Abstract
Mamassian & de Montalembert (2010) made direct comparison between the vertical and horizontal components of simple figures, an inverted ‘T’, a horizontal ‘T’, an L-shape and a cross. They proposed a model suggesting two components, anisotropy which makes a vertical line look 6% longer than a horizontal and a bisection effect which makes a bisected line 16% shorter. At VSS (2011) we reported a study of the Oppel-Kundt illusion in which we did not find any bisection effect. We have now examined anisotropy and bisection effects in separate experiments. Comparing the lengths of simple vertical and horizontal lines we find that verticals are perceived to be 7% longer, a figure close to that of Mamassian & de Montalembert. We have extended our observations to the same group of figures matching only components of the same orientation. Our results show no evidence for Mamassian’s bisection component. Rather we propose a 3-component model to explain line-length estimation in these simple figures: anisotropy, abutting and crossing. Abutting occurs when the end of one line touches another. We used the method of constant stimuli to generate psychometric functions from which PSEs were determined. We varied the length of a vertical or horizontal comparison line between 5.2-7.0 degrees of visual angle in a single interval 2AFC judgment. In an experiment comparing the vertical line in an inverted ‘T’ figure with a vertical comparison line, the former was judged 9% longer by virtue of this abutting effect. In an experiment to compare the vertical line in a cross figure with a vertical comparison line, the former was judged to be 7% shorter by virtue of the crossing effect. Finally, we confirmed our model by testing additional simple figures, such as the horizontal ‘T’ and L-shape.
Meeting abstract presented at VSS 2012