And, actually, I think you will find reflections a strictly linear relationship. The reason why it appears reflected light appears to drop off more quickly on a darker object has to do, I believe, with the non-linear response of the human eye, and with perceptual cues of surroundings and contrast. 3d is currently a bit hampered by the way caustics and global illumination are treated separately -- this both for historic reasons, and for the simple reason that machines are not yet fast enough for a true physical optics solution. Global illumination/radiosity solutions are generally heat-transfer models, with or without occlusion. Caustics are generated specifically on reflective surfaces within direct view of a light source. And of course in true physical optics, Subsurface Scattering, fog/haze diffusion, "visible" light cones, and glow are all interrelated aspects of one single system. In 3d, we are still trying to ape the real optics with various simulations. In some places, there are no optimal work-arounds (for instance, the prism effect which is in large part responsible for the look of cut crystal and many gemstones).