Marker-Passing and Microfeatures

Abstract

The cognitive modeling community is presently divided between two different approaches to the spread of activation through networks. One school holds that symbolic information must be progagated, the other that numeric weights are used and activation spreads in a more analog manner. In this paper we describe a mechanism which allows the two processes to be merged via the introduction into the symbolic network of a defining-characteristic link which is affects the spread of the symbolic information in a manner resembling local connectionist computations. We demonstrate that the combined system is more powerful than either of the separate models alone.