Selective luminance induction on bright and dark regions in textures

It is thought that the “ON” and “OFF” channels, used for detecting luminance increments and decrements, are perceptually inseparable when we process complex textures (e.g. Solomon, Sperling and Chubb, 1993). Here we show that this is not the case. Methods: a 256 grey-levels texture patch (0.5 × 0.5 deg.), with the characteristic second order statistics of natural images, was surrounded by a uniform luminance region of various sizes (up to 4.0 × 4.0 deg.). We have modulated the luminance of the surrounding region sinusoidally in time (at 0.6Hz) either above or below the mean luminance of the central texture. We found that this modulation induced changes in the perceived contrast of the central patch. Subjects were asked to null this induction by adjusting the depth of modulation of either the bright or the dark regions of the central texture. Results: when uniform surround luminance was modulated above the mean luminance of the central patch, subjects chose to null the perceived induction by primarily altering the modulation of the bright regions of the texture. Conversely, when uniform surround luminance was less than the mean of the central patch, subjects chose to null the induction by primarily altering the modulation of the dark regions of the texture. This implies that, in contradiction to previous reports, our perception of bright and dark regions in complex textures is mediated by perceptually segregated ON- and OFF-channels