Abstract
When we see natural scene images, we experience richly colored scenes. However, when color is removed from a region of the scene images, and then the images are briefly presented, observers often fail to detect the achromatized region (scotoma) (Balas & Sinha, 2007). This blindness to the absence of color can be demonstrated even if a large scotoma is located in a central as well as peripheral region. Kimura & Takahashi (VSS2017) reported that observers missed the scotoma more frequently when the scene images were in normal rather than in complementary color. This finding was apparently consistent with the account that the blindness is mediated by color completion based on color statistics of natural scenes. However, analysis using signal detection theory revealed that a strong bias in the normal-color condition led to the blindness. The sensitivity to the scotoma (d’) was actually higher in the normal- than in the complementary-color condition. These findings suggested that a perceptual/response bias, rather than color completion, mainly mediates the blindness to the absence of color. This study aimed to provide further support for the bias hypothesis. We reasoned that if the bias hypothesis is correct, observers would exhibit better performance for normal color scenes when the contribution of the bias is reduced by using a forced-choice paradigm instead of a yes-no paradigm. However, if the color-completion hypothesis is correct, the performance would be worse for normal color scenes regardless of the experimental paradigm. Results supported the bias hypothesis. When we tested the detection of the achromatized scotoma in natural scene images with a 4-alternative forced-choice paradigm, the observer’s performance was better in the normal- than in the complementary-color condition. However, when we tested the detection of the scotoma in the same images with a yes-no paradigm, the performance was worse in the normal-color condition.