ajax character encoding issue solved, but WHY?

I had a character encoding issue that I finally solved, but I don't understand why the fix works. I'm hoping someone can explain this to me!

The issue was that non-ascii chars were appearing as junk BUT only when retrieved via ajax calls. Otherwise, they displayed fine. The junk display was due to them being interpreted as ISO-8859-1, but I could not figure out why the browser was interpreting that way. All my data is handled as UTF-8.

The problem was fixed by calling utf8::decode on the data prior to sending back via ajax. BUT WHY?

I am using the JSON view to render ajax responses, and it sets the charset header correctly to UTF-8. Of course, even when you decode, perl still represents as "internal" utf8. But why should this be necessary?

Thanks!