If you know anything about network security and IT management all this stuff is encypted and even Sony doesn't know what the question and password is. They don't want to know all they can do when you forget a password or they think it might of been compremised is reset your password to some randomanly generated thing sent you your email. Which will be the likely course of action that when PSN comes up everybody will start getting emails to change their password it has been reset.
Somewhat true, but not quite...
Most enterprise systems store user accounts and passwords in either an LDAP type server (Active Directory (MS), Open Directory (Mac), OpenLDAP, Oracle DSEE, et al) or an RDBMS system (Oracle, IBM DB2, MySQL, Postgres, et al). There are probably a few cases of bespoke systems -- which also leads to the argument that an RDBMS implementation is essentially a bespoke implementation of an authN/authZ system).
Anyway, a standard method of storing user passwords would be to use a one-way hash algorithm. Most LDAP systems, for example use a SHA variant. RDBMS systems generally encrypt fields with a two-way encryption, but can also store hashed values in a field.
Depending on how Sony implemented their identity system for authN, generally speaking, once you get the password file, you can run dictionary attacks on it to figure out the cleartext version of the hashed values. This is a brute force method that works because you don't need to go through the front-end systems which log and may indicate an attack to the service provider.
For symmetrically encrypted data, if you have access to their database files, you may also have access to their code. Once you have access to their code, you can get the symmetric key by examining the code (even in compiled format).
As for challenge questions and answers, you generally need to store that in a manner you can present the end-user. That means typically you store it plain-text, or even if you encrypt it, it's usually a symmetric encryption so that you can retrieve it. Even if it isn't, and it's hashed, again... a brute force attack on challenge answers would be very simply, because they are mostly based on real words... I typically treat even my challenge questions as passwords and use mixed numbers, letters, cases, and symbols -- with long lengths. I also never use the same questions/answers on any sites, and make up answers for questions. I'm paranoid, but I work in identity management these days.
Simply put... if your security is based mostly upon the concept of securing your servers from unauthorized login access, when you lose that security, you've lost the whole house.