Memory Organization and Recall Research Paper

1. Definition

The term organization has been used in a number of ways in the context of recall and memory. One basic distinction is that between primary and secondary organization (Tulving 1968). Primary organization refers to the effects that are attributable to task characteristics. Secondary organization refers to organization that is imposed on the to-be-remembered information by the individual. It is this latter type of organization that has been of primary interest in trying to understand information processing and memory. On the basis of this Voss (1972) formulated one definition of organization that seems to capture the essence of what many memory researchers mean by organization in this context. ‘Organization is a process that intervenes between input and output in such a way that there is not a 1:1 input–output relation’ (Voss 1972, p. 176). This definition focuses on a particular output order that reflects a higherorder grouping of the to-be-remembered information and it emphasizes that the subject adds a structure to the information presented, which makes input and output different. A similar flavor is contained in the definition proposed by Tulving (1968): ‘organization occurs when the output order of items is governed by semantic or phonetic relations among items or by the subject’s prior extra-experimental or intra-experimental acquaintance with the items constituting a list’ (Tulving 1968, p. 16).

2. Historical Roots

Organization theory of memory had its heydays during the 1960s and 1970s. However, the importance of organization in relation to learning and memory had been emphasized much earlier in Gestalt psychology. Although the main focus of the Gestalt psychologists was that of perceptual groupings (e.g., similarity, proximity), the notion was that this grouping had implications also for memory and retention. As noted by Postman (1972, p. 4), Wolfgang Kohler stated that those factors that are active in organization of primary experience should also affect recall. Postman (1972) noted four basic principles in the Gestalt conceptualization of organization that are of relevance for how organization came to be used by organization theorists during the 1960s and 1970s. First, he claimed that the Gestalt view was that organization is largely established by the initial perception of the events to be remembered. Second, the form of organization is determined by the relations among the component units, such as proximity and similarity, indicating that there is a natural or optimal organization. Third, the availability of prior experiences for recall over time depends on the temporal stability of the memory traces laid down by these experiences. Fourth, the accessibility of available traces for recall is a function of the similarity between these traces and current stimuli.

Quite a different influence of how organization came to play a major role in memory research dates even further back than that of the Gestalt psychologists. This is the way organization was used in the context of text recall. The role of organization in text processing was first specified early in the twentieth century and was later refined by Bartlett (1932). The term used by Bartlett was ‘schema,’ which he defined as an active organization of past experience. The concept of schema and later the concept of script was taken up by many researchers in the area of text recall and discourse processing (e.g., Anderson et al. 1977, Kintsch and Vipond 1979, Schank and Abelson 1977).

3. Research Issues

Many research problems were invented in studies of organization and recall during the 1960s and 1970s. It was extensively explored whether there was an optimal number of categories to organize for a maximal recall to occur, whether the number of items per category was critical for clustering, and whether blocked vs. random presentation of category items made a difference.

In particular one topic was of major interest in this research field, namely that of developing measures or indices of organization. Several such measures were developed during the 1960s and 1970s. One commonly used way to classify these measures is whether organization is imposed by the experimenter or by the subject. For measures based on experimenter-based organization only one study trial is usually required, whereas for the subject-based organization, a multitrial design is required.

For experimenter-based measures, the most frequently used paradigm is that the items to be remembered belong to two or more different semantic categories. The words from the different conceptual categories usually appear in a random order in the list presented, but can, as mentioned, also appear in a blocked fashion in the study list. Even if no specific instruction to the subjects is given about organizing the words into semantic clusters, the output is typically organized such that the words recalled appear in clusters. This form of organization is referred to as categorical organization or categorical clustering. Word lists used in studies of organization and recall can also be composed of an associative rather than a categorical relationship.

Subject-based measures are used in experiments where the words comprising a list are unrelated, conceptually or associatively. That is, subjects are free to organize the words at output in whatever other order they prefer. Organization in this case is based on idiosyncrasies of the subjects and it is determined by the extent to which they recall the items of a list in the same order at two successive trials.

Although the basis for organization of the study items differ between experimenter and subject-based measures, it is generally assumed that both types of measures are indicating the same underlying psychological process of organization. Several different indices of each form of organization have been developed.

Bousfield (1953) developed several such indices for categorical clustering. The basic rationale for these different indices was to relate the number of successive repetitions from a given category in a response protocol to the total number of words recalled. A repetition in this case is the occurrence of two items from the same category directly after one another. One problem with some of the first indices developed was that they did not take into account the transition from one category to another. As such transitions are crucial for perfect clustering of a list of words with more than one category, a proper measure was not found until the item-clustering index (ICI) was developed. ICI took this aspect of clustering into account, but failed in other ways. It did not take into consideration the possibility that clustering can be perfect although not all items of all categories are recalled. The adjustedratio-of-clustering (ARC) score proposed by Roenker et al. (1971) took this into account. The formula developed to obtain a quantitative measure for organization by means of ARC is:

                               ARC =  R ‒ E(R)/maxR ‒ E(R)                                         (1)

where R is the total number of observed category repetitions, E(R) is the expected number of category repetitions, and maxR is the maximum possible number of category repetitions. E(R) is computed as the ratio between the sum of the squared number of items recalled from each category and the total number of items recalled, minus unity; maxR is the difference between the total number of items recalled and the number of categories represented in the response protocol. A comprehensive review of clustering measures was presented by Murphy (1979), who concluded that the ARC score is one of the best measures available for assessing clustering.

Measures of categorical clustering, e.g., ARC, were frequently used in memory research during the 1960s and 1970s. Today, these measures are less common in articles on memory, but are sometimes used as a dependent variable in its own right. For example, in studies of aging and memory, organization, as measured by means of the ARC score, has been used as a dependent variable to estimate cognitive deficits as a function of age in much the same way as memory deficits as a function of age is estimated by recall scores (e.g., Backman and Wahlin 1996). Backman and Wahlin (1996) concluded that organizability of the study list in semantic categories could serve as a cognitive support for recall and that the ability to utilize this support remained relatively constant across the age range that these authors studied, 75–96 years.

There are two main measures of subject-based organization. For one of these, proposed by Bousfield and Bousfield (1966), the critical index is the number of intertrial repetitions (ITR). This index is determined by constructing a matrix for each subject, with the items of the word list represented along both rows and columns of the matrix. Rows represent the nth word recalled and columns represent the (n 1)th word recalled. This matrix is then used for making a tabulation of the frequency with which given pairs occur adjacent to one another on two successive trials (Shuell 1969). This frequency value is the observed ITR value. The measure of subjective organization according to Bousfield and Bousfield (1966) is this observed ITR subtracted by an expected ITR

                                     E(ITR) = c(c ‒1) hk                           (2)

where c is the number of items common to two recalls, h is the number of words recalled on trial n, and k is the number of words recalled on trial n + 1.

The other measure of subject-based organization, proposed by Tulving (1962), is called subjective organization. Computing subjective organization this way, the data are organized in 2 x 2 contingency tables, with recall and nonrecall on trial n on one axis and recall and nonrecall on trial n 1 on the other axis. Tulving argued that the component for recall on two successive trials increased logarithmically over trials, while the component for nonrecall and recall on the two trials remained essentially constant with practice. Subjective organization is expressed quantitatively as:

                                     SO = Σnij lognij /Σni logni                  (3)

where nij represents the numerical value of the cell in the ith row and the jth column, and ni represents the marginal total of the ith row (see Shuell 1969).

The measures of subjective organization by Bousfield and Bousfield (1966) and Tulving (1962), respectively, are still used when subjective organization constitutes a dependent variable in recall experiments. The measures are of comparable power, but the ITR measure by Bousfield and Bousfield (1966) may be a bit more commonly used, simply because it is easier to compute.

One theoretical issue of great concern during the heydays of research on organization and recall was about the locus of organization. To this end it should be stated immediately that clustering is an output phenomenon from which some sort of organizational process is inferred. However, organization of the to-be-remembered information can equally well take place at the time of encoding. In a review article Shuell (1969) concluded that this was still an open question. He argued that a prerequisite for solving this issue was that a measure of organization was developed that was independent of recall. Such a measure has not yet been developed and the basic issue of the locus of organization at encoding or retrieval is still an open question. Even in later reviews, for example those collected in Puff (1979), empirical data and theoretical statements appear to be rather mixed on the topic of the locus of organization.

Another research question examined in the area of organization and recall was the relationship between the phenomenon of organization and the notion of memory as an entity of limited capacity. Although the concept of organization was of great interest in itself, its popularity or central role in memory theory was also due to the fact that it provided, and still provides, a nice illustration of the notion of a capacity-limited memory. The basic finding is that recall performance is improved whenever an organizational structure is imposed on the to-be-remembered materials compared to cases when no organizational principle can be applied. This has been demonstrated in experiments comparing recall performance in organizable and nonorganizable lists of words. Thus, the use of organization in this way can be said to be an extension of the principles of chunking proposed by Miller (1956). However, organization has also been used in a broader sense, namely as a means for compensating for a limited capacity of memory for larger sets of to-be-remembered information. This type of organization need not be limited to clustering of words in categorized lists, but can be based on other principles. For example, the type of imagery structure of the materials to be remembered in mnemonic techniques, like the loci method, is one form of organization used in this broader sense.

Still another research question discussed was whether the organization processes underlying category clustering and subjective organization were the same or different, and whether the organization processes underlying clustering according to semantic category were same or different as organization processes behind grouping on the basis of other nonconceptual or nonassociative principles. In a vast majority of the studies carried out on organization and recall, the underlying organization principles has been based on semantic properties of the to-be-remembered materials. However, in a series of studies during the 1970s it was demonstrated that organization of the to-be-remembered materials can also be organized according to modality of presentation. Mixed lists of auditorily and visually presented words were presented and output in an immediate free recall test tended to be in clusters of auditorily presented words and visually presented words (Nilsson 1973). In a later study by Nilsson (1974), the list presented contained words from different semantic categories. Half of the words in each category was presented auditorily and the other half visually. The results showed that organization by modality was dominating in an immediate free recall test, whereas in a delayed final free recall test, the words in the recall output were ordered according to semantic category.

4. Current Role of Organization in Memory Research

As stated initially, the concept of organization played a major role in memory theory during the 1960s and 1970s. Many research questions were invented and many empirical variables were explored. However, progress in solving these questions and mapping the variables into existing organization theory was not overwhelming at the time (Shuell 1969). To some extent, this state of affairs might have been due to the fact that organization theory, as it was referred to at the time, was not really a theory in a formal sense. It was more of a framework or a point of view emphasizing higher-order cognitive processes beyond simple associationistic principles of learning and memory that had dominated the scene for many years.

Looking back now at the issues explored in the 1960s and 1970s, it can be concluded that not much has happened in answering the specific questions that were asked then. The question of whether there is some optimal number of categories in a list for maximal organization to occur has not been given a final answer. The question about the locus of organization at encoding or retrieval has not yet been answered in a way that has reached consensus in the scientific community. The interest in solving these and the other questions in focus of attention at the time seems to have declined although single papers on this topic have appeared in recent years (e.g., Kahana and Wingfield 2000).

Focus and interest changed from organization during the 1960s and 1970s to interactions between storage and retrieval. Experiments on recall and recognition of individual, unique word events became fashion during the 1970s. Perhaps the strongest expression of this is the great impact that the levels-of-processing framework (Craik and Lockhart 1972) had in reorienting the research community in memory to different research questions than before. Another approach to memory research that contributed to a shift in orientation was the new interest in structural and functional properties of episodic and semantic memory (Tulving 1972) and subsequently in the issue of memory systems in general. This reorientation meant a considerable broadening of the research issues of primary interest and focus. For one thing, this wider approach to studies of memory meant that many new disciplines got involved in a collaborative enterprise to understand how mind and brain process information. Technological inventions for brain imaging (positron emission tomography and functional magnetic resonance imaging) currently play a major role in studying memory. Future will tell whether the concept of organization will be brought back into research focus again. Perhaps by using these brain imaging techniques as a means to obtain independent measures of organization.

Bibliography:

1. Anderson R C, Reynolds R E, Schallert D L, Goetz E T 1977 Frameworks for comprehending discourse. American Educational Research Journal 14: 367–81
2. Backman L, Wahlin A 1996 Influences of item organizability and semantic retrieval cues on word recall in very old age. Aging and Cognition 2: 312–25
3. Bartlett F C 1932 Remembering. Cambridge University Press, London
4. Bousfield A K, Bousfield W A 1966 Measurement of clustering and of sequential constancies in repeated free recall. Psychological Reports 19: 935–42
5. Bousfield W A 1953 The occurrence of clustering in the recall of randomly arranged associates. Journal of General Psychology 49: 229–40
6. Craik F I M, Lockhart R S 1972 Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Beha ior 11: 671–84
7. Kahana M J, Wingfield A 2000 A functional relation between learning and organization in free recall. Psychonomic Bulletin & Re iew 7: 516–21
8. Kintsch W, Vipond D 1979 Reading comprehension and readability in educational practice and psychological theory. In: Nilsson L-G (ed.) Perspectives on Memory Research: Essays in Honor of Uppsala University’s 500th Anniversary.Erlbaum, Hillsdale, NJ, pp. 329–65
9. Miller G A 1956 Language and Communication. McGraw-Hill, New York
10. Murphy M D 1979 Measurement of category clustering in free recall. In: Puff R C (ed.) Memory Organization and Structure. Academic Press, New York, pp. 51–83
11. Nilsson L-G 1973 Organization by modality in short-term memory. Journal of Experimental Psychology 100: 246–53
12. Nilsson L-G 1974 Further evidence for organization by modality in immediate free recall. Journal of Experimental Psychology 103: 948–57
13. Postman L 1972 A pragmatic view of organization theory. In: Tulving E, Donaldson W (eds.) Organization of Memory. Academic Press, New York, pp. 3–48
14. Puff R C 1979 (ed.) Memory Organization and Structure. Academic Press, New York
15. Roenker D L, Brown S C, Thompson C P 1971 Comparisons of measures for the estimation of clustering in free recall. Psychological Bulletin 76: 45–8
16. Schank R, Abelson R 1977 Scripts, Plans, Goals, and Understanding. Erlbaum, Hillsdale, NJ
17. Shuell T J 1969 Clustering and organization in free recall. Psychological Bulletin 72: 353–74
18. Tulving E 1962 Subjective organization in free recall of ‘unrelated words.’ Psychological Re iew 69: 344–54
19. Tulving E 1968 Theoretical issues in free recall. In: Dixon T R, Horton D L (eds.) Verbal Beha ior and General Beha ior Theory. Prentice-Hall, Englewood Cliffs, NJ, pp. 2–36
20. Tulving E 1972 Episodic and semantic memory. In: Tulving E, Donalson W (eds.) Organization of Memory. Academic Press, New York, pp. 381–403
21. Voss J F 1972 On the relationship of associative and organizational processes. In: Tulving E, Donalson W (eds.) Organization of Memory. Academic Press, New York, pp. 167–94