Abstract
Introduction: The hollow-face illusion is a depth-inversion illusion: the concave mask is misperceived as convex. Earlier studies employed mostly static stereo pairs of two-dimensional (2-D) images (e.g., Liu et al. 2000) and 3-D physical models (e.g., Yellott & Kaiwi, 1979). There are two extreme renderings: (1) hollow masks with random-dot texture are perceived as hollow (Georgeson, 1979); (2) hollow masks painted as realistic faces appear convex (Papathomas & Bono, 2004). We sought to combine these renderings and modulate the strength of the illusion. Methods: Random-dot (noise) texture and a computer generated texture of human features were digitally scaled and applied onto the 3-D geometry of a hollow mask. These textures were mapped onto the mask as two different layers using weight F(0≤F≤1) for the featural texture and R=1-F for the noise texture. The strength of the illusion increases monotonically as F varies from 0 to 1. The mask was presented in stereo using red-green glasses. Results: Experiment 1: To validate the stimuli, we used several variants of hollow masks (against a blank background, or “anchored” on a frontoparallel plane) and replicated the inversion effect; namely the illusion was significantly weaker when the mask was turned upside-down, as compared to the upright mask, as documented in previous studies (Hill and Bruce 1993, 1994; Papathomas & Bono, 2004). Experiment 2: We tested a hollow mask with F values ranging from 0.0 to 1.0). Our results indicate that the illusion strength increases monotonically, indeed, as F increases, starting with no illusion at F=0.0 and creating the strongest illusion at F=1.0. Conclusions: We have a method for obtaining a psychometric function for the hollow-mask illusion that can be adapted for staircase procedures because of its monotonicity. The method is well suited for perceptual studies with pathological populations, such as in schizophrenia research.
Meeting abstract presented at VSS 2015