Purchase this article with an account.
Ellis Gootjes-Dreesbach, Peter Scarfe, Andrew Glennerster; Detecting 3D location change in the presence of grouping cues. Journal of Vision 2018;18(10):503. doi: https://doi.org/10.1167/18.10.503.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Blindness to image change is well documented, but detection of 3D location change (as opposed to 3D motion) is less well characterised. Very large changes in the size of a scene can go unnoticed (Glennerster et al, 2006) but this is not something that a computer vision system could detect from the input images alone. Here, we measured the detectability of changes in the 3D location of an object relative to other, static objects when observers did not know which object was going to move. 8 spheres were presented in virtual reality (arranged in two groups at distances between 2.5-7.5m); the task was to detect which one moved (by 2m, towards or away from the observer). We have shown (VSS 2017) that joining pairs of spheres with lines ('dipoles') dramatically reduces the ability of participants to detect the moving object if (and only if) the dipoles switch between intervals to join different pairs of spheres. This could reflect a general image change or else be due to a change in grouping of the spheres. We tested this using colour to define groups rather than dipoles. Half the spheres had one colour, half another. The colours were either reversed between intervals (no grouping change) or the colour-defined groupings changed. All colour changes were irrelevant to the 3D location change task. We find (both at individual and group level) that these colour changes, of either type, have no significant effect on performance. This is true despite 95% a priori power to detect an effect of the same size as in the original experiment. Our results in the 3D domain are closely related to findings using 2D images (Jiang et al., 2004, showed that rotation of elongated axes drawn through objects, but not colour changes, disrupted performance).
Meeting abstract presented at VSS 2018
This PDF is available to Subscribers Only