Quantity matters: The frequency of deception influences automatic memory retrieval effects

Abstract

We investigated automatic retrieval of the knowledge of having lied or having told the truth to a question, depending on (a) the quality of the statement (true vs. false response) and (b) the overall proportion of (dis-)honest responses. We therefore manipulated the proportion of lies and truths being told in an oral interview. Automatic retrieval of this meta-knowledge was assessed with a categorisation task, where the probe words dishonest and honest had to be classified, while questions from the interview served as task-irrelevant prime stimuli. Results revealed an automatic retrieval of knowledge about having lied to a question only for participants who had told few lies in the interview, but not for those who had told many lies. No retrieval effects were obtained regarding questions that had been answered truthfully. These findings suggest a combined influence of quality and quantity of dishonest statements on automatic memory retrieval, thereby being in accordance with recent accounts of action control.

Keywords

Lying automatic processes instance-based learning distinctiveness

Recent research provided first insights into a basic mechanism that keeps track of the lies a person has told in response to certain questions. Specifically, as soon as a person encounters a situation that is similar to the one in which the lie was told before knowledge about having lied is automatically retrieved from memory (Koranyi et al., 2015). The approach has its roots in theories of instance-based automatisation of behaviour (Logan, 1988; see also Denkinger & Koutstaal, 2009; Hommel, 1998, 2004; Rothermund et al., 2005), which propose that the execution of an action in a specific situation is stored as an episodic unit in memory together with certain situational cues (alternatively called an event file, Hommel, 1998, 2004, instance, Logan, 1988, or stimulus-response episode, Rothermund et al., 2005). Encountering a similar situation again leads to an automatic activation of this episodic unit, enabling people to react fast and consistently across different situations. This episodic binding and retrieval perspective can explain a vast range of research findings across various paradigms in the field of cognition and action control and automatisation (e.g., Frings, Hommel, et al., 2020; Frings, Koch, et al., 2020; Schmidt et al., 2016). This model can also be applied to the field of deception, where it has been argued that the knowledge about having lied to a specific question is expected to be retrieved automatically from long-term memory when the same question is encountered again (Koranyi et al., 2015).

In the current study, we aim to extend the investigation of an automatic retrieval of knowledge about having lied by also taking into account contextual factors. Specifically, we are interested in whether the overall relative frequency of lies and truthful statements modulates the retrieval of a specific lie (or truthful statement). Such an effect of the contextual frequency of lies versus truthful statements has important theoretical as well as practical implications.

From a theoretical perspective, contexts play an important role for retrieval processes (e.g., Frings, Hommel, et al., 2020; Frings, Koch, et al., 2020): Contexts can either serve as direct retrieval cues themselves (Tulving & Thomson, 1973), or they can modulate retrieval by increasing or decreasing the distinctness of an episode. Research has repeatedly shown that stimuli and episodes that deviate from their surroundings are perceived as being more distinct and tend to attract attention (Duncan & Humphreys, 1989), which in turn facilitates memory encoding and retrieval (so-called “von Restorff effect,” Hunt, 2006; von Restorff, 1933). On the contrary, high similarity with the context reduces distinctness and has a negative effect on memory for a specific stimulus or episode. Demonstrating such an influence of the contextual frequency of lies versus truths on related binding and retrieval effects for questions and the truth values of the corresponding answers thus provides another test of episodic binding and retrieval models in the domain of deception and memory for one’s own lies. Demonstrating such an influence extends the scope of application of episodic binding accounts to these contexts, which can be considered to reflect evidence for a bottom-up modulation of episodic encoding, binding, and/or retrieval (Frings, Hommel, et al., 2020; Frings, Koch, et al., 2020). Furthermore, it would also provide further evidence for an episodic binding and retrieval account of memory for lies (Koranyi et al., 2015; Schreckenbach et al., 2019).

Variations in the frequency of lies versus truths are also interesting from a more practical or applied perspective. Differences in the default frequencies of lies can reflect personal traits (e.g., relating to honesty/dishonesty), but they may also reflect specific situational settings (e.g., different standards, norms, expectations, or temptations with regard to telling the truth or a lie, or acting in socially desirable vs. authentic ways). If our assumptions are correct, then these differences in the frequency of lying may also influence memory for one’s own (and other people’s) lies. For instance, frequent liars should have more difficulty in remembering their own lies, and will consequently run into difficulties with regard to keeping up a consistent image of themselves. It may also be more difficult to track one’s own (or another person’s) lies if this lie occurred in a context that is characterised by a high frequency of lies (e.g., in contexts in which deviations from what is socially desirable are sanctioned).

To test automatic memory for lies, Koranyi et al. (2015) developed a paradigm in which participants took part in an oral interview and were prompted to tell the truth to half of the questions being asked but to lie to the other half. To assess automatic retrieval of knowledge about having lied, a simple categorisation task was used in which the probe words honest (in German: ehrlich) and dishonest (in German: gelogen) had to be identified by pressing a corresponding key. In this task, the questions of the interview were presented as task-irrelevant primes to test whether identification of a probe word was facilitated by a corresponding prime question. As expected, the presentation of a question that had been answered dishonestly led to faster identification of the probe word dishonest compared with honest, suggesting an automatic retrieval of the knowledge of having lied to a question when the question is encountered again. Importantly, no comparable congruency effect was obtained for those questions to which participants had responded honestly during the interview, indicating that an automatic retrieval of information about the truth-value of previous statements was restricted to lies.

The aim of the current article is to extend this approach to the question of whether the frequency of lies has an influence on automatic memory processes. According to current theorising on automatic binding and retrieval processes, these episodic memory processes are influenced by contextual factors reflecting involuntary bottom-up processes (e.g., salience, figure/ground asymmetries; Frings & Rothermund, 2011, 2017). With our present study, we want to investigate whether these boundary conditions also have an influence on memory for lies versus truthful statements. Before describing the specific design and hypotheses of our study, we first introduce two different properties of lies which might influence these processes, namely their quality and their quantity.

Quality of lies

According to models of lie production, the asymmetry in encoding and retrieving lies versus truths could stem from qualitative differences between the two kinds of statements. Various lines of research suggest that giving a lie is qualitatively different from telling the truth. First, it was suggested that lying is more cognitively demanding than telling the truth (e.g., Sporer & Schwandt, 2006, 2007; Vrij et al., 2010; Walczyk et al., 2003, 2014). There are several reasons for this assumption: Because liars have to think of a novel answer, they need to suppress the truth, activate the deceptive answer deliberately (while the truth is often activated automatically), they have to remind themselves to role-play, and they tend to monitor their own behaviour and the targets’ reactions to a greater extent than do truth-tellers.

Walczyk et al. (2014; see also Walczyk et al., 2003) go one step further within their activation-decision-construction-action theory (ADCAT), where they specify the particular processes that have to occur before a lie can be successfully delivered. The authors assume that the default mode of responding is to answer honestly to a question, with the truth being activated in an automatic fashion, whereas deceiving requires additional controlled processes to take place. These deliberate processes contain the actual decision to lie, its construction, and its execution (the action component) while controlling one’s behaviour and monitoring the target’s reactions to the lie.

Models that distinguish true and false statements in terms of qualitative features provide a straightforward answer to explain the asymmetry in memory for truths and lies: Investing effort, deliberation, resources, and/or cognitive control should lead to deeper levels of processing, better encoding, and more efficient retrieval of lies compared with truths.

Quantity of lies

Another account to explain this asymmetry focuses on external rather than internal features: Recent research has invested a lot of effort to gain insights into the frequency of lies. Based on a diary study, DePaulo et al. (1996) concluded that people tell approximately one lie a day. From a moral point of view, this looks like a lot but given the mean amount of social interactions a day (being defined by DePaulo et al. as “any exchange between you and another person that lasts 10 min or more . . . in which the behavior of one person is in response to the behavior of another person.”) and the fact that during one of these interactions we typically tell more than one truth, lies are still substantially less frequent than telling the truth. Furthermore, recent findings suggest that the finding of people lying once per day on average might be based on a skewed distribution containing a few prolific liars telling several lies a day while most people indeed do not lie every day (see Serota et al., 2010).

Being less frequent renders lies more distinctive, and according to Hunt (2006), distinctiveness is a factor that facilitates memory retrieval (von Restorff effect; von Restorff, 1933). A low relative frequency of occurrence alone thus might already account for an enhanced memory for lies compared with truths, due to distinctive processing. Importantly, this explanation does not draw on specific qualities of lies (e.g., cognitive effort, controlled construction), but explains the memory asymmetry for lies and truths solely on the basis of their relative quantity.

This explanation implies that the asymmetry of retrieving lies versus truths is not an inherent feature of memory, but that it can be reversed by changing the relative proportion of lies and truths: If the low quantity of lies (i.e., their high distinctiveness) is what causes their being stored and retrieved, then it should be possible to reverse the asymmetry by making lies the default and turning true statements into a rare exception.

The current study

To test for the influence of quality and quantity on memory processes for truths and lies, we conducted an experiment that used the same paradigm as Koranyi et al. (2015). In addition to the manipulation of having to give true versus false answers to certain questions, we manipulated the overall quantity of lies and truths in our current study. Some participants had to tell only few (i.e., 25%) lies and many truths in the oral interviews, replicating the standard frequency asymmetry of everyday lies and truths. For another group of participants, however, this asymmetry was reversed (i.e., 75% lies, 25% true answers), which should render true statements more distinct.

Combining the above considerations, three possible predictions are possible: (a) only the quality of lies (i.e., low vs. high effort in producing honest vs. dishonest answers) is responsible for memory processes; (b) only the relative quantity (i.e., high vs. low distinctiveness) of a certain type of statement is what drives memory processes; and (c) a combination of both quality and quantity is necessary for memory encoding and retrieval (i.e., only distinct lies are encoded). Each prediction relates to a specific set of statistical hypotheses (see Figure 1, for a visual illustration of the predicted patterns):

(1) According to the quality account, a lie has an increased chance of being encoded into episodic memory due to the processes that are characteristic for lies (cognitive capacity, resources, depth of processing), independently of the context (see Figure 1a). In statistical terms, this prediction corresponds to a simple two-way interaction of the type of the prime question (answered truthfully vs. dishonestly) and the probe word (honest vs. dishonest). This interaction effect should be driven entirely by questions to which one has lied—no effect should be obtained for questions that were answered truthfully. Furthermore, the pattern of prime question × probe word congruency effects should not differ between the rare lies and the rare truths conditions, resulting in an absence of the three-way interaction.

(2) According to the quantity account, a statement is encoded into memory if it is distinct, that is, if it deviates from the context in which it is given: Accordingly, not only lies, but also truths can be stored and retrieved from memory if they are made in a context in which they are rare (see Figure 1b). Statistically, this pattern would also produce a two-way interaction between prime question and probe word. This interaction should be present for both the rare lies and the rare truths contexts, but these two interaction effects should be driven by different components of the prime factor: Dishonestly answered prime questions should produce faster responses for the probe word dishonest, compared with the probe word honest, whereas prime questions for which an honest answer had been given during the interview should not affect responding to the probe words. In the rare truths condition, however, a reversed pattern should emerge, with honestly answered prime questions having a facilitative effect on the probe word honest compared with the probe word dishonest, whereas no such effects should occur for prime questions to which one had lied. In statistical terms, this asymmetry corresponds to an additional interaction effect between probe word and context (in the rare lies context, responding should be faster for the probe word dishonest, whereas in the rare truths condition, responses should be faster for the probe word honest). Like the quality account, however, the quantity account does not predict a three-way interaction.

(3) According to the combined account, both properties (the quality of having told a lie, and the fact that lies are rare) are needed for episodic memory storage and retrieval. Neither truths (due to their lack of drawing on cognitive resources, resulting in shallow processing) nor lies that are given in a context of many other lies (due to their lack of distinctiveness) are stored and automatically retrieved from memory (see Figure 1c). Statistically, this prediction would result in a three-way interaction between prime question, probe word, and context: The combined account predicts an effect of the prime question × probe word interaction within the rare lies context (which should be driven by the questions to which one had lied). However, there should be no interaction between prime question and probe word within the rare truths condition.

Figure 1.

Predicted response time patterns for the qualitative ((a) cognitive effort during lie production causes episodic binding/retrieval), quantitative ((b) distinctness causes episodic binding/retrieval), and combined ((c) lack of resource investment or distinctness prevent episodic binding/retrieval) models as a function of context (rare lie vs. rare truth), prime question (lie vs. truth), and probe word (honest vs. dishonest).

Because the presence of a null effect (as it is predicted for the rare truths condition in Hypothesis 3) requires additional assumptions regarding the strength of a to-be-detected effect when using frequentist statistics, we additionally calculated Bayes factors for all three hypotheses, which also provides information on the likelihood of the null hypothesis in relation to the alternative hypotheses.

Method

Participants and design

The study has been conducted in accordance with ethical standards and was approved by the Ethical Commission of the Faculty of Social and Behavioural Sciences of the University of Jena. Sample size was determined relying on the effect size of $η_{p}^{2}$ = .08, found by Koranyi et al. (2015) in a similar analysis for a three-way interaction.¹ Accordingly, a power analysis for a two-sided F-test was conducted based on this value, setting α = .05 and 1 − β = .80, which led to a proposed sample size of 96. To account for possible exclusions, we recruited 99 students from a German University. The data of eight subjects had to be excluded because of extremely high error rates (>30%; average error rates: 3.7% without excluded participants). In addition, two subjects were excluded because of high error rates in control trials (>15%, average error rates: 3.9% without excluded participants) indicating that they did not read the prime questions carefully. The final sample thus consisted of 89 students (34 female) with an average age of M = 23.5 years (SD = 3.56). All participants gave informed consent by a keypress at the beginning of the experiment. They took part in exchange for €3 and a bar of chocolate. We used a 2 × 2 × 2 factorial design with the within-subjects factors prime question (lie vs. truth) and probe word (dishonest vs. honest) and the between-subjects factor frequency (rare lie vs. rare truth).

Procedure and materials

Upon arrival, participants were seated in front of a computer and received instructions on the screen. First, they were informed that they would participate in an interview, where they would be asked two questions about each of four different topics (university life, leisure time, money, and family; see Supplementary Material for a complete list of questions). Each question referred to an important issue but at the same time should not be too intimate to prevent participants from answering untruthfully even if they had to tell the truth (see Figure 2 for general procedure).

Figure 2.

General procedure of the experiment.

Counterbalanced across participants, frequency context was induced by first telling them that their general task was either to be honest or to lie to all questions during the interview, except for the questions stemming from one of the four topics. When these questions were asked, they should always act against their general rule, that is, tell a lie or the truth, respectively. For example, instructions in the rare lie condition would look like this: “Please answer all questions honestly, except for the questions from the topic university. When these questions are asked, please tell the interviewer a lie.” Participants were also prompted not to reveal their dishonesty to the interviewer and to act so as to convince him of the truthfulness of all their statements. To achieve this goal, participants were instructed to always wait for some seconds before providing their answer.

Participants were then guided into a separate room where the interview took place. After a short introduction, the interviewer asked the questions in each topic successively, with the order of topics in the interview being randomised. The interviewers were blind with regard to the frequency condition of participants as well as the distinct topic. All interviewers were female. The interviews had a duration of approximately 5 min.

After the interview, participants returned to their computers, where they had to perform the same priming task as in Koranyi et al. (2015). As primes, we used the questions from the interviews; six additional questions which had not been posed in the interviews were presented during filler trials to conceal the intention of the study. Participants were asked to decide as fast as possible whether a presented probe word was the word ehrlich (English: honest) or gelogen (English: dishonest) by pressing the J key for honest and the F key for dishonest. To ensure that probe words were encoded not only on a perceptual but also on a semantic level, they were degraded by inserting special characters between the letters (e.g., $eh§rl#ic%h instead of ehrlich). The locations of the special characters within the probe word were determined separately in each trial on a random basis. Each trial began with the presentation of a fixation cross (500 ms), followed by a serial word-by-word presentation of the prime question using RSVP (rapid serial visual presentation) with a base duration of 250 ms per word, plus an additional 25 ms per letter. Right after the last word of the sentence, the probe word appeared and remained on screen until the participant’s response. The next trial was initiated after an inter-trial interval of 1000 ms (see Figure 3).

Figure 3.

Trial structure of the experiment.

The priming task comprised 144 experimental trials with the order of trials randomised individually. To ensure comparable reliability of response time (RT) estimates for both the rare and the frequent conditions, the two prime questions in the distinct condition were presented 24 times each, while the remaining six prime questions of the frequent condition were shown 8 times each, half of the times preceding dishonest as probe word and half of the times honest as probe word. To ensure semantic processing of the prime questions, 28 experimental trials (randomly chosen out of the 144 trials) comprised an additional memory task that had to be performed directly after classifying the target word (see Wiswede et al., 2013). In the memory task, participants saw a question on the screen which was either the same (50%) or different from the prime question and participants had to answer the question, “Is this the question that you’ve just seen?” Participants were not informed in advance whether a trial comprised the additional memory task or not, so they had to process the prime questions in each trial.

To ensure that the honest and dishonest keys maintained their full semantic meaning across the experiment, an additional 32 filler trials were randomly intermixed into the experimental trials that required a genuine true/false decision (see Wiswede et al., 2013). In the filler trials, a true (50%) or false (50%) assertion (e.g., “Einstein was a musician”) was presented as a prime instead of a question. The assertion was followed by the question, “honest or dishonest?” which was presented as a response cue instead of the probe words. Participants had to respond to this question by pressing the same keys that were also used for the honest/dishonest decision of the experimental trials. The experiment had a duration of approximately 35 min.

Results

Data treatment

All response latencies that were more than one and a half interquartile ranges above the third quartile of an individual’s reaction time distribution were categorised as outliers (Tukey, 1977) and discarded (4.8% of all responses). All response latencies below the threshold of 250 ms were discarded (0.1%), as well as erroneous responses (3.5% of all responses). We calculated average response times for each participant and combination of the factorial design (see Figure 4 for the pattern of means).

Figure 4.

Mean response time (RT) as a function of context (rare lie vs. rare truth), prime question (lie vs. truth), and probe word (honest vs. dishonest). Error bars represent 95% CIs calculated for mixed IM-RM effects as suggested in Jarmasz & Hollands, 2009.

Analyses of variance

To test our assumptions, average response latencies were submitted to a 2 (context: rare lies vs. rare truths) × 2 (prime question: lie vs. truth) × 2 (probe word: dishonest vs. honest) ANOVA with repeated measures on the factors prime question and probe word.² Results revealed a main effect of probe word, F(1,87) = 6.74, p = .011, $η_{p}^{2}$ = .07 (90% CI = [.01, .17]), indicating that participants classified the probe word dishonest (M = 644 ms) faster than the probe word honest (M = 654 ms). In addition, an interaction effect of context and prime question emerged, F(1,87) = 8.52, p = .004, $η_{p}^{2}$ = .09 (90 CI = [.02, .19]), which was further qualified by a significant three-way interaction, F(1,87) = 4.50, p = .037, $η_{p}^{2}$ = .05 (90% CI = [.001, .14]), as predicted by the combined model.

To further disentangle this interaction, we ran separate analyses for both frequency conditions. For the rare lies condition, the results revealed no significant main effects for prime question, F(1,87) = 3.21, p = .077, $η_{p}^{2}$ = .04 (90% CI = [.00, .12]), and probe word, F(1,87) = 3.03, p = .085, $η_{p}^{2}$ = .03 (90% CI = [.00, .12]). Importantly, the predicted interaction between prime question and probe word reached significance, F(1,87) = 2.89, p = .047, $η_{p}^{2}$ = .03 (90% CI = [.00, .11]; one-tailed),³ with the descriptive pattern pointing in the predicted direction (see Figure 4). Correspondingly, follow-up tests showed that identification of the probe word dishonest (M = 626 ms) was faster compared with the probe word honest (M = 643 ms), t(44) = 3.61 p < .001, d = .54 (95% CI = [.22, .85]), after the presentation of a lie prime, whereas no such difference could be found for truth primes (t < 1). These results resemble the findings of Koranyi et al. (2015), indicating that those participants who regularly had to tell the truth and lied only to two questions during the interview showed the same congruency effect as was found before. Analyses of the rare truths condition only yielded a significant main effect for prime question, F(1,87) = 5.45, p = .022, $η_{p}^{2}$ = .06 (90% CI = [.01, .15]), but no significant interaction effect, F(1,87) = 1.69, p = .197, $η_{p}^{2}$ = .02 (90% CI = [.00, .09]), BF₀₁ = 2.28. Closer inspection of the descriptive pattern reveals two important things: First, there is no difference between probe words after the presentation of a lie prime (which would have been predicted by the quality account), and second, the difference between probe words after the presentation of a truth prime points in the opposite direction to what would have been predicted by the quantity account (i.e., faster identification of the probe word dishonest compared with honest.⁴

Bayesian statistics

To compare the probabilities for our hypotheses given the data, we also computed Bayes factors, using the R package Bain (Gu, 2016; Gu et al., 2018). Implemented in Bain is the approximate adjusted fractional Bayes factor, which can be used for the evaluation of informative hypotheses (Hoijtink, 2012). The Bayes factors and the posterior probabilities (computed assuming equal prior probabilities) are displayed in Table 1. Constraints for Hypotheses 1 (corresponding to the quality account), 2 (corresponding to the quantity account), and 3 (corresponding to the combined account) were defined in a way that they match the described data patterns in Figure 1. Note that all hypotheses are compared to an H_a, which assumes unconstrained means in all conditions. As can be seen, H₃ is supported more than both H₁ and H₂. A Bayes factor of 18.679 is usually seen as strong evidence in favour of H₃. The Bayesian error probability associated with preferring H₃ equals .281.

Table 1.

Bayes factors and posterior probabilities for Hypotheses 1–3, each in comparison to H_a which assumes unconstrained means in all conditions.

Hypothesis	BF.a	Posterior probability
H₁	4.968	.191
H₂	1.325	.051
H₃	18.679	.719
H_a		.039

Discussion

The aim of the present study was to test whether the quantity and/or quality of honest and dishonest responses to questions have an influence on automatic retrieval of the knowledge of having lied to a question before. We compared the predictions of three different models, where only quantity, only quality, or the combination of them determines the storage and automatic retrieval of an episodic unit.

All models predicted the same interaction effect of prime question and probe word in the rare lies condition. This prediction was confirmed in our data, replicating the findings reported by Koranyi et al. (2015). The models differ, however, with regard to their predictions for the rare truths condition, with the quality and quantity account predicting a two-way interaction of prime question and probe word also for this condition, resulting in an absence of a three-way interaction, whereas the combined model predicts an absence of a congruency effect within the rare truths condition, which should result in a significant three-way interaction.

Our results match with the predictions of the combined model: The three-way interaction of prime question, probe word, and context turned out to be significant, and there was no longer a significant interaction of prime question and probe word within the rare truths condition. Bayesian statistics yield additional support for the combined model. These results speak against both the quality and the quantity account. On one hand, having told a lie is not sufficient for an encoding and retrieval of knowledge about having lied, because this effect is neutralised when lies are the dominant response during the interview. Although telling lies is more cognitively demanding, this feature alone is not sufficient to produce the automatic memory retrieval effect for lies if lies are no longer distinct. On the other hand, having answered truthfully to a question does not produce a significant retrieval effect even when truthful answers are rare, and thus distinct. Thus, this finding speaks against an explanation of the retrieval effects in terms of distinctness.

Apparently, then, encoding and retrieval of knowledge about having lied to a question requires both, investment of cognitive effort (the qualitative feature of lies), and distinctness (the quantitative pre-condition of the effect). None of the two features in isolation is sufficient to produce the effect. Framed differently, both the lack of cognitive effort during response production and the lack of distinctness of the type of response that is produced to a question prevent the storage in and/or retrieval of corresponding question-response episodes from memory.

In the study by Koranyi et al. (2015), inverse efficiency scores (IES) were used as a measure to combine the effects of speed and accuracy. IES have been criticised since then for putting unequal weights on speed and accuracy depending on the level of accuracy (Liesefeld & Janczyk, 2019). This leads to unbalanced comparisons between conditions with different accuracy levels, which is why we decided to consider only RTs in this study. However, additional analyses of IES show no significant effect. This might be due to the described problems of this indicator, but also due to differences between the participants’ response behaviour in both studies. When looking at the accuracy rates, it becomes apparent that our sample shows much larger standard deviations (SD ranging from 3.5 to 14.2) than it was described in Koranyi et al. (SD ranging from 3.2 to 8.9). As it is hard to predict in advance whether effects will mainly map onto reaction time or accuracy data, one promising account in the future might be the use of a different combined measure, like the balanced integration score (BIS), which is recommended by Liesefeld and Janczyk (2019).

One might argue that our findings may be driven by a carry-over effect of the experimental instruction from the interview to the priming task. According to this alternative explanation, when reading a lie question during the experimental task, participants might remember the instruction that they should lie in response to this question because it refers to the topic for which lying is requested, rather than retrieving a specific episode in which they had lied to this specific question. For the following reasons, we think that this is an unlikely explanation for our findings. First, we tested whether merely instructing to lie to a certain question can produce a similar pattern of priming effects in a previous study (Koranyi et al., 2015). In the second experiment of this study, we used additional questions as primes in the RT task which related to the same topics to which the instructions referred but which were never presented during the interview so that participants had not lied or told the truth to these questions during the interview. As these questions did not elicit any congruency effect in the priming task, we concluded that instruction effects are not responsible for our findings. Furthermore, an instruction account cannot explain why (a) priming effects were obtained only for questions to which one had lied but not for questions that were answered truthfully, nor can it explain why (b) priming effects emerged only in the condition in which lies were rare. It thus seems highly unlikely that specific processes like we predicted them for the automatic memory of having lied before resulted from mere instruction effects. For these reasons, we assume that the pattern of findings that we obtained reflects automatic storage and retrieval processes with regard to having told a lie to a specific question in our design and do not think that instruction effects played a significant role here.

To conclude, our study provides a non-trivial explanation for the well-known fact that telling lies regularly typically incurs a high chance of being detected at some point. Our study reveals that at least part of the effect seems to be due to the lack of distinctness of the lies that are told, preventing an efficient memory retrieval of knowledge of having lied, which is a prerequisite for maintaining consistency in the fabric of lies that one has told to others. Our findings thus highlight that although one might get away with an occasional lie, one should definitely avoid becoming a liar.

Supplemental Material

QJE-STD-19-137.R2-Supplementary_Material – Supplemental material for Quantity matters: The frequency of deception influences automatic memory retrieval effects

Supplemental material, QJE-STD-19-137.R2-Supplementary_Material for Quantity matters: The frequency of deception influences automatic memory retrieval effects by Franziska Schreckenbach, Klaus Rothermund and Nicolas Koranyi in Quarterly Journal of Experimental Psychology

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Franziska Schreckenbach

Supplementary Material

The Supplementary Material is available at

Notes

References

Denkinger

Koutstaal

(2009). Perceive-decide-act, perceive-decide-act: How abstract is repetition-related decision learning? Journal of Experimental Psychology: Learning, Memory, and Cognition, 35, 742–756. https://doi.org/10.1037/a0015263

DePaulo

B. M.

Kashy

D. A.

Kirkendol

S. E.

Wyer

M. M.

Epstein

J. A.

(1996). Lying in everyday life. Journal of Personality and Social Psychology, 70, 979–995. https://doi.org/10.1037/0022-3514.70.5.979

Duncan

Humphreys

G. W.

(1989). Visual search and stimulus similarity. Psychological Review, 96(3), 433–458. https://doi.org/10.1037/0033-295X.96.3.433

Frings

Hommel

Koch

Rothermund

Dignath

Giesen

. . .Philipp

(2020). Binding and retrieval in action control (BRAC). Trends in Cognitive Sciences, 24, 375–387. https://doi.org/10.1016/j.tics.2020.02.004

Frings

Koch

Rothermund

Dignath

Giesen

Hommel

. . .Philipp

(2020). Merkmalsintegration und Abruf als wichtige Prozesse der automatischen Handlungssteuerung—eine Paradigmen-übergreifende Perspektive [Feature integration and retrieval in action control-a perspective across paradigms]. Psychologische Rundschau, 71, 1–14. https://doi.org/10.1026/0033-3042/a000423

Frings

Rothermund

(2011). To be or not to be. . .included in an event file: Integration and retrieval of distractors in stimulus-response episodes is influenced by perceptual grouping. Journal of Experimental Psychology: Learning, Memory, and Cognition, 37, 1209–1227. https://doi.org/10.1037/a0023915

Frings

Rothermund

(2017). How perception guides action: Figure-ground segmentation modulates integration of context features into S-R episodes. Journal of Experimental Psychology: Learning, Memory, and Cognition, 43, 1720–1729. https://doi.org/10.1037/xlm0000403

(2016). Bayesian evaluation of informative hypotheses [Doctoral dissertation]. University Utrecht. https://informative-hypotheses.sites.uu.nl/books-and-theses/

Mulder

Hoijtink

(2018). Approximated adjusted fractional Bayes factors: A general method for testing informative hypotheses. British Journal of Mathematical and Statistical Psychology, 71, 229–261. http://doi.org/10.1111/bmsp.12110

10.

Hoijtink

(2012). Informative hypotheses: Theory and practice for behavioral and social scientists. Chapman and Hall/CRC.

11.

Hommel

(1998). Event files: Evidence for automatic integration of stimulus-response episodes. Visual Cognition, 5, 183–216. https://doi.org/10.1080/713756773

12.

Hommel

(2004). Event files: Feature binding in and across perception and action. Trends in Cognitive Sciences, 8, 494–500. https://doi.org/10.1016/j.tics.2004.08.007

13.

Hunt

R. R.

(2006). The concept of distinctiveness in memory research. In Hunt

R. R.

Worthen

J. B.

(Eds.), Distinctiveness and memory (pp. 3–25). Oxford University Press.

14.

Jarmasz

Hollands

J. G.

(2009). Confidence intervals in repeated-measures designs: The number of observations principle. Canadian Journal of Experimental Psychology, 63(2), 124–138. https://doi.org/10.1037/a0014164

15.

Koranyi

Schreckenbach

Rothermund

(2015). The implicit cognition of lying: Knowledge about having lied to a question is retrieved automatically. Social Cognition, 33, 67–84. https://doi.org/10.1521/soco.2015.33.1.67

16.

Liesefeld

H. R.

Janczyk

(2019). Combining speed and accuracy to control for speed-accuracy trade-offs(?). Behavior Research Methods, 51, 40–60. https://doi.org/10.3758/s13428-018-1076-x

17.

Logan

G. D.

(1988). Toward an instance theory of automatization. Psychological Review, 95, 492–527. https://doi.org/10.1037/0033-295X.95.4.492

18.

Maxwell

S. E.

Delaney

H. D.

(1990). Designing experiments and analyzing data: A model comparison perspective. Wadsworth.

19.

Rothermund

Wentura

De Houwer

(2005). Retrieval of incidental stimulus-response associations as a source of negative priming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 482–495. https://doi.org/10.1037/0278-7393.31.3.482

20.

Schmidt

J. R.

De Houwer

Rothermund

(2016). The Parallel Episodic Processing (PEP) model 2.0: A single computational model of stimulus-response binding, contingency learning, power curves, and mixing costs. Cognitive Psychology, 91, 82–108. https://doi.org/10.1016/j.cogpsych.2016.10.004.

21.

Schreckenbach

Rothermund

Koranyi

(2019). And remember the truth that once was spoken: Knowledge of having disclosed private information to a stranger is retrieved automatically. Experimental Psychology, 66, 12–22. https://doi.org/10.1027/1618-3169/a000427

22.

Serota

K. B.

Levine

T. R.:.

Poster

F. J.

(2010). The prevalence of lying in America: Three studies of self-reported lies. Human Communication Research, 36, 2–25. https://doi.org/10.1111/j.1468-2958.2009.01366.x

23.

Sporer

S. L.

Schwandt

(2006). Paraverbal indicators of deception: A meta-analytic synthesis. Applied Cognitive Psychology, 20, 421–446. https://doi.org/10.1002/acp.1190

24.

Sporer

S. L.

Schwandt

(2007). Moderators of nonverbal indicators of deception. A meta-analytic synthesis. Psychology, Public Policy, and Law, 13, 1–34. https://doi.org/10.1037/1076-8971.13.1.1

25.

Tukey

J. W.

(1977). Exploratory data analysis. Addison-Wesley.

26.

Tulving

Thomson

(1973). Encoding specificity and retrieval processes in episodic memory. Psychological Review, 80, 352–373. https://doi.org/10.1037/h0020071

27.

von Restorff

(1933). Über die Wirkung von Bereichsbildungen im Spurenfeld [The effects of field formation in the trace field]. Psychologische Forschung, 18, 299–342. https://doi.org/10.1007/BF02409636

28.

Vrij

Granhag

P. A.

Porter

(2010). Pitfalls and opportunities in nonverbal and verbal lie detection. Psychological Science in the Public Interest, 11, 89–121. https://doi.org/10.1177/1529100610390861

29.

Walczyk

J. J.

Harris

L. L.

Duck

T. K.

Mulay

(2014). A social-cognitive framework for understanding serious lies: Activation-decision-construction-action theory. New Ideas in Psychology, 34, 22–36. https://doi.org/10.1016/j.newideapsych.2014.03.001

30.

Walczyk

J. J.

Roper

K. S.

Seemann

Humphrey

A. M.

(2003). Cognitive mechanisms underlying lying to questions: Response time as a cue to deception. Applied Cognitive Psychology, 17, 755–774. https://doi.org/10.1002/acp.914

31.

Wiswede

Koranyi

Müller

Langner

Rothermund

(2013). Validating the truth of propositions: Behavioral and ERP indicators of truth evaluation processes. Social Cognitive and Affective Neuroscience, 8, 647–653. https://doi.org/10.1093/scan/nss04

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB