A CONTRAST-DRIVEN AND LUMINANCE-DRIVEN MULTISCALE NETWORK MODEL OF BRIGHTNESS PERCEPTION

A neural network model of brightness perception is developed to accoun t for a wide variety of data, including the classical phenomenon of Ma ch bands, low- and high-contrast missing fundamental, luminance stairc ases, and non-linear contrast effects associated with sinusoidal wavef orms. The model builds upon previous work on filling-in models that pr oduce brightness profiles through the interaction of boundary and feat ure signals. Boundary computations that are sensitive to luminance ste ps and to continuous luminance gradients are presented. A new interpre tation of feature signals through the explicit representation of contr ast-driven and luminance-driven information is provided and directly a ddresses the issue of brightness ''anchoring''. Computer simulations i llustrate the model's competencies.