Abstract
Theories of visual attention hypothesize that neural codes activated while people view objects can also become active when they anticipate an object's appearance, supported by prior findings of attention-driven neural activity in brain regions that process visual features. Using functional magnetic resonance imaging (fMRI) with multi-voxel pattern analyses, we have decoded object-identity, as well as the color and shape, of subjects’ attentional targets after asking them to detect one of four fruits and vegetables within blocks of visual noise ending with a hidden object. Analyzing only the time-points without visual information on-screen, we find that the simultaneous emergence of feature-specific codes in early visual cortex co-occurs with decodable object-identity information later in the visual stream. We first report that a left anterior temporal region can decode object identity during preparatory attention, with neural patterns that match the activity patterns generated when subjects actually view examples of the targeted fruit or vegetable. Second, a classifier trained to differentiate time-points when people anticipate items that differ by shape can successfully distinguish a different set of attentional targets that vary in the same way using data from the lateral occipital cortex. In contrast, V4 activity patterns can generalize to new attentional targets varying by both shape and color. Third, blocks that were characterized by converging successful decoding of shape (e.g., spherical) and color (e.g., orange) in the lateral occipital and V4 regions were accompanied by successful decoding of object identity (e.g., tangerine) in the left anterior temporal lobe during visual attention to that target. Finally, the temporal lobe patterns of subjects showing strong relationships between simultaneous feature-decoding and successful object identity decoding, more closely resembled patterns generated from passively viewing the targets. Our findings support the idea that top-down processes activate various specialized representations in the visual hierarchy during preparatory attention.
Meeting abstract presented at VSS 2013