It's different people. One person reads the question and decides to answer it. They may or may not upvote it. A second person comes along and says "what a ridiculous question! It's too broad, or too localized, or offtopic, or horribly written" or whatever. They downvote it. A third person comes along and doesn't vote on the question, but on reading a few answers thinks "yes, that's good advice, or a good point, or accurate history" or whatever so they upvote the answer. And there are a varying number of people who do each of those things.
The overall effect is the same as if one person downvoted the question, answered it, and upvoted another answer, and sure it makes no sense for one person to do that. But a group of people can be contradictory. I think the underlying reason is that we have three quaity ranges - upvote, don't vote, and downvote. Since the borders vary, some people will downvote the question but not the answers, while others will upvote the answers but not the question. The result is the apparently nonsensical score combination.