2026/05/18 by Damian Koevoet · 1 voice
Computer Science · Neuroscience · #Visual Attention and Saliency Detection #Visual perception and processing mechanisms #Neural and Behavioral Psychology Studies
paper · doi:10.33540/3471
openalex publication_date 2026/05/18 · openalex created_date 2026/05/19 · openalex updated_date 2026/05/21
Humans move their eyes 3-4 times per second to select visual information. Therefore, the decision of what to select is one of the most frequent decisions the brain is faced with. What determines what is selected? Extending extant, tripartite models that consider selection to be the integrative outcome of goal-, stimulus- and history-based factors, I propose a quadripartite model that additionally predicts intrinsic attentional costs to drive selection. In a series of studies, I quantified the intrinsic attentional costs between different eye movements using pupillometry. Crucially, such costs predicted later eye-movement behavior – even when salience attracted gaze and during natural viewing. Moreover, a combination of pupillometric and neural decoding analyses demonstrated that preparatory overt attentional shifts were more costly than covert shifts, likely because an additional neural process is recruited during the former. To test a full quadripartite model, I also investigated the interplay between goal- and history-driven selection, and the neural dynamics underlying goal-driven selection using neural decoding techniques. Besides cost-driven spatial selection, a costs account also predicts that the intensity of attention is considered during cognitive tasks. Using working memory as a test case, I showed that pupil size tracked the intensity of distinct attentional processes. This makes pupil size a promising tool to investigate whether the intensity of attentional processing is titrated to minimize costs. Together, this dissertation builds the foundation for a quadripartite model of visual selection that considers cost-driven selection.