Mismatches between Represented Science Content and Unmet Expectations as a Mechanism of Model Revision