Purchase this article with an account.
Lester C. Loschky, George W. McConkie, Jian Yang, Michael E. Miller; The effects of eccentricity-dependent image filtering on saccade targeting in natural images. Journal of Vision 2002;2(7):170. doi: https://doi.org/10.1167/2.7.170.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We investigated the intersection of spatial vision and attentional selection during free viewing of complex natural scenes. Our method involved the use of a gaze-contingent multi-resolutional display (GCMRD) in which high image resolution is placed only at the center of gaze. Photographic GCMRD images were filtered as a function of their contrast, spatial frequency content, and retinal eccentricity, based on a model of visual sensitivity that had previously been tested using sinusoidal grating patches. This enabled us to understand how attentional selection, as measured by saccade targeting, is affected by the interplay between the physical characteristics of an image, its contrast and spatial frequencies, and the reduced resolution of the human visual system with increased eccentricity. Loschky & McConkie (2001) have shown that reducing image resolution in peripheral vision using a GCMRD can decrease the likelihood of sending the eyes there. However, the GCMRD had a discrete resolution drop-off (i.e., a resolution step function). Thus, our study asked 3 questions: (1) Was the observed saccade shortening with discrete resolution drop-off GCMRDs simply due to the eyes being drawn to the texture boundary between the high- and low-resolution areas? (2) If not, can both detection of peripheral filtering and its effects on saccade targeting be predicted by the contrast sensitivity function (CSF) at varying eccentricities? (3) If image filtering affects saccade targeting, what does it remove that is critical to the saccadic system? We used a GCMRD with a continuous resolution drop-off (i.e., smooth image degradation) and found that saccade lengths were still shortened. However, while detection of peripheral filtering was well predicted by eccentricity-dependent CSFs, effects on saccade targeting had a much higher threshold. Furthermore, saccade targeting was not affected until filtering had removed information close to the peak of the CSF at a given eccentricity.
This PDF is available to Subscribers Only