Introduction

Most people believe that they truly are morally good (Aquino & Reed, 2002; Molouki & Bartels, 2017; Strohminger, Knobe, & Newman, 2017), and most people believe that they are superior to both the average person and specific peers in terms of their virtue, compassion, and other positive moral traits (Alicke & Govorun, 2005; Alicke & Sedikides, 2009; Alicke, Vredenburg, Hiatt, & Govorun, 2001; Batson & Collins, 2011; Tappin & McKay, 2017). And yet, people commit moral transgressions with surprising frequency in their everyday lives. In fact, in a recent 3-day study using ecological momentary assessment, participants reported 13.6% of the time that they had “committed, were the target of, witnessed, or learned about” an immoral action within the past hour (Hofmann, Wisneski, Brandt, & Skitka, 2014).

If people do commit moral transgressions with such frequency in their everyday lives, why do they continue to adamantly believe that they are morally good? One possible answer to this question involves autobiographical memory, a collection of representations of one’s personal past that are of relevance to the self (Conway, 2005; Rubin, 1986). Although psychological research on memory has focused predominantly on passive features of forgetting (e.g., Baddeley, 1990; Schacter, 1999), mounting behavioral and neuroimaging research suggests that there are also active, motivational factors that influence forgetting. That is, people can and do actively undermine their retention of certain unwelcome memories (Anderson & Hanslmayr, 2014; Anderson & Levy, 2009; Bergström, Anderson, Buda, Simons, & Richardson-Klavehn, 2013; Hu, Bergström, Bodenhausen, & Rosenfeld, 2015). Recently, theorists have proposed that people strategically forget their own unethical past behaviors in order to avoid psychological distress and to maintain a morally good self-concept (Kouchaki & Gino, 2016; Stanley & De Brigard, 2019). Specifically, Kouchaki and Gino (2016) suggest that “[people’s] memory of their past unethical behavior becomes less clear, less detailed, and less vivid over time than their memory of their ethical actions and of actions unrelated to ethics [emphasis added]”—a phenomenon the authors refer to as unethical amnesia.

Kouchaki and Gino (2016) suggest that unethical amnesia impacts both memory accuracy (that people are more prone to errors related to memory of their unethical actions) and the phenomenology of remembering (that unethical actions, relative to other kinds of memories, are remembered less clearly and vividly). Several studies have examined the former – i.e., memory accuracy – for past unethical actions. For instance, participants who cheated on a task had less accurate memory for the content of an honor code that they had read before the task (Shu, Gino, & Bazerman, 2011). In another study, consumers tended to forget the unethical attributes of the products they purchased (e.g., made from endangered trees or manufactured using child labor); they also regarded such ignorance as more morally permissible than making the purchase while being aware of those unethical attributes (Reczek, Irwin, Zane, & Ehrich, 2018). People do not only forget details of their own past immoral actions to foster their own morally good self-concepts – some evidence suggests that when the rememberer benefits from the unfair actions of others, the rememberer tends to forget critical details about those immoral actions (Bell, Schain, & Echterhoff, 2014).

Although there have been several investigations into memory accuracy for ethical and unethical actions, fewer studies have investigated the phenomenological characteristics of remembering unethical actions. In the current two studies, we investigate the relationship between the moral severity of past moral transgressions and the phenomenology of remembering them. Moral severity refers to the degree to which a behavior is considered to be more or less morally wrong (Powell & Horne, 2016). For example, even though premeditated murder for no reason and shoplifting are both considered to be morally wrong, premeditated murder for no reason is typically considered to be a much more severe transgression than shoplifting. Our more severe moral transgressions should pose a greater threat to our morally good self-concepts than more minor transgressions. So, memories of more severe moral transgressions should be more likely to become obfuscated, relative to more minor transgressions. However, other literature reviewed below might imply the opposite prediction, namely that more severe moral transgressions might be less likely to become obfuscated relative to more minor transgressions. In what follows, we investigate three possible effects of moral severity on how people remember and judge their past transgressions, and we attempt to adjudicate between seemingly conflicting predictions implied by different literatures.

First, we investigate a possible moral severity effect on the frequency of recalling moral transgressions. According to Kouchaki and Gino (2016), “people are less likely to think about [memories of unethical actions]”; frequently recalling our past moral transgressions arouses negative feelings of distress and discomfort and threatens a morally good self-concept. Because memories of more egregious past moral transgressions arouse stronger negative feelings (Escobedo & Adolphs, 2010; Stanley, Henne, Iyengar, Sinnott-Armstrong, & De Brigard, 2017) and pose a greater threat to a morally good self-concept (relative to memories of more minor moral violations), we might expect people to recall their more severe moral transgressions less frequently than their more minor transgressions. Some evidence, however, might suggest an alternative possible pattern of results. In a study with offenders of violent crimes, 46% experienced intrusive memories (which are “recurrent … and involuntarily triggered”) about their crimes, and 36% reported ruminating on their crimes (Evans, Ehlers, Mezey, & Clark, 2007a, 2007b). These studies suggest that particularly severe moral transgressions might spontaneously and frequently emerge in consciousness with high frequency. However, these studies provide no direct evidence that these convicted criminals actually believed their crimes were morally wrong. So, it remains unclear whether memories of these crimes are damaging to their morally good self-concepts.

Second, we investigate a possible severity effect on the vivacity, detail, and sense of reliving with which past transgressions are recollected. If people selectively obfuscate their memories of their own moral transgressions to mitigate distress and to maintain a belief in a morally good self-concept (Kouchaki & Gino, 2016), we might expect that more severe unethical actions will be the primary targets of obfuscation, because they represent a greater threat to this desired self-concept. Hence, we might expect that more severe moral transgressions, relative to more minor transgressions, will be remembered with less vivacity, fewer sensory details, and a reduced sense of reliving. In contrast to this expectation, however, Evans and collaborators (2007a, 2007b) report that criminals described their intrusive memories of crimes involving harm as “vivid, clear, detailed, intrusive, unchanging, and very distressing.” In a related study with criminals convicted of homicide, “murder [was] recalled with a high level of detail, vividness, and sensory elements” as compared to less severe non-homicide offences and positive experiences (Woodworth et al., 2009). These studies suggest that people might recall their severe moral transgressions vividly and with rich detail.

Finally, we investigate a possible moral severity effect on the perceived personal importance and centrality of participants’ past moral transgressions. Past experiences remembered as highly central to one’s life play significant roles in shaping personal identity, the life story, and self-understanding (Berntsen & Rubin, 2006; Bluck, 2003; Rubin, Schrauf, & Greenberg, 2003). Because people believe they truly are morally good and regard morally good traits as particularly central to their personal identities (e.g., Aquino & Reed, 2002; Fernandez-Duque & Schwartz, 2016; Heiphetz, Strohminger, & Young, 2017; Strohminger et al., 2017; Strohminger & Nichols, 2014), we might expect people to report that their particularly severe past transgressions are not central or important to who they are or who they wish to be. If these severe transgressions were important and central to constructions of personal identity, they might threaten a morally good self-concept. Nevertheless, some evidence suggests that extremely negative events – even those that threaten desired self-concepts – are sometimes particularly central to constructions of personal identity (Rubin, Dennis, & Beckham, 2011; Rubin et al., 2003). A hallmark characteristic of these central negative events is that they tend to be recalled frequently (Rubin et al., 2011). If more severe moral transgressions are, in fact, recalled more frequently than more minor transgressions, they might also be more central to constructions of personal identity.

Study 1

In Study 1, each participant recalled and described six of their own past moral transgressions, and they provided several ratings for each memory. This allowed us to investigate the effects of moral severity on how participants remember and judge their past transgressions.

Method

Participants

A total of 138 individuals voluntarily completed this study via Amazon Mechanical Turk (AMT) for monetary compensation. Qualified participants were individuals living in the USA who had completed over 50 HITs with an approval rating above 90%. One participant reported not paying attention (see attention check below) during the study and another participant failed to follow instructions, so data were analyzed with the remaining 136 participants (Mage = 33.53 years, SD = 8.51, age range = [21, 66]; 61 females).Footnote 1 Informed consent was obtained from each participant in accordance with a protocol approved by the Duke University Campus Institutional Review Board.

Procedure

This study consisted of a single, self-paced session. Participants were assured that their names would not be linked to their responses and that all data would be securely saved and de-identified, indefinitely. Participants were asked to recall a total of six events from their personal pasts that occurred when they were older than 15 years of age, and to provide several ratings for each memory. Specifically, participants first recalled and described in two to five sentences one particular event in which they personally did something that they believed to be at least slightly morally wrong. Participants reported the month and year when the event occurred and rated the severity of the transgression (1 = slightly morally wrong; 7 = very morally wrong).

Participants then provided ratings on eight different phenomenological characteristics of remembering or memory judgments. All eight items were adapted from the Memory Characteristics Questionnaire (MCQ; Johnson, Foley, Suengas, & Raye, 1988) and the Autobiographical Memory Questionnaire (AMQ; Rubin et al., 2003). Specifically, and directly related to our primary research questions, participants reported their frequency of recalling their moral transgressions, in terms of both voluntary and involuntary retrieval (items 1–2 in Table 1).Footnote 2 Participants also rated the vivacity and detail of their memories along with their sense of reliving (items 3–6 in Table 1). Moreover, participants rated the personal importance and centrality of the events (items 7–8 in Table 1). Finally, to address several exploratory questions, participants made ratings on other phenomenological and memory characteristics (see Table S1 in the Supplementary Material for the full list of characteristics). This same process was then repeated for each of the remaining five events, one at a time. After that, participants provided basic demographic information.

Table 1 Memory characteristic ratings for primary analyses (items 1–8)
Full size table

At the end, participants were asked the following attention check: “Do you feel that you paid attention, avoided distractions, and took the survey seriously?” They responded by selecting one of the following: (1) no, I was distracted; (2) no, I had trouble paying attention; (3) no, I didn’t not take the study seriously; (4) no, something else affected my participation negatively; or (5) yes. Participants were ensured that their responses would not affect their payment or their eligibility for future studies. Only those participants who selected (5) were included in the analyses. This same attention check has been used in recent published work (e.g., Stanley, Yin, & Sinnott-Armstrong, 2019).

Results

Participants described diverse kinds of moral transgressions that varied widely in severity: failing to keep promises to family and friends, damaging the property of other people, infringing upon other people’s privacy, shoplifting, cheating on tests and exams, cheating on significant others, shooting voyeuristic films, and hurting animals.

A series of linear mixed-effects models (LMEM) were computed using the ‘lme4’ software package (Bates, Mächler, Bolker, & Walker, 2015) in R. The judged severity of the remembered transgression was modeled as the fixed-effect term predicting each memory characteristic in separate models. Participant was modeled as a random effect in each model. Significance for fixed-effects was assessed using Satterthwaite approximations to degrees of freedom, and 95% confidence intervals around beta-values were computed using parametric bootstrapping (500 simulations).

We addressed each of our three main questions in the order that they were presented in the introduction. Table 2 summarizes all LMEM results that address all three main questions (additional exploratory results can be found in Table S2).Footnote 3 Table S3 displays the relationships between all memory characteristics. First, we investigated the effect of moral severity on the frequency of voluntary and involuntary retrieval, respectively. Moral transgressions judged to be more severely morally wrong were recalled more frequently – both voluntarily (b = 0.19, p < .0001, 95% CI [0.13, 0.26]) and involuntarily (b = 0.23, p < .0001, 95% CI [0.17, 0.29]) – than moral transgressions judged to be less morally wrong.

Table 2 Linear mixed-effects models (LMEMs) fixed-effects of judged moral wrongness for Study 1
Full size table

Second, we investigated the effect of moral severity on reported vivacity, detail, and sense of reliving. Moral transgressions judged to be more severely morally wrong were remembered more vividly (b = 0.19, p < .0001, 95% CI [0.12, 0.25]) and with more visual (b = 0.28, p < .0001, 95% CI [0.21, 0.35]) and other sensory details (b = 0.15, p = .0005, 95% CI [0.08, 0.23]). Remembering more severe moral transgressions also elicited a greater sense of reliving (b = 0.25, p < .0001, 95% CI [0.19, 0.32]).

Third, we investigated the effect of moral severity on the perceived personal importance and centrality of the moral transgressions. Participants regarded events involving their more severe moral transgressions to be more important (b = 0.29, p < .0001, 95% CI [0.21, 0.35]) and more central (b = 0.24, p < .0001, 95% CI [0.17, 0.31]) to their personal identities, as compared to their more minor transgressions.

Because the phenomenology of remembering past events and the judged morality of the actions could differ as a function of how much time has elapsed since the events occurred (Stanley et al., 2017), we statistically controlled for the number of months passed (hereafter referred to as Months), starting with the remembered events that occurred in the same month as the experimental session coded as 0 (for a similar methodological approach, see Stanley, Henne, & De Brigard, 2019).Footnote 4 An initial LMEM revealed that Months had a significant effect on the judged severity of the remembered transgressions (b = .004, p < .0001, 95% CI [.003, .005]) such that remembered behaviors judged to be more morally wrong occurred in the more distant past. Hence, we statistically controlled for the effect of Months by including Months as a fixed-effect term in addition to the judged severity of the remembered transgression to predict each memory characteristic in separate models.

The fixed-effect of judged moral severity on all eight characteristic ratings remained significant after statistically controlling for Months. Relative to moral transgressions judged to be less morally wrong, moral transgressions judged to be more severely morally wrong were recalled more frequently – both voluntarily (b = 0.20, p < .0001, 95% CI [0.14, 0.27]) and involuntarily (b = 0.24, p < .0001, 95% CI [0.19, 0.30]); were remembered more vividly (b = 0.20, p < .0001, 95% CI [0.12, 0.26]) and with more visual details (b = 0.27, p < .0001, 95% CI [0.21, 0.33]) and other sensory details (b = 0.14, p = .0005, 95% CI [0.06, 0.22]); elicited a greater sense of reliving while remembering (b = 0.25, p < .0001, 95% CI [0.18, 0.31]); and were regarded as more important (b = 0.29, p < .0001, 95% CI [0.21, 0.37]) and more central (b = 0.24, p < .0001, 95% CI [0.17, 0.31]) to their personal identities.

Discussion

The results from Study 1 suggest that, even though more severe moral transgressions presumably pose a greater threat to morally good self-concepts than more minor transgressions, more severe moral transgressions were recalled more frequently, with more details, with greater vivacity, and with a heightened sense of reliving. More severe moral transgressions were also more important and central to participants’ personal identities.

Study 2

In order to more directly compare the phenomenology of remembering transgressions of varying degrees of moral severity, in Study 2 we explicitly cued participants to provide a memory of a slightly morally wrong past transgression and a very morally wrong past transgression (or the most morally wrong transgression that they had committed). We then directly compared phenomenological characteristics and memory judgments as a function of these two cues.

Method

Participants

A total of 171 individuals voluntarily completed this study via AMT for monetary compensation. Qualified participants were individuals living in the USA who had completed over 50 HITs with an approval rating above 90%. Six participants reported not paying attention during the study (see attention check below), and three other participants failed to follow instructions, so data were analyzed with the remaining 163 participants (Mage = 34.20 years, SD = 9.61, age range = [19, 69]; 72 females). Informed consent was obtained from each participant in accordance with a protocol approved by the Duke University Campus Institutional Review Board.

Procedure

This study consisted of a single, self-paced session. Participants were assured that their names would not be linked to their responses and that all data would be securely saved and de-identified, indefinitely. Participants were asked to recall two specific events that happened when they were older than 15 years of age and to provide ratings for both events. Specifically, participants recalled and described in two to five sentences an event in which they personally did something that they believe to be slightly morally wrong, and an event in which they personally did something they believe to be very morally wrong (or the most morally wrong thing they have done since they were 15 years old). For each remembered event, participants indicated the month and year that the event occurred and provided a moral judgment (1 = very morally wrong; 7 = very morally right). Participants also provided ratings on the exact same items as in Study 1 (see Table 1 and Table S1). Then, participants reported basic demographic information.

Finally, as in Study 1, participants were asked whether they paid attention, avoided distractions, and took the survey seriously. They responded by selecting one of the following: (1) no, I was distracted; (2) no, I had trouble paying attention; (3) no, I didn’t not take the study seriously; (4) no, something else affected my participation negatively; or (5) yes. Participants were ensured that their responses would not affect their payment or their eligibility for future studies. Only those participants who selected (5) were included in the analyses.

Results

Participants described diverse kinds of moral transgressions in response to both memory cues. Slightly morally wrong behaviors included making graffiti, not returning extra change, shoplifting small items, cheating on exams, and lying to family and friends for selfish reasons. For very morally wrong behaviors, participants reported cheating on significant others, inflicting physical harm on others, drunk driving, and using drugs, among other things.

In order to confirm that our memory cue manipulation worked as expected, we conducted an initial paired-samples t-test. This indicated that slightly morally wrong-cued memories (Mslightly = 3.04, SD = .76) were judged to be less morally wrong than very morally wrong-cued memories (Mvery = 1.94, SD = 1.59; t(162) = 8.77, p < .0001, 95% CI [0.85, 1.35], Cohen’s d = .69).

We then computed paired samples t-tests to compare each memory characteristic as a function of Memory Cue (i.e., slightly morally wrong versus very morally wrong). Table 3 summarizes the paired samples t-tests results addressing our three main questions that were presented in the Introduction (additional exploratory results can be found in Table S4). (Table S5 and Table S6 provide correlation matrices for all relationships between all measured memory characteristics.) First, we investigated the effect of moral severity on the frequency of voluntary and involuntary retrieval, respectively. Past behaviors judged to be very morally wrong (i.e., severe moral transgressions) were more frequently recalled – both voluntarily (t(162) = 4.97, p < .0001, Cohen’s d = 0.39) and involuntarily (t(162) = 9.11, p < .0001, Cohen’s d = 0.71) – relative to behaviors judged to be slightly morally wrong (i.e., minor moral transgressions). Figure 1 graphically depicts the differences in frequency of recall as a function of the severity of the moral transgression.

Table 3 Paired-samples t-test results for differences in memory characteristics as a function of cue (slightly vs. very morally wrong) for Study 2 (items 1–8)
Full size table
Fig. 1
figure 1

Box plots are depicted for the frequency of voluntary recall and involuntary recall as a function of memory cue (slightly morally wrong vs. very morally wrong)

Full size image

Second, we investigated the effect of moral severity on the judged vivacity, detail, and sense of reliving. Severe moral transgressions were remembered more vividly (t(162) = 2.84, p = .0050, Cohen’s d = 0.22) and with more visual details (t(162) = 6.59, p < .0001, Cohen’s d = 0.52) and other sensory details (t(162) = 5.78, p < .0001, Cohen’s d = 0.45) than minor moral transgressions. Remembering severe moral transgressions also elicited a greater sense of reliving (t(162) = 4.01, p < .0001, Cohen’s d = 0.31). See Fig. 2 for differences in Vivacity, Visual Details, Other Details, and Reliving as functions of moral severity.

Fig. 2
figure 2

Box plots are depicted for the vivacity, detail, and sense of reliving as a function of memory cue (slightly morally wrong vs. very morally wrong)

Full size image

Third, we investigated the effect of moral severity on the perceived personal importance and centrality of their past moral transgressions. Participants considered their severe moral transgressions to be more personally important t(162) = 8.11, p < .0001, Cohen’s d = 0.64) and more central (t(162) = 7.66, p < .0001, Cohen’s d = 0.60) to their personal identities than their minor moral transgressions. See Fig. 3.Footnote 5

Fig. 3
figure 3

Box plots are depicted for the perceived personal importance and centrality of past moral transgressions as a function of memory cue (slightly morally wrong vs. very morally wrong)

Full size image

As in Study 1, we statistically controlled for Months (the number of months since the event occurred) in Study 2. A paired samples t-test revealed that slightly morally wrong-cued memories (Mslightly = 55.13, SD = 81.32) occurred in more recent past than very morally wrong-cued memories (Mvery = 106.89, SD = 111.09; t(162) = 5.56, p < .0001, 95% CI [33.39, 70.13], Cohen’s d = 0.44). Hence, we statistically controlled for the effect of Months by running a series of ANCOVAs with Memory Cue (i.e., slightly morally wrong vs. very morally wrong) as the factor and Months as the covariate for each memory characteristic rating. The effects of Memory Cue on all eight characteristic ratings remained significant after statistically controlling for Months. Compared to minor moral transgressions, severe moral transgressions were more frequently recalled – both voluntarily (F(1, 324) = 21.1, p < .0001, η2 = .06) and involuntarily (F(1, 324) = 65.4, p < .0001, η2 = .17); were remembered more vividly (F(1, 324) = 7.3, p = .0074, η2 = .02) and with more visual details (F(1, 324) = 25.6, p < .0001, η2 = .07) and other sensory details (F(1, 324) = 15.6, p < .0001, η2 = .05); elicited a greater sense of reliving (F(1, 324) = 8.5, p = .0038, η2 = .03) while remembering; were regarded as more personally important (F(1, 324) = 51.8, p < .0001, η2 = .14) and more central (F(1, 324) = 40.2, p < .0001, η2 = .11) to their personal identities.Footnote 6

We also performed more conservative non-parametric Wilcoxon signed-rank tests for paired samples (see Table S4). We obtained the same pattern of results with parametric and non-parametric tests (Figs. 1 and 2 and Fig. S1).

Discussion

These findings from Study 2 corroborate our results from Study 1 using a different cueing procedure. In Study 1, LMEMs revealed that remembering more severe moral transgressions, relative to less severe moral transgressions, is associated with higher ratings on memory characteristics. In Study 2, when we explicitly cued participants to provide two different kinds of memories differing in terms of the judged severity of the transgressions (slightly morally wrong vs. very morally wrong), we obtained the same pattern of results for the memory characteristics of interest.

General discussion

In an influential recent study, Kouchaki and Gino (2016) found evidence that people's memories of their own past unethical behaviors become obfuscated over time; they refer to this phenomenon as unethical amnesia. The authors offer an explanation for this unethical amnesia phenomenon: unethical past behaviors become obfuscated so that people can avoid psychological distress and protect against threats to their morally good self-concept. As we have argued, more severe moral transgressions elicit stronger negative feelings (Escobedo & Adolphs, 2010; Stanley et al., 2017) and should pose a greater threat to our morally good self-concepts (relative to more minor transgressions). For this reason, memories of more severe moral transgressions should be more likely to become obfuscated relative to more minor transgressions. However, our results instead suggest that recollecting more severe moral transgressions, relative to less severe moral transgressions, occurred more frequently, was experienced with greater vivacity, detail, and sense of reliving, and was judged to be more important and central to constructions of personal identity. Our findings are more consistent with studies investigating how convicted criminals remember their crimes (e.g., Evans et al., 2007a, 2007b; Woodworth et al., 2009), according to which convicted criminals tend to frequently ruminate on their crimes, their crimes tend to be remembered vividly, and the experience of remembering them tends to be accompanied with strong negative emotions.

The frequent retrieval, heightened vivacity, and substantial detail of memories involving our more severe moral transgressions are likely threatening to our morally good self-concepts. Nevertheless, the vast majority of people do still believe that they are fundamentally and essentially morally good (Aquino & Reed, 2002; Strohminger et al., 2017). They believe that they exemplify desirable moral traits, qualities, and virtues (Aquino & Reed, 2002; Stanley & De Brigard, 2019) and that these traits, qualities, and virtues are essential to their constructions of their selves and identities (Molouki & Bartels, 2017; Strohminger et al., 2017). How is it, then, that so many people hold these beliefs despite readily and frequently recalling their severe moral transgressions and remembering them vividly and with considerable detail? We have suggested that other mechanisms involving autobiographical memory might still help people to support their beliefs in their morally good self-concepts (Stanley & De Brigard, 2019). For example, people strategically compare their current selves to their past selves in order to facilitate an impression of moral improvement over time (Stanley & De Brigard, 2019; Stanley et al., 2019, 2017). So, frequently and vividly remembering our severe past transgressions might help us to continually contrast our more morally upstanding current selves with our less morally upstanding past selves. In this way, we can come to believe that we have changed for the better since committing those transgressions, and that who we are now is different than who we were then.

Readily, frequently, and vividly remembering our particularly severe past transgressions with a high degree of detail might also serve other functions. For example, memories of specific past events often serve a directive function (Pillemer, 2000, 2003). That is, our memories often serve as guides and reference points for forming intentions and making decisions. In one study, for example, Mackey et al. (2001) provided a detailed report of adolescent’s memory involving a particular interaction with her parents. This remembered event came to mind frequently and continued to direct social problem solving in many new circumstances after the initial occurrence. Similarly, remembering our particularly severe past transgressions readily, frequently, and vividly might be helpful for forming of intentions to behave differently in the future and for actually making better decisions in the future – even if those severe transgressions tend to be especially unpleasant and distressing to remember and ruminate upon.

A notable feature of our studies is that we utilized real-world moral transgressions that actually occurred in the past. In contrast, a relatively large proportion of studies in moral psychology have relied on artificial vignettes describing fictional dilemmas (e.g., the trolley problem or other sacrificial dilemmas) to probe people’s moral judgments and decisions. However, the kinds of moral transgressions that people actually commit in the world rarely match the kinds of transgressions that are described in the vignettes frequently used in the literature (Hofmann et al., 2014). Furthermore, recent empirical evidence (e.g., Bostyn, Sevenhant, & Roets, 2018; FeldmanHall et al., 2012) even suggests that moral judgments and decisions made in hypothetical vignettes fail to accurately predict similar kinds of real-life moral behaviors. However, there has been some debate over the conclusiveness of evidence provided by Bostyn et al. (2018) and over the value of research using hypothetical vignettes (Białek, Turpin, & Fugelsang, 2019; Bostyn & Roets, 2019; Evans & Brandt, 2019; Plunkett & Greene, 2019). In our studies, we obtained participants’ autobiographical memories of specific past moral transgressions to gain insights into how they remember and think about those events.

Several studies have investigated the objective accuracy of memories involving past moral and immoral actions (e.g., Reczek et al., 2018; Shu et al., 2011). However, the phenomenology of remembering these actions and the judgments people make about them have remained largely unexplored (Stanley, Yang, & De Brigard, 2018). This is surprising given considerable evidence for dissociations between phenomenological characteristics (e.g., vivacity and detail) and objective accuracy (Berntsen & Thomsen, 2005; Roediger & McDermott, 1995; Talarico & Rubin, 2003, 2007). Our studies provide some insights into the phenomenology of remembering past moral transgressions.

Limitations and conclusions

One limitation of our approach is that, because we did not experimentally manipulate the severity of the moral transgressions, we could only obtain a correlative understanding of relationships between judged moral severity and memory characteristics. This entails that we could not draw strong conclusions about the direction of our effects, nor could we examine whether third variables are casually responsible for other identified bivariate relationships. We also relied on self-reports, as is the case for most research on autobiographical memories for real-life events, which has certain inherent limitations (see Garcia & Gustavson, 1997). Furthermore, it is possible that participants found certain questions ambivalent. For instance, in responding to the Importance and the Centrality questions (i.e., “How important is the event to you personally (it involves an important episode in your life)?” and “Is the event in your memory a central part of your life story?”), participants might view a moral transgression to be important and central to one’s life story either because one thinks the event is indicative of one’s current identity or because one has learned from the event and since become a better person.

Our findings leave also open the possibility that only a few of people’s moral transgressions are frequently and vividly remembered while most memories of past immoral behaviors become obfuscated with time. Given the relatively limited quantities of memories our participants reported (six memories in Study 1 and two memories in Study 2), our results do not rule out this possibility. In order to test the proportion of clearly remembered memories of moral transgressions, future work may need to collect a larger quantity of autobiographical memories of moral transgressions per participant. To that end, researchers may find that showing the participants common words as cues (see Escobedo & Adolphs (2010) for a similar approach), a variant of the Galton cue-word technique (Rubin & Schulkind, 1997). Alternatively, researchers could ask participants to provide more specific kinds of moral transgressions (e.g., lying, causing emotional harm; see Stanley et al. (2017) for a similar approach) to facilitate autobiographical memory retrieval.