Perspectives on Persuasion and Visual Information Processing

Barbara Loken, University of Minnesota
[ to cite ]:
Barbara Loken (1984) ,"Perspectives on Persuasion and Visual Information Processing", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 81-83.

Advances in Consumer Research Volume 11, 1984      Pages 81-83


Barbara Loken, University of Minnesota


A very common theme from past persuasion research is that the effects of experimental variables on measures of change are not always linear. A classic example is found in the fear appeal research, in which 3 "moderate fear" appeal may induce greater change than a "low fear" or a "high fear" appeal. This implies that we can give our respondents "too much of a good thing," and, consequently, decrease the effectiveness of our message.

Although the three articles presented here today are different from one another in many respects, I think that the: are also alike in that each employs independent variables that, in large quantities, may reduce message effectiveness. The Mann and Hill study was specifically designed to test the conditions under which reactance will occur. A harshly-worded message was found to induce more reactance than a politely-worded message. The Yalch and Sternthal study was designed to circumscribe the conditions under which self-referencing induces change. Yalch and Sternthal's findings imply that certain self-referencing strategies do induce change, but that when two strategies are combined the effects are not additive. Since a single underlying dimension of self-referencing is proposed for each strategy, the paper argues, in essence, that a nonlinear relationship exists between the amount of self-referencing and subsequent change.

Smith, Houston and Childers did not analyze change per se, but one wonders if a nonlinear relationship also occurs between the level of prior knowledge a consumer has and the amount of visual processing that he or she engages in. Schematics were found to be more likely than aschematics to process information visually. It seems likely, however, that highly experienced consumers may, in certain cases, revert back to simple rules of thumb (Bettman 1979) or abstract principles, which simplify the decision-making process. The Smith et al. study pertained to the case of script processing, which I'll return to later, but consider for a moment the case of processing information about a product category. A consumer who has no knowledge or experience with a product category, say, a home computer, may have difficulty visualizing either the product or usage of the product. However, a consumer who has recently purchased a home computer should have no difficulty visualizing one, and most likely, that consumer will visualize the particular home computer that he or she now owns. But what happens when a consumer has extensive knowledge or experience with several brands within a product category? In response to an experimenter's probe about the product category, would the consumer visualize one or more brands within the product category, or would the consumer answer on the basis of a more abstract representation of the product category? And do we, after repeated experiences with a product or product decision-making, retrieve visual or non-visual information about the product when reporting judgments about it, or do we simply retrieve our previously formed attitudes? Furthermore, in script processing, the content of the information contained in the script may be important. As the Smith et al. paper suggests, the effects of experience on visualization of base actions may be different from the effects of experience on visualization of idiosyncratic actions. In either case, if highly experienced consumers revert back to simple heuristics, then the possibility exists that here, too, a nonlinear relationship may be found between the level of the experimental variable (i.e., a priori experience) and the dependent measure (amount or type of visual processing).


Assuming that we do find a nonlinear relationship in a persuasive communication study, we can ask whether this relationship is really due to the moderating influence variable specified. Indeed, one aspect of the persuasive communication process that is often overlooked is the nature of the moderating influences underlying the experimental variables. There are two questions, in particular, pertaining to the experimental variables and the moderating influences underlying them that could be addressed by the researcher. First, one needs to ask whether the experimental variables did, in fact, vary along the dimension specified. This condition is necessary in order to argue that the dependent variable did or did not vary as a function of level of the dimension portrayed by the independent variable. For example, in doing research on fear appeals, it would be important to show that a "high fear" appeal did, in fact, generate more fear than a "moderate fear" appeal. Analogously, in the Yalch and Sternthal study, it would be important to verify first, that each self-referencing strategy did, in fact, induce self-referencing, and second, that the two strategies combined induced a greater amount of self-referencing than either strategy singly. If this were not the case, of course, then one could not argue that high amounts of self-referencing decreased message effectiveness. While neither the Yalch and Sternthal nor the Mann and Hill papers reported manipulation checks, and the Smith et al. study used an individual difference variable (rather than manipulating level of experience), I suspect this issue would not be a major concern in any of the three studies. Post-hoc manipulation checks would probably support the use of the labels used to describe the dimensions varied in each study. The one exception, perhaps, might be the "obtrusiveness" of the trash cans in the Mann and Hill study, in which the "obtrusive" trash can was one shaped like a bird. Since the design of it sounds quite unusual, it seems conceivable that some people might not correctly categorize this as a receptacle, and if this were the case the decorative trash can would actually be less obtrusive than the plain one.

A second question a researcher needs to ask is whether the postulated moderating influences did, in fact, occur. This is an age-old question that has plagued many persuasion theories. The most notable, perhaps, is cognitive dissonance theory, which postulates change as a function of the level of dissonance induced. In the case of reactance theory, the moderator is the amount of threat to one's decision freedom. Since such moderating influences are difficult to measure, they typically are not, and leave open the possibility of other interpretations for data results. The extent to which the proposed moderators in the three studies presented today are in fact present poses an interesting dilemma. I'd like to propose alternative hypotheses as to who change occurred in these studies.

In the Mann and Hill study, the postulated moderating influence is a threat to one's decision freedom. Specifically, reactance theory proposes that people are less likely to comply with a request when they perceive compliance as a threat to their freedom than when they do not perceive it as a threat to their freedom. Yet, an important unanswered question is whether people's perceived freedom was threatened. If someone demands that I dispose of litter, it is possible that I may feel co-opted from making up my own mind about the issue, particularly if the issue is an important one to me. However, in the case of an unreasonable demand, that appears to have no justifiable basis, a person may react to the demand not only because of a loss of perceived freedom. Consider the situation posed by Mann and Hill in several cells of their design, in which people were told that they must dispose their handout in the trash can in front of Woolco, even though this trash can was on the other end of the mall and more available trash cans were probably in sight. People must have wondered at the reason for such an unlikely request, for such a simple task as disposing of trash. They also may have wondered at the reason for disposing the handout in the one receptacle shaped like the state bird. The credibility of the person distributing the handout might also have been questioned. Therefore, alternative moderating influences under the demand/specific, obtrusive cells may exist, such as disbelief, source derogation, or indignance. Physiological measures have occasionally been used in the past to try to measure possible moderating influences, although in this case it might be more useful in measure the beliefs or moderating cognitive responses to the message. Petty, Ostrom and Brock (1981) observe in relation to reactance effects that "...the subject's own antagonistic cognitive responses may be so much more persuasive than the message's arguments that the subject may come away with an attitude opposite to that advocated (referred to as boomerang)" (p. 13). Of course, the design of the Mann and Hill study did not permit them to measure cognitive variables, but I do think that the study environment was sufficiently artificial and contrived that further analysis of the moderating influences would be useful.

One other observation about the Mann and Hill study pertains to their first hypothesis that the specificity of the request is important for inducing compliance. The authors give an example of this in their paper, i.e., that common anti-litter messages such as "Keep your environment clean" may be less effective than messages identifying specific actions, such as "Please return this can to your grocery store for recycling". In their study, the authors argue that they found evidence, too, that an abstract request induced less compliance than a polite, specific request. However, this depends on how we define "compliance". In the "abstract" condition, in which people were told, "Please dispose of properly", the appropriate measure of compliance would be disposal of the handout in any receptacle. That is, we can only measure whether people are disposing of the handout if we have data from all possible receptacles. Since the measure of compliance consisted of only two trash receptacles, out of the many available in the mall area, we're not able to say whether or not people complied with the request. In contrast, people in the polite/specific condition were asked to "Please dispose of in the trash can in front of Woolco", so that in this case it would be appropriate to measure handouts disposed in this receptacle. Therefore, while the Mann and Hill study shows the importance of agreement in specificity of measures, it does not show that people in the abstract condition were less likely to comply. In fact, it is interesting that the number of handouts recovered from the receptacle at Location 4 was nonsignificantly different for the abstract and the polite/specific conditions. Any recoveries here indicate compliance with the abstract request, but, strictly speaking, noncompliance with the specific/polite request.

The Yalch and Sternthal paper also postulates moderating influences, in this case the availability of self-referencing information and the subsequent interference of this information. More specifically, when subjects engage in self-referencing, references to self become more available. At excessive levels, then, subjects' thoughts are "flooded" with self-referencing information. Presumably, the amount of information available that is relevant to the target message is diminished when the person's memory for self-relevant information is flooded in this way. While this hypothesis is appealing in several respects, the interference notion is particularly provocative. It occurs to me that another type of interference mechanism may be important here. Specifically, when a person receives both strategy manipulations, i.e., when a person is told to relate the product to themselves and their experiences and told that they are the type of person who cares about taste, this person may, in fact, function in the role of a food critic. In such a capacity, the person may become more critical of the beverage tasted, as compared to the person who does not evaluate the beverage relative to past experiences with other beverages. Without inducing the instructions variable, the "critical" perceptions elicited by labelling may not occur. The actual taste of the beverage may be a determining factor in the evaluation. Devoting more thought to a mediocre tasting beverage may induce a more negative evaluation.

Another observation worth mentioning is that the type of "self-referencing" used in the Yalch and Sternthal study is clearly different from the use of personal references used in advertising, i.e., the use of "you". We clearly have need for- a cataloguing of the different sorts of self-referencing and their differential effects: Do all of these types of self-referencing induce the same effects? Do some induce other processes as well, or different amounts of self-referencing?


Finally, both the Smith et al. and Yalch and Sternthal papers raise some interesting possibilities concerning the manner in which people process information, the most notably whether people have dual modes of processing, and whether these modes function independently. Both papers draw upon Lord's (1980) findings which suggest that the effects of "schematic" processing can be separated from the effects of "imaginal" processing. Yalch and Sternthal argue that Lord's findings can be interpreted with respect to their "availability" hypothesis, that under high levels of self-referencing, self-relevant thoughts interfere with processing of the target information. Smith et al., in contrast, attempt to show the interrelationship between schematic and imaginal models of processing. One of the major contributions of both papers is their discussion of the possibilities for further elaboration of the Lord findings and their implicit or explicit challenges to the notion of dual processing modes.

Smith, Houston and Childers argue that scripts represent one type of schema, and describe findings which show that in the realm of scripts the two modes of processing are not necessarily independent. Those who had had experience with a job placement service - i.e., who had a previous "schema" about the service - were likely to use (vivid) visual information when reporting the script or sequence of actions one engages in at the job placement service.

I think these authors are addressing am issue that is becoming increasingly important to psychology and consumer behavior researchers. However, one of the frustrating aspects of this research area is the inability to delimit the elements which compose a schema. Although scripts are Often called a form of schema, as the Smith et al. paper argues, it is also the case that early definitions of script often makes reference to its dual nature. Abelson (1976) describes a script as a series of "vignettes", where a vignette is an "encoding of an event of short duration, in general including both an image (often visual) of the perceived event and a conceptual representation of the event" (p. 34, as quoted in Lord 1980). So, by its very nature, the script has been defined with respect to both schematic and imaginal properties. The Smith et al. results demonstrate, further, that those who have well-defined scripts in memory are likely to rely upon the visual elements contained in them.

With respect to persuasion, the specific findings from these studies have both practical and theoretical implications. The Yalch and Sternthal findings suggest that strategies such as "labelling" and "self-referencing" may not be additive in their effects on people's product evaluations, and furthermore, that we do need to carefully evaluate our definitions and measures of self-referencing. Perhaps we cannot always advocate self-referencing as an effective marketing strategy. Another important moderating variable noted in their paper, but not measured, is recall and could be investigated in future studies. The Smith et al. findings suggest several questions relevant to the per suasion process. For example, does the type of processing determine the nature of the information that people attend to? Does the type of processing (visual and nonvisual) mediate belief or attitude change? While the answers to these questions are no doubt complex, they warrant further attention in consumer behavior research, and certainly have strong implications for advertising and marketing strategy


Bettman, James R. (1979), An Information Processing Theory of Consumer Choice, Reading, MA: Addison-Wesley.

Lord, Charles G. (1980), "Schemes and Images as Memory Aids: Two Modes of Processing Social Information," Journal of Personality and Social PsYchology, 38 (February), 257-269.

Petty, Richard E., Ostrom, Thomas & Brock, Timothy C. (1981), "Historical Foundations of the Cognitive Response Approach to Attitudes and Persuasion," in Cognitive Responses in Persuasion, eds., Richard E. Petty, Thomas Ostrom, and Timothy C. Brock, Hillsdale, N.J.: Lawrence Erlbaum. 5-29.