Purchase this article with an account.
Axel Grzymisch, Cathleen Grimsen, Udo Ernst; Attentional Effects in Contour Integration in Dynamic Scenes. Journal of Vision 2015;15(12):326. doi: 10.1167/15.12.326.
Download citation file:
© 2017 Association for Research in Vision and Ophthalmology.
Contour integration (CI) is an integral part of visual information processing requiring the combination of aligned edge configurations into coherent percepts. Subjects are efficient at detecting contours, reaching peak performances for stimulus presentation times of 100-200ms. CI research usually employs flashed static stimuli, making this an artificial situation compared to the continuous observation of everyday scenes. We addressed this discrepancy between commonly employed stimuli in CI and natural vision by employing novel dynamic stimuli comprising slowly rotating Gabor elements. Contours formed at predefined times and locations when 10 Gabor aligned. We expect these stimuli to better approximate natural vision as they require sustained observation and accounting of dynamic changes to generate a coherent picture of a visual scene in order to detect the appearance of a contour. Since CI is believed to be a pop-up effect we expected similar performances for brief and extended presentations. However, results show a statistically significant (t(8)=8.83, p< 0.001) drop in performance of 19.5% for extended presentations of varying lengths, between 1840ms and 3680ms, when compared to a peak performance of 87% for a presentation time of 230ms. This dramatic decrease in performance led us to hypothesize that contour perception in extended presentations is a more demanding task, hence, that it might draw higher benefits from attention than brief presentations. By employing distinct single cues, and their combinations, we evaluated the capabilities of the visual system to use and combine independent information about a target. Cues targeting feature-based attention and location-based attention lead to a 6.7% and a 5.5% improvement in performance, respectively, when compared to a no-cueing condition in extended presentations. Their combination lead to an apparent additive effect, significantly (t(8)=3.3, p< 0.05) improving performance by 12.2%, thus restoring most of the decrease in performance seen between the peak condition and the extended condition.
Meeting abstract presented at VSS 2015
This PDF is available to Subscribers Only