Abstract
The McCullough effect (McCullough, 1965) can be induced in several minutes by viewing vertical black and green stripes in alternation every few seconds with horizontal black and red stripes. After induction, achromatic stripes appear black and pink if they are vertical or black and greenish if horizontal. When viewed after induction, different stimuli are differentially effective at reducing the effect. Explicit experiments involving natural viewing conditions (Skowbo et al, 1974) as well as the general persistence of the McCullough effect over time show that natural viewing conditions are not especially effective at reducing the effect, which is surprising if the effect is due to recalibration in a functionally important mechanism (Humphrey, 1998). Several features of natural viewing may contribute to this result, such as low power at the relevant spatial frequency and orientation or the presence of multiple frequencies and orientations. We tested two sets of natural images as post-induction stimuli. The two sets were created by finding 38 image pairs. The images within a pair were matched for overall contrast energy, but one image was balanced for contrast energy at vertical and horizontal orientations and the other was strongly imbalanced. We hypothesized that “strongly oriented” natural stimuli would be more effective as reducers of the McCullough effect. We also tested achromatic gratings and white screen stimuli for comparison. Effect strength was measured by color nulling before induction, after induction, and after 20 minutes of exposure to the post-induction stimulus. Results for the two sets of natural images did not differ greatly, so additional experiments with greater power are underway. Natural stimuli were better reducers than the white screen, in contradiction to Skowbo et al (1974), but in agreement with the hypothesis that natural images are what normally keep the mechanism calibrated.