The effects of deception on memory: a comparative study of actors and eyewitnesses accounts

Abstract

Deception is a common occurrence in daily life and has been shown to impair memory. This study investigated the memory-undermining effects of deception in a simulated daily life scenario, focusing on the potential moderating effect of the liars’ role (i.e., actor vs. eyewitness). In a 2 × 2 between-subjects design, 128 participants were randomly assigned to one of four conditions: actors who told the truth, actors who lied, eyewitnesses who told the truth, and eyewitnesses who lied. Participants engaged in a simulated shopping task, followed by an interview where they either provided truthful or deceptive responses about the items they bought (actors) or observed (eyewitnesses). Two days later, participants completed a series of memory tests assessing item memory, source memory, destination memory, and non-believed memories. Results showed that deception impaired source memory, with a greater impairment observed for actors than eyewitnesses. Deception also led to more non-believed memories and impaired item and destination memory, regardless of the liars’ role. These findings suggest that the role of liars moderates the effects of deception on memory, with self-related deception (actors) leading to greater memory impairments than other-related deception (eyewitnesses).

Introduction

Deception, or lying, is a deliberate behavior that aims to provide false information and induce false beliefs [1, 2]. Despite its prevalence in daily life [3, 4] and its essential role in social interactions [5], deception can have consequences for liars’ memory [6]. This study aimed to understand the effect of deception on memory better.

Extensive research has investigated the effect of deception on memory and the factors that can modulate the effect. It has been demonstrated that the effects vary depending on the type of deception employed. The three main types of deceptive strategies are false denial (denying events or details that happened), feigning amnesia (claiming memory loss for events or details that one truthfully remembers), and fabrication (fabricating events or details that did not happen) [7]. Studies on false denial have consistently shown that it leads to more omission errors [8–13] and creates more non-believed memories [14, 15] compared to telling the truth. It was demonstrated that feigning amnesia can result in more omission and commission errors than truth-telling [16–20]. Fabrication has also been found to bring more false memories and memory distortions than honesty [14, 21–24]. It also has been demonstrated that repeated lying [22] and more lies [25] can increase the disruptive effect of deception on memory. Furthermore, internally motivated lies can lead to a larger impact of deception on memories than externally motivated lies [26], but different internal motivational lies do not lead to significant differences in memory impairment [27]. Individual differences, such as the cognitive theory of mind [28] and psychopathic traits [29], can also modulate the memory-undermining effect of deception. In short, deception can disrupt lars’ memories.

The memory and deception (MAD) framework [7] posits that the cognitive demands of lying influence how deception affects memory. It has been suggested that deception involves more cognitive control than truth-telling [30]. Lying requires more mental effort and resources, leaving less capacity for encoding and retrieving truthful information [31]. It has been argued that false denial is less cognitively demanding, feigning amnesia requires more resources, and fabrication is the most resource-intensive [7]. This gradation in cognition demand leads to varying cognitive loads, impacting memory differently.

The enthusiasm for studying the memory-undermining effect of deception stems largely from its strong implication for the legal context. About 70% of articles in this field are interested in simulating real legal contexts where individuals lie in an experimental setting [32]. Many of these studies have employed various methods to simulate legal contexts, such as mock crimes or watching crime videos, to investigate the effect of deception on memory. Accordingly, participants in previous studies could be considered as offenders, victims or eyewitnesses. Offenders often lie in real legal situations and forensic contexts, and prior studies consistently found that participants lying after conducting a mock crime impairs their memory of the crime [12, 33, 34]. It has been suggested that eyewitness memory can be disrupted, resulting in false memories when eyewitnesses lie [35]. Indeed, participants asked to watch crime videos serve as eyewitnesses, and their memories of the crime can be undermined when they lie [13, 17, 18, 22, 36, 37]. However, recent work did not observe that lying eyewitnesses had worse memories of informal conversation and assault films compared to their honest counterparts [38]. Thus, the evidence for the effect of deception on eyewitness memory is mixed.

Given the inconsistencies in previous findings and the potential differences in cognitive processes underlying self-related and other-related deception, it is essential to directly compare the memory-undermining effect of deception between actors and eyewitnesses. This exploratory research aimed to investigate whether the impact of deception on memory would be modulated by the liars’ role, specifically focusing on the differences in the memory-undermining effect of deception between actors and eyewitnesses. Previous research has primarily investigated the memory-undermining effect of deception in actors (such as offenders) and eyewitnesses separately, with limited direct comparisons between the two roles. While studies on actors have consistently demonstrated memory impairment after lying about their mock crime [12, 33, 34], findings on the impact of deception on eyewitness memory have been mixed [13, 17, 18, 22, 36–38]. This discrepancy in findings may be attributed to the differences in cognitive processes underlying self-related and other-related deception. An actor’s deception can be considered self-related, as it involves lying about one’s actions and experiences. In contrast, an eyewitness’s deception can be classified as other-related, as it pertains to lying about the actions and experiences of others observed during the event. The cognitive neuroscience of deception has provided insights into the distinct mechanisms involved in different types of lies [39], with self-related lies potentially requiring more self-monitoring and conflict-resolution processes than other-related lies [40]. Event-related potential (ERP) studies found that self-related lying elicited a larger late positive component (LPC) compared to other-related lying, suggesting that self-related deception may demand more cognitive resources and engage more elaborative processing [41].

The present study manipulated the roles (actors vs. eyewitnesses) and responses (truth-telling vs. lying) as independent variables. In this study, we used the term ‘actors’ to refer to participants actively engaged in the simulated shopping task instead of ‘eyewitnesses’ who observed the task. The manipulated variables resulted in four experimental groups: actors who told the truth, actors who lied, eyewitnesses who told the truth, and eyewitnesses who lied. To comprehensively assess the effects of deception on different aspects of memory, we tested item memory, source memory, and destination memory. Item memory refers to the ability to remember the specific items encountered during the event, while source memory involves remembering the context in which the information was acquired [42]. On the other hand, destination memory refers to the ability to remember to whom one has conveyed information [43]. Liars must keep track of their lies and those who have lied to prevent discrepancies and uphold the deceit [44, 45].

Moreover, this study used error rates as the primary measure of memory performance. Error rates were calculated as the proportion of incorrect responses (i.e., false alarms and misses) out of the total test items. This measure provides a straightforward and easily interpretable index of memory accuracy, allowing for comparisons between the different experimental conditions. We also assessed participants’ subjective memory ratings and beliefs about their memory performance. These variables were included to provide insight into participants’ metacognitive assessments and explore how deception might influence their subjective remembering experiences. Based on the observations in previous research [8–24], we predicted that participants in the lying condition would report more errors and lower ratings of memory quality and vividness compared to those in the honest condition. As for beliefs about memory performance, we predicted that participants in the deceptive condition would report lower levels of confidence in their memory compared to those in the honest condition. Moreover, cognitive neuroscience research suggests that self-related deception may require more cognitive resources and elaborative processing than other-related deception [39–41]. Thus, we further hypothesized that the memory-undermining effect of deception in the final memory tests would be more pronounced for actors than eyewitnesses. We hope this exploratory research can reconcile the inconsistencies in previous findings and provide valuable insights into the factors that modulate the relationship between deception and memory in legal contexts.

Methods

Participants

A priori power analysis was conducted using G*Power software 3.1 [46] to determine the required sample size for this study. The power analysis was carried out for the main and interaction effects in a 2 × 2 between-subjects ANOVA. Based on an expected medium effect size (f = 0.25), an alpha level of 0.05, and a desired power of 0.80, the analysis indicated that a minimum sample size of 128 participants (32 per group) was required. Initially, 136 participants were recruited, but eight did not attend the second session. Thus, 128 participants (56 male) aged 18–24 (M = 21.65, SD = 1.31) completed the study. All participants provided informed consent and received compensation.

Design

The study applied a 2 × 2 between-subjects design. The independent variables were roles (actors or eyewitnesses) and responses (truth-telling or lying). Participants were randomly assigned to one of four experimental conditions: actors providing truthful responses (n = 33), actors providing deceptive responses (n = 33), eyewitnesses providing truthful responses (n = 31), and eyewitnesses providing deceptive responses (n = 31). Participants were randomly assigned to each group. The dependent variables were error rates, memory ratings, belief ratings, and the number of non-believed memories in the memory tests.

Procedure

The study adopted a daily life paradigm consisting of two sessions, following prior research [25, 26, 47, 48].

Session 1

A small mock store was set up on the 14th floor of a research institution, selling 20 different products: chewing gum, toothpaste, cookies, bread, Coca-Cola, pie, chocolate, instant noodles, soap, toothbrushes, instant coffee, tissues, water, hangers, seaweed, laundry detergent, garbage bags, N95 masks, towels, and swabs.

Two participants took part simultaneously, with one assigned to the role of actor to complete a simulated shopping task and the other assigned to the role of eyewitness to observe the shopping activities. Actors were asked to buy ten items without time limitations and to scan a payment QR code to simulate a transaction without real currency. Eyewitnesses were allowed to move freely but must maintain a distance of at least three meters from the actors. After the shopping task, all participants completed a 5-minute distractor task (playing Tetris).

After the distractor task, participants took a baseline memory test. In this test, actors recalled items they bought, and eyewitnesses recalled items that actors bought during the simulated shopping task. After recalling each item, participants were asked to rate their memory for each item on an 8-point scale. The specific prompt was as follows: Please rate your memory for this item on a scale from 1 (no memory at all) to 8 (very complete and vivid memory). Participants were also asked to rate their belief for that item on an 8-point scale from 1 to 8. The specific prompt was as follows: Please rate your belief for this item on a scale from 1 (no belief) to 8 (strong belief). These scales were adapted from the Autobiographical Belief and Memory Questionnaire [49, 50]. A five-minute filler task (playing Tetris) followed.

Two interviewers unfamiliar with the participants questioned them in a room on the 13th floor. The interviewers obtained the shopping lists while participants took their baseline memory test and selected five items from the shopping lists for the interview. Additionally, they asked questions concerning five items (milk, mirror, water cup, dried beef, and shampoo) not sold in the mock store. A pre-experiment revealed a floor effect when different interviewers randomly asked the interview items. Thus, interviewers were asked to inquire about items from various sources separately. One of the interviewers asked questions about the items in the shopping lists, while the other interviewer asked questions about the items not sold in the store. The interviewers used a fixed question structure: “Did you buy ….?” for actors and “Did he/she buy ….?” for eyewitnesses, alternating between the two interviewers. Half of the actors and eyewitnesses were instructed to respond honestly to the interview questions, while the other half were asked to respond deceptively. That is, participants in the honest condition were asked to give positive responses (“Yes, I/she/he bought this item”) for the items in the shopping lists and to provide negative responses for those not in the shopping lists (“No, I/she/he did not buy this item”). Conversely, participants in the lying condition were asked to give negative responses for the item in the shopping lists (“No, I/she/he did not buy this item”) and to provide positive responses for those not in the shopping lists (“Yes, I/she/he bought this item”). The interviewers gave no feedback to participants during the interview.

Session 2

Two days after the first session, participants were asked to honestly complete several memory tests in the following order: item memory, source memory, and destination memory. In the item memory test, like the baseline memory test, participants were asked to recall the items in the shopping list and rate their memory and belief for each item using the same scale. Twenty items were provided in the source memory test: ten items in the shopping list, the five items not sold in the mock store but were asked about in the interview and another five items that were not sold in the mock store and were not asked about in the interview (kettle, milk tea, peanut, washbowl, and cereal). Participants were asked to indicate if they were asked about these items during the interview and to rate their memory and belief for each item in the source memory test. In the destination memory test, the items asked about in the interview were provided, and participants were asked to identify the specific interviewer asked about each item and to rate their memories and beliefs for each item.

Participants were debriefed after the memory test, emphasizing that lying should not be promoted.

Scoring

The scoring methods were similar to previous research [26]. Memory tests were scored based on performance. The second author scored the tests, double-checked by a student assistant unaware of the study’s design. Correct responses received one point; incorrect or missing responses received zero points. Maximum scores were 10 for the baseline, item, and destination memory tests and 20 for the source memory test. Error rates were calculated by subtracting the obtained score from the maximum possible and dividing by the maximum. Average belief and memory ratings were calculated by summing and dividing by the number of items. Error rates and average ratings were analyzed across participant groups.

Results

In line with recent research [25, 26, 47, 48], data analysis was performed using the lme4 package in the R system. Belief and memory ratings were analyzed using a linear mixed-effects (LMEs) model, while error rates were examined using a generalized linear mixed-effects (GLMEs) model. Both models incorporated participants and items as crossed random effects [51]. In the LMEs and GLMEs models, the output includes the estimated coefficient (b), which represents the change in the dependent variable (belief and memory ratings or error rates) for a one-unit change in the corresponding independent variable, holding other variables constant. The standard error (SE) associated with each estimated coefficient measures the variability or uncertainty in the coefficient estimate. The t-value (t) (LMEs model) or z-value (z) (GLMEs model) is calculated by dividing the estimated coefficient by its standard error (t or z = b / SE) and follows a t-distribution (LMEs model) or a standard normal distribution (GLMEs model) under the null hypothesis that the coefficient is zero. Following the methodology of previous studies [51, 52], |z| or |t| values exceeding 1.96 were considered statistically significant, with higher values indicating larger effects. Descriptive and inferential statistics are presented in Tables 1 and 2, respectively.

Table 1.

Mean error rates and ratings in the memory tests for each condition

Memory Tests	Roles	Responses	Error Rates (%)	Memory Ratings	Belief Ratings
Baseline Memory Test	Actor	honest	10.61(1.7)	7.28(0.07)	7.84(0.03)
		lying	12.58(1.89)	7.31(0.07)	7.82(0.04)
	Eyewitness	honest	13.94(1.91)	7.24(0.08)	7.77(0.05)
		lying	17.42(2.16)	7.23(0.08)	7.73(0.05)
Item Memory Test	Actor	honest	9.41(1.61)	6.65(0.11)	7.26(0.11)
		lying	14.85(1.96)	6.66(0.12)	7.71(0.06)
	Eyewitness	honest	12.57(1.83)	6.38(0.12)	7.59(0.07)
		lying	15.81(2.08)	6.24(0.13)	7.52(0.08)
Source Memory Test	Actor	honest	25.91(1.71)	5.93(0.09)	5.98(0.09)
		lying	42.26(1.99)	5.51(0.11)	5.87(0.11)
	Eyewitness	honest	26.52(1.72)	6.16(0.11)	6.33(0.09)
		lying	33.88(1.91)	5.54(0.11)	5.66(0.11)
Destination Memory Test	Actor	honest	38.18(2.68)	4.82(0.15)	4.89(0.18)
		lying	49.68(2.84)	4.23(0.17)	4.04(0.19)
	Eyewitness	honest	38.49(2.68)	4.61(0.17)	4.76(0.17)
		lying	50.32(2.84)	2.94(0.15)	2.99(0.15)

Note: The standard errors of means are shown in parentheses

Table 2.

Statistical effect for the memory tests

		Error Raters				Memory Ratings				Belief Ratings
		b	SE	z		b	SE	t		b	SE	t
Baseline Memory Test	Intercept	2.33	0.26	9.01		7.27	0.08	86.59		7.79	0.05	162.97
	Responses	0.21	0.21	1.03		-0.02	0.14	-0.13		0.03	0.07	0.45
	Roles	0.38	0.21	1.91		0.09	0.14	0.61		0.08	0.07	1.28
	Roles × Responses	-0.12	0.41	-0.29		0.01	0.02	0.01		-0.01	0.13	-0.12
Item Memory Test	Intercept	2.28	0.24	9.41		6.48	0.16	40.59		7.68	0.04	160.82
	Responses	0.19	0.21	0.93		0.51	0.25	2.06		0.03	0.07	0.45
	Roles	0.42	0.21	2.11		0.08	0.27	0.31		0.08	0.07	1.28
	Roles × Responses	0.28	0.39	0.71		0.31	0.49	0.60		-0.02	0.13	-0.12
Source Memory Test	Intercept	-0.75	0.08	-9.44		5.75	0.13	43.37		5.92	0.13	46.34
	Responses	-0.57	0.09	-6.56		0.72	0.22	3.93		0.59	0.21	2.79
	Roles	0.17	0.09	1.93		0.14	0.22	0.66		0.31	0.21	0.59
	Roles × Responses	-0.41	0.17	-2.31		-0.19	0.42	-0.44		-0.39	0.41	-0.95
Destination Memory Test	Intercept	-0.28	0.09	-3.22		4.08	0.17	23.81		3.99	0.19	21.26
	Responses	-0.51	0.14	-3.59		0.91	0.27	3.33		1.09	0.31	3.49
	Roles	-0.02	0.14	-0.16		0.96	0.27	3.51		0.78	0.32	2.48
	Roles × Responses	0.02	0.28	0.09		-1.16	0.53	-2.21		-0.74	0.61	-1.22

Note: Bold numbers indicate significant values

No unexpected responses were observed during the interview, and no data were excluded from analyses.

Baseline memory test

The baseline memory test was conducted to examine whether participants remembered the items in the shopping list. A marginally significant main effect of roles on error rates was observed (p = 0.057), suggesting that actors had better memories for shopping lists than eyewitnesses. No other differences reached a significant level in the baseline memory test.

Item memory test

The item memory test aimed to determine if deceptive responses disrupted memories of items in shopping lists. As expected, a main effect of roles on error rates was obtained, indicating that eyewitnesses made more errors on shopping lists than actors. Moreover, a main effect of responses on memory ratings was observed, suggesting that honest participants had clearer and more complete memories of their shopping lists than their lying counterparts, as anticipated. No other significant differences were observed.

Source memory test

The source memory test investigated whether participants remembered which items were asked about in the interview. As anticipated, significant main effects of responses on error rates, memory and belief ratings were observed, suggesting that lying could undermine liars’ memory compared to honest responses. More importantly, a significant effect of interaction on error rates was observed. Simple effect analysis revealed that honest actors made fewer errors than lying actors on the test (b = 0.78, SE = 0.12, z = 6.35, p < 0.001), as expected. Simple effect analysis revealed that lying eyewitnesses made more errors than honest eyewitnesses, but the effect size of deception on memory was smaller than their actor counterparts (b = 0.38, SE = 0.15, z = 2.49, p < 0.05), as anticipated.

The items presented on the source memory test were divided into four categories: items in the shopping lists that were asked about in the interview (category A), items in the shopping lists that were not asked about in the interview (category B), items not in the shopping lists that were asked about in the interview (category C), and items not in the shopping lists and not asked about in the interview (category D). Comparisons among groups on the four categories were conducted, with descriptive and inferential statistics shown in Tables 3 and 4, respectively.

Table 3.

Mean error rates and ratings for each item category and condition on the source memory tests

Item Categories	Roles	Responses	Error Rates (%)	Memory Ratings	Belief Ratings
Category A	Actor	honest	31.52(3.63)	6.15(0.17)	6.12(0.18)
		lying	43.23(3.99)	5.85(0.21)	6.06(0.22)
	Eyewitness	honest	21.82(3.23)	6.59(0.17)	6.82(0.16)
		lying	26.45(3.55)	5.83(0.18)	5.97(0.18)
Category B	Actor	honest	31.52(3.36)	5.31(0.17)	5.24(0.19)
		lying	50.97(4.03)	4.49(0.24)	5.13(0.24)
	Eyewitness	honest	38.18(3.79)	5.59(0.17)	5.72(0.21)
		lying	43.87(3.99)	4.92(0.21)	4.99(0.22)
Category C	Actor	honest	30.91(3.61)	6.38(0.16)	6.52(0.17)
		lying	32.26(3.77)	6.13(0.17)	6.25(0.21)
	Eyewitness	honest	31.52(3.63)	6.44(0.18)	6.64(0.18)
		lying	38.07(3.91)	5.81(0.19)	5.73(0.19)
Category D	Actor	honest	9.71(2.31)	5.89(0.17)	6.03(0.18)
		lying	42.58(3.98)	5.28(0.22)	5.85(0.23)
	Eyewitness	honest	14.55(2.75)	5.94(0.21)	6.08(0.18)
		lying	27.11(3.85)	5.51(0.22)	5.81(0.21)

Note: The standard errors of means are shown in parentheses. Category A, B, C, and D represent items in the shopping lists that were asked about in the interview, items in the shopping lists that were not asked about in the interview, items not in the shopping lists that were asked about in the interview, and items not in the shopping lists and were not asked about in the interview, respectively

Table 4.

Statistical effect for item categories on the source memory test

		Error Raters				Memory Ratings				Belief Ratings
		b	SE	z		b	SE	t		b	SE	t
Category A	Intercept	-0.89	0.11	-7.89		6.12	0.15	40.38		6.25	0.17	37.73
	Responses	-0.42	0.21	-2.08		0.56	0.24	2.31		0.47	0.23	2.03
	Roles	0.67	0.21	3.34		-0.26	0.24	-1.09		-0.37	0.23	1.62
	Roles × Responses	-0.31	0.39	-0.76		-0.42	0.47	-0.88		-0.75	0.46	-1.64
Category B	Intercept	-0.36	0.11	-3.21		4.95	0.19	26.73		5.16	0.19	27.79
	Responses	-0.59	0.18	-3.35		0.69	0.31	2.23		0.46	0.31	1.53
	Roles	-0.01	0.17	-0.03		0.08	0.31	0.26		0.14	0.29	0.49
	Roles × Responses	-0.59	0.34	-1.75		-0.19	0.59	-0.33		-0.76	0.58	-1.32
Category C	Intercept	-0.74	0.14	-5.41		6.17	0.19	31.25		6.28	0.19	32.54
	Responses	-0.18	0.19	-0.95		0.48	0.23	2.04		0.65	0.25	2.59
	Roles	-0.15	0.19	-0.79		0.19	0.23	0.82		0.24	0.25	0.97
	Roles × Responses	0.26	0.38	0.68		-0.46	0.46	-0.99		-0.53	0.49	-1.07
Category D	Intercept	-1.53	0.23	-6.69		5.36	0.17	32.49		5.91	0.16	36.75
	Responses	-1.54	0.28	-5.55		0.79	0.33	2.39		0.44	0.31	1.42
	Roles	0.18	0.27	0.66		-0.02	0.34	-0.06		0.11	0.32	0.36
	Roles × Responses	-1.33	0.53	-2.49		0.25	0.65	0.41		-0.05	0.61	-0.08

Note: Bold numbers indicate significant values

As anticipated, the memory-undermining effect of lying was observed across the item categories. For items in category A, participants in the lying condition needed to make false denial responses. As shown in Tables 3 and 4, a significant main effect of response on error rates was observed, suggesting the presence of the DIF (denial-induced forgetting) effect, with liars showing a higher error rate than honest participants, as expected. Moreover, a main effect of response on memory and belief ratings was observed for items in categories A and C, indicating that liars had worse memories and weaker beliefs than honest participants, as anticipated. For items in categories B and D, we observed the main effects of response on error rates and memory ratings, suggesting that liars made more errors and had more unclear memories than honest participants, as anticipated. More importantly, a significant interaction effect between the independent variables on error rates was found for the item in category D. Simple effect analysis revealed that the memory-undermining effect of lying was larger for actors (b = -2.02, SE = 0.36, z = -5.68, p < 0.001) compared to eyewitness (b = 1.08, SE = 0.46, z = 2.34, p < 0.05), as expected.

Destination memory test

The destination memory test examined if participants remembered the interviewer who asked about a specific item and determined if liars had worse memories about the interviewer they lied to compared to honest participants. A main effect of response on error rates was observed, indicating that honest responses led to fewer errors than lying, as anticipated. The main effects of response and role on belief ratings were found, suggesting that honest responses resulted in stronger belief ratings, and actors had stronger beliefs than eyewitnesses on the destination memory test. Moreover, a significant effect of interaction on memory ratings was obtained. Simple analysis revealed that lying and honest conditions did not differ for actors (b = 0.43, SE = 0.36, t = 1.18, p > 0.05), while honest eyewitnesses had better memories than lying eyewitnesses (b = 1.39, SE = 0.41, t = 3.39, p < 0.05), which is not anticipated.

Non-believed memories

Following previous studies, non-believed memories were defined as items with memory ratings at least two scale points greater than belief ratings regardless of response accuracy [9, 25, 26, 48, 53]. We counted the number of non-believed memories on memory tests for each group (see Table 5).

Table 5.

Number of non-believed memories for each group in memory tests

	Honest Responses			Lying Response
	Actors	Eyewitnesses		Actors	Eyewitnesses
Baseline Memory Test	5	1		2	8
Item Memory Test	11	7		5	4
Source Memory Test	29	16		24	56
Destination Memory Test	34	23		32	21

We defined the non-believed items as 1 and other items as 0. Thus, GLMEs could be used to analyze the differences among groups on memory tests. We obtained a main effect of response on the source memory test, suggesting that lying groups had more non-believed memories than honest groups (b = 1.05, SE = 0.51, z = 2.06, p < 0.05).

Discussion

The present study investigated whether memory performance differs among participants with different roles when engaging in deception, examining the potential moderating effect of roles on memory impairment caused by deception. Previous research on the effects of deception on eyewitness memory has yielded mixed results [13, 17, 18, 22, 36–38]. The present study addresses the inconsistencies by directly comparing the memory-undermining effects of self-related deception (actors) and other-related deception (eyewitnesses) in a more ecologically valid setting.

Our findings contribute to the resolution of the mixed evidence regarding the effect of deception on eyewitness memory. The results showed that eyewitnesses who engaged in other-related deception experienced memory impairments in source memory, particularly for information unrelated to the deceptive act (Category D items). This finding supports the notion that deception can disrupt eyewitness memory, as suggested by previous studies [13, 17, 18, 22, 36, 37]. More importantly, the memory impairment observed in eyewitnesses was less pronounced than that of actors who engaged in self-related deception. This difference in the magnitude of the memory-undermining effect may help explain the inconsistent findings in previous research, as the cognitive demands and mechanisms underlying self-related and other-related deception may vary [39–41]. The cognitive neuroscience research suggests that self-related and other-related deception involve distinct neurophysiological mechanisms [40, 41]. Our results align with this idea, as actors who engaged in self-related deception experienced greater impairments in source memory than eyewitnesses who engaged in other-related deception. This finding suggests that self-related deception may require more cognitive resources and engage more elaborative processing compared to other-related deception [39–41], potentially leading to more significant memory disruptions.

The memory and deception (MAD) framework [7] and the Activation-Decision-Construction-Action Theory (ADCAT) of deception [54] provide further support for the role of cognitive demands in moderating the memory-undermining effect of deception. In the present study, actors had more contextual information about the shopping lists than eyewitnesses and thus might have required more cognitive effort and resources to suppress truthful memories successfully, leading to greater forgetting of the source memory items.

The observed denial-induced forgetting (DIF) effect in the source memory test is consistent with previous research [9, 13–15], indicating that false denials led to more forgetting in the deception group. This effect can be attributed to the increased cognitive load associated with deception, as liars must suppress the truth, construct a plausible alternative, and monitor their behavior to appear credible [55]. Consequently, participants in the deception group had fewer cognitive resources available to encode and store the information related to the interview items, leading to greater forgetting of these items than the honest group.

In addition to the effects on source memory, the present study also found that deception led to more non-believed memories and impaired item and destination memory, regardless of the liars’ role. These findings align with previous research showing that deception, regardless of the liar’s role, can create more non-believed memories [14, 15, 25, 48] and disrupt item and destination memory [25, 26, 44, 45, 47]. Cognitive dissonance may arise when an individual’s deceptive statement conflicts with their original memory. As a result, the individual may begin to doubt the accuracy of their original memory, forming non-believed memories. Moreover, memory impairment may also be due to the cognitive demands associated with constructing and maintaining a deceptive narrative, which can interfere with the encoding and retrieving of accurate information [55, 56].

The significant interaction effect on memory ratings in the destination memory test suggests that the impact of deception on the subjective quality and vividness of destination memory varies depending on the individual role (actor or eyewitness) in the event. The unanticipated nature of this interaction effect highlights the complex relationship between deception and the subjective experience of destination memory. This finding has important implications for situations where the accuracy and credibility of eyewitness accounts are crucial, such as in legal settings. Further research is needed to better understand the factors contributing to how deception affects destination memory differently for actors and eyewitnesses. Strategies should also be developed to assess and improve the reliability of eyewitness testimony when deception is involved.

It is important to note that the observed differences in baseline memory test performance between actors and eyewitnesses are consistent with the study design and do not undermine the interpretation of our findings. Actors had access to visual and tactile information during the shopping task, while eyewitnesses only had visual information. Additionally, actors were closer to the items than eyewitnesses, which may have allowed them to encode more detailed information. These factors may have contributed to the actors having a better initial memory of the shopping list. Previous research has established that the quantity and quality of sensory information are crucial for forming clear contextual memories [57]. However, our primary focus was on within-role comparisons, investigating the memory-undermining effect of lying separately for actors and eyewitnesses. Therefore, the baseline differences between roles do not confound the interpretation of the lying effects within each role.

This research has several theoretical and practical implications. Theoretically, the study contributes to the literature on deception and memory by demonstrating that the liar’s role can moderate the effects of deception on memory, helping to reconcile the mixed findings in previous research. The findings suggest that self-related deception may be more cognitively taxing than other-related deception, leading to greater memory impairments, especially in source memory [8]. These findings are also crucial for legal and forensic settings, emphasizing the importance of legal professionals assessing the influence of deception on eyewitness accounts and developing evidence-based interviewing methods to minimize its negative effects on memory.

Several limitations to this study need to be acknowledged. The simulated shopping task and controlled interview setting may not fully represent real-life deception, particularly in emotionally charged and consequential contexts such as criminal cases. Eyewitnesses in real-world criminal cases often experience significant emotional distress and trauma from witnessing violent or life-threatening events [58], which can influence their cognitive processes, including attention, perception, and memory [59]. Moreover, the consequences of deception in real-world legal and forensic contexts are often more severe than those in a controlled experimental setting, as eyewitnesses may face intense pressure to provide accurate testimony [60]. Given these differences, it is important to approach the extension of the findings to legal or forensic contexts with caution. Future research should aim to bridge this gap by investigating the impact of deception on memory in more emotionally charged and consequential contexts.

Another limitation of the current study is that we did not measure or control for the amount of time participants spent on the shopping task or the distance between the actor and eyewitness. These variables could vary naturally to maintain ecological validity and reflect real-world contexts’ inherent variability and unpredictability. However, the lack of measurement and control over these variables may introduce potential confounds and sources of noise in our data. The varying shopping time or distance between the actor and eyewitness could have influenced how deception affects memory in ways not considered in our analysis. Recognizing the trade-offs between maintaining ecological validity and ensuring experimental control is important. Future research should balance competing factors by combining naturalistic and controlled experimental designs or using advanced statistical methods to analyze the impact of these variables on the memory-undermining effects of deception.

While this study provides valuable insights into the effects of deception on memory, it is important to consider the nature of the memory measures employed. It is important to note that the observed differences in memory performance may not necessarily reflect differences in actual memory but could also be influenced by response biases. Future research could benefit from using more diverse and sensitive memory measures to capture the nuances of how deception influences memory processes. Such measures could help clarify the distinction between actual memory impairments and potential response biases that may arise from deception. By employing a wider range of memory assessments, future studies could provide a more comprehensive understanding of the complex relationship between deception and memory.

Conclusion

The present study demonstrates that deception can impair memory in daily life, with the extent and nature of this impairment varying based on the role of the liars and the type of memory assessed. Actors engaging in self-related deception experienced greater source memory impairments than eyewitnesses engaging in other-related deception. Deception also led to more non-believed memories and impaired destination memory, regardless of the liar’s role. These findings highlight the cognitive mechanisms underlying the memory-undermining effects of deception and have implications for legal and forensic contexts.

Abbreviations

EF

Executive Function

DIF

Denial-Induced Forgetting

Author contributions

YL designed the experiments and collected the data with help from ZL. ZL analyzed the data. YL wrote the manuscript with critical comments from ZL. The authors have no conflicts of interest.

Funding

This study received no funding.

Data availability

Data Availability Statement The data that supports the findings of this study is available in the Figshare repository: https://figshare.com/s/f5ede1923841c85152b1.

Declarations

Ethics approval and consent to participate

This study was performed in line with the principles of the Declaration of Helsinki. The School of Education and Psychological Science Ethics Committee at Sichuan University of Science and Engineering granted approval. Written informed consent was obtained from all individual participants included in the study.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.