The Structural Memory: A network model for human perception of serial objects

Summary The Structural Memory, a network model for human perception of serial objects, such as series, is presented. Our theoretical assumptions originate from the Structural Information (Theory (SIT) (Buffart & Leeuwenberg, 1983; Leeuwenberg, 1969), a theory concerning the perceptible structures in an object and the human preference for one of these structures. A symbolic notation of such a perceptible structure in an object is called a representation. For a given maximum number of symbols we can generate all representations automatically. From this procedure we define G(eneration)-relations between the representations. We also define S(tructure)-relations based upon the structures described by the representations. The representations and relations can be seen as respectively the nodes and links in a network. This network is the basis for the Structural Memory. We therefore assign an activation value to each representation in the network, expressing the strength of the preference for the described structure at a certain moment. By means of a process model we are able to make predictions for the strength of the preference for a perceptible structure in an object. The process is only based upon the network structure, because of a relation found between the preference measure, used in SIT, for a perceptible structure in an object and the number of G- and S-relations of the corresponding representation in the network. It is shown that with two of these process models some experiments by Van Leeuwen and Buffart can be simulated.