Incrementality presents some challenges however
in terms of the learning algorithm. In particular, the
order in which learning takes place (e.g., whether
rules or exceptions are showed first) may affect the
outcome. If enough information is available (i.e., if
either the learning period is long enough or the
information provided is rich enough), then the system
is likely to reach a state consistent with its actual
environment, even though it may have been
inconsistent at earlier times, as in the above example