Mammography to tomosynthesis: examining the differences between two-dimensional and segmented-three-dimensional visual search.

BACKGROUND: Radiological techniques for breast cancer detection are undergoing a massive technological shift-moving from mammography, a process that takes a two-dimensional (2D) image of breast tissue, to tomosynthesis, a technique that creates a segmented-three-dimensional (3D) image. There are distinct benefits of tomosynthesis over mammography with radiologists having fewer false positives and more accurate detections; yet there is a significant and meaningful disadvantage with tomosynthesis in that it takes longer to evaluate each patient. This added time can dramatically impact workflow and have negative attentional and cognitive impacts on interpretation of medical images. To better understand the nature of segmented-3D visual search and the implications for radiology, the current study looked to establish a new testing platform that could reliably examine differences between 2D and segmented-3D search. RESULTS: In Experiment 1, both professionals (radiology residents and certified radiologists) and non-professionals (undergraduate students) were found to have fewer false positives and were more accurate in segmented-3D displays, but at the cost of taking significantly longer in search. Experiment 2 tested a second group of non-professional participants, using a background that more closely resembled a mammogram, and replicated the results of Experiment 1-search was more accurate and there were fewer false alarms in segmented 3D displays but took more time. CONCLUSION: The results of Experiments 1 and 2 matched the performance patterns found in previous radiology studies and in the clinic, suggesting this novel experimental paradigm potentially provides a flexible and cost-effective tool that can be utilized with non-professional populations to inform relevant visual search performance. From an academic perspective, this paradigm holds promise for examining the nature of segmented-3D visual search.

Predicting Airport Screening Officers' Visual Search Competency With a Rapid Assessment.

Objective The study's objective was to assess a new personnel selection and assessment tool for aviation security screeners. A mobile app was modified to create a tool, and the question was whether it could predict professional screeners' on-job performance. Background A variety of professions (airport security, radiology, the military, etc.) rely on visual search performance-being able to detect targets. Given the importance of such professions, it is necessary to maximize performance, and one means to do so is to select individuals who excel at visual search. A critical question is whether it is possible to predict search competency within a professional search environment. Method Professional searchers from the USA Transportation Security Administration (TSA) completed a rapid assessment on a tablet-based X-ray simulator (XRAY Screener, derived from the mobile technology app Airport Scanner; Kedlin Company). The assessment contained 72 trials that were simulated X-ray images of bags. Participants searched for prohibited items and tapped on them with their finger. Results Performance on the assessment significantly related to on-job performance measures for the TSA officers such that those who were better XRAY Screener performers were both more accurate and faster at the actual airport checkpoint. Conclusion XRAY Screener successfully predicted on-job performance for professional aviation security officers. While questions remain about the underlying cognitive mechanisms, this quick assessment was found to significantly predict on-job success for a task that relies on visual search performance. Application It may be possible to quickly assess an individual's visual search competency, which could help organizations select new hires and assess their current workforce.

Attending globally or locally: Incidental learning of optimal visual attention allocation.

Attention allocation determines the information that is encoded into memory. Can participants learn to optimally allocate attention based on what types of information are most likely to change? The current study examined whether participants could incidentally learn that changes to either high spatial frequency (HSF) or low spatial frequency (LSF) Gabor patches were more probable and to use this incidentally learned probability information to bias attention during encoding. Participants detected changes in orientation in arrays of 6 Gabor patches: 3 HSF and 3 LSF. For half of the participants, an HSF patch changed orientation on 75% of the trials, and for the other half, an LSF patch changed orientation on 75% of the trials. Experiment 1 demonstrated a change probability effect and an attention allocation effect. Specifically, change detection performance was highest for the probable-change type, and participants learned to use a global spread of attention (fixating between Gabor patches) when LSF patches were most likely to change and to use a local allocation of attention (fixating directly on Gabor patches) when HSF patches were most likely to change. Experiments 2 and 3 replicated these effects and demonstrated that an internal monitoring system is sufficient for these effects. That is, the effects do not require explicit feedback or point rewards. This study demonstrates that incidental learning of probability information can affect the allocation of attention during encoding and can therefore affect what information is stored in visual working memory. (PsycINFO Database Record

Visual Search: You Are Who You Are (+ A Learning Curve).

Not everyone is equally well suited for every endeavor-individuals differ in their strengths and weaknesses, which makes some people better at performing some tasks than others. As such, it might be possible to predict individuals' peak competence (i.e., ultimate level of success) on a given task based on their early performance in that task. The current study leveraged "big data" from the mobile game, Airport Scanner (Kedlin Company), to assess the possibility of predicting individuals' ultimate visual search competency using the minimum possible unit of data: response time on a single visual search trial. Those who started out poorly were likely to stay relatively poor and those who started out strong were likely to remain top performers. This effect was apparent at the level of a single trial (in fact, the first trial), making it possible to use raw response time to predict later levels of success.

Binding global and local object features in visual working memory.

When briefly presented with global and local visual information, individuals report global information more quickly and more accurately than local information, a phenomenon known as the global precedence effect (GPE; Navon, 1977). We investigated whether a bias toward global information persists in visual working memory (VWM) and whether the VWM representations for global and local features include information bound to their hierarchical levels and to each other. Navon figures, in which a larger (global) letter is composed of smaller (local) letters, were presented, and participants performed a change detection task that required participants to remember features only (either a global or local letter changed to a new identity); features bound to their hierarchical levels (the global and local letters within an object swapped levels); or features bound to each other within an object (2 letters from the same level swapped between objects). Performance suggested that there was a GPE in VWM (new global letters were more accurately detected than new local letters) and that although global and local features were not necessarily bound together in VWM, they were bound to their corresponding hierarchical levels. These results indicate that level binding in VWM occurs more readily than binding specific object features together. These findings further our understanding of how hierarchical objects are represented in VWM.

The effects of incidentally learned temporal and spatial predictability on response times and visual fixations during target detection and discrimination.

Responses are quicker to predictable stimuli than if the time and place of appearance is uncertain. Studies that manipulate target predictability often involve overt cues to speed up response times. However, less is known about whether individuals will exhibit faster response times when target predictability is embedded within the inter-trial relationships. The current research examined the combined effects of spatial and temporal target predictability on reaction time (RT) and allocation of overt attention in a sustained attention task. Participants responded as quickly as possible to stimuli while their RT and eye movements were measured. Target temporal and spatial predictability were manipulated by altering the number of: 1) different time intervals between a response and the next target; and 2) possible spatial locations of the target. The effects of target predictability on target detection (Experiment 1) and target discrimination (Experiment 2) were tested. For both experiments, shorter RTs as target predictability increased across both space and time were found. In addition, the influences of spatial and temporal target predictability on RT and the overt allocation of attention were task dependent; suggesting that effective orienting of attention relies on both spatial and temporal predictability. These results indicate that stimulus predictability can be increased without overt cues and detected purely through inter-trial relationships over the course of repeated stimulus presentations.

Changing target trajectories influences tracking performance.

People have the ability to attentively select and successfully track several moving objects, a process known as multiple-object tracking (MOT; Pylyshyn & Storm Spatial Vision 3: 179-197, 1988). Various factors have been known to influence MOT performance, such as speed, number of distractors, and proximity, while recent work has suggested that object trajectories may also be a factor (Fencsik, Kleiger, & Horowitz Perception and Psychophysics 69: 567-577, 2007). Meanwhile, unexpected changes in motion information have been demonstrated to be a critical facet for attracting attention Howard & Holcombe Attention, Perception & Psychophysics 72: 2087-2095, (2010). Therefore, we suggest that unexpected changes in target trajectories are an important factor in tracking performance. The research presented here controlled for spatial proximity while manipulating the number of instances in which an object changed trajectory. We found that spatial proximity had no effect on tracking performance but, rather, as the number of trajectory changes increased, tracking performance suffered. Results imply that the ability to track multiple moving objects is limited by unexpected changes in direction.

Reallocating attention during multiple object tracking.

Wolfe, Place, and Horowitz (Psychonomic Bulletin & Review 14:344-349, 2007) found that participants were relatively unaffected by selecting and deselecting targets while performing a multiple object tracking task, such that maintaining tracking was possible for longer durations than the few seconds typically studied. Though this result was generally consistent with other findings on tracking duration (Franconeri, Jonathon, & Scimeca Psychological Science 21:920-925, 2010), it was inconsistent with research involving cuing paradigms, specifically precues (Pylyshyn & Annan Spatial Vision 19:485-504, 2006). In the present research, we broke down the addition and removal of targets into separate conditions and incorporated a simple performance model to evaluate the costs associated with the selection and deselection of moving targets. Across three experiments, we demonstrated evidence against a cost being associated with any shift in attention, but rather that varying the type of cue used for target deselection produces no additional cost to performance and that hysteresis effects are not induced by a reduction in tracking load.