In this All Hallow's Eve special, we'll focus on a well-established semantic universal: conservativity. Namely, the fact that "every ghost is spooky" is truth-conditionally equivalent to "every ghost is a ghost that's spooky". All quantificational determiners in natural language exhibit this property and kids can't even learn novel quantifiers that don't. But despite its importance and robustness, we still don't have a satisfactory account of where conservativity comes from. I'll argue that some of our recent work on the mental representation of "every" suggests a new way forward.