A maximum likelihood, nonmetric multidimensional scaling procedure for word sequences obtained in free-recall experiments

A maximum likelihood nonmetric multidimensional scaling procedure is developed for word sequences obtained in free-recall experiments in order to spatialy represent the structure of semantic memory. A Monte Carlo simulation showed that this procedure can reproduce given configulations, and a tentative application to empirical data (free-recall data of historical personages by the author as the subject) gave promising results. The procedure has two aspects: mathematical modeling of free-recall and the application of the constant utility model (Luce & Suppes, 1965) to multidimensional scaling. Relations to other maximum likelihood MDS’s and the SAM model for memory retrieval (Raaijmakers & Shiffrin, 1981) are discussed.