Two models, one originating from an artificial-intelligence paradigm and the other from database research, that incorporate connectionist techniques into their knowledge representation and reasoning processes are described. The first approach, called evidential reasoning, is based on semantic networks and focuses on solving inheritance and recognition queries using a rich internal structure. The second approach, called the associative relational database, provides a query language to manipulate knowledge stored in simple uniform structures. In addition to solving ordinary information retrieval, associative databases support robust retrieval with imprecise queries, which is impossible in traditional databases. The two modeling techniques are compared.