Three-Dimensional Disadvantage

Article: Three-Dimensional Disadvantage
Source: University of California, Santa Barbara, via ScienceDaily
Published: March 16, 2021

“We’re good at making technology, but sometimes we don’t really connect with it that well,” a senior author of the study comments, “And we don’t know that we don’t connect with it that well.” Researchers in the field of visual search recently conducted a study exploring how human vision processes 2D versus 3D images. The assumption in developing increasingly sophisticated imaging technologies is that detection success should increase with the additional information. However, this is not necessarily the case. The researchers found that we are actually worse at finding small targets in 3D image stacks than in single 2D images. According to the study, observers searching through 3D renderings had higher small target miss rates and significantly decreased decision confidence, though the observers also overestimated how much of the image they explored. Eye-tracking software, for example, showed that subjects conducting the 3D search were looking through only about half of the search area while reporting up to more than 80% image exploration.

This discrepancy lies in part in how we use central and peripheral vision. When searching through 2D images, observers tended to rely on their fovea, which offers the sharpest vision and is used for fixation, while they tended to rely on peripheral vision when searching through the composite images of 3D renderings, and move their eyes less. The combination of eye movement under-exploration, reliance on peripheral vision, and a bit of self-limiting search strategy on the part of radiologist participants in the study resulted in a high number of small target errors in 3D searches. The opposite was found for large targets, where detection improved in the 3D searches; this finding is consistent with use of peripheral vision. In illustrating the gaps between the availability of technology and how our eyes and brains process vision, the study is valuable in informing a better use of technology, for example, to use 2D images for small-target searches and 3D images for large-target searches. Additionally, implementing systems that don't possess these limitations, such as computer vision (e.g., deep learning neural networks), could improve detection success.

My rating of this study: ⭐⭐⭐

Lago MA, Jonnalagadda A, Abbey CK, et al. "Under-exploration of Three-Dimensional Images Leads to Search Errors for Small Salient Targets." Current Biology. 19 January 2021. https://doi.org/10.1016/j.cub.2020.12.029