Pavlovian associations are forever: On classical conditioning and extinction