As far as I can tell, we have no consistent stance or policy on this one.
Per our proposed/staged review guidelines:
Edits that change an answer's explanation or code, with a supposedly "better" alternative. Even if the proposed solution is better, it should be added as a comment, or a separate answer. Reject as "radical change" or "invalid edit". If the edit summary specifically mentions that these changes are from a comment by the author of the post, the edit can be accepted.
Regarding edits that change the author's point of view, Shog9 argues that it's okay to make technical corrections if the author generally got it right. (Less so on questions because it's much harder to tell whether the correction in question is part of the OP's problem.) But wrong answers should not be rewritten.
In a similar but slightly differently colored question, Shog9 takes a more conservative stance and argues to consider the author's inadvertency v. intention. In both of Shog9's answers the message seems to be if the author was wrong (in editor's view), let it be and downvote and comment. But the vast middle ground of the author was partly right but perhaps didn't consider this new viewpoint that may improve an answer - see my answer on which this happened today - is unaddressed.
Per some random question here, it's okay to approve substantive changes.
We edit the shiznit out of questions because they're often really bad until we do so. Usually everyone's okay with this, since the edit is a one-way improvement that only increases the question's chance of receiving support.
Per the FAQ, of course, other people can, in fact, edit your posts.
It's been discussed in this thread, but I can't seriously consider it authoritative since it's > 4 years old and we've had a lot of movement on this topic since then.
Normally I would propose a solution here but... I've seen both cultures multiply on meta.SO. I lean toward answers are wikis and should be improved and it's better to have fewer cohesive answers than many split ones, when people are willing to act reputation-blindly. And when reviewing edits I've just skipped many of these, and given that I'm 5k on meta and 10k on SO, you'd think I'd be more informed of our policy. So I at least need guidance and I'd to ensure there's a consistent policy laid out.
For reference, here's a discussion on the question-side of things, although in practice we seem to be on the same page that questions can be rewritten until they are of quality, probably just because everyone wins so there's no need for theoretical discussion and justification.