This would be dependent on the receiving mail system.

If you send an email to my server from an email address that does not exist, then my mail server rejects the connection to send the email.

For example, if you send an email from user1@domain1.com to user2@domain2.com, the domain2.com mail server actually asks domain1.com if it has a user by the name of "user1" and if it does not, domain2.com rejects the connection request.

This process also works a bit like a tar-pit because it slows the acceptance of emails -- thus flushing out many of the spammers.

Of course, if the spammer injects a valid "from" address into the email, then this can still get through and that's where your spam filters kick in...

Just some more food-for-thought.