Purchase this article with an account.
Jianwei Lu, Neuroscience Program, Laurent Itti; Perceptual consequences of feature-based attention. Journal of Vision 2004;4(8):265. doi: https://doi.org/10.1167/4.8.265.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Feature-based attention has been shown to enhance the representation of attended visual attributes throughout the visual field. We investigate the consequences of feature-based attention onto visual perception, using dual-task human psychophysics. Subjects simultaneously performed two pattern discriminations on drifting Gabors presented bilaterally to fixation: the first task presumably triggered a first set of feature-based attention effects; the second task triggered a second set and, of interest here, possibly benefited from the first set. Stimuli were horizontal (H) or vertical (V), drifting slowly (S) or faster (F). Tasks were orientation (O) or drift speed (D) discriminations, using a dual two-interval forced-choice paradigm. We measured dual-task discrimination thresholds (75% correct) with a dual staircase procedure for the 64 combinations: primary side, left or right; tasks, O or D; stimuli, HS, VS, HF or VF. We define stimulus orientation as relevant to the orientation discrimination task while drift speed is irrelevant, and conversely, then focus on how primary tasks modulated secondary thresholds. Under same-task conditions, feature-based attention benefited secondary thresholds when stimuli shared task-relevant features, but did not benefit further from additionally sharing task-irrelevant features; however, when no task-relevant feature was shared, a small benefit was observed from sharing a task-irrelevant feature. When tasks differed, secondary thresholds were improved equally as soon as any relevant or irrelevant feature was shared, compared to when nothing was shared. Our systematic study suggests a model with three components: engaging attention onto a stimulus enhances task-relevant features strongly and task-irrelevant features weakly; performance at a pattern discrimination benefits strongly from task-relevant features and weakly from irrelevant features; feature enhancement and feature benefit combine multiplicatively, with the final benefit dictated by the largest of those products.
This PDF is available to Subscribers Only