Smith, E. E., Shoben, E. J. & Rips, L. J. (1974). Structure and process in semantic memory: A featural model for semantic decisions. Psychological Review, 1, 214-241.
According to the probabilistic view concept is a summary description of features of all objects representing the given concept. The given object described by the concept may have the given feature of this concepts with a given probability. In other words the probabilistic view assumes the concepts are the sets of some features associated with the probability of their occurrence.
Smith, Shoben, Rips following their researches assumed that the critical features may be specified among all features of the given concept. Critical features are fundamental for recognition of the object belonging to the given concepts (categorization) because they constitute a “core” of the concepts and all objects belonging to the given concepts may be characterized with those critical features. Authors assumed also existence of the characteristic features indicating how typical object is.
The second issue raised by the authors is the problem of the comparison between the category and a particular specimen. According to authors such comparison is made within two steps. In the first step the category and the specimen are compared entirely. It means that all features: critical and characteristic are taken into account. If the specimen has all or almost all critical features and more of characteristic once then the positive answer is given for the question on similarity. In the second step assumes detailed comparison consisting in comparison of the critical features only. The idea of this step is to check the proportion of critical features of the specimen towards all critical features of the category compared.
The article is the classic one on the grounds of cognitive psychology. The most valuable point of the article is observation on definitions of the categories, as consisting in set of features. Probabilistic view underline one more important thing, probabilistic relations between the category and its features. That symbolic version of connectionist or semi-connectionist weights between chunks is just those probabilistic relations.
The early work described above in perceptual and conceptual categorization assumed that categories had critical features and that category membership could be determined by logical rules for the combination of features. More recent theories have accepted that categories may have an ill-defined or “fuzzy” structure and have proposed probabilistic or global similarity models for the verification of category membership. It was difficult to believe that there is any biological distinction between some features describing object indeed.