I don't know how it is in other fields, but in Mathematics and Theoretical Computer Science, any questions pertaining to the paper should be addressed to the corresponding author, that is if you don't have particular reasons for contacting any of the other authors.
And whether you should contact the corresponding author depends on many things. If there is no other way, or if it is, e.g., much more invloved, to get an answer to your question, then yes, contact the corresponding author. However, if the information is easily available out there, then perhaps you will just be wasting the time of the corresponding author.
If you are fairly certain you've found an error in the paper, then I believe you should definitely try to get in touch with the corresponding author.
Also, be prepared to not get a reply. Corresponding authors are often too busy.