Accession Number : ADA259860
Title : Part 1: An Explanation of How Settling Time and Harmony Can Be Used as Measures of Prior Knowledge. Part 2: Progress on Other Parts of the Project
Descriptive Note : Quarterly rept.
Corporate Author : GEORGIA UNIV ATHENS DEPT OF PSYCHOLOGY
Personal Author(s) : Britton, Bruce K
Report Date : 22 Dec 1992
Pagination or Media Count : 8
Abstract : We measure the person's mental representation by asking them how closely associated are all possible pairs of the important terms in the subject matter area. (2) This gives us an association matrix. In this matrix, we can imagine all the terms listed across the top and along the side, and an entry in each intersection indicating the closeness of association. Another way to imagine this matrix is as a diagram of an associative network, with each term represented as a node, and each association represented by a link between nodes. Each link can be either excitatory, which leads to a tendency for one node to go on when the other is on, or one node to go off when the other is off; or else inhibitory, leading to a tendency for one node to go off when the other is on. (3) We then use this association matrix to simulate the person thinking about the subject, using the spreading activation program as our working model of the thinking process. (4) After the simulation stops thinking, we measure how long it had to think to reach a stable state, and how well its final state satisfies the constraints in its association matrix. If they match well, it has high harmony. (5) Finally, the critical thing for the ONR contract is that there are relatively few ways for the subject to provide a set of links that satisfy the constraints, especially in a large matrix. One of them--the one we are interested in--is by having a well-developed and therefore consistent view of the subject matter area from which the terms come, and this corresponds to a high level of prior knowledge. Another way to get a high harmony is by chance, happening to luck into a consistent set of ratings, but this probability is precisely calculable.
Descriptors : *SIMULATION , *COGNITION , *ASSOCIATIVE PROCESSING , ACTIVATION , PROBABILITY , RATINGS , TIME , NODES , MODELS , NETWORKS
Subject Categories : Psychology
Distribution Statement : APPROVED FOR PUBLIC RELEASE