Since, in order to survive, we must be able to move about effectively, perception must to a certain degree achieve stable and veridical representations. It must tell us how matters stand out there. But the universe is in constant flux. We move about and other things also move. Day turns into night. Sound sources approach and recede. How can perceptual stability be achieved in the face of the ongoing flux?
We can perhaps formulate a better question by asking what aspect of the universe most needs stability. For example, is it the differences or the proportions and ratios that need to remain constant in perception? Apparently it is the proportions—the ratios. When we walk toward a house, the relative proportions of the house appear to remain constant: the triangular gable looks triangular from almost any distance. A photograph portrays the same picture whether we view it under a bright or a dim light: the ration between the light and the shaded parts of the photograph seems approximately the same even though the illumination varies. The perceived relations among the sounds of speech remain the same whether the speech is soft or highly amplified. In other words, the perceptual domain operates as though it had its own ratio requirement—not a mathematically rigid requirement, as in physics, but a practical and approximate requirement.
The usefulness of perceptual proportions and relations that remain approximately constant despite wide changes in stimulus levels is immense. Think how life as we know it would be transformed if speech could be understood at only a single level of intensity, or if objects changed their apparent proportions as they receded, or if pictures became unrecognizable when a cloud dimmed the light of the sun.
By making the perceived aspects o stimuli depend on power functions of the stimulus dimensions, nature has contrived an operating mechanism that is compatible with the need for reasonable stability among perceptual relations.
Stanley Smith Stevens, Psychophysics: Introduction to its Perceptual, Neural, and Social Prospects, New Brunswick, 1986, pp. 18-19