Here is why (sorry if I was vague in my explaination before). With SD channels, the receiver is receiving a 4:3 frame, it can then stretch this frame when you have it set to stretch mode and thus fill your whole screen.
On HD channels though, the broadcasters send a full 16:9 frame, so the letterboxing and pillars you see on most of the HD channels are actually coming right from the broadcaster, not from your receiver. This is very apparent on some stations like NBC that actually add their own logoed pillar bars, but the same effect is on the other channels just they leave them black.
So your box thinks that it is getting a full 16:9 picture and thus cannot stretch it like it can when it gets a 4:3 picture. It does not know that it is black bars, gray bars, or whatever.
Broadcasters do this so that their content is shown in its proper aspect. In your situation though with MTV, that is just because MTV does not broadcast anything in HD yet and thus is just mirroring their SD feed over to the HD channel and adding the pillar bars.
Understand?