The problem considered in this paper is one of extracting perceptually relevant information from groups of objects based on their descriptions. Object descriptions are qualitatively represented by feature-value vectors ...