Abstract
Attention is allocated preferentially according to object boundaries; regions of an attended object enjoy an attentional benefit over non-attended objects. Two theoretical mechanisms have been proposed to account for object-based attentional benefit: (1) Attentional Spreading hypothesis, proposing spreading of attention within an attended object; and (2) Attentional Prioritization hypothesis, suggesting that this advantage stems from increased prioritization given to locations within an already attended object, and that this prioritization is mediated by certainty about the visual scene. Some evidence favoring attentional prioritization has been observed with a flanker paradigm using a cross-like display measuring the degree of interference exerted by flanking letters as a function of whether flankers appear on the same- or different-object (Shomstein and Yantis, 2002). Recent evidence, using a data-limited (accuracy around 70%) cross-like display, provided further support for attentional spreading, arguing that reaction time measures may not have been sensitive to object-based flanker interference effects (Ho, 2011). Here we argue that uncertainty is the factor that drives object-based attentional selection via attentional prioritization, rather than attentional spreading, modulated by segmentation cues signaling object boundaries. We tested different levels of target-based identity uncertainty and its influences on the flanker interference effect by dividing the data-limited cross-like paradigm into three levels of uncertainty: high, accuracies of 60-80%; moderate, accuracies 70-80%; and low, accuracies 80-100%. Additionally, segmentation was manipulated by varying colors of objects with an expectation that segmentation cues should modulate flanker interference effects to a greater extent. We observed that flanker interference was modulated by object representation only under high levels of uncertainty and only when segmentation cues were available. Taken together, these findings indicate that when uncertainty regarding target identity is high, object representations influence attention, however, once uncertainty is eliminated; objects no longer constrain attentional allocation, consistent with the attentional prioritization account.
Meeting abstract presented at VSS 2014