A while back I had the opportunity to hear Kevin Harrigan speak about CLOE, an Ontario-based feeder repository system for MERLOT. The group in attendance had just finished a quick rehash of the old and not terribly illuminating "what is a learning object" debate. Kevin (and CLOE) have a refreshing take on the whole mess.
CLOE focuses not so much on defining learning objects as on defining learning objectives. As Kevin put it, every discipline has certain "stutter points"--certain topical areas where students tend to get stuck. In evolutionary biology, for example, students often hold onto Lamarckian ideas (e.g., the giraffe has a long neck because its ancestor stretched to reach leaves in the high trees) long after they have learned the abstract concept of natural selection. Learning objects are simply educational tools that are created to address those particular stutter points. A learning object is a thingie designed to meet a learning objective.
To my mind, this is a crisp and clean way of describing learning objects so that they are obviously useful to faculty members.