Rips (1989) is one of the most widely cited demonstrations of the relevance
of critical features in conceptual structure. A critical feature will impo
se constraints on how an object is classified, regardless of other factors,
such as overall similarity. We review and clarify critical features propos
als in terms of whether an assumption is being made about necessary feature
s, sufficient features, or both (i.e. a definition). On the basis of this,
we scrutinize Rips' investigation and present our own empirical re-evaluati
on of this work. The results reported here do not support any strong view o
f necessary or sufficient features.