Ideal-observer models of perceptual contrast enhancement

摘要: In standard ideal-observer models of sensory cue integration [1], the perceptual estimate resulting from combination two cues lies in interval bounded by estimates each separately. For example, this type model accounts well for psychophysical result - observers give an in-between haptic-alone and visual-alone estimates, when asked to ridges height with both vision touch [2]. Nevertheless, a class illusion is supposedly not accounted model, namely contrast illusion, such as size-weight [3,4]. weight objects same mass but size, larger lighter. Using models, we showed that it possible account provided statistical correlation between taken into account. Our argument based on inference linear minimum-variance unbiased estimation, maximum posteriori least relative surprise estimation. This general long deals physical quantity proportional another also available cue, volume given material illusion.