I understand from RegDwight answer that there are technical limitations for this, but if there wasn't, I think these counts shouldn't be hidden from users, even from those with 1 rep.
The reason for this is that seeing the distinction between upvotes and downvotes gives you an indication regarding the potential disagreement within the community on that subject. In turn, this gives readers a hint that they should dig deeper and find other sources of information, instead of assuming that the highest voted answers are more likely to be generally correct.
An +2 answer can come from 2 upvotes or 4 upvotes and 2 downvotes. You could have 4 upvotes coming from voters who know a bit about the subject and think the answers sounds right, combined with 2 downvotes from more expert users. (This can easily happen on specific subjects that have low read counts.)
As a newcomer to a subject, it's useful to be able to see there isn't unanimity.
Even for more expert users, it can be useful. Firstly, the expansion of SE sites to more specific areas can also lead to lower reps in each sub-field (since they're not gathered together). Secondly, expert users who are completely new to SE also have to start with 1. If you're an actual expert in field and you see incorrect answers that are highly upvoted on a site, it gives a bad image of the entire site (why bother joining a community where people don't seem to know what they're talking about). Seeing how votes are spread can help convince those experts it's worth joining.