Hess R F, Dakin S R, Badcock D
McGill Vision Research, Department of Ophthalmology, McGill University, Montreal, Quebec, Canada.
Vision Res. 1994 Sep;34(18):2439-51. doi: 10.1016/0042-6989(94)90288-7.
A spatial localization task was used to determine the accuracy of localization of random element clusters about their centroid. The random element clusters were generated within a circular region and the observers' task was to decide whether the cluster lay to the right or left of a reference line, defined by two vertically separated reference elements. Both the reference elements and the cluster elements themselves were comprised of spatially narrowband stimuli (gabor patches), and were presented at a constant suprathreshold level. The thresholds for localization of the element cluster were measured with varying element number, and for a number of different sizes of circular region and element. The relationship between localization threshold and element number was not monotonic. Thresholds were found to rise with increasing element number up to about six elements within a region, and to then fall with further increase in element number. Asymptotic thresholds at high element number were indistinguishable from those obtained for filled circles of the same size. The results do not conform to any one of a number of models based on a centroid analysis alone. By manipulating the spatial and orientational properties of the elements comprising the cluster to be localized it was determined that the more central mechanism underlying localization receive input from different spatial and orientation filters.
