OO is one toolkit for producing useful representations. It is a useful one (else it would not have become so popular), but it is not the only one, and it is not always the right one.

If a tangent may be allowed here. I think discussing how OO can usefully model the problem space is the wrong angle. Very few problem spaces actually map directly or naturally to an OO model (simulation being one reasonable exception). OO has more to do with how we model our programs than how we model our problem space. Another way, OO is more about how we structure and manage our models, than about how we model.