Research

Memory during visual search. Several tasks in our every day life involve visually locating targets of interest (finding the milk in the refrigerator, finding a friend in the crowd or a pen in a cluttered drawer). The BVC lab is interested in understanding how visual attention and memory are involved in this process. Through the use of eye tracking technology and cognitive modeling, research from the BVC lab has provided evidence that observers remember where attention has been allocated (I already looked in on the right side of the top shelf for the milk."), but not to remember what has been examined ("I know I have looked at the orange juice, but I have still not found the milk"). Furthermore, BVC Lab research suggests that visual search is made efficient not only by remembering where search is going, but also by remembering a plan for where attention will be directed as the search process proceeds (see Beck, Peterson, Boot, Vomela, & Kramer, 2006; Peterson, Beck, & Wong, 2008; Peterson, Beck, & Vomela, 2007; Beck, Peterson, & Vomela, 2006). This research bridges the gap between theory and practice by informing the design of GUIs and training techniques for operators in jobs that require visual search tasks (e.g., air traffic controllers, operators detecting mines in sonar scans of the ocean floor, and pilots locating targets in digital maps).

Updating visual representations. As we look around our visual environment our brain must construct a stable representation of the visual world. This requires updating the representation as eye movements are made and attention is shifted approximately every 350 milliseconds. The BVC Lab has employed change detection tasks and eye movement tracking technology to determine the types of information that are maintained over time and the limits of this ability. For example, research from the BVC Lab has demonstrated that observers' expectations about the stability of visual information over time greatly influence their ability to detect a change in the visual world from one glance to the next. Furthermore, the visual representation of an initial view is easily overwritten by a second view unless attention is used to stabilize the first representation. In addition, by measuring participants' eye movements while they completed a change detection task, it was concluded that when changes in the visual world are not detected, it is largely due to a failure to direct focused attention to the visual world before and after the change rather than to a failure at the perception, awareness, or decision levels of the process (see Beck, Angelone, & Levin, 2004; Beck, & Levin, 2003; Beck, Peterson, & Angelone, 2007). Research in the BVC lab is currently investigating research questions related to the role of attention in feature binding and the relationship between short-tem and long-term memory representations.

Visual metacognition. Of particular importance when considering attention and memory strategies that observers may adopt in the hope of improving task performance are the observers' beliefs about their visual attention and memory abilities. This is referred to as visual metacognition. Research from the BVC Lab has examined the extent to which participants can accurately predict their performance on visual change detection tasks. Interestingly, participants tend to overestimate their performance, suggesting that they can be unaware of when additional resources may be needed to improve visual memory and attention performance (see Levin, & Beck, 2004; Beck, Levin, & Angelone, 2007; Levin, Drivdahl, Momen, & Beck, 2002).

Optimizing Naval Pilot's Visual Search Performance. Critical to the performance of operators using charts to navigate is the ability to find a target quickly. For example, if a pilot is flying low and needs to identify high altitude obstacles in the flight path, the quicker and more accurately these obstacles are identified in the chart, the faster a pilot can direct the course to prevent a collision. Research in the BVC lab focuses on identifying the factors that allow targets to be found quickly and what factors add to clutter in the charts, thereby impairing the ability to effectively locate a target and identifying design techniques for the development of computer interface mapping systems such that the operator can use them more effectively. This work is currently in progress and is being completed in collaboration with Maura Lohrenz of the Naval Research Laboratory (NRL) at Stennis Space Center. From research using a Color-Clustering Clutter (C3) algorithm designed to quantify the amount of clutter in a visual display (Lohrenz & Gendron, 2008) it has been determined that the C3 is a good predictor of subjective ratings of clutter (Lohrenz, Trafton, Beck, & Gendron, 2009). In addition, C3 values are also predictive of visual search performance (Beck, Lohrenz, & Trafton, in press). When participants searched through aeronautical charts for a target that varied in the amount of global clutter (the clutter of the whole chart) and the amount of local clutter (the amount of clutter surrounding the target), search reaction time (RT) was slower as the amount of global clutter increased. This effect was strongest when the target was in a high local clutter region. Beck et al. (in press) also examined the impact of target and distractor saliency. The effects of clutter were strongest when the target was non-salient and the distractors were salient. In addition, eye movements were measured and indicated that the increase in RT for higher levels of clutter was caused by an increase in the number of fixations that occurred before the target was found. This indicates that increasing clutter increases the number of areas in the charts competing with the target for attention. We concluded from these studies that search performance can be improved if the amount of the clutter immediately surrounding the target is minimized and the target is salient. If the target features and location are unknown, then limiting the overall amount of global clutter is also a good design technique for improving search performance. Current studies are examining different training techniques for improving search performance in cluttered aeronautical charts. In addition, the C3 algorithm is being applied to the task of detecting mines in sonar scans of the sea floor.

Development and Specificity of Expertise. This line of research focuses on how expertise leads to better attention strategies. Specifically, we are interested in the ability to learn effective attention strategies and the degree to which these effective attention strategies can be generalized to non-expertise related tasks. For example, one study currently underway in the BVC lab focuses on radiologists' ability to detect visual changes in radiographs. Expertise in reading radiographs requires the ability to successfully identify abnormalities in radiographs and changes from pre- to post-treatment radiographs. Of particular interest is how much and what type of training is necessary to develop these abilities. First through fourth year LSU School of Veterinary Medicine graduate students participated in the experiment. Students were asked to detect changes between pre- and post-treatment radiographs. Expertise level performance was not found until the students had completed their rotation in the radiological clinic. This demonstrated that the technical knowledge gained in the classroom was not sufficient to improve performance. Rather extended "eyes-on" training was necessary. In a follow-on study Veterinary doctors at LSU's School of Veterinary Medicine completed the radiographchange detection task. A significant correlation between level of expertise (years in profession and hours per week reading radiographs) and performance on the change detection task was not found, indicating that once expertise has been obtained, performance does not continue to improve.