The Effects of Repetition and Levels of Processing on Learning and Attitudes

Alan G. Sawyer, The Ohio State University
ABSTRACT - Two empirical studies about repetition and levels of processing are analyzed. Suggestions for future research are discussed.
[ to cite ]:
Alan G. Sawyer (1982) ,"The Effects of Repetition and Levels of Processing on Learning and Attitudes", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 439-443.

Advances in Consumer Research Volume 9, 1982      Pages 439-443


Alan G. Sawyer, The Ohio State University

[I thank Carl Obermiller for his helpful comments.]


Two empirical studies about repetition and levels of processing are analyzed. Suggestions for future research are discussed.


Two studies (Saegert and Young 1982; Poiesz 1982) have focused on the effects of repetition. I reviewed both these papers earlier and thus many of my and two other reviewers' comments have been already incorporated in their final drafts. However, a few remaining comments seem pertinent. This paper discusses these experiments and then briefly addresses suggestions about appropriate future directions for research about repetition and information processing.


Saegert and Young's (S & Y's) experiment has some important strengths. It uses actual print advertisements (shown via slides) for well-known national brands, is carefully controlled and analyzed, and, as I've previously advocated (Sawyer 1981; Sawyer and Ward 1979), examines delayed effects of repetition in addition to the immediate effects. Essentially, this experiment adds repetition (one or two exposures) and time of measurement (immediate or 24 hour delay) to the previous levels-of-processing experiment by Saegert (1978).

Although I have no specific criticisms about S & Y's research method and analysis, I do worry about the possibility that a reader might infer some overly simplistic conclusions from S & Y's experimental results. Perhaps future research in this area can address some of my speculative concerns which, if appropriate, may limit the external and/or construct validity of S & Y's research.

First, although I cannot dispute their results, I question the validity of S & Y's hypothesized explanation. They write:

Subsequent repetitions of ads may simply serve as opportunities for consumers to perform more of the type of elaborative activities hypothesized to occur in the "deep processing" operationalization. If this is so, it should be found that a second repetition of a brand name will increase memory following deep processing more than following shallow processing.

I would have predicted (incorrectly, I guess, based on these data and that of Jacoby, Bartz, and Evans (1978) that, with deep processing, a great deal of meaningful processing occurs during a single exposure, whereas, with shallow processing, very little meaningful processing occurs. [Like S & Y, I'll use "shallow" and "deep" even though I agree with them and others (e.g., Olson 1978) who argue that these terms are too simple.] Thus, I would predict that repeated exposures would lead to much less incremental processing under deep processing conditions than for those limited to shallow processing- a result opposite to S & Y's prediction and results. Krugman's (1962) initial speculations about repetition and low involvement (which I assume would more likely lead to shallow than deep processing) suggests that repetition is crucial for low involvement situations but not for high involvement. Later, Krugman (1972) implies chat only three exposures are necessary for complete effect -- if attention and processing are sufficiently intense (presumably a high involvement situation).

A possible explanation of S & Y's results is that the operational manipulation of deep processing (e.g., "'Have you heart of the brand name before?") restricted subjects from more elaborative processing. Focusing on brand name recognition may interfere with more elaborative cognitive processing or responding such as, "Do I agree with these statements?", "There is little merit in those claims," or "I wonder whether that could be true?" This interference might occur if the focusing on the brand name restricts subjects' attention and diverts them from the message itself. More elaborative processing might produce my hypothesized ceiling effect or, at least, cause it to occur at lower exposure levels. Also, perhaps, as Krugman suggested, the benefits of repetition continue during two or three exposures before the end in incremental effect is reached. Furthermore, it may take many (e.g., five or ten) exposures under shallow processing to get much incremental recall or recognition (Krugman's preferred measure).

As a die-hart skeptic, I worry about the external validity of S & Y's results under more mundane real viewing conditions. First, as speculated above, the results may be quite different with higher exposure levels. I would expect that, with a greater number of exposures, the difference between deep or elaborate processing and shallow or narrow processing would lessen. Of course, deep processing should never become worse than shallow processing at any level of repetition. (However, I would guess that shallow processing would less likely lead to tedium. If so, shallow processing might yield more positive results at very high exposure levels. That is, "wearout" (Axelrod 1980; Greenberg and Suttoni 1973) might occur sooner for deep processing than for shallow processing. [Obviously, we need more research about what causes boredom or irritation in advertising. In survey research, repetition is cited as a prime source by consumers (Bauer and Greyser 1967), but there may be factors such as "soft sell" vs. "hard sell" (Ray and Sawyer 1971; Silk and Vavra 1974) that interact with advertising repetition in terms of tedium/ irritation and information processing.]) However, unless a "real world" way to elicit deep or elaborate processing is found (such as has been attempted by Reid and Soley (1980) and Mowen (1980), the advertiser does not have the choice about how consumers attend or process an advertisement. Thus, the advertiser may have to settle for low level shallow processing. S & Y do not say otherwise, but I do fear that a reader might incorrectly draw such a conclusion from their results.

Part of the basis for my intuitive hypothesis contrary to that of S & Y may be based on the fact that I assume that processing is likely to change at different levels of exposure, if not tightly controlled experimentally. For example, Thomas Smith's (1885) Hints to Intending Advertisers speculated that the following might take place.

The first time a man looks at an advertisement, he does not see it.

The second time he does not notice it.

The third time he is conscious of its existence.

The fourth time he faintly remembers having seen it before.

The fifth time he reads it.

The sixth time he turns up his nose at it.

The seventh time he reads it through and says, "Oh bother"'

The eighth time he says, "Here's that confounded thing again"'

The ninth time he wonders if it amounts to anything.

The tenth time he thinks he will ask his neighbor if he has tried it.

The eleventh time he wonders how the advertiser makes it pay.

The twelfth time he thinks it may be worth something.

The thirteenth time he thinks it must be a good thing.

The fourteenth time he remembers that he has wanted such a thing for a long time.

The fifteenth time he is tantalized because he cannot afford to buy it.

The sixteenth time he thinks he will buy it some day.

The seventeenth time he makes a memorandum of it.

The eighteenth time he swears at poverty.

The nineteenth time he counts his money carefully.

The twentieth time he sees it, he buys the article, or instructs his wife to do so.

In addition to the fact that today no wise husband tells his wife to do anything, levels (or spread) of processing theory suggest a different way to explain Smith's and my speculations. Initially, depending upon the situation, processing might be either shallow or deep. (Alternatively, processing might be both shallow and deep. Nelson, Reid, and McEvoy (1977) and Bransford et al. (1979) fount similar evidence of both semantic and non-semantic processing with only one exposure.) Middle-levels of exposure would be more likely deep and elaborative whereas, at very high levels, processing likely becomes neither shallow nor deep but a kind of middle level recognition like retrieval of past processing. This speculation may be akin to Krugman's (1972) hypothesized three stages of "What is it?", "What of it?", and "Ho hum, I've seen that before." Whatever the pattern, it seems unlikely that shallow processing will remain shallow as repetition increases. Likewise, initially deep processing will likely not remain deep at-all exposure levels.

Finally, let me turn from independent variables to dependent ones, I would like to know more about the validity or appropriateness of recall versus recognition. I think the correct answer depends, as S & Y suggest, on the advertiser's needs and goals. However, these goals should, in turn, depend on what type of "memory behavior" is adopted by the consumer (or what type can be prompted by the ad copy). Bettman (1979) discusses in-store and out-of-store processing which describe when the consumer activates the memory trace. These types of processing may better be measured by, respectively, recall (semantic memory) and recognition (episodic memory). Larry Jacoby of McMaster University has coined the term "perceptual fluency" to denote a third type of learning - one that may be most valid of all for other than a few exposures. If the phrase "more research is needed" were ever appropriate, it would seem to be so for thiS question of an appropriate measure of recall - or, more likely--the appropriate measures of recall under various communication strategies and situations.


Poiesz (1982) [It's very appropriate that someone with such an unpronounceable name as Poiesz join in the investigation of the exposure-affect relationship made famous by Zajonc (that's "Zionts" not "Zajonk"). Peter Wright humorously pointed out this morning that Zajonc could use his own name in a mere exposure study. However, Peter did not realize that Zajonc and Rajecki (1969) did use both surnames in a mere field exposure study. Surely, Poiesz has the potential to continue this tradition.] has perhaps personified Jerry Olson's (1982) presidential address plea to entertain creative hypotheses about important topics. He has conceived the notion of "functional exposure" which he believes can offer an explanation of Zajonc's mere exposure results that is an alternative to those co only cited (e.g., response competition, expectancy arousal, and positive habituation and tedium). As I'll briefly discuss later, I don't think that his functional exposure explanation can explain all past mere exposure results. However, neither can any of the other explanations (see Harrison, 1977; Sawyer 1977, 1981). Despite the fact that Poiesz's cognitive explanation seems to ignore Zajonc's (1980) recent conceptualization, it does seem very worthwhile to carefully consider Poiesz's interesting ideas.

Poiesz argues that a positive frequency-affect relationship depends upon the subject's perception of the experimental task. An expectancy of a "performance task" (such as memory) will likely lead to a positive effect whereas expectation of a "non-performance" task (such as a like/ dislike evaluation) will likely result in no frequency-affect relationship. Poiesz suggests that variables manipulated in past experiments--such as type of stimulus and time-interval between exposure and ratings - are confounded by differences in subject's perception of the task expectation.

McGuire (1969) points out that there is often a research life cycle to an artifact in which it is first unknown, than noticed but perceived as a nuisance that must be controlled, than ultimately studied because artifactual behavior like all behavior is interesting in its own right. Accordingly, it seems appropriate to try to understand what psychological process underlies Poiesz's functional exposure concept -- even if it is an experimental artifact.

Although I am not confident I fully understand Poiesz's conceptualization of functional exposure, his operational manipulations suggest to me that a parsimonious explanation of his construct involves information processing. Craik and Lockhart's (1972) original conceptualization of the levels of processing framework for memory research cited as evidence the consistent superiority of recall under intentional learning compared to incidental learning. Poiesz's functional exposure construct seems very similar to past manipulations of intentional versus incidental learning.

It seems very likely that the operational manipulations in the various experimental conditions do not just manipulate uncertainty or functional exposure but also motivation to process and/or extent of processing. Indeed, these concepts may be irrevocably intertwined. A brief examination of Poiesz's manipulations may be enlightening. Condition 1 ("hard/easy task") likely manipulates motivation to process the repeated stimuli. Condition 2 ("non-performance task") seems to manipulate extent or level of processing. In fact, the instructions "all we want you to do is to give your opinion as to the quality of the slides" is similar to Bower and Karlin's (1974) manipulation of deep or elaborative processing. It also may lead the subjects to engage in what Peter Webb (1979) calls "central processing." Most theories of mere exposure focus on "initial" processing of the exposure and not on any subsequent evaluative thoughts. Since this subsequent "thinking" will likely dominate mere exposure effects -(c.f. Grush 1976), there is a low likelihood of any frequency-affect relationship in this condition. Finally, condition 3 (low interest) seems certain to lower attention in comparison to the instructions in conditions 4 and 5.

Of course my speculations are just that speculations. Without manipulation checks (Kidd 1976, 1977), I cannot assess just what is being manipulated in the various conditions. However, unfortunately, neither can Poiesz. In my review of his first draft, I suggested the need for manipulation checks. In this draft, Poiesz answers my suggestion by stating:

It was decided to refrain from manipulation checks. These either would have increased the exposure rating interval, thus confounding manipulations, or would reflect the joint effects of the uncertainty and frequency manipulation.

I disagree with Poiesz's rationale. Manipulation checks are needed to assess the adequacy of the experimental operations. Moreover, Poiesz's reasons for not including the manipulation checks are incorrect. Instead of preventing a confounding of the experimental conditions by omitting manipulation checks, Poiesz's design is in fact confounded because he does not include a manipulation check.

Using Cook and Campbell's (1976) symbols to diagram a research design, I have described Poiesz's experiment in Figure l. Poiesz did include a type of manipulation check question in one condition--condition 5 (the mere exposure condition). In that condition, he asked subjects, just after exposure but before the affect ratings, to "indicate their perceived chance of having to do a performance or task and the perceived chance of a nonperformance task (chances adding to 100%). " However, no similar question (or any other) intervened between the exposures and the affect ratings which were the dependent variable in the other four experimental conditions. Thus, the omission of any similar question in conditions 1, 2, 3 and 4 means that the "mere exposure" condition 5 is confounded by the longer interval between exposures and ratings in condition 5.

Is such a confound likely to be important? Can this confound explain the experimental results? I believe the answer is "yes" to both questions. First, note that the only condition which showed an exposure effect was condition 5. [I reject the analysis that omitted the 15 frequency level as a post hoc fishing expedition for statistical significance. The data are what are important not the TyPe I error probability (see Sawyer and Peter 1982).] Second, remember that Poiesz himself referred to Stang's (1974) review which indicated that even very short delays between exposure and measurement of affect increased the likelihood of a positive frequency-affect relationship. Therefore, Poiesz, by omitting the manipulation checks which would yield useful information, has unintentionally biased his treatments to yield results favorable to his hypothesis.



Let me briefly discuss the second experiment in Poiesz paper. He admirably tried to "explore a more consumer-oriented task." However, the relationship between perceived risk and the functional exposure construct is not clear. Furthermore, Poiesz's operational manipulations of perceived risk seem to also manipulate product quality (higher in the low risk conditions since, in the high risk conditions, the product was described as having low demand "due to their unusual "swamp-and-jungle ingredients"). I don't understand why the "pre-instruction" condition produced more positive results than the post-instruction. Since, in the latter condition, there was a longer interval between exposures and affect ratings, I would have predicted the opposite result. My guess (considerably aided by the Monday morning advantage of being able to see the results) is that the level or extent of processing was varied between the two types of instruction in some important way. Again, manipulation checks might be very informative.

In summary, I think that, although Poiesz's hypothesis about functional exposure is a very interesting one, his two experiments do not adequately test this theory. Moreover, like other mere exposure explanations, his theory does not seem to have the potential to explain all past exposure results. Functional exposure cannot explain results of positive affect formation independent of cognition (e.g., Matlin 1971; Wilson 1979). Nor can it easily explain positive effects in field experiments (e.g., Crandall 1972; Zajonc and Rajecki 1969); nor can it explain negative (Cantor 1968) or inverted U-curve frequency-affect results (Zajonc, Shaver, Tavris, and Van Kreveld 1972). However, I applaud Poiesz's creative hypothesis generation and encourage him to work further in this area. (I have a file-drawer full of repetition experiments that forced me back to the drawing board.)


In addition to further research about how to obtain more elaborate cognitive processing in ways controllable by advertisers, future studies involving levels of processing ought to examine attitude measures as well as learning. The learning results are interesting and useful, but they would be especially so if they were linked as explanations of the process by which attitude effects occur. Carl Obermiller, a doctoral student at Ohio State who will join the marketing faculty at the University of Washington, currently is working on a dissertation that manipulates spread of processing or encoding elaboration (distracted processing, structural processing, semantic processing, associative processing, and affective processing) and exposure (0, 1, 2, 3, or 6). Measurements of both recognition and evaluation will be included. Later experiments (not in the dissertation) may involve delay of measurement as well as exposure intervals more akin to mundane real levels than that of S & Y and most other repetition studies. Given the controversy about whether cognition mediates the effects of repeated exposures on affect formation (cf. Zajonc 1980) and the priority of interest by advertisers in attitudes instead of recall, it seems inappropriate to me to restrict information processing research in consumer behavior to only cognitive measures. Although cognition is of interest by itself, I am more interested in its role (or non-role) in affect formation and / or choice behavior.

Finally, I hope that we learn more about how people process repeated exposures outside of the lab under various "natural" conditions. Does the level or extent of processing vary with repeated exposures? If so, how does it differ under different advertising situations? Belch (1981) and Calder and Sternthal (1979) have studied exposure conditions much more mundane real than in the usual laboratory setting. Similar research that tries to assess the extent and type of information processing under these conditions would be very useful.

I hope future research will build on my recommendations and the important empirical work of Saegert and Young and Poiesz.


Axelrod, J. ML (1980), "Advertising Wearout," Journal of Advertising Research, 20, pp. 13-18.

Bauer, R. A., and Greyser, S. A. (1968), Advertising in America: The Consumer View, Division of Research, Graduate School of Business Administration (Boston: Harvard University).

Belch, G. E. (1981), "An Examination of Comparative and Noncomparative Television Commercials: The Effects of Claim Variation and Repetition on Cognitive Response and Message AcceptAnce," Journal of Marketing Research, 18, pp. 333-341.

Bettman, G. J. (1979), "Memory Factors in Consumer Choice: A Review," Journal of Marketing, 43, pp. 37-53.

Bransford, J. D., Franks, J. J., Morris, C. D., and Stein, B. S. (1979), "Some Genera-l Constraints on Learning and Memory Research," in Levels of Processing in Human Memory, L. S. Cermak and Fergus F. I. M. Craik, (eds.) (Hillsdale, NJ: Lawrence Erlbaum & Associates)e pp. 331-353.

Bower, G. H., and Karlin, M. B. (1974), "Depth of Processing Pictures of Faces and Recognition Memory," Journal of Experimental Psychology, 103, pp. 751-757.

Calder, B. J., and Sternthal, B. (1980), "Television Commercial Wearout: An Information Processing View," Journal of Marketing Research, 17, pp. 173-186.

Cantor, G. N. (1968), "Children's 'like-dislike' ratings of Familiarized and non-familiarized visual stimuli," Journal of Experimental Child Psychology, 6, pp. 651-657.

Carlsmith, J. N., Ellsworth, P., and Aronson, E. (1976), Methods of Research in Social Psychology (Reading, MA: Addison-Wesley Publishing Company).

Cook, T. D., and Campbell, D. T. (1979), Quasi-Experimentation: Design and Analysis Issues for Field Settings (Chicago: Rand McNally).

Craik, F. I. M., and Lockhart, R. S. (1972), "Levels-of-Processing: A Framework for Memory Research," Journal of Verbal Learning and Verbal Behavior, 11, pp. 671-684.

Crandall, R. (1972), "Field Extension of the Frequency Affect Findings," Psychological Reports, 31, pp. 371-374.

Greenberg, A., and Suttoni, C. (1973), "Television Commercial Wearout," Journal of Advertising Research, 13, pp. 45-57.

Grush, J. E. (1976), "Attitude Formation and Mere Exposure Phenomena: A Non-artifactual ExpLanation of Empirical Findings," Journal of Personality and Social Psychology, 33, pp. 281-290.

Harrison, A. A. (1977), "Mere Exposure," in Advances in Experimental Social Psychology, L. Berkowitz, (ed.). 10, (New York: Academic Press), pp. 39-83.

Jacoby, L. L., Bartz, W. R., and Evans, J. D. (1978), "A Function 1 Approach to Levels of Processing," Journal of Experimental Psychology: Human Learning and Memory, 4. pp. 331-346.

Kidd, R. F. (1976), "Manipulation Checks: Advantage or Disadvantage?" Representative Research in Social Psychology, 7, pp. 160-165.

Kidd, R. F. (1977), "Manipulation Checks: Some Further Considerations," Representative Research in Social Psychology, 8, pp. 94-97.

Krugman, R. E. (1965), "The Impact of Television Learning: Learning Without Involvement," Public Opinion Quarterly, 29, pp. 349-356.

Krugman, R. E. (1972) "Why Three Exposures May be Enough," Journal of Advertising Research, 12, pp. 11-14.

Matlin, ML W. (1971), "Response Competition, Recognition, and Affect," Journal of Personality and Social Psychology, 19, pp. 295-300.

McGuire, W. J. (1969), "Suspiciousness of Experimenter's Intents" in Artifact in Behavioral Research, R. Rosenthal and R. Rosnow, (eds.), (New York: Academic Press).

Mowen, John C. (1980), "The Availability Heuristic: The Effect of Imagining the Use of Product Perceptions," Proceedings of the American Marketing Association.

Nelson, D. L., Reed, V. A., and McEvoy, C. L. (1977), "Encircling Strategy and Sensory nd Semantic Interference," Memory and Cognition, 5, pp. 462-467.

Olson, J. C. (1982), "Presidential Address," in Advances in Consumer Research, A. MitcheLl (ed.), 9, (Ann Arbor, MT! Association for Consumer Research).

Olson, J. C. (1978), "Theories of Information Encoding and Storage: Implications for Consumer Behavior," in The Effect of Information on Consumer And Market Behavior, A. Mitchell, (ed.), (Chicago: The American Marketing Association), pp. 49-60.

Poiesz, T. (1982), "Functional Exposure and Consumer Behavior: Introducing an Alternative Hypothesis for Frequency-Affect Relationships," in Advances in Consumer Research, A. Mitchell, (ed.) 9, (Ann Arbor, MI: Association for Consumer Research).

Ray, M. L., and Sawyer, A. G. (1971), "A Laboratory Technique for Estimating the Repetition Function for Advertising Media Models," Journal of Marketing Research, 8. pp. 20-29.

Reid, L. N., and Soley, L. C. (1980), "Levels-of-Processing in Memory and the Recall and Recognition of Television Commercials," in Current Issues and Research in Advertising, (eds.), J. H. Leigh and C. R. Martin, (Ann Arbor, MI: Division of Research, Graduate School of Business Administration, University of Michigan), pp. 135-146.

Saegert, J. (1978), "A Demonstration of Levels-of-Processing Theory in Memory for Advertisements," in Advances in Consumer Research, W. Wilkie, (ed.), 6, (Ann Arbor, MI: Association for Consumer Research, pp. 82-84.

Saegert, J., and Young, R. R. (1982), "Comparison of Effects of Repetition and Levels of Processing in Memory for Advertisements," in Advances in Consumer Research, A. Mitchell, (ed.), 9. (Ann Arbor, MI: Association for Consumer Research).

Sawyer, A. G. (1977), "Repetition and Affect: Recent Empirical and Theoretical Development," in Foundations of Consumer and Industrial Behavior, A. G. Woodside, P. Bennett, and J. Sheth, (eds.) (New York: American Elsevier).

Sawyer, A. G. (1981), "Repetition, Cognitive Response, and Persuasion," in Cognitive Responses to Persuasion, R. E. Petty, T. M. Ostrom, and T. C. Brock, (eds.), (New York: Lawrence Erlbaum Associates), pp. 237-262.

Sawyer, A. G., and Peter, J. P. Jr. (1982), "The Insignificance of Statistical Significance Tests in Marketing Research, unpublished working p per, (The Ohio State University).

Sawyer, A. G., and Nard, S. (1979), "Delayed Effects of Advertising," in Research in Marketing," 2, (Greenwich, CT: Jai Press), pp. 259-315.

Silk, A. J., and Vavra, T. G. (1974), "The Influence of Advertising's Affective Qualities on Consumer Response," in Buyer/Consumer Information Processing, (eds.) G. D. Huges and M. L. Ray, (Chapel Hill: University of North Carolina Press).

Smith, T. (1885), Hints to Intending Advertisers,(London).

Stang, D. J. (1974), "Methodological Factors in Mere Exposure Research," Psychological Bulletin, 81, 12, pp. 1014-1025.

Webb, P. H. (1979), "Consumer Initial Processing in a Difficult Media Environment," Journal of Consumer Research, 6, pp. 225-236.

Wilson, W. R. (1979), "Feeling More Than We Can Know: Exposure Effects Without Learning," Journal of Personality and Social Psychology, 37, pp. 811-821.

Zajonc, R. B. (1980), "FeeLing and Thinking: Preferences Need No Inferences," American Psychologist, 35, pp. 151-175.