Neural Mechanisms of Object-Based Attention

Cerebral Cortex April 2015;25:1080–1092 doi:10.1093/cercor/bht303 Advance Access publication November 11, 2013 Neural Mechanisms of Object-Based Attention Elias H. Cohen and Frank Tong Psychology Department and Vanderbilt Vision Research Center, Vanderbilt University, Nashville, TN 37240, USA Address correspondence to Dr Elias H. Cohen. Email: Keywords: fMRI, fusiform face area, human visual cortex, multivariate pattern analysis, parahippocampal place area, visual attention Introduction According to prominent theories of object-based attention, the attentional system is predisposed to select entire visual objects during top-down enhancement (Duncan 1984; Kahneman et al. 1992; Baylis and Driver 1993; Blaser et al. 2000; Driver et al. 2001; Scholl 2001). The ability to enhance the visual representation of entire objects, even in the presence of spatially overlapping distractors, may be particularly useful for distinguishing objects in cluttered real-world scenes (Peelen et al. 2009; Cohen et al. 2011). For example, consider a predator attempting to identify its prey hiding in a thicket of ferns. In such situations, object-based attention could be used to selectively enhance the relevant portions of the image belonging to the partially hidden animal, and to suppress information from competing objects, such as the leafy branches that lie before or around the attended object. Most neural investigations of object-based attention have relied on simple stimuli, such as intersecting lines, simple shapes, or overlapping sets of moving dots, which can be readily segmented and perceptually organized based on their spatiotemporal continuity. These studies suggest that top-down feedback to early visual areas is important for the attentional selection of simple objects or perceptual groups (Roelfsema et al. 1998; ValdesSosa et al. 1998; Blaser et al. 2000; Muller and Kleinschmidt 2003; Schoenfeld et al. 2003; Fallah et al. 2007; Ciaramitaro et al. 2011; Hou and Liu 2012). © The Author 2013. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: However, real-world stimuli such as people, vehicles, or buildings are far more complex in their featural and spatial characteristics. Correspondingly, a more sophisticated mechanism appears necessary to explain how top-down attention can enhance the representation of a complex object when it appears in the presence of a competing overlapping distractor. In this case, object-based selection would need to be informed by high-level knowledge regarding the detailed visual structure of the attended object; otherwise, there would be little basis for distinguishing the features of one object from those of another under conditions of spatial overlap (see Fig. 1a). Only a few studies have investigated this more challenging form of objectbased attentional selection, focusing on the modulatory effects of attention in high-level object areas and the activation of frontoparietal control networks during this top-down selection process (O’Craven et al. 1999; Serences et al. 2004; Furey et al. 2006). However, recent work by Al-Aidroos et al. (2012) has provided evidence to suggest that feedback to early visual areas may also contribute to the attentional selection of complex objects. They found that the functional connectivity between category-selective object areas and early visual areas was reliably modulated, depending on whether participants were attending to faces or scenes presented under conditions of spatial overlap. These findings suggest a possible role for early visual areas in the attentional selection of complex objects; however, it is unclear what types of visual signals might be enhanced in these early areas to mediate this selection process. The goal of our study was to determine whether objectbased attention might rely on pattern-specific feedback to early visual areas to selectively enhance the set of low-level features corresponding to the attended object. Although early visual areas are primarily tuned to local features and insensitive to complex object properties, we hypothesized that attending to 1 of 2 overlapping objects may depend on selectively enhancing the visual representations of the local features corresponding to the attended object. This hypothesis leads to the following predictions. First, when covert attention is directed toward 1 of 2 overlapping objects, activity patterns in early visual areas should be biased toward the pattern that would result if the attended stimulus were presented in isolation. Such a prediction can be viewed as an extension of the biased competition model (Desimone and Duncan 1995). Second, if feedback to early visual areas contributes to the attentional selection of object-relevant signals, then the strength of this pattern-specific attentional bias signal in early visual areas should be predictive of the strength of attentional modulation found in high-level object areas. Such functional coupling would imply that early-stage attentional filtering can determine the quality of object-selective information that ultimately reaches higher level visual areas. Finally, we predicted that attentional modulation in early visual areas should be reliant upon high-level object knowledge, such that relevant features What neural mechanisms underlie the ability to attend to a complex object in the presence of competing overlapping stimuli? We evaluated whether object-based attention might involve pattern-specific feedback to early visual areas to selectively enhance the set of lowlevel features corresponding to the attended object. Using fMRI and multivariate pattern analysis, we found that activity patterns in early visual areas (V1–V4) are strongly biased in favor of the attended object. Activity patterns evoked by single faces and single houses reliably predicted which of the 2 overlapping stimulus types was being attended with high accuracy (80–90% correct). Superior knowledge of upright objects led to improved attentional selection in early areas. Across individual blocks, the strength of the attentional bias signal in early visual areas was highly predictive of the modulations found in high-level object areas, implying that pattern-specific attentional filtering at early sites can determine the quality of objectspecific signals that reach higher level visual areas. Through computational modeling, we show how feedback of an average template to V1-like units can improve discrimination of exemplars belonging to the attended category. Our findings provide a mechanistic account of how feedback to early visual areas can contribute to the attentional selection of complex objects. Materials and Methods Participants A total of 10 healthy observers, aged 23–32, participated in one or more of the following experiments, with 6 observers in Experiment 1 (observers 1, 2, 3, 4, 5, 6), 5 observers in Experiment 2 (1, 2, 4, 5, 7), 5 observers in Experiment 3 (1, 3, 7, 8, 9), and 5 obse (...truncated)