Abstract

The perceptual changes (PCs) associated with viewing an ambiguous figure like the Necker cube (NC) can be considered as a series of independent events in time. The time between successive PCs is then percept duration, and the frequency distribution of percepts plotted against duration is adequately modelled by a gamma distribution (Borsellino et al, 1972 Kybernetik 10 139 - 144). A logarithmic transformation on the duration scale enabled us to model the subsequent distribution with a Gaussian. The cumulative Gaussian fitted to the normalised ogive allowed simple comparison of observer performance in different conditions. We found no differences between ogives generated for the standard 'cube-up' and 'cube-down' 3-D perceptual alternatives. When observers recorded 2-D perceptual alternatives, the ogives were shallower and shifted to the right. This suggests that the neural mechanisms governing PCs between 3-D and 2-D perceptual alternatives differ from those that govern the PCs between the opposing 3-D perceptual alternatives of 'cube-up' and 'cube-down'. Further, we show that the same analysis can be applied to reaction times generated in an inhibition-of-return (IOR) task. We argue that this form of analysis is superior to other more commonly reported methods for both attentional and PC data.