As shown above the self is linked to memory (Kihlstrom and Klein, 1994; Wilson and Ross, 2003; Conway, 2005; Klein, 2012), but also to its morality via self-conscious emotions of, for example, guilt. A self-conscious, compared to, for example, a basic, emotion is related to the self because it can be elicited only by the self: “… self-awareness… can affect the nature of emotional experience… it can bring self-evaluative states such as shame, guilt, and pride” (Silvia and Eddington, 2012, p. 426). Guilt is, in other words, elaborated by the cognitions of the self; by the “I” and “me” self-reflections and evaluations (Tracy and Robins, 2004), in relation to some personal and/or cultural standard (Lewis, 2008). Its adaptive, reparative, function is to motivate the psychological agent to avoid socially undesirable behavior (Baumeister et al., 1994; Cristofari and Guitton, 2014); that is, not “to do wrong”.

In that sense, guilt is part of the morality of the self, the autobiographical memory (Körner et al., 2016), due to its psychological role of blaming one’s behavior: “I did that horrible thing” (Tangney and Tracy, 2012, p. 448). Several previous findings in emotion research have also indicated that one of the roles of negative emotions is to regulate behavior (Frijda et al., 1989; Scherer and Wallbott, 1994). It has also been suggested that proneness to guilt may be a trait that prevent people from thinking, feeling and behaving unethically (Cohen et al., 2011, 2012). In other words, “guilt proneness is a personal trait indicative of a predisposition to experience negative feelings about personal wrongdoing” (Cohen et al., 2012, p. 355). It has been shown to differ, for example, between people who have not been diagnosed with antisocial personality disorder and those with antisocial personality disorder (Haidt and Kesebir, 2010). Finally, previous neuroimaging studies have also indicated neural correlates between moral functioning and autobiographical memory (Han, 2017) as well as between moral emotions and self-related processing (Han et al., 2016); thus, suggesting associations between the self and its morality at the neural level.

As mentioned above, remembering is a reconstructive process. This means that recalled personal, autobiographical, memory can transitorily reconstruct one’s sense of the self. For example, Gino and Desai (2012) showed that asking participants to recall childhood memories triggered the concept of “moral purity” (“Children are innocent and virtuous.”); which in turn affected their prosocial behavior. This suggests that our sense of the self may be influenced and momentarily re-structured by moral involvements (Bergman, 2002; Hardy and Carlo, 2005; Heiphetz et al., 2016), indicating a moral self-regulating mechanism (Escobedo and Adolphs, 2010); an essential moral self, suggesting that “Moral traits are considered more important to personal identity than any other part of the mind” (Strohminger and Nichols, 2014, p. 168).

It can, according to Blasi (1984), be experienced phenomenologically as the “real me”: “…the authentic self, and the deepest principle that guides the individual” (Stets and Carter, 2011, p. 194). In line with this, some studies have indicated that recalling good past moral actions can motivate a consistency in subsequent moral behavior (Reed et al., 2007; Shao et al., 2008), but others have shown that good past moral actions can also license to act immorally (Monin and Miller, 2001; Sachdeva et al., 2009; Merritt et al., 2010; Conway and Peetz, 2012).

Given the above, we investigated one aspect of the reconstructive personal, autobiographical memory, namely, the past self-defining moral memories in relation to present moral intentions of willingness to do wrong and feelings of guilt in social and economic situations. We posed the following questions: how do we remember our own moral deeds, and do they have any impact on our current morality in similar contexts? In general, we expected the self-defining moral memory to be coherent, vivid and easy to recall due to its first-person perspective and the self-conscious, emotional constituents (Moll et al., 2005; Welzer and Markowitsch, 2005; Escobedo and Adolphs, 2010; Sutin and Robins, 2010), as well as involving the feeling of “seamless whole” (Prebble et al., 2013). In line with we-are-motivated-to-act-consistent-with-our self-view perspective (Blasi, 1980; Aquino and Reed, 2002), we also predicted that the moral intention of willingness to do wrong will be lower in participants when they experienced greater feelings of guilt, due to the latter’s reparative function (Tangney et al., 1996; Gino and Pierce, 2009).

The two types of situations were included because previous research has indicated social (Kortenkamp and Moore, 2006; Greenwood, 2011; Knez, 2013, 2016b) and economic/resource (Hardin, 1968; Agerström and Björklund, 2009) dimensions of moral dilemmas; in the words of Aquino et al. (2009, p. 124): “…situational factors may activate a person’s moral identity or they may activate alternative facets of identity, thereby increasing or decreasing the current accessibility of the moral self-schema within the working self-concept.” Some previous research has also indicated that morality and self-conscious emotions may differ across cultures (Markus and Kitayama, 1991; Wang, 2008; Wong, 2009). For example, Vauclair et al. (2014) investigating four-country/two-culture samples reported some type of relativism in morality dimensions across cultures.

Finally, we measured 10 aspects of self-defining moral memory (vividness, coherence, accessibility, time perspective, sensory detail, emotional intensity, visual perspective, sharing, distancing, and valence) previously identified as the phenomenological dimensions of autobiographical memory (Sutin and Robins, 2007). Phenomenology is a crucial feature of autobiographical memory providing experiences, thoughts and emotions (Singer and Salovey, 1993; Sutin and Robins, 2007) as the base for the sense of self (Prebble et al., 2013).

In total, 116 subjects with varying levels of education and life experiences and a mean age of 25.9 (range = 17–67) participated in the present study. They (90 women and 26 men) were randomly recruited from a university (N = 63 students, females = 51, males = 12, mean age = 23.7, age range = 18–38), an upper secondary school (N = 32 students, females = 28, males = 4, mean age = 18.2, age range = 17–20), and from a private company (N = 21 employees, females = 11, males = 10, mean age = 44.4, age range = 26–67). The subjects received a cinema ticket in appreciation of their participation.

The temporal ordering of the tasks began with the judgment and guilt estimations of moral dilemmas followed by the estimations of subjects’ most important self-defining moral memories; memories of high “moral identity centrality” (Aquino et al., 2009; Dinh and Lord, 2013) grounding the self (Conway, 2005).

First, participants were told that their participation was voluntary and that they could leave the study whenever they wanted. Second, participants completed the moral judgment task (willingness to do wrong and anticipated guilt) including four dilemmas, two representing a social and two representing an economic situation (see below for details). Third, they were asked to recall and write down their most important self-defining moral memory (by using one the four dilemmas as a frame of reference for their self-defining moral memory; this was done, in order to relate estimations of moral intention and guilt to self-defining moral memory) with the following instruction:

Fourth, they were asked to estimate their retrieved self-defining moral memory (see above) on 63 scales, measuring the phenomenological dimensions of vividness, coherence, accessibility, time perspective, sensory detail, emotional intensity, visual perspective, sharing, distancing, and valence. In the present study we will report only quantitative data; that is, the phenomenology estimations (63 scales) of the self-defining moral memory.

Finally, an ethical approval was not required for this study in accordance with national and institutional guidelines. Though, we conducted it in accordance with APA’s (American Psychological Association) ethics code. Accordingly, participants were informed about: (1) objectives of the present study; (2) their right to withdraw from the participation at any time without any consequences; (3) how long it will take to complete the tasks and information about the types of tasks involved; (4) confidentiality; (5) that they will not be financially compensated for their participation; and (6) whom to contact about any questions related to this study.

In line with Agerström and Björklund (2009), behavioral moral intentions (willingness to do wrong) and anticipated moral guilt (feelings of guilt) were measured with four moral dilemmas; two (nr. one and three, see below) depicting social, and two (nr. two and four, see below) depicting economic situations. Moral dilemmas represented a disagreement between altruistic and egoistic motives (Agerström and Björklund, 2009; Knez, 2016b) in that the participants faced a conflict between unselfish motives in terms of greater well-being or gain for one or several other persons (at the cost of a smaller personal discomfort or financial lost) and selfish (pure hedonic) motives in terms of a smaller personal comfort, convenience, or gain (at the cost of a greater utility, well-being or pro-environmental values for one or several other persons). Thus, dilemmas depicted a clash between universal and non-universal norms/principles/values (Hare, 1981; Hauser, 2006). Accordingly and yielding all dilemmas, participant faced a conflict between two alternatives; one implying worse consequences for one or several other people, and one implying better outcomes compared to the alternatives (Singer, 2011).

Participants were asked to indicate their willingness to do wrong and feelings of guilt on a nine-point Likert scale, where 1 indicated “not likely at all” and “no guilt at all” and 9 indicated “very likely” and “extreme guilt.” The following four dilemmas and subsequent two questions, measuring moral intention and moral guilt respectively, were administrated:

“The Memory Experience Questionnaire” (Sutin and Robins, 2007), henceforth MEQ, was used to estimate subjects’ most important self-defining moral memories. This measure contains 63 scales/statements, defining the 10 phenomenological dimensions of vividness, coherence, accessibility, time perspective, sensory detail, emotional intensity, visual perspective, sharing, distancing, and valence. Estimations were made on a five-point Likert scale ranging from 1 “strongly disagree” to 5 “strongly agree.” The recalled self-defining moral memories were categorized as belonging to either a “self-defining moral memory related to a social situation (e.g., betrayal of a desperate friend)” or a “self-defining memory related to an economic situation (e.g., tax evasion).”

The 10 phenomenological dimensions of MEQ have previously shown an adequate reliability, Cronbach alphas of 0.72 to 0.97 (Sutin and Robins, 2007). According to these authors, vividness (Cronbach alpha 0.85) refers to the visual clarity and intensity of a memory. Coherence (Cronbach alpha 0.79) implies the degree to which the memory is remembered as a logical story in a certain time and place rather than fragments of an experience or a mixture of similar memories. Accessibility (Cronbach alpha 0.83) indicates the easy of a memory to be retrieved. Time perspective (Cronbach alpha 0.85) refers to the perceived time-related clarity. Sensory detail (Cronbach alpha 0.72) measures the sensory details re-experienced. Emotional intensity (Cronbach alpha 0.86) refers to the intensity of memory-related emotions. Visual perspective (Cronbach alpha 0.87) indicates the perspective from which the memory is “seen”; higher scores indicate that the person views the experience through her/his own eyes (a first person perspective) as opposed to being an observer watching oneself (a third person perspective). Sharing (Cronbach alpha 0.89) measures the extent to which a memory is shared with others. Distancing (Cronbach alpha 0.87) measures how much the person distances him-/herself from the retrieved memory, that is, from “having been” that person in the memory. Valence (Cronbach alpha 0.97) indicates the positive/negative valance of a memory.

The within-subject-independent variables were: (1) moral-related intention in social vs. economic situations; (2) moral-related guilt in social vs. economic situations; and (3) 10 phenomenological dimensions of the self-defining moral memory. The between-subject-independent variable was the self-defining moral memories categorized as belonging to either a “self-defining moral memory related to a social situation (e.g., betrayal of a desperate friend)” or a “self-defining memory related to an economic situation (e.g., tax evasion).

The dependent variables were the estimations of willingness to do wrong (moral-related intention), feelings of guilt (moral-related guilt), and phenomenology of the self-defining memory.

Five types of analyses was performed: (1) Effects of self-defining moral memory (memories of social vs. economic moral situations) and moral-related intention (willingness to do wrong in social vs. economic moral situations) on moral-intention-estimation; (2) Effects of self-defining moral memory (memories of social vs. economic moral situations) and moral-related guilt (feelings of guilt in social vs. economic moral situations) on moral-guilt-estimation; (3) Associations between feelings of guilt and moral intention (willingness to do wrong) in social vs. economic moral situations; (4) Effects of self-defining moral memory (memories of social vs. economic moral situations) and 10 phenomenological dimensions on mean phenomenology of self-defining memory; and (5) Associations between moral-related intention (willingness to do wrong) and guilt, and the 10 phenomenological dimensions of self-defining moral memory, across moral situations. ANOVAs were performed in sections (1), (2), and (4). Regression analyses were performed in sections (3) and (5). Note: no significant correlations between age and estimations of moral intention and guilt were found; therefore the age was excluded from further analyses.

Additionally, and following Sutin and Robins (2007) a descriptive interpretation of the mean phenomenology (estimation) for each dimension was made. Thus, and in addition to the parametric statistics, we interpreted the mean estimation within each dimension as signifying more (mean score 3-5) or less (mean score 3-1) type of phenomenology (10 dimensions). For example, a self-defining moral memory was interpreted as more easy to access (accessibility) and share with others (sharing) if the mean estimation was 3-5 and as less easy to access and share with others if the mean estimation was 3-1. In other words, the mean score of 3 on a five-point Likert scale was regarded as a demarcation line of a more or less type of a phenomenology of self-defining moral memory content.

No significant result involving self-defining memory was obtained. However, a main effect of type of moral-related intention (willingness to do wrong), Greenhouse–Geisser = 2.98 (MSE), F(1,114) = 74.57, p < 0.01, η² = 0.4, showed that participants were more willing to do wrong in social compared to economic moral situations (M = 5.47, SD = 1.84 vs. M = 3.51, SD = 1.91).

As above, no significant result involving self-defining memory was obtained. A main effect of type of moral-related guilt, Greenhouse–Geisser = 2.14 (MSE), F(1,114) = 26.25, p < 0.01, η² = 0.19, showed however, that participants felt more guilty when behaving immorally in economic vs. social moral situations (M = 6.43, SD = 1.76 vs. M = 5.45, SD = 1.72).

Given the two results above, regression analyses were performed to check for the associations between guilt and moral intention (willingness to do wrong) in the respective moral type of situation. As can be seen in Table 1, guilt was a stronger predictor for willingness to do wrong in economic compared to social moral situations. Thus, the association the higher the level of guilt the less willingness to do wrong was accounted for more by the economic than by the social moral situation (explained variance 50% vs. 24%, see Table 1).

A main effect of the 10 phenomenological dimensions, Greenhouse–Geisser = 7.75 (MSE), F(9,1026) = 143.48, p < 0.01, η² = 0.14, showed that participants’ self-defining moral memory contained mostly details of accessibility (M = 4.05, SD = 3.67) and least of sharing (M = 2.08, SD = 0.93), t(115) = 5.59, p < 0.01 (see Figure 1). This indicates that the self-defining moral memory is generally easy to retrieve, but rarely shared with others (p < 0.01). (For all post hoc comparisons between the 10 phenomenological dimensions see Table 2.)

However, if we interpret descriptively (see Materials and Methods) the mean phenomenology within each dimension (signifying 3-5 as more and 3-1 as less), it can tentatively be suggested that (see Figure 1) the self-defining moral memory is relatively: (1) more easy to access (accessibility), logically structured (coherence), vivid, seen from the first-person perspective (visual perspective), real (sensory detail), and (2) less positive (valence), emotionally intense, chronologically clear (time perspective), in agreement with the present self (distancing), told about to others (sharing).

Additionally, an interaction effect between the type of self-defining moral memory and the 10 phenomenological dimensions, Greenhouse–Geisser = 143.87 (MS), F(9,1026) = 18.67, p < 0.01, η² = 0.14, was obtained. According to the follow-up one-way ANOVAs significant differences between social and economic category of self-defining moral memory were shown for valance, F(1,116) = 19.22, p < 0.01, η² = 0.14, emotional intensity, F(1,116) = 10.02, p < 0.01, η² = 0.08, sharing, F(1,116) = 10.26, p < 0.01, η² = 0.08, and coherence, F(1,116) = 8.95, p < 0.01, η² = 0.07. As can be seen in Figure 2, this indicates that the self-defining social compared to economic moral memory is more negative, emotionally intense and talked about, but less coherent.

Regression analyses were performed to check for the associations between moral-related intention (willingness to do wrong) and guilt, across moral situations, and the 10 phenomenological dimensions of self-defining moral memory. It was shown that the higher willingness to do wrong the less the accessibility of self-defining moral memory, and the greater the feeling of guilt the more real (sensory detail) self-defining moral memory (see Table 3).