Abstract
Numerous visual effects suggest that scene statistics are internalized by the human visual system and priors appear to be updated constantly (Adams et. al. 2004). Burge et al. (2008) found a statistical relationship between convexity and metric depth in natural scenes and showed that the visual system behaves as if it exploits this relationship. If the statistical relationship between convexity and depth changes does the visual system learn that change? Can haptic feedback adapt the influence of convexity on visual depth perception? To investigate, we designed an experiment with three phases: a pre-adaptation phase in which the effect of convexity on depth percepts was measured, an adaptation phase in which haptic information about the depth step was provided and a post-adaptation phase in which any change in the effect of convexity on depth perception could be assessed. In the pre-adaptation phase we replicated the results of Burge (2008). Observers were shown different pairs of consistent and inconsistent displays in a 2IFC paradigm and were asked to select the interval in which they saw the less separation between the near and far regions. For the adaptation phase (monocular), our subjects were divided into two groups: group A received haptic feedback indicating that the concave region was always near, while group B received haptic feedback indicating that the convex region was always near. Afterwards, we re-measured the visual effect of convexity on depth perception and found that the adaptation with concave region near reduced the effect (i.e. convexity was now worth less disparity) while the adaptation with convex region near increased the bias (i.e. convexity was now worth more disparity). After one day both adaptations were still present. We conclude that haptic feedback changes the visual perception of depth and that if the statistical relationship between convexity and depth changes this change is learned.