Independent effects of lighting, orientation, and stereopsis on the hollow-face illusion

Three experiments were conducted to investigate factors contributing to the 'hollow face' illusion. A novel method was employed in which the distance from the mask at which the illusion became apparent or disappeared, when retreating or approaching, respectively, was taken as a measure of the strength of the illusion. In all the experiments an effect of direction of observer's movement was found, demonstrating the stability of the initial percept. Upright orientations were compared with inverted ones to investigate if the illusion reflects a bias towards a familiar percept. The direction of lighting was also varied. Independent main effects of orientation and lighting were found to be consistent with preferences both for upright faces and for top lighting. However, inverted faces also produced the illusion to some extent, suggesting a general preference for convexity. The role of stereopsis in resolving the illusion was tested by comparing monocular with binocular viewing conditions. Monocular viewing conditions gave rise to shorter distances, suggesting that the retinal disparities available with binocular viewing are important in disambiguating the illusion at small distances. The results are interpreted within the framework of a 2 1/2-dimensional sketch derived from independent modular processing of visual information.