Does the Emperor Ride Again?

Joel B. Cohen, University of Florida
[ to cite ]:
Joel B. Cohen (1984) ,"Does the Emperor Ride Again?", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 367-348.

Advances in Consumer Research Volume 11, 1984      Pages 367-348

DOES THE EMPEROR RIDE AGAIN?

Joel B. Cohen, University of Florida

With apologies to Mike Birnbaum for the similarity in titles, I fear the children's story about the emperor's new clothes may be somewhat applicable to the papers on this session. As you recall, while the emperor paraded around in supposed new and quite excellent clothes he was, in fact, stark naked. Moreover, the emperor seems a frequent visitor to our conferences and has an active typewriter to boot. Careful readers of our literature should have no trouble spotting the emperor as he strolls about. More worrisome is the fact that he has so many willing clothiers.

We have somewhat of a penchant in our field to find, or better yet, coin new terms with which to better package old ideas, and also to believe that by giving something a nice sounding name we are, at the same time, explaining it. If our efforts to understand consumer behavior phenomena even partially matched the splendor of our labels this would be a grand field indeed.

With that as a backdrop, let us turn to the papers presented during this session. Both have some interesting and useful points to make, and neither deserves to be singled out to absorb the brunt of the above criticism. They do, however, illustrate some of the shortcomings alluded to above and so the criticism is appropriate, however widely shared.

VERBAL AND INTERACTIVE PROTOCOLS

This is almost two papers in one. The first part of the paper reviews many of the problems and limitations of verbal protocols as a "window" on information processing and decision making. The paper incorporates many of the other critical reviews of this methodology and summarizes most of the earlier-made points. This section of the paper, then, would be a good place to begin for someone considering the use of verbal protocols in their research who would also benefit from being guided to other papers on the subject.

The author, at several points, seems to suggest that "interactive" protocols will help overcome some of the limitations of verbal protocols as well as a "psychologistic bias" which is "largely a function of the predominance of cognitive psychology in the study of consumer information processing and decision making." This competitive pairing of protocols from individual consumers and protocols "from two or more consumers engaged in decision making activities" is probably a mistake. It is not really a question of which one is best, nor does the presence of more than one person overcome the limitations of verbal protocols. In fact, interactive protocols would appear to be best suited to studying different phenomena (e.g., group or husband-wife decision making, bargaining), and they would also appear to have an added set of limitations related to interpersonal influences on individual behavior (including verbal behavior).

This paper would have been better with an expanded discussion of the contributions one might expect from interactive protocols in looking at particular research issues and topics. Though the author subsequently arrives at an assertion that the "unique character of conversation" makes this technique useful for the study of individual information processing, there is little convincing evidence or argument advanced for this. Statements such as, "cognitive activity is social in nature and origin," unless severely qualified are open to considerable debate. Information processing is, in and of itself, a cognitive activity (despite a person having multiple goals) and so it seems entirely appropriate that research methods developed in cognitive psychology be applied to study it. The author's concerns regarding a "psychologistic bias" due to a reliance on cognitive psychology might be better directed to consumer decision making to the extent he believes interpersonal processes are slighted during the search and evaluation phases. Hopefully the author will subsequently demonstrate that the use of interactive protocols will aid our understanding of such interpersonal dynamics.

Before leaving this paper, I have to say that there is a certain air of imperial presence in the term "interactive protocols." It holds out such promise that I felt surely there is either a conceptual or methodological advance in the paper I was about to read. While the term seems quite proper, to learn that it refers to simply a verbal protocol when two or more people are interacting is a bit of a comedown. Little did I know that I had been gathering interactive protocols for years in the focus groups I have run. "May I take your cloak. your highness?"

SCHEMATIC PROCESSING OF INFORMATION

This paper introduces the concept of "stereotype or schema transfer," an interesting idea stimulated in some measure by the work of Susan Fiske and her associates (see Fiske, 1982). Essentially, the authors propose that if an appropriate schema to process new information is not available, an existing and somehow related (e.g., physical proximity) schema could be "transferred" to process the input. Such a process might link up fairly well with work in concept identification and especially categorization as well as analogical reasoning and problem solving. It's an interesting question; how does one proceed when he doesn't have the specific knowledge necessary to comprehend and respond to a stimulus but has an available repertoire of related semantic and procedural knowledge? What dimensions of similarity between new input (conceivably products) and familiar concepts does the consumer rely on to make judgments and inferences? This "schema transfer" notion might be quite promising.

It was with considerable disappointment, then, that the translation of this concept was simply that, given a product about which little is known, consumers will use available information about a person who endorses it to answer questions about the product. The evaluative aspects of this effect are not at all new. They have their origin in balance and congruity theories: if a strongly positive source endorses a product, people are more likely to infer that the product is good. Do we really need a term called "schema transfer" merely to explain why a consumer might believe that an unknown magazine endorsed by a glamorous actress is more likely to contain articles and features of interest to her than to a politician, football player or someone else with stereotypically different tastes and preferences? The consumer is not, in any meaningful sense, "transferring" anything: he is making plausible inferences when requested to make judgments for which there is no direct evidence. Surely the "transfer" of a schema must involve more than that -- or these are truly insubstantial garments.

As is becoming the rule, this paper uses the term "schema" in several ways and as a result the exact meaning of the term is not always clear. One way the authors use the term is in the context of schema-based processing. Used in this fashion, the term refers to a theory which contains some important propositions about the effects of knowledge structures on cognitive activities. Using the framework so nicely laid out by Alba and Hasher (1983), we can summarize these by saying that schemas are supposed to be: selective (i.e., admitting certain "good-fitting" elements into memory but ignoring others), abstractive (i.e., eliminating irrelevant detail and surface structure while retaining the gist or essence of the information), interpretive (i.e., adding implications and inferences which relate the presented information to the person's knowledge structures) and integrative (i.e., combining new and old information together into a single integrated memory representation).

Though this is not discussed in the paper, these have been very controversial issues, but the weight of evidence seems pretty clearly to be counter to such a theory as is implied above. When, for example, a different perspective is supplied at the time of retrieval than was used at encoding (e.g., Pichert and Anderson, 1977), presumed selectivity effects can be largely eliminated. Less important detail is often inaccessible in a recall test simply because it has undergone less processing, but the fact that it has been encoded can usually be substantiated using a recognition test. Similarly, that old knowledge is used to help interpret new information is not a proposition unique to schema theory. What is unique is the assumption that the two become woven together (as well as with associated inferences), meaning that the separate identities are lost. The evidence for such encoding and comprehension effects is just not very strong. It is worth noting that schema theories of information processing inevitably predict distortions in memory as opposed to accuracy. While distortions may be inherently more "interesting" than accuracy, and apparent instances of distortion may therefore be memorable, we would probably be wise to pay some homage to accuracy and in particular to note that people's memories can be quite accurate when it is important that they be correct. Until the conditions which lead to various types of "schematic distortions" (e.g., which make newly presented information indiscriminable from old knowledge) are more carefully laid out and methodological issues associated with generating the information to be retrieved and relied upon are more adequately dealt with (see Alba and Hasher, 1983), it would be best to avoid uncritical borrowing of schema theory assumptions.

There is a second way the concept of a schema is being used, and that is far less problematical. To talk about memory schema for a particular domain of semantic or procedural knowledge seems merely a shorthand way of referring to that part of a person's cognitive structure. No schematically-based information processing assumptions are necessarily involved in that usage of the term. Whether this de-mystification of the term takes too much of the fun away from using it is possible, but unlikely.

The paper under discussion uses the term "schema" in both ways. Schema theory processing views are advanced in the front part of the paper. However, the schema notion that is crucial to the authors' hypotheses stops a little short of actually implying distortions in memory traces through either a loss of new information during encoding or through an integration of new and old information. Rather a schema is considered to be that portion of one's memory network activated by new information, and it is asserted only that "information gaps" in the new information can be filled by such a schema. The authors do not address whether in so doing the person loses the ability to distinguish new from old information or information from different sources or the degree to which the two are inextricably bound together. These are, of course, important schema theory issues.

The authors should be clearer on these issues so we know to what extent they believe subsequent encoding is influenced by pre-existing knowledge structures. From my point of view, what they are looking at in the present research are logical inferences between two closely related inputs: knowledge regarding various sources/endorsers and knowledge about the product being endorsed. When the latter information is meager, we fill in the information gaps made salient by the researchers' questions using the closely related knowledge, in this case either an actress "schema" or a politician "schema." The authors do not investigate whether or not their subjects could distinguish between the information conveyed directly or through such inference processes or the degree to which these inferences were part of a conscious problem solving strategy instigated by the questions they were to answer (as opposed to a more automatic aspect of encoding and comprehension). They, unfortunately, appear to assume the operation of the more fundamental processes without really considering the probable effects of task requirements (e.g., the lack of direct knowledge and the need to make inferences of this type to respond to the questions).

Looked at a slightly different way, research designs of this type create demand effects by asking subjects for information which the experimenter has provided in another form. Subjects understand that they are to use this other information (why else have they been given it?), and so a bias in favor of responses consistent with such information is created. This is especially likely when the form of the responding instrument is itself directive, in the sense of providing response/ retrieval cues that suggest reliance on the other information. Less direct dependent measures can sometimes help in alleviating this problem. The authors have given less than adequate attention to such methodological sources of confounding in this study. A considerable amount of revision in their study procedures is indicated for subsequent work in this area.

Above all what the field needs is a careful thinking through of the processes through which different knowledge structures influence each other and subsequent use of acquired information. More precise hypotheses need to be formulated which specify the nature, timing and source of the expected effects and the degree to which these are context dependent. Much of the difference between the consumer information processing domain and more traditionally studied information processing topics is undoubtedly due to context-related effects. We need to more adequately conceptualize and operationalize the nature of the consumer contexts which are most important to information encoding, storage and retrieval. Because this is so essential to progress in our field we should avoid the temptation to use terms such as "schema" as a cloak to hide fuzzy thinking on information processing issues, and we should not be hesitant to look carefully at all such trappings to see of what substance they are made.

REFERENCES

Alba, J. W. and Hasher, L. (1983), Is Memory Schematic? Psychological Bulletin, 93, 203-231.

Fiske, S. T. (1982), Schema Triggered Affect: Applications to Social Perception. In M. S. Clark and S. T. Fiske (Eds.), Affect and Cognition, Hillsdale, N. J.: Lawrence Erlbaum Associates.

Pichert, J. W. and Anderson, R. C. (1977), Taking Different Perspectives on a Story. Journal of Educational Psychology, 69, 309-315.

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