Karakatsani, Maria Eleni. Cognitive influences in the perceptual brain experiments and computational models for the Ames window illusion. Retrieved from https://doi.org/doi:10.7282/T3CC0XQW
DescriptionThe current study is part of a more comprehensive project that aims to explore potential differences between schizophrenia (SZ) patients and healthy controls in perceiving depth-inversion illusions (DII). Previous work with two types of DII, namely the hollow mask and the reverse perspective illusions, has indicated that SZ patients tend to rely less on experience and stored knowledge, in this case the experience with faces and linear perspective, than healthy controls. The present study explores how healthy controls perform on variants of the “Ames window illusion” that uses humans’ experience of viewing rectangles. The Ames window is based on a rotating trapezoid, typically rotating about a vertical axis that is located in the middle between the vertically oriented long and short bases. Because the trapezoid is perceived as a slanted rectangle, viewers perceive the Ames window illusion, which is a type of DII: the window appears to oscillate back and forth even though it rotates continuously in the same direction. The most plausible explanation is that viewers perceive the inverse depth when the short base is closer than the large base, because of prior experience in viewing slanted rectangles. We investigated the strength of the illusion by using nine computer-generated windows that were displayed on a screen. The nine windows were designed to vary systematically three key parameters: (1) The long-to-short base ratio; (2) the height-to-short base ratio, and (3) the presence or absence of shadows. These stimuli were used in two experiments to assess illusion strength using two measures: (A) Asking observers to report which base was in front at selected instances, signaled by auditory beeps; (B) Asking observers to indicate reversals in rotation direction. The two measures produced results that had a high degree of correlation, thus confirming the validity of the methods. The data were fed to an optimization algorithm for a model that was based on a linear combination of the weights of the three parameters. The model produced results that were significantly correlated with the experimental data. The next phase will involve experiments with SZ patients, based on the results of the present study.