Sufficient eye movement coverage of the 2D image plane might mediate under-exploration in 3D search
Year: 2023
Type: article
Abstract: Introduction: Radiological and airport security decisions often require searching and scrolling through a 3D volume, rendered as a stack of 2D images ("slices"). Humans under-explore 3D volumes, potentially causing many search errors for small targets that are salient when foveated but difficult to see in the visual periphery (Lago et al., 2021). It remains unclear why observers under-explore during 3D but not 2D search. We hypothesize that the proportion of area explored in the 2D image plane leads to the termination of the 3D search. Methods: Six observers completed the 3D search (50% prevalence) for a small (0.04 dva) and large (0.44 dva) target embedded in 1/f noise in three conditions with the same volumetric area but different sized dimensions: 24.4 x 18.2 dva, 100 slices; 12.2 x 18.2 dva, 200 slices; and 12.2 x 9.1 dva, 400 slices. Observers also completed a 2D search task with the same rectangular areas as the 3D conditions. To compute search coverage, we derived the radius of a circular Useful Field of View (UFOV) for each target by measuring retinal eccentricity that produced 93% detection (large = 3.3 dva, small = 1.2 dva). Results: The proportion of the 2D image plane covered by the observers’ UFOV was comparable between the 2D and 3D searches for both targets across the three conditions. For the large target, the mean ratio (3D / 2D), across observers, of the proportion of the planar area covered for the 100, 200, and 400 slice conditions were: 1.13 (standard error = 0.06), 1.13 (0.05), and 1.1 (0.02), respectively. The means for the small target were: 1.34 (0.13), 1.25 (0.09), and 1.2 (0.08). Conclusion: Humans might be using the area covered in the 2D plane to terminate their 3D search, leaving much of the volume unexplored.
Source: Journal of Vision
Authors: Devi Klein, Miguel P. Eckstein
Institution: University of California, Santa Barbara
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: gold
Estimated APC paid: $1,850
OpenAlex ID: W4386247609