Normative uncertainty for non-cognitivists

Normative judgments involve two gradable features. First, the judgments themselves can come in degrees; second, the strength of reasons represented in the judgments can come in degrees. Michael Smith has argued that non-cognitivism cannot accommodate both of these gradable dimensions. The degrees of a non-cognitive state can stand in for degrees of judgment, or degrees of reason strength represented in judgment, but not both. I argue that (a) there are brands of noncognitivism that can surmount Smith??s challenge, and (b) any brand of non-cognitivism that has even a chance of solving the Frege?CGeach Problem and some related problems involving probabilistic consistency can also thereby solve Smith??s problem. Because only versions of non-cognitivism that can solve the Frege?CGeach Problem are otherwise plausible, all otherwise plausible versions of noncognitivism can meet Smith??s challenge.