The role of memory in the construction of linear orderings

Abstract

Two experiments examined the construction of linear orderings (e.g., ABCD) from presented pairs of adjacent items (e.g., AB, BC, CD) in an attempt to determine why some orders of presentation are more difficult than others. Experiment 1 found order effects when participants constructed orderings but not when they simply recalled presented pairs, suggesting that order effects reflect processing difficulties encountered during construction. Experiment 2 used several different orders and successive and simultaneous visual presentation of pairs. Results showed that order effects can be attributed to the memory load imposed by certain pairs, the type of processing demanded, the need to reorder previously stored information, and the loss of information with continued rehearsal. Results strongly support the Foos, Smith, Sabol, and Mynatt (1976) model of linear construction.Several recent studies have presented individuals with information about adjacent items in a linear ordering (e.g, AB, BC, CD) and asked the individuals to construct the ordering (e.g., ABeD). Performance is affected by the order in which information is presented (Foos, Smith, Sabol, & Mynatt, 1976), the strategy that participants adopt in representing presented information (Mynatt & Smith, 1977), the completeness of the underlying ordering (Foos, 1980), the number of items in the ordering (Foos, 1980), the relationship between a sentence adjective and an item's placement (Potts & Scholz, 1975), and various combinations of the above (Foos, 1980). The present paper examines the effects produced by presentation order in an attempt to determine whether these effects can be attributed to the need for different processing for different inputs or simply to differences in the amount of information that must be held in store pending recall. Foos et al. (1976) describe three major effects produced by the order in which adjacent pairs of items are presented to persons attempting to construct the underlying ordering. The first of these occurs when pairs of items, after the initial pair, contain one old (i.e., previously presented) and one new item. These are called match (M) pairs, since they allow the individual The authors wish to thank an anonymous reviewer for helpful suggestions in preparation of the final version of this article and to acknowledge the support of National Science Foundation GrantBNS to match the input pair with previously stored information via the old item. If an individual has stored some information (e.g., Be), there are two kinds of M pairs that can be presented. In one case, the input will add a new item to the end of the previous information string (e.g., CD to produce BCD), and in the other case, to the beginning (e.g., AB to produce ABC). The former is called Process M1, and the latter, Process M2. Presentation orders produce better performance as the number of M1 pairs exceeds the number of M2 pairs.The second effect of presentation order occurs when all the pairs are not M pairs. When one presents a pair of new, and thus nonmat...

The construction of linear orderings under conditions of increased memory load

“…In sum , the present results support the hypothesis (Foos & Sabol, 1981) that differences between match and nonmatch presentation orders are due primarily to memory load , whereas other order effects are due primarily to processing differences . In addition , the results suggest that, under conditions of increased memory load , individuals find it easier to construct in a single direction (beginning or end) and more difficult to construct in two directions (beginning and end ).…”

“…The other differences described above are due primarily to processing. Studies that have reduced memory load (Foos & Clark, 1983;Foos & Sabol, 1981) have shown that such reductions reduce matchnonmatch differences but do not significantly alter the other differences assumed to be due to processing. The present study attempted to address this question by increasing, rather than reducing, memory load.…”

“…Nonmatch orders produce lower performance than match orders in large part because of the increase in memory load that occurs when a sentence presents two new items that cannot be immediately added to a single ordering. Reducing memory load by keeping all presented information available to the individual (e.g., on a CRT) reduces the difficulty of non match orders (Foos & Sabol, 1981).Differences between the two nonmatch orders (1-3-2 and 3·1-2) are found in some instances but not in others. Such differences appear to result from the loss of separate storage for the nonmatching information presented in the second sentence.…”

“…Nonmatch orders produce lower performance than match orders in large part because of the increase in memory load that occurs when a sentence presents two new items that cannot be immediately added to a single ordering. Reducing memory load by keeping all presented information available to the individual (e.g., on a CRT) reduces the difficulty of non match orders (Foos & Sabol, 1981).…”

Constructing Schemata While Reading Simple Stories

“…Unfortunately, Baker compared the mean of the chronological sequence (1-2-3) with the means of the two flashback sequences (1-3-2 and 2-1-3), but, as mentioned previously, the 1-3-2 flashback sequence should have been no more difficult than the chronological sequence if the information remained available (Foos & Sabol, 1981). A comparison of the subject's performance for the chronological sequence (1-2-3) with that for the diacult flashback sequence (2-1-3) might have demonstrated an effect of input sequence on story-reading time.…”