. The first part deals with understanding the role of high-level factors in scene perception using the change blindness (flicker) paradigm. The second part focuses on the role of the emotional context of scenes and personality traits on change blindness performance. In the first study, we were interested in understanding whether there are high-level scene factors that influence change detection. Our experiment was motivated by several facts. Firstly, scene perception research shows that the human visual system has a tendency to attend predominantly to informative and important parts of a scene (Buswell, 1935; Henderson, 2003; Loftus & Mackworth, 1978; Yarbus, 1967), and that in active tasks such as change detection, attending to information with high visual saliency (i.e. spatial frequency, color, and intensity) is inhibited (Henderson, 2003; Henderson & Hollingworth, 2003; Turano, Geruschat, & Baker, 2003). Therefore attention is rather dependent on high-level factors (Beck et al., 2004; Hollingsworth & Henderson, 2000; Rensink et al., 1997). Taken together, the goal of the first part of the dissertation was to explore what high-level scene information is considered by the human visual system informative and important. In the experiment, observers inspected various sets of two real-world scene images which featured a changed aspect. The images were shown alternating in sequential order (i.e., the flicker task) and the observers were asked to click on the location where the change had occurred. The results demonstrated that not all changes present in a scene have the same likelihood to be detected. The changes that seem to be dominant in attracting attention were found to be changes occurring in the central region, context-related changes, probable changes, and changes occurring within a figure. Our results are consistent with results of other studies. For instance, research on scene perception has shown that changes in the central region are detected faster than those in the marginal (Rensink et al., 1997), changes inconsistent with the scene context are detected faster than those consistent with the scene context (Hollingsworth & Henderson, 2000), probable changes faster than improbable ones (Beck et al., 2004), changes in a figure are detected to a greater extent than ones in the background (Mazza et al., 2005; Turatto et al., 2002). Interestingly, the detection of changes occurring within close proximity to a figure was most difficult. This indicates that when searching for changes in scenes, parts of the scenes close to the most powerful attractors are being shadowed, and therefore seem to be ignored by selective attention. This could be ascribed to the role of expectations in change detection tasks. Therefore, we believe that specifically in explicit change detection tasks, an individual might use a certain heuristic that helps her scan the scene. The study discusses the significance of the categories under which a stimulus falls in terms of its non-visual or semantic properties. The goal of the second study fo