Abstract
Objects that are highly visible can be rendered unrecognizable in the presence of nearby features, a phenomenon termed crowding. The real-world significance of crowding is often illustrated with hazard detection failures. However, real objects usually occur at differing depths, which has been shown to alleviate crowding in some studies but increase it in others. This lack of consensus on the importance of depth on crowding limits our ability to answer the question: is crowding significant in the real world? We developed a multi-depth plane display consisting of three screens at different depth planes (near = 40cm, mid = 1.26m, far = 4m). Stimuli were aligned either with the observer’s cyclopean point or their dominant eye. Observers adjusted the orientation of a foveal Landolt-C to match that of a randomly oriented C presented at 10° eccentricity and flanked by a 4-sector ring at the same or different depth plane. Crowding was quantified from the circular variance of matching error. Crowding significantly increased with flanker proximity at all depth combination and there was no release from crowding in any of the depth combinations tested. There was however a trend for more crowding when flankers were presented behind target depth compared to in front of, or at target depth. However, overall, this trend was only significant for cyclopean-aligned stimuli. Follow up experiments revealed that the increased crowding by far flankers for cyclopean-aligned stimuli was the results of perceptual occlusions. The results demonstrate that crowding impairs performance across depth and is therefore significant in the real world.