I know that it does in PHP, and I'm pretty sure it does in Java. I haven't used the latest versions of .NET, so I won't speak for them. It seems very awkward, but I was wondering if there was an underlying reason for this.
One reason is compatibility - anyone who has done 'check for existence' knows to exclude directories; changing that behaviour may confuse those who rely on that behaviour.
Secondly, the underlying code often does a check on the operating system for existence in a catlog of filesystem entries - to the OS, a directory is the same as a file. In other words, it's looking for an entry of 'xyz' in the catalog not a file with name 'xyz' in the catalog.
Backwards compatability is the main reason, I suspect.