In search of the M-Gram: The structure of organization in free recall

This study examined the effects of category structure and category size on categorization reaction time, artificial categories being used to obtain greater experimental control than is usually found in semantic memory research. Four artificial categories varying in structure (hierarchical and nonhierarchical) and size (8 or 16 instances) were introduced to 20 adolescent subjects over a 12-week period by means of stories, exercises, and discussions. Significantly longer categorization reaction times were required for instances from the hierarchical categories, but no set size effect was found. The application of clustering and multidimensional scaling procedures to subjects' free recall data revealed that subjects had acquired the hierarchical structures but imposed their own structures on the other categories. These findings were interpreted within a spreading activation framework.Research in semantic memory has relied heavily on the natural memories of subjects, and relatively few studies have attempted to use artificial information. The few include Smith, Haviland, Buckley, and Sack (1972), who used learned artificial facts in order to examine the confounding in the Collins and Quillian (1969) memory organization research between noun-property frequency and number of hypothesized deductive steps, and Potts (1976), who examined the ability of subjects to correctly accept or reject assertions based on previously learned artificial logical relationships. Recently, Murphy and Smith (1982) used artificial category names in order to control for both the length and familiarity of natural category names. None of these experiments, however, has examined the spontaneous generation of a semantic memory structure for unfamiliar stimulus material over a lengthy training period.Awareness of problems associated with the use of natural memories has led to both caution and debate over the control of relevant variables. Landauer and Meyer (1972) discussed the problems encountered in determining word frequency and relative sizesof semantic categories, as well as the confounding introduced by This paper is based on an MA thesis submitted to the University of Newcastle by the first author and supervised by the second author. We wish to thank the subjects for their generous participation in the project for a period of almost 12 months. words with distinctive phonetic markers (e.g., "ology") and categories with an unusually large proportion of short or long words. Anderson and Bower (1973) doubted if factors such as word frequency, conjoint propositional freq uency, recency, and concreteness could be adequately controlled.The debate over measurement of category size illustrates the seemingly intractable control problem using natural memories. Methods of measuring category size can be summarized as being either analytic or subject-based (Wilkins, 1979). The analytic methods are illustrated by the use of nested pairs of categories, as in Landauer and Freedman (1968), and the counting of exemplars in some compendium of English words, s...

Section: Resultsmentioning

confidence: 81%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Categorization reaction time, category structure, and category size in semantic memory using artificial categories

Casey

Heath²

1983

“…While subjective organization measures can indicate when an effect has occurred, they are of little use in determining how to produce a stronger one. One possibility would be to construct a spatial mapping (Friendly, 1977) of the subject's organizational structure from the data available from all previous trials. The materials could then be presented in a way that is maximally consistent with this structure.…”

Section: Resultsmentioning

confidence: 99%

Interactive presentation in multitrial free recall

Friendly

Franklin

1980

This paper explores whether free recall performance and organization can be facilitated using computer-controlled interactive presentation, where stimulus presentation on Trial N+1 depends on the individual subject's prior recall history. A series of experiments investigated the effects of two types of response-contingent presentation order manipulations and of selective tagging manipulation designed to highlight the presentation of previously nonrecalled items. An initial experiment using 20-word lists showed no differences among groups, due to a ceiling effect. Two subsequent experiments, using longer lists (30 and 40 words), found that recall and subjective organization were increased by selective tagging and by presentation orders that preserved the subject's prior order of recall. The results demonstrate that subjects' previous recall histories contain information that can be used to facilitate recall. The potential of other interactive manipulations is discussed. This paper describes a series of initial experiments using an interactive free recall procedure to study memory organization and the retrieval of information from memory. In standard multitrial free recall (SFR), the learner is presented a list of items to memorize, usually in random order, one item at a time. The learner then attempts to recall as many items as possible, in any convenient order. Interactive free recall (IFR) differs from this description principally in that what is pre-

“…One reason is that the use of trees in the representation and recognition of two-and three-dimensional patterns has received considerable attention in the automatic pattern recognition field (e.g., Pavlidis, 1977, Chapter 10), indicating their potential sufficiency as models. Another reason is that trees generally require less "memory capacity" than do networks to represent the same information (Friendly, 1977), although it might take more time to retrieve the information from a tree. A third reason is that it is usually better to resort to a complicated model only after a simpler version of the model has been shown to be inadequate.…”

mentioning

confidence: 99%

Trees as memory representations for simple visual patterns

Cunningham

1980

The form of the memory representation for visual patterns consisting of a few line segments was investigated with three tasks. One task was the reproduction of remembered patterns using pencil and paper. The order in which the subparts of those patterns were reproduced was used to make suggestions about possible memory representations for the patterns. The second task involved rating the "goodness" of subparts of the patterns. These ratings were also interpreted as reflecting the underlying representations of the patterns. In the third task, the time required to recognize that a small pattern was part of a larger remembered pattern was used as further evidence about the memory representations of the patterns. The results from all three tasks are consistent with the theory that the memory representations are tree structures. Methods for deriving and comparing tree structures are discussed. This paper has two overall goals: (I) to give empirical evidence that memory representations of simple visual patterns can be modeled by a particular kind of network, namely, a tree, and (2) to illustrate a methodology that has recently been developed that can be used to discover optimal tree representations in many research domains.