Abstract
One's ability to locate a target decreases as a function of similarity to the environment, such as when a target is camouflaged (e.g., Bravo & Farid, 2004; Neider & Zelinsky, 2006; Wolfe et al., 2002). Short-term training improves the ability to detect both trained and novel camouflaged targets in artificial scenes (Boot, Neider, & Kramer, 2009; Neider et al., 2013). However, it is less well known whether training in camouflage detection can be improved in natural scenes and over longer training durations. Here we present manual response and oculomotor data from a larger study in which we assessed the extent to which participants could be trained to better detect camouflaged targets in natural scenes over 14 training sessions, and whether that training transferred to novel nature scenes with similar targets. We utilized three training groups: adaptive camouflage (staircase method with changes in target size), massed camouflage (blocked sessions, target size decreasing across sessions), and control (T/L search, increasing set-size). All reported results are specific to the sessions assessing transfer of training to camouflaged targets in novel search scenes. We found a benefit of training for massed and adaptive conditions in response time and accuracy. To better understand the dynamics underlying training benefits we used a median split to divide participants into groups of learners and non-learners (i.e., steep vs. shallow slopes). We found that learners in the massed training group improved in accuracy and response time at transfer more so than adaptive or control. Furthermore, adaptive trainees displayed a significant shift in strategy after training as reflected by longer, but less frequent fixations. For participants in the massed training group, those identified as learners had a reduction in initial fixation duration. Overall, training improved search performance through changes in oculomotor strategy, but training type affected the changes in strategy.
Meeting abstract presented at VSS 2018