Mapping Conscience: Network Analysis Into the Differences in Maturation of Offending and Non-Offending Adolescents

Abstract

Conscience is a diagnostically relevant concept in forensic psychiatry, but often misinterpreted as an all-or-none phenomenon. We conceptualize the conscience as a psychic function in which elements like empathy, self-conscience emotions such as shame, guilt and pride, and moral orientation work together. The differences in conscience functioning can be described in terms of developmental levels of integration. We conducted network analyses on data collected via a questionnaire survey held among 52 offending and 243 non-offending juveniles. We displayed two networks: One representing the non-offenders’ normative and one representing the offenders’ defiantly maturing conscience. As was hypothesized, in the non-offenders network, almost all elements clustered into one clinically meaningful network, indicating integration of the different elements of the normative maturing conscience. In the offenders network, the correlations between the elements were sporadic, indicating a lack of integration of the defiantly maturing conscience. The difference between the two networks was more prominent for empathy and moral orientation than for self-conscious emotions. This research supports the theory of differences in maturation of conscience instead of being an all-or-none phenomenon and calls for further research, taking a deeper look at the significance of integration of the conscience and its implications for offending behaviour.

Keywords

conscience delinquency empathy guilt moral reasoning network analysis shame

Introduction

Conscience is a clinically relevant concept in forensic psychiatry. Forensic psychiatric or psychological reporters are asked to describe the functioning and level of development of a suspect’s conscience. Research into forensic practice has shown that a malfunctioning conscience is the main reason to plead for mandatory treatment of an individual in the Netherlands (Duits, 2006; Le Sage, 2005; Ten Berge et al., 2009). Furthermore, insight into the functioning and development of the conscience is essential for formulating effective individual treatment goals and can contribute to the formulation of interventions that create the circumstances under with the conscience can develop favorably. Despite its relevance, there is no consensus on the precise definition of the term conscience nor do we have a comprehensive theory for describing and assessing its function (Le Sage, 2005). Some define conscience as the level of cognitive or emotional moral development (Eisenberg & Farbes, 1998; Gibbs, 2019), others as the capacity for cognitive and affective empathy (Hoffman, 2000; Jolliffe & Farrington, 2004), or the superego (Carveth, 2013).

In a series of papers, Schalkwijk (2015, 2018; Schalkwijk, Stams, Dekker, et al., 2016; Schalkwijk, Stams, Stegge, et al., 2016; Verkade et al., 2020) proposed a theory of the conscience as a psychic function for balancing the self. Self-states fluctuate and are evaluated continuously, a process which is regulated by the conscience. Most of the time, the conscience functions in the background of consciousness, keeping self-esteem stable. Like a thermostat, it remains on stand-by when self-esteem is in equilibrium, but it comes into action when it receives and has to process information indicating that the homeostasis of the self is threatened. In Schalkwijk’s theory, there are three necessary domains for the functioning of conscience: Empathic activity, experiencing self-conscious emotions, and making moral judgments (Figure 1). The more the three domains overlap, as shown in the middle of the figure, the more they contribute optimal to functioning of the conscience. Each of these domains includes two or more distinct aspects that develop during one’s youth. For example, the domain empathy contains the aspect of affective and of cognitive empathy.

Figure 1.

Domains of the conscience. This Venn diagram represents the relationship between the different psychological functions, empathy, moral reasoning and self-conscious emotions, and the conscience.

The level of maturation¹ of the conscience can be inferred from the level of the integration of the components of conscience that gradually appear in early childhood development. The earliest element of conscience to develop is the capacity for affective empathy, generally evolving in the first 2 years. Empathic ability is not only fundamental in interaction with others, but it is also important for the emergence of feelings of shame and guilt about (fantasized) interactions with imaginary others (Cuff et al., 2016; Schalkwijk, 2015). Then, a character-like proneness to experience self-conscious emotions, such as shame, pride, and guilt develops, alongside with characteristic mechanisms for coping with these emotions. Simultaneously, cognitive empathy starts to evolve, and this ability, together with the capacity for moral judgement, expands strongly after the age of about 7 (Schalkwijk, Stams, Stegge, et al., 2016). The integrated accumulation of these developmental processes ensures that self-esteem can be kept in balance: the conscience becomes an internal working model for self-esteem regulation with unique personally preferred reaction responses (Tangney et al., 2007).

When in a clinical assessment of conscience one of the components appears to be completely missing or is grossly deficient, the clinician will have to conclude that the conscience lacks an essential component and that this hinders good functioning of the conscience. However, this is the case in only a very small percentage of delinquent adolescents. More often, a relative deficit is visible and the clinician will then describe the overall functioning of the conscience, including the influence of the relative deficit on overall functioning.

As mentioned earlier, meta-analyses have shown that insufficient empathic capacities (Van Langen et al., 2014), low levels of guilt and shame (Spruit et al., 2016), and lower stages of moral judgment (Stams et al., 2006) are positively correlated with criminal behavior. However, the way these domains relate to one another in the functioning of the conscience is still unknown. Research is needed to find out whether they work independently or influence each other mutually.

The present study builds on the research of Schalkwijk, Stams, Stegge, et al. (2016), comparing a group of adolescent offenders and non-offenders. We found moderate to small statistically significant differences between the two groups in the levels of affective or cognitive empathy, proneness to experiencing and handling self-conscious emotions, and moral developmental level. But beings clinicians as well, we questioned the significance in real life of these differences. We concluded that the idea of an all-or-nothing difference between the two groups was not feasible and rejected the idea of actual deficits. Instead we clinically gave meaning to the relative differences and concluded that the most likely interpretation was a difference in developmental level and integration of the components. This hypothesis fits well with the fact that almost all delinquent adolescents relinquish delinquency after the age of 23, when their focus shifts to work, relationships, and housing (Schalkwijk, 2022).

In the present study we tried to find new patterns in the existing data that could give a tentative explanation for the differences in maturation of the two groups. We therefore decided to focus on the quality of integration of the different aspects that are component parts of the different domains of the conscience. We used network analysis to study the relationships between the different items on empathy, self-conscious emotions, and morality without assuming them to have any a-priori relations. Our expectation was that for the non-offenders, the post-hoc grouping of the items into communities would reflect the theoretical a priori domains of the conscience and that all communities would be interconnected. This would indicate that the interrelated domains are indeed different parts of a mature conscience. Secondly we expected that for offenders, taking as assumption that they have a lower level of maturation of conscience, the different items would be related differently.

Our study aims at delving further into, and consolidating the integrative theory on the conscience as one integrated function incorporating empathy, moral reasoning, and self-conscience emotions. We expect that increased knowledge in this field will contribute to more consensus on the understanding of the conscience

Method

Participants

A total of 334 adolescents between 13 and 18 years of age participated. The non-offender group consisted of 275 adolescents (54% male and 46% female), attending different types of schools with an overrepresentation of schools for lower levels of educational achievement. The offender group consisted of 59 adolescents (80% male and 20% female) who completed the questionnaire while being psychologically evaluated before trial. Of this group, 39 adolescents were in pre-trial psychological assessment, and were therefore, “suspects” in the eye of the law. The other 20 adolescents were post-trial, in other words: convicted offenders. Throughout this paper we will use the word “offenders” for the sake of international harmonization of concepts. The offenses the adolescents were charged with included theft and robbery, either with or without violence. There were no charges of sexual offenses. The average age of the offender group was 1 year higher than the non-offender group (M_age and SD: offenders 15.53 (1.21), non-offenders 14.52 (1.07)). For further specification of demographic characteristics we refer to Schalkwijk, Stams, Stegge, et al. (2016).

Procedure

The non-offenders were found by asking schools to participate in the project, whereas the offenders were found via forensic psychologists called in to examine juveniles awaiting trial or sentencing. Offenders completed the questionnaires on a voluntary basis at home, at school, or in prison. They received €7.50 as a token of gratitude.

For all participants informed consent was obtained from both parents and the adolescent participants. The research was introduced to the participants as “a survey on the opinions of adolescents to help teachers to better understand adolescents.” In the questionnaire, references to delinquency were not made, but after all the questions had been answered, additional biographical questions on social-cultural background, school career, and delinquency were asked. The non-offenders completed the questionnaires in their classroom during a regular lesson. Approval was obtained from the Ethics Review Committee of the University of Amsterdam.

Questionnaires

Interpersonal Reactivity Index (IRI)

The IRI (Davis, 1983) was devised to measure empathy. Using 5-point Likert-type scales, four seven-item scales each tap a separate item of the global concept empathy: “Perspective Taking” (PT), “Empathic Concern” (EC), “Fantasy” (F), and “Personal Distress” (PD). Perspective Taking, referring to spontaneous attempts to adopt the perspective of others, assesses the cognitive aspect of empathy. Empathic Concern, designating one’s feelings of warmth, compassion, and concern for others, measures the affective aspect of one’s empathic capacity (Jolliffe & Farrington, 2004). The Fantasy-scale, relating to the tendency to identify with characters in movies, novels, plays, and other fictional situations is also related to the traditional notion of empathy, as has been shown by Pulos et al. (2004). Finally, the fourth scale, Personal Distress, represents a separate factor that might refer to the physical aspect of empathic capacity and functions as a pro-social motive (Hoffman, 2000). Personal Distress seems to be more of a focus on one’s own emerging (negative) emotions in a situation of heightened emotion.

Test of Self-Conscious Affect for Adolescents (TOSCA-A)

The TOSCA (Tangney et al., 1989) was devised to measure the disposition or proneness to experience shame, guilt, and pride. In this research, the version for adolescents was used, the TOSCA-A (Tangney & Dearing, 2002). A basic tenet of the emotion theory underlying the TOSCA-A is that the self-conscience emotion is elicited when some aspect of the self is scrutinized and evaluated with respect to moral standards (Tangney et al., 2007). In the questionnaire the adolescent indicates his or her propensity to experience shame and guilt in scenarios about different situations on a 5-point Likert-type scale. The Guilt scale is a measure of mild and adaptive forms of guilt, whereas the Shame scale is a measure of maladaptive aspects associated with shame (Luyten et al., 2002). The TOSCA-A also measures Alpha Pride and Beta Pride; the first reflecting positive inward gratification, the second reflecting behavior that contributes to hubris. However, reliability of the latter two scales tends to be rather modest, as they are based on only few items each (Tangney et al. 2007). The TOSCA-A also contains two scales that tap coping styles for dealing with self-conscious emotions: externalization of blame and detachment from the situation.

Compass of Shame Scale (CoSS)

This instrument assesses individual differences in coping with shame and has shown satisfactory construct validity (i.e., close model fit in confirmatory factor analysis) and high internal consistency reliability (Capinha et al., 2021; Elison, Lennon, & Pulos, 2006; Elison, Pulos, & Lennon, 2006; Schalkwijk, Stams, Dekker, et al., 2016; Vagos et al., 2019). The adolescent is asked how he or she handles shame when he or she experiences it. The CoSS consists of four 10-item scales representing the poles of Nathanson’s (1992) compass of shame theory. Of these “Attack Self” represents inward-directed anger and self-blame, “Avoidance” stands for the tendency to hide or withdraw when shamed, “Denial” reflects disavowal and emotional distancing or minimization of the situation into something neutral or positive, “Attack Other” represents outward-directed anger (i.e., aggression) and blaming others. A fifth scale assesses adaptive responses to shame, with a minimum of distortion of the shame emotion: “Adaptive” captures the acknowledgement of shame and motivation to apologize and/or make amends. All in all the basic tenet of the CoSS is that it measures emotion-focused coping. In the internalizing copings styles Attack Self and Avoidance shame is acknowledged and the self is experienced as failing. In the externalizing coping styles Denial and Attack Other shame is not acknowledged and the shame feeling is diminished. Similar to the TOSCA-A, all scales of the CoSS, relate to different items of the domain of self-related emotions.

Moral Development Measure (MOM)

The MOM (Stams et al., 2008) is an 18-item self-report measure, assessing the adolescent’s inclination to a Punishment or Victim-based moral development. The two types represent different levels of moral development; punishment-based reflects a pre-conventional level whereas victim-based indicates functioning on a conventional level of moral reasoning (Gibbs, 2019). The questionnaire presents nine moral situations, each including a perpetrator and a victim. The respondent is asked to evaluate how serious the consequences are for the victim and how serious the punishment is for the perpetrator on a 4-point Likert-type scale, varying from “not serious” to “very serious.” If a respondent tends to consider the consequences for the victim as very serious, showing consideration for or identification with the victim, he or she is said to have a strong victim-based development. If a respondent most frequently considers the consequences for the perpetrator as very serious, he or she is said to have a strong punishment-based development. In this way, both types of moral development reflect a different aspect within the domain of moral reasoning as part of the conscience.

Statistical Analyses

Network analyses were conducted to assess the interrelations between the items of the IRI, TOSCA-A, CoSS, and MOM. Keep in mind that “item” refers to the sum scores of the subscales of the questionnaire. A network consists of nodes, representing the items, and edges, representing connections between node-pairs. A network therefore represents the (patterns of) cohesion between the various items believed to be aspects of the conscience. From this perspective, an unintegrated and malfunctioning conscience is characterized by the loss of desired connections and the formation of undesirable connections. Hence, one would expect the conscience networks of the non-offenders and offenders to chiefly share their structure while differing in a number of topological locations. Hence, for that reason network extraction was based on targeted fused ridge estimation of scaled partial correlation matrices (Bilgrau et al., 2015; Van Wieringen & Peeters, 2016). This approach explicitly takes into account that there are two groups of interest (offenders and non-offenders) “for which the shared network structures should be fused whereas the unique network structures should be distinguished” (De Leeuw et al., 2017). This approach can also deal with high variable to observation ratios. The resulting networks are conditional independence graphs. In such graphs the nodes represent the items and the connections between node-pairs represent a substantive partial correlation, that is, an association that cannot be conditioned away by conditioning on the remaining items.

The specifics of network extraction are as follows. Complete cases were retained: 52 offenders and 243 non-offenders. The data was Gaussianized by a nonparanormal transformation, a nonparametric extension of the normal distribution (Liu et al., 2009). Subsequently the networks were extracted according to the method mentioned above. This method, in this instance, depends on three penalty parameters: two ridge penalties (determining the rate of shrinkage of the offender and non-offender partial correlation matrices toward the identity matrix) and one fusion penalty (determining the similarity between the offender and non-offender partial correlation matrices). Optimal values for these penalties were determined by k-fold cross-validation with k set to 30. The resulting partial correlation matrices were sparsified by thresholding, choosing an absolute value ≥0.2 as the cut-off value. In this way, only those edges were retained of which the corresponding absolute partial correlation equaled or exceeded .2. The networks were subjected to further analysis, more specifically community-detection: the identification of closely connected subnetworks that are believed to represent functional modules. Community-detection was based on the Girvan-Newman algorithm (Girvan & Newman, 2002).

Results

Our expectation as far as the non-offenders were concerned was that the post-hoc grouping of the items into communities would reflect the theoretical a priori domains of the conscience and that all items would be interconnected. This would confirm that the interrelated domains were, in fact, different parts of a mature conscience. In Figure 2, the non-offender network, using the Girvan Newman algorithm, is presented.

Figure 2.

Non-offender network using the Girvan Newman algorithm. A solid edge represents a positive partial correlation, a dashed edge, and a negative partial correlation.

In the non-offender network, all items were connected and as such part of one interconnected network, with the exception of the two items related to the proneness to experiencing pride. The network consisted of three closely related communities. The community shown on the middle right represents all items of maladaptive shame coping (all part of the CoSS), with the two internalizing shame coping items related to proneness to shame and the two externalizing shame coping items interrelated. This community stands in a positive relationship to the community shown on the middle left that represents the items relating to the proneness to experiencing shame and guilt or showing a defense reaction against experiencing self-conscious emotions (all part of the TOSCA-A). The community shown at the bottom of Figure 2 is more complex, containing all the items on empathy and moral development and the item on adaptive shame coping. All items in this community are positively correlated to one another except for Punishment-based Moral Orientation which is directly negatively associated with both Victim-based Moral Orientation and Adaptive Coping of Shame.

For offenders, assumingly with a lower level of maturation of conscience, we expected the items to be related to a lesser extent, or at least in a different way. As can be seen in Figure 3, representing the offender network using the Girvan Newman algorithm, this is a more erratic network with less connections between the different items.

Figure 3.

Offender network using the Girvan Newman algorithm. A solid edge represents a positive partial correlation, a dashed edge, and a negative partial correlation.

This lack of connection may indicate that, when one aspect of the conscience is activated, other items are not automatically activated as well. As with the non-offenders, the proneness to experience self-conscious emotions falls into two communities: On the one hand items connected with the proneness to experience shame and guilt or defense against experiencing self-conscious emotions (bottom right) and on the other proneness to experiencing pride (upper left). Maladaptive shame coping items here fall apart in two separate communities: internalizing (bottom left) versus externalizing (middle left) coping strategies. Empathic Concern and Fantasy form one community, both related to affective empathy. Victim-based Moral Orientation is isolated within its own community (upper right). In the middle right a more extended community stands out, showing a negative relationship between adaptive coping and punishment orientation and a positive relationship between adaptive coping and perspective taking. Interestingly, we found a positive partial correlation between Punishment-based moral orientation and Personal Distress, the physical aspect of empathy.

Discussion

In the first study on this dataset (Schalkwijk, Stams, Stegge, et al., 2016) we were interested in differences between the group of offenders and non-offenders and we came to the conclusion that the domains of conscience seemed to differ moderately in maturation and not in an all-or-nothing fashion. In the present, more exploratory study, we focused on the relationships of the different aspects and domains of conscience within the different groups and tried to get insight in the quality of the conscience in the different groups. Naturally, we cannot escape comparing the two groups and their differences will be presented also. In our presentation of the results, we will assume that the findings for the non-offending adolescents are a model for a normative maturing conscience and those for the offending adolescents a model for the defiantly maturation of conscience.

The Normative Maturing Conscience

In the non-offenders network, almost all items were seen to cluster into one network of three connected communities, supporting the hypothesis that a normative maturing conscience can indeed be understood as a psychological function, integrating different domains (Schalkwijk, Stams, Stegge, et al., 2016). The isolated position of the two pride-items is most likely related to the low reliability of these items. For the time being, we will leave them out of our discussion.

As can be seen in Figure 2, three clinically meaningful communities came to the fore: one for maladaptive shame coping, one for the proneness to or the defense against experiencing shame and guilt, and one including empathy, moral orientation, and adaptive shame coping. The first two communities roughly replicate the TOSCA-A and the CoSS, showing that the items of each of the questionnaires did indeed measure different but related aspects of self-conscience emotions, namely, the proneness to experiencing them and the way to cope with them. This finding suggest that diagnostically, a clinician should differentiate between the two. Interestingly, adaptive coping has no direct correlation with the different maladaptive coping strategies of shame (CoSS). This is in line with earlier work done by Schalkwijk, Stams, Dekker, et al. (2016), who found that adaptive and maladaptive coping related to different sets of psychological functioning. When developing the Compass of Shame, Nathanson (1992) acknowledged that subjects can use all four different maladaptive coping strategies in different situations and that they may even overlap to a certain extent. For interventions aiming at integrating the conscience, this might implicate that focusing on changing maladaptive coping strategies may be less effective than focusing on adaptive coping, as adaptive coping is related to empathic skills and moral development.

Interestingly, the two communities for self-conscience emotions are connected by two items representing externalizing: CoSS’s Attack Other and TOSCA-A’s Externalization. This indicates that externalizing versus internalizing is an important factor in the functioning of conscience. Externalizing locates the source of one’s emotions outside oneself and can thus legitimate potential offending (aggressive) behavior. Moreover, externalization is negatively associated with Emotional Concern, an aspect of affective empathy. This triad is clinically stating the obvious: the more the adolescent externalizes, the less is his or her capacity for affective empathy, and vice versa.

In the third community empathy and moral reasoning are associated, replicating the a-priori questionnaires IRI and MOM, supplemented with adaptive shame coping (CoSS). This is in line with Hoffman (2000) placing empathy at the core of morality. Furthermore, the fact that this community in particular differs markedly from the non-offenders network, makes this cluster of items a point of extra interest.

In this community, all empathy items were connected (see the upper line in the bottom community in Figure 2, read from left to right).

The negative correlation found between punishment and victim-orientated moral orientation in this cluster, fits Kohlberg’s theory of psychological development in different phases (Kohlberg, 1981). The orientation on punishment (the pre-conventional level) wanes the moment moral development leads to internalizing victim-orientation (the conventional level).

Now, let’s take a purely clinical perspective on the circle of the five items in the third community, imagining it to be a core aspect of a healthy maturing conscience (the community shown at the bottom of Figure 2), starting on the upper left and proceeding clockwise. The capacity for affective empathy (Empathic concern, EC) is positively associated with the capacity for cognitive empathy (Perspective taking, PT). Adolescents who have developed this capacity seemed to be able to cope with shame in an adaptive way (Adaptive coping, CA). That is, they owned their feelings of shame as a temporary emotion telling them something about their self-esteem. The negative correlation with the preconventional level of moral orientation (Punishment-based moral orientation, MP) shows that as they take responsibility for the emotions of others and own their self-conscious emotions they steer clear of behavior that would bring them in a position which might invoke punishment. Instead, they focus on the consequences their violation might have for the victim (Victim-based moral orientation, MV). They keep the victim in mind, as can be deduced from the positive correlation with affective empathy (Empathic concern, EC).

The Defiantly Integrated Conscience

We will now discuss the findings for the offending adolescents, assuming them to be a model for arrest in the maturation of the conscience. The network consisted of seven small communities, none of them connected, implying less cohesion between the items in general. Our a-priori theoretical framework, clustering the items into three specific domains was absent. In point of fact, the coherence among items of the original questionnaires (known to have high internal consistency reliability), was small in this group. These findings clinically indicate less psychological integration of the conscience in offending juveniles.

We described above that in the network of the non-offending adolescents, a strong and clinically meaningful community of empathy, moral development, and adaptive shame coping was found. The network of the offenders also showed such a community, but in a very barren form with a positive correlation between cognitive empathy (Perspective taking, PT) and adaptive shame coping (Adaptive, CA), as well as a negative correlation between adaptive shame coping and the pre-conventional level of moral orientation (Punishment-based moral orientation, MP). Again, this fits the clinical notion that owing one’s shame as a temporary emotion (adaptive shame coping) developmentally goes together with the ability to reason about someone else’s motives (Goffin et al., 2020). On this developmental level, punishment based moral orientation has been outgrown and relinquished. However, in the offender network, a positive correlation between punishment-based moral orientation (MP) and personal distress (PD) was found. Both items are known to correlate with an earlier developmental phase in moral orientation and in empathy, respectively (Hoffman, 2000; Schalkwijk, 2022). Personal distress is associated with altruism. In a paper on the difference between empathy and altruism, De Waal (2007) proposed that what is called altruism might in fact be the lifting of one’s own personal distress when seeing that another, be it animal or human, is in a perilous situation. The role of mirror neurons might account for this contagion by another’s distress. This finding might indicate that in offenders, this physical aspect of empathy is experienced not when the wellbeing of someone else is threatened, but when punishment is at bay. Our findings seem to support this theory of an earlier developmental level of empathy in offending adolescents. It has to be noted, however, that an alternative explanation for this correlation could be the current situation of the participants in the study. As you will recall, they were in the course of being examined for trial; and this could have made them prone both to identifying with a fictitious person receiving punishment and to reporting higher levels of personal distress.

As just mentioned, in the offender network the items didn’t cluster with the other items of the questionnaire, but instead, formed separate communities. The finding that concerning empathy, the items Fantasy and Emotional Concern are clustered into a separate community, might indicate that for adolescents with a less matured conscience, experiencing warm feelings toward others is preserved for fictional, distant others.

As far as proneness to experience self-conscious emotions is concerned, the two pride items and the other four TOSCA-A items are regrouped into two communities in a similar way as in the non-offending network. The maladaptive coping items (CoSS) were gathered into two communities for either externalizing or internalizing coping, indicating that these two coping styles exist side by side rather than integrated. The lack of correlation between items of one questionnaire also counted for the two levels of moral reasoning. As was just mentioned, the pre-conventional level of moral development (Punishment-based moral orientation, MP) did correlate with other items, but the item reflecting the conventional level of moral orientation (Victim-based moral orientation, MV) had no correlation with any other item. This further supports the theory of a lack of integration of the domains in the functioning of conscience in the offender group. Clinically, this finding suggests that offenders might not experience an ambivalence conflict when confronted with moral dilemmas.

The Difference Between the Normative and the Defiantly Integrated Conscience

Lastly, we will dwell briefly on the differences between the two groups from the viewpoint of the a-priori theoretical domains empathy, moral orientation, and self-conscious emotions.

When trying to explain the differences between the conscience of non-offenders and offenders, and determine the protecting and facilitating factors for criminal behavior, we found a slight indication that the domains empathy and moral development are of greater influence in its functioning than the domain of self-conscious emotions. Meta-analyses for predicting criminal behavior showed a large effect size for moral development (d = 0.76), a medium effect size for cognitive empathy (d = 0.43), and small effect sizes for guilt (r = −.28) and affective empathy (d = 0.19; Spruit et al. 2016; Stams et al. 2006; Van Langen et al., 2014). Furthermore, in a post-hoc regression analysis preformed on this dataset (Figure 4), using a gradient boosting machine (GBM), we found that the item Punishment Orientation, representing the pre-conventional level of moral reasoning, explained by far the highest amount of variance.

Figure 4.

Logistic regression using a gradient boosting machine (GBM) using the existing data (variables and co-variables), the GBM created a model to predict whether a participant would belong to the offender or the non-offenders group with a predictive value 90%. The relative influence of the different variables within this model are presented.

On the other hand, in our study, the normative maturing conscience was characterized by interaction between three communities: one for proneness to experience or defend against self-conscious emotions, another for maladaptive coping with shame, and the third encompassing a clinically meaningful association of empathy, moral orientation, and adaptive coping. The conscience of offenders on the other hand was characterized by a lack of interaction between the communities, which we interpreted as a lack of integration of the domains in the functioning of conscience in the offender group. Could it be that regarding offenders, the more the conscience is immature, that is, its component elements are less integrated, the bigger the impact of lower levels of moral orientation and (to a lesser extent) empathy are? This could explain the large and medium effect sizes for items within these domains in predicting criminal behavior, when compared with the small effect sizes of the aspects of self-conscience emotions as mentioned previously. Although outside the scope of our study, we might also extrapolate our findings to more neurotic problems in (non-offending) adolescents. If it be so that in a more mature conscience, the integration of self-conscience emotions takes effect and its influence within the conscience increases, could this account for the suffering of adolescents with a too strict and exacting conscience, reporting higher levels of shame and guilt?

Focusing on empathy, we saw an intriguing difference between the two groups. For non-offenders, all items were positively correlated within one domain, in line with what is known about different developmental phases of this faculty (Figure 2). For the offending adolescents, however, the empathy items were dispersed in two communities (Figure 3) with only one (unexpected) direct positive connection. Clinically this indicated that in the normative maturing conscience different aspects of empathy were integrated, while in the less maturing conscience such integration was not present. Furthermore, in the non-offending group, both the emotional and cognitive aspects of empathy were correlated to other items, indicating a central role within the conscience for these aspects of empathy. In the offenders network however, with its direct correlation with the preconventional level of moral orientation, we found a more prominent role for personal distress, the most basic level of empathy.

Focusing on moral development, we saw that the negative correlation between punishment and victim orientation for non-offending adolescents indicated that the first phase had been superimposed by the conventional phase. As we noted earlier, this is in line with Kohlberg’s notion of phases of moral development, where the later phase largely integrates and overwrites the former phase but that under stress regression to an earlier developmental phase could occur (Gibbs, 2019). In offending adolescents, it is clinically interesting to see that victim orientation stands alone and has no partial correlation with any other item. This indicates that, in offenders, when other aspects of the conscience are activated, for instance while transgressing, the other is not taken into account at all. It is important to note that this finding does not suggest that offenders never keep anybody in mind, only that the offending adolescent is more inclined to be self-oriented and feel personal distress when punishment threatens.

Focusing on self-conscious emotions, we saw that in the normative maturing conscience, almost all items were found in two connected but different communities: one encompassing the items related to the proneness to experience or defend against self-conscience emotions and one containing maladaptive forms of shame coping. As mentioned earlier, this indicates that proneness to an experience and coping with the experience reflect clinically different psychological mechanisms. Unexpectedly, the CoSS-item for adaptive shame coping was not associated with any item for maladaptive shame coping. As noted before (Schalkwijk, Stams, Dekker, et al., 2016), this suggest that in the CoSS, adaptive and maladaptive shame coping really are different constructs. Clinically, this indicates that the ability to cope adaptively with shame in one situation does not (directly) prevent us from responding less adaptively to shame in other situations. In the offending group there was less integration between the different items of maladaptive coping of shame.

Taken together, in the 2016 study, we focused on the lower developmental levels of the three domains of conscience functioning to differentiate between offending and non-offending adolescents. In the current exploratory study, we focused not per se on the differences between these two groups but on the characteristics within the groups. We operationalized maturation not by the height of item scores, but by the quality of connectivity between the scores. This approach does more justice to the common knowledge that there are many adolescents with lower levels of item scores that do not commit offenses. Adolescents with higher levels of item scores may even become offenders, due to a lacking integration of the separate domains.

Concerning treatment of offenders, this exploratory study seems to indicate that treatment might need to focus on integrating the domains of the conscience rather than focusing on further development of separate domains. As the functioning of conscience is emotion based, the treatment should focus on recognizing on owning one’s emotions. The more the emotions are owned, the more the use of self-serving cognitive distortions will disappear. After all, self-serving distortions became active when the emotions pertaining to identity were too threatening.

Limitations

There are some limitations to this study. The first concerns the limited number of participants and the relatively small number of offenders compared to the non-offending participants. The use of targeted fused ridge estimation of scaled partial correlation matrices enabled us to produce the two presented networks based on partial correlations with an absolute value ≥0.2 as the cut-off value. More participants would have increased the certainty that the networks are indeed an accurate representation of the structure of conscience. Also, the small number of participants did not allow us to study possible effects of maturation within the two groups. The second limitation lies in the assumption that detention on suspicion of offending behavior is an accurate outcome measure for a less mature conscience. Even though the direct connection between the two is generally accepted in literature, the link is in our view, at best partial. It needs little imagination to envisage a person with an insufficient conscience who will never be convicted of a crime. A possible selection bias in the group of offenders might have been that 39 adolescents were included that were selected for evaluation before trial on the basis of suspicion of psychological disturbances. We would be interested to see whether the use of a different outcome measure, for instance behavioral problems, would lead to a different interpretation and possibly better understanding of the conscience. A third limitation of our study is related to the data collection. Due to practical considerations we used self-report questionnaires to probe the different domains of conscience. As with all self-reports, this brings the disadvantage of subjectivity and socially desirability responses with it. Furthermore, the two groups were in different settings whilst completing the questionnaires. While the non-offending juveniles completed them collectively in their classrooms at school, the offending juveniles were asked to fill them out solitarily, while they were being psychologically evaluated before trial. Moreover, only offenders received money for participating in the study while all non-offenders participated voluntarily. Possibly the difference in partial correlations found in the two networks could be attributed to a lack of motivation among the offenders to answer the questionnaire accurately, leading them to tick off the form randomly.

To our knowledge, this is the first time this type of analysis has been used to investigate the connectivity of different psychological functions in relation to offending behavior. Our research supports the hypothesis of the conscience as an integrated psychological function and calls for a more central, prominent role for empathy and moral development within it. Furthermore, we suggest that the degree of integration and the superimposition of developmental phases should be a focus of interest for further research. This is especially so, if we think about treatment and prevention of criminal behavior. More specifically, last year we started a research project translating the results of this study at group level into an instrument for diagnosing the functioning of the conscience on an individual level. This instrument will contribute to the prevention of recidivism in as much as it may be used to choose more specific interventions for criminal behavior.

Footnotes

Declaration of Conflicting Interests

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

Funding

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

ORCID iD

Frans Schalkwijk

Notes

References

Bilgrau

A. E.

Peeters

C. F. W.

Eriksen

P. S.

Bøgsted

Van Wieringen

W. N.

(2015). Targeted fused fidge estimation of inverse covariance matrices from multiple high-dimensional data classes. Journal of Machine Learning Research, 21(26), 1–52.

Capinha

Rijo Matos

Pereira

(2021). The Compass of Shame Scale: Dimensionality and gender measurement invariance in a Portuguese sample. Journal of Personality Assessment, 103, 807–817. https://doi.org/10.1080/00223891.2020.1866587

Carveth

D. L.

(2013). The still small voice: Psychoanalytic reflections on guilt and conscience. Karnac Books.

Cuff

B. M. P.

Brown

S. J.

Taylor

Jowat

D. J.

(2016). Empathy: A review. Emotion Review, 8, 144–153.

Davis

M. H.

(1983). Measuring individual differences in empathy: Evidence for a multidimensional approach. Journal of Personality and Social Psychology, 44, 113–126.

De Leeuw

F. A.

Peeters

C. F. W.

Kester

M. I.

Harms

A. C.

Struys

E. A.

Hankemeier

van Vlijmen

H. W. T.

van der Lee

S. J.

van Duijn

C. M.

Scheltens

Demirkan

van de Wiel

M. A.

van der Flier

W. M.

Teinissen

C. E

. (2017). Blood-based metabolic signatures in Alzheimer’s disease. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring, 8, 196–207.

De Waal

F. B. M

. (2007). Putting the altruism back into altruism: The evolution of empathy. Annual Review of Psychology, 59, 279–300.

Duits

(2006). Kwaliteit onderzoek Pro Justitia van jongeren [Quality of pre-trial mental health evaluation of youth]. van Gorcum.

Eisenberg

Fabes

R. A.

(1998). Prosocial development. In Damon

(Ed.), Handbook of child psychology (pp. 701–778). John Wiley.

10.

Elison

Lennon

Pulos

(2006). Investigating the compass of shame: The Development of the Compass of Shame Scale. Social Behaviour and Personality, 34, 221–238.

11.

Elison

Pulos

Lennon

(2006). Shame-focused coping: An empirical study of the Compass of Shame Scale. Social Behaviour and Personality, 34, 161–168.

12.

Gibbs

J. C.

(2019). Moral development and reality. Beyond the theories of Kohlberg, Hoffman, and Haidt. Oxford University Press.

13.

Girvan

Newman

M. E. J.

(2002). Community structure in social and biological networks. Proceedings of the National Academy of Sciences of the USA, 99, 7821–7826.

14.

Goffin

K. C.

Kochanska

Yoon

J. E.

(2020). Children’s theory of mind as a mechanism linking parents’ mind-mindedness in infancy with children’s conscience. Journal of Experimental Child Psychology, 193, 104784.

15.

Hoffman

M. L.

(2000). Empathy and moral development: Implications for caring and justice. Cambridge University Press.

16.

Jolliffe

Farrington

D. J.

(2004). Empathy and offending: A systematic review and meta-analysis. Aggression and Violent Behavior, 9, 441–476.

17.

Kohlberg

(1981). The philosophy of moral development. Moral stages and the idea of justice. Harper & Row.

18.

Le Sage

. (2005). De gebrekkige gewetensontwikkeling in het Jeugdstrafrecht. Implicaties voor de toerekening en behandeling. [The inadequate conscience development in juvenile criminal law. Implications for accountability and treatment]. SWP Publishers.

19.

Liu

Lafferty

Wasserman

(2009). The nonparanormal: Semiparametric estimation of high dimensional undirected graphs. Journal of Machine Learning Research, 2009, 2295–2328.

20.

Luyten

Fontaine

J. R. J.

Corveleyn

(2002). Does the test of self-conscious affect (TOSCA) measure maladaptive aspects of guilt and adaptive aspects of shame? An empirical investigation. Personality and Individual Differences, 33, 1371–1387.

21.

Nathanson

D. L.

(1992). Shame and pride. Norton.

22.

Pulos

Elison

Lennon

(2004). The hierarchical structure of the Interpersonal Reactivity Index. Social Behavior and Personality, 32, 355–360.

23.

Schalkwijk

(2015). Conscience and self-conscious emotions in adolescence: An integrative approach. Routledge.

24.

Schalkwijk

(2018). A new conceptualization of the conscience. Frontiers in Psychology, 9, 1863.

25.

Schalkwijk

(2022). Morele ontwikkeling en jeugddelinquentie (Moral development and juvenile delinquency). Boom.

26.

Schalkwijk

Stams

G. J.

Dekker

Peen

Ellison

(2016). Measuring shame coping: The validation of the Compass of Shame Scale. Journal of Social Behavior and Personality, 44, 1775–1792.

27.

Schalkwijk

Stams

G. J.

Stegge

Dekker

Peen

(2016). The conscience as a regulatory function: Empathy, shame, guilt, pride and moral orientation in delinquent adolescents. International Journal of Offender Therapy and Comparative Criminology, 60, 675–693.

28.

Spruit

Schalkwijk

van Vugt

Stams

G. J.

(2016). The relation between self- conscious emotions and delinquency: A meta-analysis. Aggression and Violent Behavior, 28, 12–20.

29.

Stams

G. J.

Brugman

Dekovic

van Rosmalen

van der Laan

Gibbs

J. J.

(2006). The moral judgment of juvenile delinquents: A meta-analysis. Journal of Abnormal Child Psychology, 34, 697–713.

30.

Stams

G. J.

Dekovic

Brugman

Rutten

E. A.

Van den Wittenboer

G. J. H.

Tavecchio

L. W. C.

Van Schijndel

(2008). The relationship of punishment- and victim-based moral orientation to prosocial, externalizing, and norm trespassing behaviour in delinquent and comparison adolescents: A validation study of the moral orientation measure. Journal of Experimental Criminology, 4, 41–60.

31.

Tangney

J. P.

Dearing

R. J.

(2002). Shame and guilt. Guilford Press.

32.

Tangney

J. P.

Stuewig

Mashek

D. J.

(2007). Moral emotions and moral behaviour. Annual Review of Psychology, 58, 345–372.

33.

Tangney

J. P.

Wagner

P. E.

Gramzow

(1989). The test of self-conscious affect. George Mason University.

34.

Ten Berge

de Beurs

Duits

van der Hoorn

. (2009). Evaluatie Indicatiestelling PIJ-maatregel. [Evaluation indications for compulsory treatment of juvenile suspects]. NIFP/EUR. https://www.narcis.nl/publication/RecordID/oai:pure.eur.nl:publications%2Fd2f37e4b-5291-4463-bee2-3c19eb51dadf

35.

Vagos

Ribeiro da Silva

Brazão

Rijo

Elison

(2019). Psychometric properties of the Compass of Shame Scale: Testing for measurement invariance across community boys and boys in foster care and juvenile detentions facilities. Child & Youth Care Forum, 42(4), 257–388.

36.

Van Langen

M. A. M.

Wissink

I. B.

van Vugt

E. S.

van der Stouwe

Stams

G. J. J. M

. (2014). The relation between empathy and offending: A meta-analysis. Aggression and Violent Behavior, 19, 179–189.

37.

Van Wieringen

W. N.

Peeters

C. F. W

. (2016). Ridge estimation of inverse covariance matrices from high-dimensional data. Computational Statistics & Data Analysis, 103, 284–303.

38.

Verkade

Karsten

Koenraadt

Schalkwijk

(2020). Conscience as a regulatory function: An integrative theory put to the test. International Journal of Offender Therapy an Comparative Criminology, 64(4), 375–395.