Philip Pettit has argued that all semantically basic terms are learned in response to ostended examples and all non-basic terms are defined via them. Michael Smith and Daniel Stoljar maintain that this “global response-dependence” entails noumenalism, the thesis that reality possesses an unknowable, intrinsic nature. Surprisingly Pettit acknowledges this, contending instead that his noumenalism, like Kant’s, can be construed ontologically or epistemically. Moreover, Pettit insists, construing his noumenalism epistemically renders it unproblematic. The article shows that construing noumenalism epistemically prevents Pettit (...) from knowing whether members of different communities respond to different properties in the world or the same properties differently. Pettit then faces a trilemma. He can construe noumenalism ontologically and confront Smith and Stoljar’s charge. He can construe noumenalism epistemically and confront the author’s charge. Or he can reject global response-dependence. After explaining why Pettit should choose the middle horn, the article closes with lessons about global response-dependence generally. (shrink)