The feature-comparison model is a psychological theory of semantic memory hypothesized by Smith, Shoben and Rips (1974). The feature-Comparison model was purposed in contrast to the semantic network approach that was purposed by Collins and Quillian. The feature-comparison model is much simpler than the Collins and Quillian network in its structure, however it is also more elaborate in its assumptions about retrieval as well.
The Feature-Comparison model proposes that humans semantic memory works by the use of feature lists. Feature lists assume that semantic memory is a collection of lists of semantic features, which are simple, one-element characteristics or properties of the concept.
An example that exemplifies the Feature Comparison model: When we (humans) think of a concept, like that of a robin, we unconsciously make a list of semantic features (for this example these would be things like animate, red-breasted, and feathered). These feature lists are then (unconsciously)ordered in terms of definingness, or priority, from top to bottom. The most defining features are located on the top of the list, while the lower features on the list are called the characteristic features, and they are features that are common but not essential to the meaning of the concept. In our example, the defining characteristics of a robin match the defining characteristics of a bird (animate, feathered, living), thus we are able to know that a robin is a bird. The more overlapping defining characteristics two things have, the quicker the retrieval will be between the two concepts. So when we are asked a question like "is a robin a bird?", we can quickly decide our answer and say yes because there are so many overlapping defining characteristics. These quick answers are called stage one responses. The reason that they are called stage one responses is because all that needs to be done to answer the question is a quick (still unconscious) feature comparison. If the two different things being compared are not so easily differentiated then a stage two comparison is needed. This stage two comparison is a comparison of defining features, in which a slower comparison is done with only defining features. An example of a question that would cause this would be "Is a chicken a bird", in which the reply is usually a slow "yes".
Evidence for Smith's Feature-Comparison Model come from his paper in which he asked people simple true/false questions and timed their decision time. The decision times matched predicted results for Smith's theories.