Abstract
Recognizing aerial scene imagery is an essential skill in a variety of work domains, yet little is currently known about how this skill forms and develops with experience. The current study aimed to elucidate the neural mechanisms underlying expertise in aerial scene recognition in order to understand the acquisition of experience with aerial imagery. We conducted an intensive six-session behavioral training study combined with multiple fMRI scans using a large set of high-resolution color images of real-world scenes varying in their viewpoint (aerial/terrestrial) and naturalness (manmade/natural). Half of the participants were trained to categorize these visual scenes at a specific-subordinate level (e.g., truss bridge, suspension bridge) and half of the participants passively viewed the same images while performing an orthogonal task. Both groups saw the same scene stimuli for five of the six training sessions; the sixth session consisted of a novel set of scenes to assess learning transfer. We found group-specific improvements in behavioral performance across training sessions and scene dimensions, including learning transfer, with greatest behavioral improvements observed for aerial scenes. In contrast, the passive-viewing group showed no major improvements despite equal exposure to the stimuli. Complementing the behavioral effects, we found experience-related neural changes in the experimental group: response magnitudes in scene-selective cortex (PPA and OPA) were correlated with improvements in behavioral performance for aerial imagery, but not in control regions (e.g., EVC, FFA). Whole-brain analyses revealed that over the course of training with aerial scenes, additional regions were recruited beyond scene-selective cortex, primarily lateral ventral-occipitotemporal cortex and posterior parietal cortex. Together these findings suggest that acquiring experience in categorizing aerial scenes entails the engagement of multiple visual areas involved in object and scene recognition, as well as the potential involvement of top-down attention mechanisms, and visuospatial processing. This research is supported by ONR BAA N00014-16-R-BA01.
Acknowledgement: ONR BAA N00014-16-R-BA01