The present experiment was designed to examine the plasticity of mental color codes. On each trial, subjects judged whether two color chips were physically identical. On primed trials in the chip condition, the prototype color chip was presented for 2 sec prior to the test pair. In the name condition, the category name was presented prior to the test pair. In one context condition, all test pairs involving moderate goodness levels were presented prior to the presentation of more extreme test pairs. In the other context condition, subjects were initially exposed only to the extremes of category membership. For same judgments in the name condition, initial exposure to the extremes of category membership produced priming effects that were restricted to good examples of the color category, whereas initial exposure to moderate goodness levels extended priming effects to all goodness levels. The relationship between priming and goodness level did not vary with test order in the chip condition. It was concluded that the nature of the mental representation generated to a category name can be readily modified by context.