Neural basis of attention to multi-part, hierarchically organized objects

Visual attention can be directed at an object as a whole (the global level) or to its parts (the local level). The cortical circuitry enabling these attentional configurations is not fully understood. This topic has been studied with hierarchical Navon figures, global letters made out of local letters. Using a novel paradigm we separated the presentation of these two levels in time. We found that seeing a shape at the global or local level momentarily blocks from awareness additional shapes from the other level, an interference not present for shapes from the same level. Using event related potentials we show that this attentional selection modulates early potentials with probable sources in visual extra-striate cortex. Moreover, by examining local activation patterns with functional MRI, we found a divergent specialization for the abstract information provided by the hierarchical figures. Information about shape (invariant to changes in level) was carried preferentially by lateral ventral-occipito-temporal cortex (VOT), overlapping object- and face-selective cortex. Conversely, information about level (invariant to changes in shape) was preferentially carried by medial VOT, and occipital areas partly covering house/scene-selective cortex. Congruently, house and scene distracters abolished the privileged access to awareness of shapes from the same level, whereas faces (or other objects) did not. This implies a shared circuitry processing scene-layout and the internal structure of multipart objects, which is exploited by attention to control the access of shapes into awareness. Thus, neural models explaining attention to hierarchically organized objects must include what is traditionally considered scene-selective areas in addition to the extra-striate cortex involved in shape analysis and the fronto-parietal control areas.

Speaker: Mitchell Valdes-Sosa, Cuban Center for Neuroscience, Havana