Do We Believe What We Remember Or, Do We Remember What We Believe?

Sharmistha Law, University of Toronto
ABSTRACT - Previous research has shown that in general, highly credible sources enhance persuasion. In this study Tulving’s (1985) "remember" and "know" procedure is used to examine what happens to the truth judgments of fictitious marketing claims when the source is falsely remembered, and when the source is forgotten. Source credibility of claims was manipulated at presentation. Perceived validity, recognition and source memory were measured at retrieval. Findings suggest that "remember" and "know" responses have a differential impact on truth judgements. Subjects also frequently reported seeing the fictitious claims before, and revealed a tendency to trust these false judgements the most.
[ to cite ]:
Sharmistha Law (1998) ,"Do We Believe What We Remember Or, Do We Remember What We Believe?", in NA - Advances in Consumer Research Volume 25, eds. Joseph W. Alba & J. Wesley Hutchinson, Provo, UT : Association for Consumer Research, Pages: 221-225.

Advances in Consumer Research Volume 25, 1998      Pages 221-225


Sharmistha Law, University of Toronto


Previous research has shown that in general, highly credible sources enhance persuasion. In this study Tulving’s (1985) "remember" and "know" procedure is used to examine what happens to the truth judgments of fictitious marketing claims when the source is falsely remembered, and when the source is forgotten. Source credibility of claims was manipulated at presentation. Perceived validity, recognition and source memory were measured at retrieval. Findings suggest that "remember" and "know" responses have a differential impact on truth judgements. Subjects also frequently reported seeing the fictitious claims before, and revealed a tendency to trust these false judgements the most.


Marketers spend large amounts of money to have a plausible spokesperson endorse their products. Michael Jordan reportedly made $36 million in 1993 alone in endorsement fees (Advertising Age). Marketers hope that through repeated exposure, the favorable attitudes towards the endorser will transfer onto the product. Indeed, past research shows that in general, highly trustworthy and/or expert sources enhance persuasion (Hovland and Weiss 1951; Sternthal, Dholakia, and Leavitt 1978; Harmon and Coney 1982). There is a rich literature in consumer behavior which investigates the man dimensions of credibility (such as source expertise, trustworthiness, country of origin) and its impact on a variety of product evaluations: brand attitudes (Homer and Kahle 1990; Stanley 1978; Goldberg and Hartwick 1990; Ratneshwar and Chaiken 1991), intention to buy versus lease (Harmon and Coney 1982), judgments of quality (White, and Cundiff 1978; Erickson, Johansson, and Chao 1984).

However, little attention has been directed to the relationship between repetition, memory for source (credible or otherwise), and its impact on belief. Yet in real life, Nike repeats its advertisement to have Michael Jordan sell its shoes. But what happens to beliefs when people remember the repeated message but forget its source (that is, forget it was Michael Jordan)? What happens if people remember the message but misremember its source (for instance, misattribute it to O. J. Simpson)? This research attempts to answer these two important questions.

The article first describes what we know about the impact of repetition on belief. It then gives an overview of the relevant literature on memory and memory measures. Following this, the design, procedure, and results obtained are discussed.


A consequence of repetition is an increase in belief. A number of studies have established that simple exposure to a claim (even when it is fictitious), enhances its perceived validity. This phenomenon is referred to as the truth effect and is remarkably robust across a variety of experimental manipulations (Hasher, Goldstein and Toppino 1977; Bacon 1979; Arkes, Hackett and Boehm 1989; Arkes, Boehm and Xu 1991; Hawkins and Hoch 1992).

Bacon (1979) was the first to suggest that familiarity (as assessed by recognition memory) is crucial for the truth effect to occur. He found that items thought to be repeated were judged more true than those thought to be new, regardless of its true repetition status. More recent studies provide conclusive support for the mediating role of recognition in the truth effect (Hawkins and Hoch 1992; Boehm 1994).

However, investigators of the truth effect have generally assumed recognition memory to be a unitary concept. But recent research in cognitive psychology has demonstrated that recognition can take one of two forms (Blaxton 1991; Tulving 1985; Gardiner 1988). An item can be recognized when its occurrence brings to mind some specific details about the previous encounter, such as the precise moment or person from whom it was learned. This is labeled recognition with recollection. Alternatively, the item may be recognized just on the feeling of familiarity. That is, recognition without recollection.

A popular paradigm for measuring the two bases of recognition (called "remember" and "know") has been developed by Tulving (1985). The procedure requires people to make judgments regarding the nature of their memories for recognized items. A "remember" experience is one for which they have vivid memories of the previous encounter with the item (perhaps by recalling the font in which it was presented, or its position on the printed page). A "know" judgment is made when subjects are certain about seeing the claim before but are unable to recall specific details about its prior occurrence. Numerous studies have shown that the distinction between "remembering" and "knowing" is meaningful to people, that people can make corresponding judgments about their memory performance, and that dissociations between "remember" and "know" responses are observed under different task conditions. For instance, levels of processing and generate-versus-read study conditions have been shown to affect "remember" responses but not "know" responses (Gardiner 1988; Rajaram 1993). Aging reduces "remember" responses but leaves "know" responses untouched (Waltr and Parkin 1990). In addition, Rajaram (1993, Experiment 3) has identified a priming manipulation which results in the opposite pattern of results: masked repetition of words generated more "know" responses without affecting "remember" responses. Furthermore, it has been established that "know" responses are not equivalent to lower confidence levels (Rajaram 1993; Gardiner and Java 1990).

Empirical evidence suggests that the distinction between "remember" and "know" might enable us to gain additional insight about the truth effect. Past research shows that marketing claims misattributed to an extra-experimental source receive the highest truth ratings (Arkes et al. 1989; Arkes et al. 1991, Boehm 1994; Law and Hawkins 1997). Arkes et al. (1989) gave subjects a recognition test in which they were required to choose one of three categories: "I have never heard this statement before"; "I have heard this statement before in this experiment"; "I have heard this statement before but not in this experiment (from a friend, newspaper, radio, class, etc.)." Note that here, identifying an item to be repeated corresponds to a "remember" response. Results showed that claims falsely attributed to a source outside the study were perceived to be the most valid. Law and Hawkins (1997), among others, have reported similar results.

Yet, the few studies that have investigated the role of source memory in the truth effect report mixed results. While Boehm (1994) reports a path analysis which shows no direct influence of source dissociation on truth ratings, Arkes et al. (1991, Experiment 1) found that source memory had only an indirect effect on truth ratings via its influence on recognition memory. A possible reason for this inconsistency is that the recognition measures used in these studies did not permit responses where subjects were confident that the item is repeated, but were unable to remember its occurrence. Specifically, the investigators of the truth effect have generally assumed that when subjects attribute a claim to an "outside" source, they are essentially falsely remembering an additional source. However, in light of the research from cognitive psychology described earlier, one might expect that the two bases of recognizing a claim will have a differential impact on beliefs.

In summary, the purpose of the current study is to further our understanding of results reported in earlier investigations of the truth effect. Specifically, this research explores what happens to truth ratings when source is forgotten. And when source is falsely remembered.


To explore what happens to truth ratings when we forget the source of a claim, and when we falsely remember it, a study was conducted which employed Tulving’s (1985) "remember" and "know" procedure in a recognition memory task. Participants were shown fictitious marketing claims paired with either a credible or not-so-credible source at learning. After a brief retention interval, the claims were re-presented together with similar, but new claims. Subjects were then required to indicate how true they thought each claim was, if they recognized seeing it before, and if so, whether they recognized it on the basis of "remember" or "know."


Thirty-eight undergraduate commerce students participated in the study in lieu of course credit. Data from four subjects could not be included in the analysis as they failed to understand or complete the dependent measure booklets.

Design and Stimuli

The study employed a completely within subject design with two levels of credibility (credible v. doubtful) and two levels of repetition (claims repeated once, "old", and tose seen once, "new"). Credibility was manipulated at presentation by dividing the list of claims into two sets of equal length: for one set, subjects were told that the statements are from dependable sources (such as Consumer Reports, Better Business Bureau, etc.) while for the other set, subjects were informed that the claims were obtained from various commercial sources (billboards, TV commercials, etc.). Data from a pilot study was used to select these sources. The contrast between the credible and doubtful source was reinforced by including obviously true filler claims in the credible block and obviously untrue filler claims in the doubtful block. The order of presentation of credibility was rotated across groups such that about half of the participants were presented with claims from a credible source first while the others were first shown claims from a doubtful source.

The stimuli were 70 fictitious marketing claims about real products taken from (Law, Hawkins and Craik 1996; Law and Hawkins 1997), such as "British Airways has flown the greatest number of transcontinental passengers." These claims had been pretested on 22 subjects to ensure that they had a mean truth rating of approximately 3.0 on a scale of 1 (Definitely False) to 5 (Definitely True), standard deviation of less than 1, and with a unimodal distribution such that they did not appear obviously true or false to some subjects. Sixty-four of these claims were randomly divided into two sets of 32 (List A and List B). One set represented the "old" claims, presented at both study and test, whereas the second set was used as "new" claims shown only at test. The remaining six claims were used as practice items presented at the beginning of the study and test lists. Because there were two study conditions, the study list was further subdivided into two sets of 16: one set presented as the credible set and the other presented as the doubtful set (see above for credibility manipulation). Five non-target obviously true or obviously false items were added to the relevant source condition to help reinforce the credibility manipulation. As in the case of source credibility, repetition status was balanced across participant groups so that each statement appeared as an "old" and "new" item about equally often.


Subjects participated in two sessions, session 1 and session 2, separated by a 15 minute unrelated filler task. They were told that the purpose of the study was to investigate the effectiveness of various marketing claims. No mention of the truth rating and recognition task was made at this point.

In session 1, each claim was presented for 10 seconds, and subjects were asked to rate its comprehensibility. During session 2, the studied claims were repeated together with similar, new claims and subjects were asked to (a) assess the validity of each claim (a 16-cm. continuous scale similar to previous studies), and (b) indicate whether they "remember" the claim (R: that is, whether seeing the claim brings back specific recollection of where it was seen earlier); whether they simply "know" the claim (K: that is, recognition without recollection of specific context); or whether they were seeing the claim for the first time ("N" for new). For the claims they "remember", subjects were required to circle the source of prior exposure from a list which had two broad options, First session and Outside of Experiment (e.g., commercial material; word-of-mouth; personal experience; News/ research reports. An "other" category was also included to allow subjects to specify a source other than the ones listed). The instructions to explain the "R" and "K" distinction followed very closely to those specified by Rajaram (1993).


The results are presented in three sections. First, ata regarding manipulation checks and counterbalancing procedures are reported. Second, the results relating to the replication of the truth effect are presented. The final section contains the analysis of the impact of memory for source and credibility of source on beliefs.

Manipulation check

A one-way ANOVA indicated that claims paired with credible sources were given higher truth ratings (=11.4) than those presented with a doubtful sources (=9.3, F(1, 33)=15.62, p< .001.

Booklet Counterbalancing

A 2 (repetition status: old vs. new) x 2 (List A vs. B) ANOVA confirmed that there was no significant main effect of list type, and no interaction between repetition status and list type with regards to the truth effect (both Fs<1).

Replication of the Truth Effect

As expected, there was a reliable main effect of repetition status, F(1, 33)=19.35, p <.001, with "old" claims given higher truth ratings compared to "new" items (11.0 vs. 10.2, respectively). This effect is the classic truth effect and demonstrates that the sheer repetition of a claim enhances its perceived truth value (Hasher, Goldstein and Toppino 1977). Also confirming prior findings, claims thought to be repeated were rated higher in truth-value (=11.1) compared to claims thought to be new (= 8.6, F(1, 33)=73.19, p<.001). This effect of perceived repetition on truth ratings also concurs with previous research (e.g., Hawkins and Hoch 1992). In addition, the effect of perceived repetition on the truth effect (= 8.6 vs. 11.1, for claims thought to be "new" vs. "old", respectively) was greater than the effect of actual repetition (=10.2 vs. 11.0, respectively). This is also in keeping with results reported by Bacon (1979) and others (Hawkins and Hoch 1992; Law 1994).





Recognition Performance

Overall, recognition accuracy was high. Table 1 displays the proportion of hits ("old" claims correctly labeled "old") minus proportion of false alarms ("new" claims labeled "old") as a function of different credibility conditions.

As the mean HR-FA scores in Table 1 shows, subjects’ recognition accuracy was appreciably high. A one-way ANOVA with credibility as a within-subject factor revealed no significant main effect of credibility on recognition memory (as measured by HR-FA ratio, F(1, 33)=1.77, p>0.1).

Impact of Loss of Memory for Source on Truth Ratings

Truth ratings for claims correctly identified as being repeated, and for which source memory is either intact or lost, is given in Figure 1. A 2 (source credibility) x 2 (source attributions: K vs. R) revealed a significant main effect of source attribution such that "K" responses were given higher truth ratings (=11.2) compared to "R" responses (=10.3, F(1, 16)=9.1, p<.01).

This main effect was qualified by a robust interaction between source credibility and source attributions, F(1, 10)=8.3, p<.02. As Figure 1 shows, recognizing a claim but forgetting the source of a claim (a "K" response) enhanced subjects’ belief in a claim when it had been learned from a doubtful source at session 1. However, the loss of source memory for a claim associated with a credible source at study did not have a significant effect on truth ratings in session 2. In other words, when source memory is lost, claims associated with a doubtful source are as credible as those learned from an "authority."

Post-hoc paired comparison t tests were computed separately for "R" and "K" responses. As expected, for !R" responses (i.e., when subjects remember the source), a significant effect of credibility was found such that claims in the credible condition were given higher truth ratings (=11.4) than were claims in the doubtful condition (=9.3), [t(60)=3.7, p<.01]. There was no difference across credibility conditions for "K" responses, [t(32)=0.7, p>.1].

These results suggest that recognition memory accompanied by source recollection has a significantly different impact on our beliefs relative to recognition memory without source recollection.

Impact of False Memory for Source (FMS) on Truth Ratings

For this analysis, claims were subdivided according to whether participants accurately remembered at session 2 that it had come from a credible or doubtful at session 1 or whether they falsely remembered having seen the claim outside the study. The mean ratings of these claim subsets are given in Table 2. A 2 (source credibility) x 2 (source attributions: first session vs. outside) ANOVA revealed a significant interaction, suggesting that FMS had a dramatic impact on truth ratings, F(1, 29)=7.27, p<.01. When subjects falsely remembered having seen a fictitious claim before, they judged the claim to be particularly true (=12.9) relative to when their source memory was accurate (= 11.2, F(1, 47)=21.38, p<.001) - presumably because they trusted the outside source more than the experimental setting.



More remarkably, amongst claims correctly recognized as being repeated, the probability of FMS was quite high (45%). In other words, these results show that not only did participants frequently commit false memory errors (i.e., they reported they had seen the statement before outside-which is not possible since they were fictitious and created for experimental purposes), but they also showed a tendency to trust these false judgments the most.


The primary results of this study can be summarized as follows: First, recognition memory with or without recollection of source had a differential impact on truth ratings. In general, "know" responses were associated with higher truth ratings relative to "remember" responses for repeated items. Second, false recognition responses for source was made quite frequently. Third, false recognition memory for source enhanced the perceived truth-value of a claim quite dramatically.

Taken together, this study showed powerful false memory effects in remembering source of message: the illusion of remembering claims that have never been seen or heard can occur quite readily. Furthermore, these "misremembered" and hence misattributed items did not just evoke a feeling of familiarity but were consciously recollected as having occurred. More surprisingly, participants perceived these "false" memories to be quite valid. In other words, it appears that people believe what they remember.

This study was designed to investigate what happens when people forget or falsely remember the source of a message. But an examination of the results suggest that a useful way to reconceptaulize this phenomenon is to consider the impact of source credibility on beliefs as an interactive function of what we learn (at the time of encoding) and what we remember (when we retrieve information) at the time of judgment. If validity judgments are tagged on to a message at the time of encoding, then information gleaned from a credible source would be believed more, regardless of whether one remembers the source from which the message was learned. In the marketing context, this means that the consumer’s belief in an advertising claim would depend entirely upon the consumer’s impression of the communicator of the message at the time of learning. However, if validity assessments are made during retrieval, then we ould expect people to believe more in those messages which they (correctly or incorrectly) attribute to a credible source at the time of retrieval. Here, the failure to remember the credible source, or mistakenly assigning it to an incorrect source at retrieval, would incorrectly inflate or deflate consumers’ belief in a claim. We are currently investigating this issue by having subjects indicate at retrieval not only whether claims remembered as having learned from the experiment had been paired with a credible or doubtful source. But also whether the outside sources were considered credible or otherwise. This measure of source memory provides a more accurate estimate of when and how source credibility impacts beliefs.

The findings reported here lend preliminary support for the position that the impact of a source’s credibility on the perceived truthfulness of an advertising claim is heavily influenced by the credibility of the source accessed at retrieval. Truth ratings for claims presented with a credible source were no different from claims paired with a doubtful source when the source was forgotten. These results have practical as well as theoretical implications. From the practical perspective, it suggests that communicating a message with the help of a credible spokesperson is not enough to enhance the persuasiveness of a message. Rather, the success of a marketer’s efforts in using a credible endorser is critically dependent upon whether the accurate endorser is retrieved at evaluation. Perhaps purchase material, reinstating the credible communicator, could be used to assist consumer’s retrieval efforts. At a theoretical level, this study demonstrates that the distinction between recognition memory with or without recollection is useful in studying the impact of source credibility on beliefs. The result also suggest that the impact of source credibility might depend more on what is remembered (or not) at the time of evaluation relative to what is learned at the time of acquisition.


Advertising Age (1993), "Whole new game without Jordan," October 11, p. 1-48.

Arkes, H. R., L. E. Boehm G. Xu (1991), "Determinants of Judged Validity," Journal of Experimental Social Psychology, 27, 576-605.

Arkes, H. R., C. Hackett and L. E. Boehm (1989), "The Generality of the Relation Between Familiarity and Judged Validity," Journal of Behavioral Decision Making, 2, 81-94.

Bacon, F. (1979), "Credibility of Repeated Statements: Memory for Trivia," Journal of Experimental Psychology: Human Learning and Memory, 5 (3), 241-252.

Blaxton, T. A. (1989), "Investigating Dissociations Among Memory Measures: Support for a Transfer Appropriate Processing Framework," Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 657-668.

Boehm, L. E. (1994), "The Validity Effect: A Search for Mediating Variables," Personality and Social Psychology Bulletin, 20 (3), 285-293.

Erickson, G. M., J. K. Johansson and P. Chao (1984), "Image Variables in Multi-Attribute Product Evaluations: Country of Origin Effects," Journal of Consumer Research, 11 (September), 694-699.

Gardiner, J. M. (1988), "Functional Aspects of Recollective Experience," Memory and Cognition, 16, 309-313.

Goldberg, M. E. and J. Hartwick (1990), "The Effects of Advertiser Reputation and Extremity of Advertising Claim on Advertising Effectiveness," Journal of Consumer Research, 17 (2), 172-179.

Harmon, R. R. and K. A. Coney (1982), "The Persuasive Effects of Source Credibility in Buy and Lease Situations," Journal of Marketing Research, 19 (May), 255-260.

Hasher, L., D. Goldstein, and T. Toppino (1977), "Frequency and the Conference of Referential Validity," Journal of Verbal Learning and Verbal Behavior, 16, 107-112.

Hawkins S. A. and S. J. Hoch (1992), "Low-Involvement Memory Without Evaluation," Journal of Consumer Research, 19 (September), 212-225.

Homer, P. M. and L. R. Kahle, (1990), "Source Expertise, Time of Source Identification, and Involvement in Persuasion: An Elaborative Processing Perspective," Journal of Advertising, 19 (1), 30-39.

Hovland, C. and W. Weiss (1951), "The Influence of Source Credibility on Communication Effectiveness," Public Opinion Quarterly, 15 (Winter), 635-650.

Law, S. and S. A. Hawkins (1997), "Advertising Repetition and Consumer Beliefs: The Role of Source Memory," in Measuring Advertising Effectiveness, William Wells, ed., New York: NY: Erlbaum, 67-75.

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

Parkin, A. J. and B. M. Walter (1992), "Recollective Experience, Normal Aging, and Frontal Dysfunction," Psychology and Aging, 7 (2), 290-298.

Rajaram, S. (1993), "Remembering and Knowing: Two Means of Access to the Past," Memory and Cognition, 21 (1) 89-102.

Ratneshwar S. and S. Chaiken (1991), "Comprehension's Role in Persuasion: The Case of Its Moderating Effect on the Persuasive Impact of Source Cues," Journal of Consumer Research, 18 (1), 52-62.

Stanley, T. J. (1978), "Are Highly Credible Sources Persuasive?" Journal of Consumer Research, 5, 66-67.

Sternthal, B., L. W. Phillips and R. Dholakia, R. (1978), "The Persuasive Effect of Source Credibility: A Situational Analysis," Public Opinion Quarterly, 42 (Fall), 285-314.

Tulving, E. (1985), "Memory and Consciousness," Canadian Psychologist, 25, 1-12.

White, P. and E. D. Cundiff (1978), "Assessing the Quality of Industrial Products," Journal of Marketing, 42 (January), 80-86.