What & When: How Objects and Uncertainty Constrain Visual Attention Open Access
Downloadable ContentDownload PDF
Selective attention is a mechanism for processing and filtering a large quantity of sensory information to accomplish task goals. Visual selective attention is guided by space, features, and objects. For this dissertation, three separate sets of experiments were conducted to assess the role of objects in attentional deployment. Specifically, two main questions were posed across these three experiment sets: (i) What attributes of objects contribute to attentional deployment? (ii) Under what conditions do objects serve as an attentional constraint? Standard assumption regarding the role of objects in attention suggests that objects within an attended location are automatically facilitated for attentional selection through a process called sensory enhancement. However, recent behavioral and neurophysiological evidence suggests that: (i) visual uncertainty is a determining principle for object-guided attention, and (ii) all objects are not alike, and thus not processed equivalently within object-selective neural regions, which may have consequence for the extent to which objects guide attention. Behavioral research into the role of visual uncertainty in attentional selection has shown that objects only guide attention when spatial information is not predictive for attentional guidance. While this evidence has demonstrated that spatial uncertainty elicits object-based guidance, visual uncertainty is not limited solely to the spatial domain. The first set of experiments presented here examines the role of non-spatial uncertainty in the context of object-based attention. This work explores two dimensions of non-spatial uncertainty: perceptual masking of target identity and varied object configuration. Object-based effects are shown here only when non-spatial uncertainty is sufficiently high, suggesting a role of general uncertainty (i.e., spatial and non-spatial) as a governing principle in attentional allocation. Most studies of object-based attention define objects according to spatially-defined object boundaries, which are processed in a bottom-up manner. However, real-world objects have additional contextual properties, including semantic meaning and inferred object size. While the role of semantic object meaning has been well-studied, the role of inferred object size has been particularly understudied. Recent neurophysiological work has suggested topographic organization for inferred object size in visual cortex. This organization may influence how real-world objects guide attention. The second set of experiments in this dissertation examines the contribution of inferred object size to attentional deployment, as top-down object properties may shape attentional engagement with objects in a manner similar to bottom-up properties such as object boundaries or retinal object size. Here, it is shown that both the focus and the shifting of attention are parametrically influenced by inferences of real-world object size, such that reaction times were slower and attentional shifts costlier within large objects. A third experiment systematically tested the influence of both uncertainty and object identity on attentional deployment. Specifically, fMRI and multivariate pattern analysis were used to probe the effect of spatial uncertainty on the robustness of neural representations of real-world object identity. As previous object-based attention literature has shown that spatial uncertainty facilitates the use of objects in attentional deployment, this experiment aimed to test how neural representation of objects is modulated by the predictive power of spatial attention. The findings from this experiment show that representations of top-down object identity within object-selective cortex are only facilitated when space does not serve as a sufficient attentional constraint, supporting the account that uncertainty governs deployment of attention to both space and objects. Overall, the series of experiments included in this dissertation provide novel hypotheses, offer experimental evidence for the proposed hypotheses, and furnish evidence necessary for re-framing current theories of attentional guidance. Using experimental evidence, I demonstrate that attention is a dynamic system contingent on the state of the visual environment and the observer, as well as both top-down and bottom-up properties of objects. Uncertainty about environmental factors including space and objects, as well as real-world object size and identity, all affect attentional allocation, indicating that new models of attention must incorporate these factors.