The absolute and spectral sensitivities to light by the human circadian system, measured through melatonin suppression or phase shifting response, are beginning to emerge after a quarter century of active research. The present paper outlines a hypothesized model of human circadian phototransduction that is consistent with the known neuroanatomy and physiology of the human visual and circadian systems. Spectral opponency is fundamental to the model, providing a parsimonious explanation of some recently published data. The proposed model offers a framework for hypothesis testing and subsequent discussion of the practical aspects of architectural lighting with respect to light and health.