Memory Research: an Examination of the Role of Recognition and Recall

Jolita Kisielius, The University of Michigan
[ to cite ]:
Jolita Kisielius (1984) ,"Memory Research: an Examination of the Role of Recognition and Recall", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 353-354.

Advances in Consumer Research Volume 11, 1984      Pages 353-354


Jolita Kisielius, The University of Michigan

Both of the papers in this session are concerned with examining the learning processes underlying consumers' responses to advertising. The first paper by Lorne Bozinoff and Victor J. Roth entitled "Recall and Recognition Memory for Product Attributes and Benefits" investigates the differences between the two learning measures of recall and recognition for product attributes and beliefs, whereas the second paper by M. Carole Macklin of "Verbal Labeling Effects in Short-Term Memory for Character/Product Pairings" focuses on recognition and its relationship to verbal labeling. Each of the papers will be examined in turn. The contributions of each paper will be discussed Followed by an analysis of various issues posed by the paper. Finally, in a concluding section, the similarities and dissimilarities between the papers will be addressed.

The paper by Bozinoff and Roth is an interesting one both to consumer behavior practitioners and academicians. The authors argue that more emphasis should be placed on examining attributes and benefits because they will have differential effects on recall and recognition. The distinction between attributes and benefits has often been ignored by consumer behavior strategists and researchers. In the planning of advertising strategies, attributes and benefits are often used synonymously by copywriters to describe consumers' needs for products and services. The sane lack of attention to the difference between attributes and benefits can be found in consumer behavior research studies that interchange attributes and benefits tn assessing consumers' attitudes towards a product or service.

A striking aspect of Bozinoff and Roth's research is the breadth with which they approached their topic. The authors attempted to examine a substantial number of issues related to their topic--the effects of attributes and benefits on recall and recognition, the transformability of attributes and benefits, the strengthening of the link between attributes and benefits, and the relationship of familiarity and interest to the recall of product attributes and benefits. It should be noted that the authors recognized the complexity of the learning process concerning attributes and benefits by employing two measures of learning of recall and recognition. These measures in other learning contexts have been found to address different aspects of the learning process.

Certain aspects of the research design were very executed. Because the stimulus consisted of new products, previous information about the products could not affect subjects' processing of the stimulus information. The order of presentation of the attributes and benefits was rotated in order to eliminate order bias. Subjects were randomly assigned to treatments and the subject sample was appropriate for the type of research being undertaken. The operationalizations of the dependent variables were appropriate. Recall was measured by a standard free recall task and recognition was assessed by including false recognition items.

To increase the contribution of the Bozinoff and Roth paper to the research area concerned with assessing the impact of attributes and benefits on recall and recognition, a number of issues may need to be considered. Perhaps the those important issue concerns the theoretical framework of their research. The authors take an information processing approach to explaining whether consumers evaluate products using attributes and benefits and how this information is stored in long term memory. They seem to adopt a schema notion of the processing of information in that a schema is assumed to organize information about a concept. A propositional representation format of a schema is suggested to explain attributes and benefits. For example, the authors state: "A product attribute conceptualization implies that these propositions are in the form of attributes, while a product benefit conceptualization implies these propositions are in the form of benefits." The authors then continue this reasoning that if the product information is stored in long term memory in the form of product attributes, then that memory should he superior. Conversely, the opposite prediction is made for benefits.

Several questions arise concerning the authors' view of memory processing. First, their theoretical view was not directly tested. It is not clear how the authors tested the notion of the structure or propositional representation of either the attributes or the benefits. No hypotheses were formally stated for the study. Second, the central concepts of the study of attributes and benefits need to be more clearly defined. Benefits are defined as "needs the products will fulfill," whereas attributes are "the physical characteristics of the products." Are these definitions discrete or are they intended to have some basis of similarity? Perhaps the concepts may differ in concreteness or on some other factors. The conceptualization of attributes and benefits as stated in the paper is not sufficient for motivating the rationale for why these concepts should have differential effects on recall and recognition.

Another issue that needs to he addressed by the authors is that their view does not account for all of the results found. Neither attributes nor benefits were found to be superior across product categories. Although the authors found some support for their investigation of attributes and benefits, the evidence for the differential effect of attributes and benefits was product specific. For example, for the video phone, benefits were recalled and recognized better than attributes. More false recognition of the benefits of video phones was also found. A similar trend reversed for the electric car. The reverse situation was found to be true for convenience food. For this latter product, attributes were better recalled and recognized in comparison to benefits. No viable explanation for the product differences was offered tn the paper.

The last issue that needs to be addressed concerns the administration and analysis of the dependent variables of recall and recognition. The authors appeared not to rotate the order tn which the recall and recognition measures were administered. It is possible that the measures may have been reactive. By asking subjects to complete the recall task before the recognition measure, the responses to the recognition measure may have been affected by the recall task. More detail is also needed on the analysis of the two measures. For example, no information is presented on the coding of the recall measure. In addition, the recognition task measured certainty of recognition as well as recognition itself. It is not clear whether this distinction was considered in the analysis.

The second paper by Carole Macklin, "Verbal Labeling Effects in Short-Terns Memory for Character/Product Pairings" deals with a very important topic for understanding age differences in consumer behavior because of labeling's pervasiveness as a strategy used in learning. The paper attempted to test the hypothesis that verba' labeling would have a differential effect on recognition for cued (first and second graders) and strategic processors (fourth and fifth graders). One of the contributions of this paper is its addition to the small but very important body of research on children's information processing. The methodology and mode of analysis included a number of appropriate procedures. The stimuli were thoughtfully created to appeal to the child subjects by using a playing card format. Subjects were randomly assigned to treatments and the seating procedure controlled for possible reactivity between subjects. A pretest was undertaken to determine the ease of labeling the stimuli. In addition, the dependent measure was appropriately designed for the experimental setting and included incorrect product choices as well. Finally, the analysis included a check for sex differences tn responses.

Despite tile positive aspects of the Macklin paper, a number of issues remain unresolved. The issue centers on the nature of the labeling process. Although the author states that differences in labeling are expected for the two age groups, a more detailed description of how and why this difference occurs would be useFul. Perhaps such a description could explain why the predicted interaction between age and labeling did not occur. In addition, since only main effects were Found, the results may be subject to alternative explanations.

Another issue concerns the use of familiar products as stimuli and their subsequent effect on the results. The finding that older children did not exhibit differences for the labeling and no-labeling conditions may have been due to the fact that the older children were more familiar with the currently existing products and were thus unaffected by labeling.

The rotation procedure of the characters and products needs to be more fully explained. This rotation may have led to character-product pairings that confused subjects. For example, for the statement "Debbie Dancer likes Bubble Yum," in the label condition subjects sere presented with a picture of a woman dancing and a picture of the actual gum. It is possible that labeling was especially important in this study because the pictures appeared to he unrelated to each other. A manipulation check on the difficulty of associating the character and product pictures in the label and no-label conditions For the two age groups would have been useFul. The pictures used as the stimuli appeared to be examples of non-interactive imagery. Would labeling have been assumed to have the same effect For pictures representing interactive imagery?

The Bozinoff and Roth and Macklin papers are both concerned with elucidating the learning process and both investigate the dependent variable of recognition. Both studies Find differences in recognition on the basis of their independent variable manipulations. However, these papers differ in a number of ways; these points of difference can serve as topics for future research on learning. Different age groups were used in the two studies. The subjects in the Bozinoff and Roth study were college students; whereas Macklin used six to eight and ten to eleven year old children. The type of product used in both studies differed in that the Bozinoff and Roth study used unfamiliar new products; whereas Macklin presented her subjects with currently existing products. The authors also intended to study different aspects of memory. Bozinoff and Roth stated that they were investigating long term memory while Macklin was interested in short term memory. However, this distinction may not be useful in classifying the studies. Bozinoff and Roth presented subjects with a distractor task before measuring recall and recognition while Macklin measured recognition immediately after the stimulus presentation. Based on their procedures it is questionable whether the onset of the dependent measures led to the investigation of different aspects of memory. Finally, both studies differed in the manner by which recognition was measured. Macklin's measure involved asking subjects to match the product to the character after subjects were exposed to the character-product pairings. Bozinoff and Roth included a certainty aspect to their measurements of recognition. Subjects were asked to state the degree to which an attribute/ benefit was mentioned on a four point scale ranging from definitely mentioned to definitely not mentioned.

In summary, although differences between the two studies can be found, the overriding conclusion to be abstracted from the studies is that recognition differences exist in the processing of information. Furthermore, these differences can be the result of the age of the processor or the type of information presented - attribute-benefit or labeled-unlabeled information.