Purchase this article with an account.
Bobby Stojanoski, Matthias Niemeier; Using Feature-based Attention to Examine the Hierarchical Structure of Visual Processing. Journal of Vision 2011;11(11):199. doi: https://doi.org/10.1167/11.11.199.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Object perception is achieved through a hierarchically organized, yet highly interconnected network of brain areas. The functionality of this hierarchy can be studied using feature-based attention. For example, directing attention to lower-level features such as contours modulates orientation-sensitive mechanisms in early visual areas. However visual areas associated with object processing can also be targeted. We have recently shown that feature-based attention to contours produced late modulations associated with ventral-occipital areas, compared to early effects in extrastriate areas when cueing attention to motion. To further elucidate the procedural structure of object perception and the role feature-based attention plays, we used a difficult object recognition task by presenting fragmented “pillows” or “flowers” among random black and white lines, and investigated the mechanisms of cueing feature-based attention to the colour or shape of the objects. We predicted that invalid colour cues will incur large perceptual costs because filtered colour lines will need to be retraced. Invalid shape cues are less costly since retracing relates to later stages of object processing. These predictions were tested in two studies. In each, we recorded event-related potentials (ERPs) while participants were cued (80% validity) to either shape or color. When cued to shape, their task was to indicate the colour of the object, and when cued to colour they were asked to report its shape. Behaviourally, we found a cueing effect in both experiments, but it was stronger when expecting colour. Our ERP and spectral analysis results revealed that attending to shape produced modulation at 250 ms. Attending to colour, however, generated effects first, from 80–160 ms, and later between 460–580 ms. These effects can be attributed to early contour integration processes during easier valid trials and large temporal costs for difficult invalid trials. Our data support a hierarchical cost model of perceptual decisions.
This PDF is available to Subscribers Only