Incomplete figure perception - the problem of invisible masking

Our previous findings (2002 Perception 31 Supplement, 116) suggested that the Gollin test of incomplete figure perception (Foreman and Hemmings, 1987 Perception 16 543 - 548) and the Mooney test of incomplete silhouette faces may be considered as signal masking problems, in this sense similar to the Poppelreuter test of contour figure extraction from noisy overlapping figures. We considered the incompleteness of a Gollin figure to be the result of its masking by multiplicative noise similar to the texture of transparent windows and opaque elements. The brightness and colour of the opaque elements are identical to the background and size of transparent windows corresponding with the size of fragments of the incomplete figure. In other words, incomplete figures are figures occluded by an 'invisible' mask. The purpose of this work has been to determine the statistical properties of the 'invisible' mask and to show their connection with thresholds of incompleteness of figures at recognition. We have made additional investigations, which show that the incompleteness thresholds depend on the level of similarity of the spatial-frequency bands of the figure and the 'invisible' mask. The thresholds are reduced when the 'invisible' mask changes to a visible one. We presented in psychophysical experiments incomplete figures with two sizes, covering the foveal or macular areas. We measured the thresholds of recognition of incomplete figures in normal participants and in neurological patients with 'filtration agnosia' and showed recognition-threshold differences for the different tests. This difference depends on the similarity between the spatial-frequency spectra of the visible (Poppelreuter) and 'invisible' (Gollin, Mooney) masks with those of the incomplete figure, and also with the localisation of visual system damage.

European Conference on Visual Perception.
Paris, France.
1-5 Sept 2003.
ECVP 2003 Abstract Supplement.