Purchase this article with an account.
Matthias Niemeier, Bobby Stojanoski; Contributions of feature-based attention to object perception. Journal of Vision 2005;5(8):1035. doi: 10.1167/5.8.1035.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Recent evidence suggests that attention helps in perceiving “low-level” visual features, such as luminance, motion and orientation. It remains unclear, however, whether and how attention enhances “higher-level” vision. Here, we studied attentional contributions to the processing of proto-objects of intermediate complexity. - We probed perception of loops that consisted of edge segments, luminance-defined gabors, using a dual-task design. Two concurrent rapid serial presentation streams of arrays of scattered gabors appeared in the left and right visual hemifield, respectively. On the primary task side, as indicated by a fixation arrow, subjects made quick responses in one of two conditions, (a) when a subset of gabors was oriented to form the contour of a loop, or (b) when the loop was defined by rotating gabors. These ”primary loops” appeared twice per trial, and coinciding with either of them, another loop was presented on the secondary task side. Perceptual thresholds of this “secondary loop” were determined in a two-alternative-forced-interval fashion by adding different amounts of orientational noise to the gabors. We found that perception of the secondary loops was better when the primary loops were easier to detect, as measured in pretests. Moreover, the condition mattered: perception was better when subjects attended to contour-defined primary loops rather than motion-defined loops. Is this advantage due to more symmetry between the primary and the secondary task? - We replaced the primary loops by s-shapes. Again, secondary loops were easier to perceive when the primary s-shapes were contour-defined rather than motion-defined. Our results suggest that feature-based attention enhances not only low-level vision but also perceptual processes that require grouping simple image structures into more complex ones. This possibly involves enhanced collateral activation of neurons in higher-level visual areas with increased cross-talk between the visual hemifields.
This PDF is available to Subscribers Only