Toward a New Understanding of the Effects of Adverstising: a Look At Implicit Memorial Processes

Ida E. Berger, University of Toronto
[ to cite ]:
Ida E. Berger (1991) ,"Toward a New Understanding of the Effects of Adverstising: a Look At Implicit Memorial Processes", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 688-692.

Advances in Consumer Research Volume 18, 1991      Pages 688-692


Ida E. Berger, University of Toronto

[The author thanks Karen Finlay for her contribution to this summary paper and gratefully acknowledges the financial support of the Social Sciences and Humanities Research Council of Canada, Grant # 410-90-1431.]

This paper summarize the contents of a special topic session (session 9.1) held on Sunday, October 7th. Papers were presented by Susan E. Heckler, Christopher P. Puto and Francoise Jaffe (University of Michigan); Carol Pluzinski (New York University); Douglas M. Stayman (Cornell University) and Frank R. Kardes (University of Cincinnati) and Ida E. Berger and Karen Finlay (University of Toronto). The-session was chaired by Ida E. Berger with Larry Percy (LINTAS: USA) serving as a discussant.


Within the last few years several consumer behavior researchers have been investigating the role of implicit memorial processes in persuasion. Working independently, these investigators have developed analytical frameworks and research methodologies to test several interesting and new hypotheses regarding the impact of non-conscious, spontaneous or implicit responses to advertising. The fundamental goal of this session was to bring these researchers together to share with each other and other ACR participants the results of their pioneering work. Taken together this body of research represents a new perspective on advertising effects.

Historically, advertising effects have been studied within the context of attitude-change paradigms. The underlying assumption of most advertising studies has been that an advertisement (or its repetition) must change attitudes in order to be effective. Much effort has consequently been expended in understanding what executional, repetition or situational variables lead to the encoding/retrieval of desired information and the formation of positive attitudes. Similarly, we have seen much progress in the development of measures of advertising effectiveness that test the conscious, controlled or explicit effects of advertising exposure (i.e. Day after Recall and Persuasion scores). While the limitations of this perspective and these measures have been discussed before, no integrated new perspective has appeared. The four papers presented in this special session did not, in and of themselves, define a new perspective, but they raised some important issues and pointed out some interesting new research directions.




The session began with a paper by Heckler, Puto and Jaffe. Heckler laid the foundation for the empirical work by pointing out the potential of recent developments in the investigation of non conscious and automatic information processing in psychology (cf. Hasher and Zacks 1979, 1984; Zajonc 1980; Zajonc and Markus 1982; Lewicki 1986). Non-conscious information processing refers to human activities which are based on a complicated set of rules that are (1) not acquired through the mediation of consciousness, (2) not operating at the level of consciousness and (3) not consciously available (Lewicki 1986). Examples from everyday life include speech production, recognition of faces, judging beauty, etc.

Heckler pointed out that the outcome of such processes may be the development of memorial associations which are utilized in subsequent conscious and unconscious processing activities. Studying these processes is not seen as an alternative to the examination of consciously controlled mental activities, but rather, as a potentially important part of a general effort to understand consumer information processing and behavior. For example, consumers may process and acquire product information without being aware of it, or decision rules and shopping procedures may utilize algorithms or memory traces acquired non-consciously.

Research in cognitive psychology (e.g., Hasher and Zacks 1979; 1984) and social cognition (e.g., Lewicki 1986; Zajonc 1980) has recently examined the effects of non-conscious processes on higher level cognitive operations such as the information of memory traces and the development of preferences. One particularly interesting series of studies (Lewicki 1986) demonstrated support for the learning of covariation information (e.g., that short haired women were "kind" and long haired women were "capable"), despite subjects' inabilities to identify such covariation when consciously evaluating the stimuli. Another set of studies showed that in the absence of memorial or other externally available information, such non-consciously acquired associations may be utilized in forming preferences or making judgments regarding similar stimulus objects. Once again, subjects were unable to consciously access the covariation information which appeared to have influenced the subsequent judgments.

In a study that carefully mirrored Lewicki's (1986) methodology but used marketing oriented stimuli, Heckler and Puto (1987) showed that memory traces which linked packaging information with product attributes (for example, tall, slender shampoo bottles with "economical" shampoos and short, rounded bottles with "conditioning" shampoos) could be non-consciously acquired. Two questions not addressed in any of Lewicki's experiments nor in the Heckler and Puto research are (1) the potential decay over time of memory traces acquired through non-conscious mechanisms and (2) the effects of "noise" on these non-conscious processes. Each of these issues is especially relevant to a consumer behavior environment. The strength of the associations over time must be examined because some research has suggested that processes"... may not last long enough for the person who views an advertisement on television, say, to get to the grocery store (Kihlstrom 1987, p. 1449). Additionally, in most consumer goods categories few associations are completely consistent (i e., not all shampoos of a certain bottle shape share the same positioning strategy).

Accordingly, Puto then presented new empirical work utilizing the experimental techniques and measures developed by Lewicki and extended by Heckler and Puto (1987). Specifically, he presented the results of two studies, one which measured the impact of the non-conscious information acquisition immediately after exposure and after a 4-7 day delay and a second that added new non-conforming attribute information ("noise") to the product descriptions.

The results showed that covariation information acquired non-consciously was accessible to influence subsequent processing even after a 4-7 day delay. However, noise present at the time of processing rendered the target covariation information inaccessible. These results demonstrated that in noise free circumstances, the effect of non-consciously acquired information on judgements is robust and impervious to the effects of time delays. The results provide insight for advertising theoreticians and practitioners regarding the way in which non-conscious information processing might affect memory and judgements of marketing stimuli.



The second paper in the session continued the "non-conscious processing" theme by exploring the notion of "partial activation". Reporting on work from her dissertation (Pluzinski 1990), Carol Pluzinski discussed the fact that current recall based measures of advertising effectiveness assume that access to information about a brand name is an all or nothing process. In other words, it is assumed that either a brand name is activated in memory or it is not. Yet consumers may have vague, transient or even feeling oriented reactions to ads that cannot be verbalized and thus cannot be captured by simple recall based responses. Contrary to existing frameworks, Pluzinski introduced the idea that there exist various levels of activation of a concept in memory and suggested that information that is only "partially activated" may have systematic effects on consumers' impressions/reactions to brands.

Partial activation refers to a state in which a concept (i.e., idea, thought, judgment, evaluation) is not fully activated in memory; it is the state in which a concept is activated, but below the threshold of conscious awareness. Pluzinski argued that partial activation may be an important element in either the process of encoding (people may not always fully attend to ads) or the process of retrieval (people do not always fully remember ads). Her study examined how certain consumer responses to advertisements, while inaccessible to conscious awareness, might still be activated (partially), and thus be accessible for non-conscious, automatic processing. Hence, a finer grained representation of reactions to advertising was provided.

Using existing advertising slogans as primes, Pluzinski's study created conditions of both full and partial activation of a brand name, and then compared the effects on subsequent processing of that brand, including evaluations. The findings showed that full activation of a brand name (correct recall following the slogan prime) facilitated subsequent processing and lead to more favorable evaluations. Partial activation, by contrast, was associated with interference, inhibition and evaluations that were based on very subjective feelings, (meta-cognitions such as strong feelings of knowing). She concluded that advertisers should not assume that a brand name below a recall threshold will not be activated at all. In fact it may be partially activated and thereby may influence subsequent judgements. However, she cautioned that unless the brand is consciously processed (i.e., totally activated), its evaluation will not be automatically activated.



The third paper (prepared by Doug Stayman and Frank Kardes and presented by Frank Kardes) focused on individual differences that lead to the formation of strong brand attitudes. The paper was based on recent findings indicating that omitting explicit conclusions from advertisements and thereby inducing consumers to infer their own conclusions facilitates the development of strong brand attitudes (Kardes 1988; Sawyer 1988). Kardes argued that this subtle indirect approach to persuasion offers many advantages over more traditional approaches. For example, with indirect persuasion: (a) counter-arguing is minimized, (b) reactance is avoided, (c) self-generated arguments are more credible and memorable than explicit arguments, and (d) attitudes based on self-generated inferences are accessible, confidently held, and exert a strong influence on subsequent judgments and decisions (for a review, see Kardes forthcoming).

Stayman and Kardes extended previous approaches by examining the role of two key inference processes in persuasion: inference generation and inference utilization. They predicted that any variable that influences the extent to which consumers elaborate on the contents of a persuasive text should affect inference generation. Following Kardes (1988), a mock ad was created and response latencies to inferential and attitudinal inquiries were used to test this hypothesis. As expected, the likelihood of spontaneous inference generation was greater for high (as opposed to low) need for cognition individuals (Experiment 1) and in high (as opposed to low) involvement conditions (Experiment 2).

Kardes further argued that consumers who generate inferences do not necessarily use these inferences as inputs for brand attitude formation in all cases. Indeed, high need for cognition individuals engage in elaborate processing simply because they enjoy the process of thinking, not because they are motivated to satisfy some extrinsic goal. However, individuals who are attuned to internal, self-generated information - such as attitudes, opinions, and inferential beliefs - are likely to use inferences in brand attitude formation, given that inferences had been formed spontaneously and are available for use. Hence, he predicted that inference utilization would be more likely for low (as opposed to high self-monitoring individuals). Consistent with this prediction, more extreme, more confidently held, and more accessible brand attitudes were formed when consumers were likely to both generate (i.e., high need for cognition individuals) and use (i.e., low self-monitoring individuals) inferences as inputs in brand attitude formation.

The experimental results were consistent with a two-stage inference process model in which variables that influence elaborate processing also affect inference generation, and variables that influence sensitivity to self-generated information also affect inference utilization. Extreme, confidently held, and accessible judgments are formed when consumers are likely to generate and utilize inferences derived from explicit product information. Kardes concluded by emphasizing the important insights that an understanding of processing effects can yield. In the Stayman and Kardes study for instance, the strategic deletion of portions of a persuasive message created a more effective and compelling message, that lead to stronger brand attitudes for some individuals in some situations.



The last paper in this session was prepared by Berger and Finlay. Berger began the presentation by pointing out that psychological and consumer behavior researchers are exhibiting growing interest in the notion that attitudes have at least two dimensions: valence and strength. Furthermore, recent evidence in the advertising repetition area (Berger and Mitchell 1989) indicates that advertising exposures can have a strong impact on at least two aspects of attitude strength (attitude accessibility and confidence) and thereby on attitude-behavior consistency. Berger and Finlay's paper introduced a Two-Stage model of how attitudes guide behavior. The model disentangles the determinants, memorial processes and behavioral consequences of high attitude confidence and high attitude accessibility.

BerBer argued that in order for a previously formed attitude to guide a subsequent decision, two processes must occur. First, the attitude must be activated from memory. This may happen in a controlled fashion, as when an individual tries to consciously "remember" how they feel about an object, or in a spontaneous fashion, as when their feelings simply "pop" into their consciousness in the presence of the object. Second, the activated attitude must be accepted as a piece of information upon which a decision can be based. An individual who has active an attitude that s/he finds "unacceptable" may not behave in accordance with this attitude. Rather, under some circumstances, the behavioral implications of the previously formed attitude may be overwhelmed by new information available in the situation.

Attitude accessibility and attitude confidence may play distinct roles in this "Two-Stage attitude-to-behavior process". Attitude accessibility may influence the likelihood - of spontaneous activation of the attitude and thereby may influence the first stage of the process. Attitude confidence, on the other hand, may influence the likelihood that an attitude is considered acceptable or relevant to the decision and thereby may influence the second stage of the process. Berger pointed out that this Two-Stage model assumes that attitude accessibility and attitude confidence represent distinct psychological constructs accomplishing unique functions in the a-b process. Furthermore, the model suggests that under some behavioral circumstances, an attitude that is highly accessible, but held with low levels of confidence, will not be very predictive of subsequent behavior.

These propositions were tested in a laboratory experiment that used repeated attitudinal expression and advertising repetition to independently manipulate attitude accessibility and attitude confidence, respectively. The results indicated that attitudes that were repeatedly expressed were indeed more accessible from memory, but were held with no more confidence. Attitudes that were based on multiple ad exposures were indeed held wit more confidence, but were no more accessible from memory. In other words, the factors expected to influence the constructs uniquely, did so, and importantly, under these circumstances, the constructs exhibited no inter-relationship. Furthermore, in a behavioral situation in which subjects had the opportunity to reconsider their attitudes, attitude-behavior consistency was moderated only by attitude confidence. There was n evidence of any influence due to accessibility.

Not only were these results consistent with the predictions of the Two-Stage model, but, as Berger pointed out, the results have direct implications for advertisers. Research on advertising has assumed that the effects of exposure can be completely captured in an ad's effect on a brand's evaluation. By contrast, this study showed that under some circumstances advertising does not influence brand attitudes but rather influences how accessible attitudes are from memory or how confidently they are held. Therefore, via its influence on attitude strength (accessibility or confidence) advertising can influence the strength of the relationship between attitudes and behavior.


The discussant for this special session, Larry Percy from Lintas: USA, commented on the applicability of the work to advertisers and advertising agencies. Although he indicated that it was interesting and important to understand advertising effects in terms of implicit memorial processes, he argued that researchers' consideration of some real-world constraints might provide more immediate utility to advertisers. In particular, he pointed out that most of the processes advertisers deal with are visually driven (eg. recognition of product in the purchase situation), so that verbal recall of brands and brand information may not be all that relevant. On the other hand, the results might be more exciting if the measures could be extended to include non-verbal stimuli and responses.

Secondly, since consumers often make purchases in a stimulus-based situation, recognition may be a better measure of advertising effectiveness than recall, particularly for low involvement products. In fact partial activation may be an indicator that a brand will likely be recognized in a purchase situation. This may be an interesting area for future research.

Thirdly, as interesting as the study of the effects of individual difference variables on processing is, if those kinds of individuals can't be isolated in the population, the results do not have immediate applicability. Direct matching of target demographics just does not seem to be affordable in today's media buying situations. From a managerial perspective, greater attention to variables that have a more "general" effect on processing might be more attractive.

Finally, Percy questioned the relevance of attitude confidence in low involvement purchase situations, but acknowledged that for high involvement products, it may make sense to use advertising frequency to drive up attitude confidence.


By way of conclusion it might be useful to highlight the three dominant themes addressed in these papers. First, all four papers focused on the accessibility, as opposed to the content of information in memory. Previous models have given advertisers the tools to influence the content of memory (attitude valence, specific beliefs) but these papers pointed out that with a better understanding of accessibility, advertisers could actually influence what associations (Heckler, Puto and Jaffe), brand names (Pluzinski), inferences (Stayman and Kardes), or attitudes (Berger and Finlay) are activated in a behavioral situation.

Secondly, all four papers concentrated on the process by which memorial information is activated or utilized. The dominant question in these papers was not whether a memorial concept was available for retrieval (i.e., covariation information following non-conscious learning, Heckler, Puto and Jaffe; brand names following a slogan prime, Pluzinski; inferences following conclusion omission, Stayman and Kardes; attitudes following advertising expression, Berger and Finlay) but whether and how this memorial information might influence subsequent processing. By understanding the variables that influence these processes, advertisers will be better equipped to influence market outcomes.

Finally, all four papers grappled with the possibility that the processes being investigated are at least partially spontaneous. Traditional advertising researchers have shed considerable light on the nature of controlled, conscious information processes, while recognizing that not all processes are of this kind. The papers in this session (particularly Heckler, Puto and Jaffe and Pluzinski) tried to understand, model processes that occur outside of conscious awareness, that cannot be articulated and often can only be inferred from resulting psychological or behavioral states. These spontaneous, implicit, uncontrolled processes represent a part of the 'black box' that advertisers are just beginning to explore.

The perspective advocated by all presenters in this session moved well beyond traditional attitude-change paradigms. According to these researchers, even without changing brand evaluations, advertising that induces non-conscious learning, or only partially activates memorial information, or induces self-generated inferences or creates confidently held attitudes can be very effective. The methods developed in these studies and the results presented should contribute to a new understanding of the effects of advertising.


Berger, Ida E. and Andrew A. Mitchell, (1989) "The Effect of Advertising on Attitude Accessibility, Attitude Confidence and the Attitude-Behavior Relationship", Journal of Consumer Research, 16 (December) 269-279.

Hasher, Lynn and Rose T. Zacks (1979), "Automatic and Effortful Processing in Memory", Journal of Experimental Psychology:General, 108, 356-388.

Hasher, Lynn and Rose T. Zacks (1984) "Automatic Processing of Fundamental Information," American Psychologist, 39 (December), 1372-1388.

Heckler, Susan E. and Christopher P. Puto (1987) "Unseen Effects of Advertising:Non-conscious Consumer Information Processing", presented at the Annual Conference of the Association for Consumer Research, Cambridge, Massachusetts.

Kardes, Frank R. (1988), "Spontaneous Inference Processes in Advertising: The Effects of Conclusion Omission and Involvement on Persuasion", Journal of Consumer Research, 15 (September), 225-233.

Kardes, Frank R. (forthcoming) "Consumer Inferences: Determinants, Consequences and Implications for Advertising", in Advertising, Exposure, Memory and Choice, ed. Andrew A. Mitchell, Hillsdale, NJ:Erlbaum, in press.

Kihlstrom, John F. (1987) 'The Cognitive Unconscious", Science, 237.

Lewicki, Pawel (1986), Non-conscious Social Information Processing, Orlando, FL: Academic Press, Inc.

Pluzinski, Carol (1990) Automatic Processes in Consumer Response to Advertising, Unpublished Doctoral Dissertation, University of Michigan, Ann Arbor.

Sawyer, Alan G. (1988) "Can There Be Effective Advertising Without Explicit Conclusions? Decide for Yourself," in Nonverbal Communication in Advertising, eds. Sidney Hecker and David W. Stewart, Lexington, MA:Lexington, 159- 184.

Zajonc, Robert B. (1980) "Feeling and Thinking:Preferences Need No Inferences," American Psychologist, 35 (February), 151-175.

Zajonc, Robert B. and Hazel Markus (1982), "Affective and Cognitive Factors in Preferences," Journal of Consumer Research, 9 (September), 123-131.