Using books to prime behavior — LessWrong