Comparing attention estimates across two tasks through formalization

Visual attention selects task-relevant elements from the visual environment, linking perception and action. Because of this intermediary role, it provides a basis for comparing performance in different tasks. In this work, we study whether model-based estimates of visual attention resources – concretely, processing speed derived from a formal theory of visual attention – are comparable across tasks. To our best knowledge, such a comparison on the process level has not been conducted yet. Thirty participants completed two tasks in a within-subject design. For each task, attention resources are estimated using the same theoretical framework, allowing a direct task-to-task comparison. The results show that model-based estimates of attentional processing speed differ between tasks. At the same time, the estimates show different patterns of variability, which suggests that systematic adaptations of one or both models could, in principle, align the parameters across tasks. These findings indicate that current model estimates are task-dependent, but not necessarily incompatible. Identifying adaptation points would move the study of attention beyond task-specific phenomenology toward a more general characterization of attentional mechanisms. Although this paper does not provide a solution to this problem, it formulates the challenge and outlines concrete directions for future research.