Comments on &Quot;Theoretical and Empirical Perspectives on Memory

Citation:

John G. Lynch, Jr. (1982) ,"Comments on &Quot;Theoretical and Empirical Perspectives on Memory", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 354-356.

Advances in Consumer Research Volume 9, 1982      Pages 354-356

COMMENTS ON "THEORETICAL AND EMPIRICAL PERSPECTIVES ON MEMORY

John G. Lynch, Jr., University of Florida

The papers at this session are almost surely as diverse as those at any session at the convention this year. All of the papers can be loosely construed to deal with the effects of consumers' prior knowledge on some aspect of information processing and decision making. However, the papers by Beattie and by Loken have highly theoretical concerns with frontier issues in consumer information processing. In contrast, the paper by Saegert and Young was motivated by a pragmatic interest in concrete problem -- to assess baseline knowledge levels among consumers so that the effects of nutrition information campaigns could be properly ascertained.

When I first received these papers, I was unconvinced that any common threads tied them together. In fact, I suspected that, as a set, they comprised a sort of ACR discussant"s Rorschach. Upon closer study, though, I have been able to discern some very interesting commonalities. I will try to develop these common themes as I discuss each separate paper.

"EFFECTS OF PRODUCT KNOWLEDGE ON COMPARISON, MEMORY, EVALUATION, AND CHOICE: A MODEL OF EXPERTISE IN CONSUMER DECISION MAKING"

At the 1979 ACR convention in San Francisco, a distinguished panel pondered the question, "Why don't consumers search more for information?", motivated by mounting evidence that consumers engage in amazingly little external search prior to purchase, even in expensive and presumably "involving" product categories. One of the ideas that was advanced at that session was that consumers often lack the requisite knowledge to be able to assimilate information about complex products. Therefore, extensive search for relevant external information would only serve to confuse rather than enlighten the consumer. The implication of this line of argument is that, if only consumers knew more about a product class, they would voluntarily seek out more information, and would more effectively assimilate the information to which they were exposed.

Since those ideas were first proposed, Bettman and Park (1980) have explored their implications for external search, and Johnson and Russo (1981) have explored their implications for the recall of product information. Beattie's paper makes a theoretical contribution that complements these empirical efforts, in that it further specifies the relationships between prior knowledge and attention. "Ideal brand" schemata serve to overcome, to some extent, the limitations of human attentional capacity by allowing new information to be chunked in terms of schema relevant concepts, and by directing the focus of attention to schema relevant features and away from schema irrelevant features. This allows experts to process more information than novices. Experts are assumed to use their spare capacity to process information about differences from an ideal as well as the similarities to the ideal. Novices, because they cannot chunk information in terms of schema-relevant concepts, have only enough capacity to process information about features of a brand that make it similar to some ad hoc ideal -- e.g., to the most expensive brand present in the display. This is quite an interesting hypothesis, and certainly deserves empirical investigation.

I have one small clarifying point about the illustration Beattie provides of how experts and novices might differ in their decision making processes. In the example given, experts engage in processes identical to those of novice consumers, except that the former give relatively greater weight to "important" dimensions, and less to "unimportant" ones in evaluating the distance of a brand from an ideal. This conveys the idea that expert consumers' "ideal brand" schemata serve to focus their attention on schema relevant attributes. Nothing in the example, though, implies that novices use only similarity information and that experts use both similarity and difference information. One might incorporate this into the formal model by saying that the probability that a given dimension will be salient (and contribute to perceived distance of a brand from the ideal) is a negative function of "objective" distance between the two on that attribute dimension for novices, but a quadratic function of this distance for experts. Some such modification must be mate if the example is to convey the idea of the proposed strategic differences between expert and novice consumers. The particular modification proposed ties in rather nicely with non-ideal-point models of overall evaluation (Fiske, 1980) that hypothesize that greater weight is given to attribute information that is extremely positive (very close to the ideal) or negative (very far from the ideal), with relatively less attention given to product information that is evaluatively neutral (moderately distant from the ideal).

Finally, I would like to call attention to one extremely interesting aspect of Beattie's model that was discussed only briefly. She alludes to the idea that novice consumers have attentional foci that are at once diffuse and shallow. Their attention is easily shifted by salient perceptual cues. Attention researchers distinguish between the breadth of the attentional focus (number of attributes considered) and the stability of that focus (i.e., the degree to which attention vacillates among the attributes that receive attention). The latter issue has not been extensively studied, but Kahneman (1973) has noted that highly labile attention has a number of dysfunctional consequences for task performance. In Beattie's context, novice consumers may experience this state of lability of attention, in which one is engaged in an involuntary search of memory for concepts relevant to encoding the information. In such a search, the external material may take contact with a number of concepts, all of which are only weakly and briefly activated. In such a state of lability, no concept makes a close enough match with the external information to gain ascendancy, and so the new material receives very little of what Mandler (1980) calls "intra-item integration." One function of well-formed schemata may be to prevent this dysfunctional state of attentional lability, so that the attended items receive sufficiently high levels of intra-item integration for retrievable traces to be stored in memory.

"MEMORY FOR LOGICALLY DEDUCED CONCLUSIONS"

Loken's paper is like Beattie's, in that it deals with the effects of a schema-like cognitive structure C in this case, the consumer's understanding of the rules of logic. However, Loken's emphasis is not on how some knowledge structure facilitates the encoding of new information, but upon how formal logic serves as an organizational principle in memory that facilitates retrieval processes. Loken predicted that logically related propositions would be organized (which I take to mean "stored") together in memory, and thus each such proposition could cue the others with which it was associated. This prediction was not supported. If anything, there was a trend for subjects to evidence superior recall of a presented "conclusion" if it was not logically related to its "premises in comparison with conditions in which the conclusion followed logically from the premises. The data did support the prediction that when a conclusion (A implies C) was not explicitly presented, it was more likely to be falsely recalled as having been presented if the subject had been exposed to premises from which it could be logically deduced. These intrusions were taken as evidence that formal logic acts as an organizational principle in memory.

Loken has already noted the most important limitation of her study. We cannot say with confidence that when subjects who have been exposed to statements "A leads to B" and "B leads to C" falsely recall and recognize "A leads to C", the latter proposition must have been stored in memory. It is possible that subjects have simply stored the constituent premises, and generated "A leads to C" at the time of the memory test. This sort of interpretive ambiguity is not peculiar to Loken's study. Indeed, the problem is endemic to memory research. We often have difficulty pinpointing a given stage of information processing as the locus of the (in this case, intrusion) effects to be explained.

The obtained results could be interpreted as indicating that subjects exposed to "A leads to B" and "B leads to C" draw the inference that "A leads to C" after exposure to the second "premise", and all three elements are stored together in memory. Alternatively, subjects may infer that "A leads to C" after exposure to the two premises, and all three assertions may be stored separately (as if they were organized in terms of order of occurrence). A third possibility is that logically related statements are stored together in memory, not because of their logical relation, but because they share common elements ("A leads to ]" and "B leads to C" share the element "B", whereas "A" leads to D" and "E leads to C" share no common elements). Assertions that are stored together may be recalled as a unit. At the time of recall, subjects may draw the inference that "A leads to C" must have also been presented. These inferences could be based on logical deduction, or they may be based upon simple guessing that A was presented together with some element that was stored together with it. Quite often, subjects may guess that it was "C" that was paired with "A", producing intrusions in the Logical, No Conclusion condition. Under this last interpretation r formal logic need not play a role in either the initial interpretation of the information, its stored organization in memory, retrieval processes, or the generation of intrusions.

This is not to say that I personally find these alternative explanations to be as compelling as that advanced by Loken. In fact, I would judge her account of her data to be considerably more plausible. I am simply trying to point out the kinds of interpretive ambiguities that attend research on such a difficult topic. Loken's a priori hypotheses seem both interesting and reasonable, her experiment was well designed, and here data are provocative. It would certainly be worthwhile for consumer researchers to pursue this topic in an attempt to sort out some of the alternative process explanations.

Perhaps the greatest contribution of Loken's work is to call consumer researchers' attention to the importance of inference processes. Although work within the multiattribute attitude tradition (Fishbein & Ajzen, 1975) has raised the issue of inferential belief formation, consumer researchers have displayed remarkably little interest in inference making, with the exception of some activity in the attribution area. Clearly, however, such processes are extremely important in areas such as deceptive advertising. With the completion of the landmark 4 A's study by Jacoby and his colleagues last year, more and more researchers are becoming concerned with consumers' miscomprehension of product information. Certainly the study of inference making will be critical to understanding this "miscomprehension", and we will have to explore both logical and nonlogical rules of inference to progress. I noticed that Phillip Dover is presenting a paper at this conference that concerns inferential belief formation. I am certain that his paper and Loken's paper will stimulate consumer researchers' interest in this very important topic

"DISTINGUISHING BETWEEN TWO DIFFERENT KINDS OF CONSUMER NUTRITION KNOWLEDGE"

I suspect that Saegert and Young were dismayed by the published title of the present session, as it is not particularly descriptive of their paper. As I understand it, their study was motivated by a desire to assess baseline levels of consumer awareness of various nutrition facts so that the effects of nutrition information campaigns could be assessed. It is interesting to note that many of the issues that were raised with respect to Loken's theoretical paper are of concern in Saegert and Young's applied research. In particular, Saegert and Young imply that when consumers answer incorrectly on a true-false test of nutrition knowledge, they may err for one of three reasons:

a. They may have no pertinent knowledge stored in memory, and either guess wrong, or say "don know";

b. They may have pertinent beliefs stored in memory that are "incorrect" from the perspective of nutrition experts;

c. They may have no specific knowledge stored in memory, but have some conceptually related cognitive structures (e.g., their attitudes toward health foods) from which they make an inference at the time of the true-false test that is scored as "incorrect".

One would expect errors of the first type (a) to be random while errors of the second two types might be reliable in a test-retest sense.

Saegert and Young note that, in the pools of items commonly used to measure nutrition knowledge, many pertain to the use of health foods. They reason that interpretive ambiguities of the sort mentioned above are especially problemative for these items. Thus, while there is no theoretical reason to distinguish "health food beliefs" and "general nutrition beliefs" that are actually stored in memory, there may be considerable practical value in separating test items pertaining to these two types of beliefs. It seems that "General Nutrition" items are likely to be scored wrong primarily because of a lack of awareness (i.e., explanation (a) above), while (b) mistaken beliefs and (c) errors based upon inference are also distinct possibilities for "Health Food Knowledge" items. This also suggests that - r education campaigns aimed at disseminating accurate Health Food information may be less efficacious than those aimed at engendering accurate General Nutrition Knowledge.

The authors were also interested in the relationships between nutrition knowledge and behavior. I could not tell whether this was because they were intrinsically interested in such a link, or because they thought it would shed some light on the hypothesis that "Health Food Knowledge" responses tend to be inferred from the consumers' attitudes toward health foods. Perhaps the idea is that for "General Nutrition Knowledge" items, beliefs that are already held might be assumed to shape the consumer attitudes toward health foods and their consumption, while for "Health Foot Knowledge" items, expressed beliefs were inferred from (caused by) attitudes. As far as I could tell, the issues with which Saegart and Young were concerned do not fall nicely within any existing attitude framework. No theory available would clearly predict any relationship between scores on the two nutrition knowledge scales and behavior.

One small suggestion that I would make for future work in this area would be to exclude the "don't know" response category on the nutrition knowledge tests, in spite of the many reasons that could be advanced to justify its inclusion. If consumers were forced to respond either "true" or "false", we could better assess whether their scores on the test exceeded that which would be expected by chance. (To minimize the effects of response biases in this approach, each "true" item should have a parallel form in which the correct answer is "false". Sets of questions would then be composed under the constraint that half be true and half be false.) As it now stands, consumers seem to be "correct" on many of these items considerably less than half of the time, but it is not clear what can be said about this. Further, the provision of a "don't know" category may cause spurious differences in consumers' (and demographic groups') scores on the nutrition knowledge tests, simply because people differ in terms of how uncertain they must be before saying "don't know" (i.e., they differ in their response biases).

SUMMARY

Taken together the papers in this session point suggest the following directions for future research:

a. We must attempt to come to grips with the methodological problem of discerning whether information "recalled" has been retrieved from storage in long term memory, or has been generated (e.g. inferred) at the time of the memory test on the basis of other related information.

b. Further work is needed to study the processes by which consumers form inferences about products on the basis of externally provided information (e.g., advertisements) and information stored in memory.

c. The role of prior knowledge in the organization of new information in memory seems likely to be a fruitful avenue for further research.

d. Students of consumer decision making ought to actively pursue the how prior knowledge guides encoding and especially attention during consumer choice.

REFERENCES

Bettman, J. R., and Park, C. W., (1980), "Effects of Prior Knowledge and Experience and Phase of the Choice Process on Consumer Decision Processes: A Protocol Analysis," Journal of Consumer Research, 7, 234-248.

Fishbein, M., and Ajzen, I., (1975) Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research, Reading, Mass.: Addison-Wesley.

Fiske, S. T., (1980), "Attention and Weight in Person Perception: The Impact of Negative and Extreme Behavior," Journal of PersonalitY and Social Psychology, 38, 889-906.

Johnson, E. J., and Russo, J. E., (1981), "Product Familiarity and Learning New Information," in R. Monroe (Ed.) Advances in Consumer Research, Vol. 8, Ann Arbor, Ml: Association for Consumer Research, 151-155.

Kahneman, D., (1973), Attention and Effort, Englewood Cliffs, NJ: Prentice-Hall.

Mandler, G., (1980), "Recognizing: The Judgment of Previous Occurrence," Psychological Review, 87, 252-271.

----------------------------------------