The intrusive influence of task-irrelevant representations on attentional selection Open Access
Downloadable ContentDownload PDF
Selective attention is the cognitive mechanism through which a subset of sensory input is selected and prioritized for further processing. Years of research have elucidated the effect of task-relevant information as well as salient events on attention. However, considering that most of the information that we are surrounded by, is in fact task-irrelevant, it is important to understand whether task-relevance modulates attentional influence. In this dissertation, three separate sets of experiments investigated the influence of non-salient, task-irrelevant information on attention. First, the influence of task-irrelevant mid-level representations on attentional shifts was examined by manipulating the retinal size of objects. The results show that attentional shift is modulated by object size, with faster shifts towards thicker than thinner objects, demonstrating that mid-level object properties modulate attentional selection when task-irrelevant. Second, whether the semantic property of a scene and its relationship with objects influences attentional selection when task-irrelevant was investigated. It is shown that high-level associations between real-world scenes and objects guide attentional allocation when task-irrelevant, with a benefit towards congruent associations and the amount of facilitation was dependent on the individual strength of association between a scene and object. This demonstrates that high-level semantic associations influences attention in a continuous manner, independent of task-relevance. Third, using neuroimaging methods, the mechanism through which task-irrelevant semantic information influences attentional selection was investigated. The results demonstrate that task-irrelevant semantic information influences spatial priority maps in the intraparietal sulcus (IPS), subsequently modulating early visual cortex (EVC) activity. Overall, the series of experiments included in this dissertation provide behavioral and neural evidence of the continuous influence of mid-level and high-level information on attentional allocation and propose that existing models of attentional guidance must be re-framed to incorporate such intrusive influence.