Abstract
After adapting to high-pass filtered (sharpened) images, subsequent images appear to be blurred. Similarly, after looking to low-pass filtered (blurred) images, subsequent images appear sharpened. Until now these two effects were assumed to reflect a response change in the same underlying mechanism. We investigated the characteristics of the adaptation curves as a function of adapting stimuli. Methods: The procedures were modified from Webster et al (2002). Observers adapted (initial 30s, top-up 3sec) to images that were digitally blurred or sharpened by varying the slope of the spatial spectrum up to ±0.50 relative to the natural slope. Observers were asked to decide whether a briefly presented (500ms) test image was perceived “too blurred or too sharp compared to what you think is normal”. Twenty seven observers were tested to determine the point of subjective neutrality (PSN the spectral slope of the image that appears normal). Observers adapted to at least one scene at a minimum of 7 adaptation levels. The PSN for each adaptation level was used to compute individual adaptation curves. Results: Adaptation curves were characterized by the slope of the sigmoid fitted curves (representing the gain of adaptation) and the asymptotes (saturation levels). All observers showed repeatable adaptation effects. Inter-observer variability in the gain of adaptation was found. In addition, within-observer differences were found in the saturation levels to low-pass and high-pass filtered images; some observers showed less adaptation (or even none) to blur than to sharp or vice versa. Conclusions: Adaptation to blurred and sharpened images varies among individuals. Asymmetry between adaptation to blurred and sharpened images suggest a different process for each of these phenomena previously considered parallel.