Imagine that you’re talking to a police officer about a suspicious person reported to have had a knife. You say: “The man who had the knife also had a gun.” But what if the same suspicious person, doing the same thing, was instead reported as having had a gun? You may have instead said: “The man who had the gun also had a knife.” Both facts are true from your subjective perspective. The man had both a knife and a gun. But what you say to the police officer—assuming, of course, that you want him to know both facts—depends on what you know about what he knows about.

Interestingly, though: If we take out of the analysis the speaker and the listener, along with their differing subjective perspectives (e.g., the speaker knowing X but the listener not), and we just imagine the scene itself (of the man with the knife and the gun) from an omniscient, objective perspective, then there’s no possible distinction between what we’d imagine between the original sentence and its flipped-around counterpart. Whether “the man who had the knife also had a gun,” or “the man who had the gun also had a knife”—whether the speaker is making one assumption about the listener or another—the two omnisciently, objectively imagined scenes are identical: There was a man of a certain kind. He had both a knife and a gun, both of them of a certain kind. And he was doing certain things, e.g. walking in a certain way.