Empathy and Utilitarian Judgment in Incarcerated and Community Control Adolescents

Empathy and Delinquent Behaviors in Adolescence

Empathy is the ability to understand and share the emotional state and the situation of another individual (Cohen & Strayer, 1996). There is a broad consensus among researchers that empathy is a multidimensional construct (Davis, 1980), involving both the cognitive ability to take the perspective of others and the affective sharing of their emotional state (Lauterbach & Hosser, 2007). Empathy-related responses are believed to be the primary prerequisites for morality (Hoffman, 1987; Miller & Eisenberg, 1988), because they promote pro-social and altruistic behavior (Eisenberg, Miller, Shell, McNalley, & Shea, 1991; Hoffman, 1987) and inhibit antisocial and aggressive behavior (Gibbs, 2014; Hoffman, 2000). Thus, empathy is considered a protective factor because coming to know the internal state of others and vicariously experiencing their distress will encourage supportive behavior and deter harmful behavior (Eisenberg & Miller, 1987; Miller & Eisenberg, 1988). Correspondingly, the lack of empathy does not encourage the inhibition of behaviors that are harmful to others (Feshbach, 1975; Jolliffe & Farrington, 2006; Miller & Eisenberg, 1988), such as criminal behavior.

Criminological research into the relationship between empathy and delinquent behavior has produced inconsistent findings. Over the last decade some meta-analytic studies (Jolliffe & Farrington, 2004; van Langen et al., 2014) have shown that while most studies found an association between empathy and aggressive or antisocial behavior (van der Helm, Stams, van der Stel, van Langen, & van der Laan, 2012), others (Lardén, Melin, Holst, & Långström, 2006; Vachon, Lynam, & Johnson, 2014) did not find this relationship. Regarding this issue, the literature on psychopathy, in which a deficit in perspective taking is one of the most relevant features, has evidenced that such a deficiency in this age period could overlap with normative characteristics of adolescent development (Piquero et al., 2012; Seagrave & Grisso, 2002). Consequently, the utility of empathy in explaining juvenile offenses, although extensively researched, has been called into question (Seagrave & Grisso, 2002).

Moreover, a review of the literature indicates that the strength of the relationship between empathy and delinquent behavior has varied in function across a number of methodological factors, such as the type of instrument used and the nature of the constructs investigated (Jolliffe & Farrington, 2004). Cognitive empathy, for example, was found to be more strongly associated with delinquency than affective empathy (Lauterbach & Hosser, 2007; van Langen et al., 2014), especially in both younger people compared with adults and men/boys compared with women/girls (van Langen et al., 2014). These findings are further corroborated by most research on juvenile justice, suggesting that significant developmental changes occur from adolescence to young adulthood in the cognitive ability of social perspective taking (Cauffman & Steinberg, 2000)—an aspect of psychosocial maturity that is also likely to affect individuals’ antisocial decision-making (Cruise et al., 2008). Moreover, the extant literature investigating individual differences in the development of empathic abilities suggested the influence of atypical social contexts such as incarceration, that appear to temporarily impact the development of psychosocial maturity, with effects increasing with age (Dmitrieva, Monahan, Cauffman, & Steinberg, 2012).

Accordingly, the status of incarceration may also be relevant with respect to empathy. It has been suggested that a repressive prison climate is negatively associated with cognitive empathy in juvenile delinquents (van der Helm et al., 2012), and might account for findings of lower empathy scores in offenders (Jolliffe & Farrington, 2004; Jolliffe & Murray, 2012). Overall, the results of these studies show that an empathic deficit, particularly a cognitive empathic deficit, may be an important factor to consider in the explanation of juvenile delinquency. An empathic deficit, indeed, could account for inadequate processing of social information as well as moral attitudes (Crick & Dodge, 1994; Lemerise & Arsenio, 2000), factors that tend to be positively associated with criminal recidivism (Piquero, Bouffard, Piquero, & Craig, 2016; van der Put et al., 2012).

Utilitarian Versus Nonutilitarian Judgment, Empathy, and Delinquent Behaviors

The investigation about UJ in moral psychology is largely based on Greene’s dual process theory (Greene, 2008; Greene, Sommerville, Nystrom, Darley, & Cohen, 2001), a model that is attracting increasing attention in the last several years. Starting from the classical philosophical distinction between utilitarianism and deontology, UJs can be defined as judgments endorsing harmful actions that promote the greater good (Greene, 2007) and as judgments that privilege aggregate welfare over that of a small number of individuals (Moll, Zahn, de Oliveira-Souza, Krueger, & Grafman, 2005). UJs are based on an evaluation of cost-benefit ratio for a given course of action: from a utilitarian perspective, sacrificing the welfare of one person for that of more than one person is considered “right”; the central aim of utilitarianism is indeed to maximize benefits for the largest number of people. Deontological judgments, on the contrary, are based on an immediate feeling that an action is intrinsically “wrong,” irrespective of its consequences, for example, a judgment that killing one person is not acceptable even if it would save the lives of several people.

Greene (2008) posited that two competing processes, supported by different neurological systems, are implicated in making moral judgments: (a) a cognitive-driven process that produces UJs, and (b) an emotion-driven process that produces deontological judgments. Much research in this field is based on scenarios derived from the “trolley dilemma” and its variants, such as the “footbridge dilemma.” In these scenarios, participants are faced with the choice of whether or not to kill one person to save five others. The trolley dilemma is an example of an “impersonal” scenario: the choice is whether to flip a switch to divert a trolley so that one man will be killed rather than five. The footbridge dilemma is an example of a “personal” scenario: the choice is whether to push a heavy man down a footbridge to stop a trolley running over five people. In most individuals, the “personal” moral scenarios evoke a dominant negative emotional response that elicits moral disapproval, whereas the “impersonal” moral scenarios are not associated with a dominant emotional response, and thus elicit rational, utilitarian decisions (Greene, 2007; Greene et al., 2004; Greene et al., 2001). However, it is important to note that both processes are involved any time an individual has to make a judgment in a moral dilemma. This means that a utilitarian response could result from both a higher engagement in cognitive deliberation with a more pragmatic cost-benefit analysis or a lower sense of harm aversion due to reduced empathy for the victim (Patil & Silani, 2014). There is an ongoing debate about this topic, and empirical evidence has been produced supporting both the associations between decreased empathy and increased deliberation with utilitarian responses in moral dilemmas (see Duke & Bègue, 2015). Relevant to the purposes of the current study, there is much evidence showing that reduction in empathic response can increase the likelihood of UJs in moral dilemmas (for a review, see Patil, 2015). However, some studies found an increase in utilitarian responses associated with a higher affective arousal (Patil, Cogoni, Zangrando, Chittaro, & Silani, 2014). When the different components of empathy were investigated in detail, results became inconclusive. Some studies highlighted the role of affective empathy (see Patil & Silani, 2014), although with some inconsistencies. For instance, Sarlo, Lotto, Rumiati, and Palomba (2014) found that utilitarian responses were associated only with personal distress, while no significant relation was seen with empathic concern and perspective taking. However, Patil and Silani (2014) found that empathic concern was the only significant predictor of utilitarian choices. On the contrary, another study (Gleichgerrcht et al., 2013) found that the tendency to make UJs is associated with a selective impairment of cognitive empathy.

With respect to the relation between utilitarianism and delinquent behavior, research is still in the early stages. We do not know of any study directly evaluating this relation. Although there have been a number of investigations into the association between UJ and psychopathy (Patil, 2015), only in one case did this involve incarcerated persons (Koenigs, Kruepke, Zeier, & Newman, 2012). Although some of these studies found that individuals with psychopathic traits were more likely to make UJs than those without these features (Bartels & Pizarro, 2011; Landon & Delmas, 2012; Patil, 2015), others failed to find this association (Cima, Tonnaer, & Hauser, 2010; Glenn, Raine, & Schug, 2009; Pujol et al., 2011). In an in-depth analysis of the construct of psychopathy, Gao and Tang (2013) found that utilitarian responses were positively correlated only with certain dimensions of psychopathy.

The association of utilitarianism with psychopathy and decreased empathy seems, moreover, to suggest that UJs do not reflect a desire to maximize everyone’s welfare. Rather, they are more likely to be the result of a reduced emotional aversion to causing “up-close and personal” harm to others (Duke & Bègue, 2015). This result is also consistent with other studies that found utilitarianism not (or negatively) associated with a number of measures representing paradigmatic markers of concern for the greater good (Kahane, Everett, Earp, Farias, & Savulescu, 2015). It is, however, important to underscore that other studies have found utilitarian choices associated with indicators of concern for the greater good, such as moral identity (Conway & Gawronski, 2013).

This issue is the focus of an ongoing debate in the literature (see Bazerman & Greene, 2010; Bennis, Medin, & Bartels, 2010; Conway & Gawronski, 2013) that recently gave rise to a new alternative “two-dimensional” model of utilitarian thinking highlighting the distinct “positive” (impartial concern for the greater good) and “negative” (permissive attitude toward instrumental harm) components of utilitarian decision-making (Kahane et al., 2017).

The Present Study

The aims of this study were as follows: (a) to compare incarcerated and community control youth with regard to empathy and UJ, and (b) to test a theoretical model in which empathy is associated with delinquent behavior directly and indirectly, through mediation of the tendency to make UJs. The overarching objective was to disentangle the complex morality-related mechanisms underlying juvenile delinquency. Most studies on this topic have been carried out using self-report measures of delinquency, allowing the detection of involvement in deviant behaviors of different grades. Not all of these trends, however, result in a criminal career, which was the developmental outcome on which we wanted to focus. We therefore decided to include as our participants a group of incarcerated youth who may be considered representative of the population of serious offenders, because in Italy youth aged less than 18 are only imprisoned for very serious crimes. Corroboration for this comes from a Ministry of Justice report that in 2015 only 449 youths were incarcerated in Italy, compared with 52,164 adults (Ministry of Justice, 2015a, 2015b). However, looking at the community control sample, it is important to consider the long line of classic longitudinal studies in criminology (Farrington, Ttofi, Crago, & Coid, 2014; Loeber & Farrington, 2012; Moffitt et al., 2002) evidencing that delinquency is also widespread in the normal population. Thus, we developed our research design considering the levels of involvement in deviant behaviors among community youth.

With respect to the first aim of the present study, we expected, in line with the literature cited above, that incarcerated delinquent youth would differ from community control youth, involved or not in deviant behaviors, showing lower scores for empathy, in particular for cognitive empathy, and a higher tendency to make UJs. With regard to the second aim, we expected on the basis of the literature cited above that (a) empathy and UJ would show a negative and positive association, respectively, with delinquent behavior; (b) empathy would be negatively associated with UJ; and (c) UJ would mediate the relationship between empathy and delinquent behavior. The hypothesized theoretical model is shown in Figure 1. Moreover, two additional models including community control youth with problematic involvement in deviant behavior were tested to ascertain whether the morality constructs we considered were related to delinquency and not to incarceration status.

OPEN IN VIEWER

Figure 1:

Hypothesized Theoretical Model

Method

Preliminary Analyses

Before testing our hypotheses, we checked our data for outliers and assessed the distribution of variables. No outliers were found, and all values of skewness and kurtosis were between ±1, which is usually considered an acceptable range for normal distribution (George & Mallery, 2010). Moreover, to detect among community youth those participants highly involved in deviant behaviors, we calculated T scores on the Rule-Breaking subscale from the YSR, finding 53 participants scoring higher than 70 (the cut-off indicating a problematic involvement in deviant behaviors; Achenbach, 2001). Therefore, we distinguished a community control group without problematic involvement in deviant behaviors comprising 186 participants (M_age = 17.23 years, SD = 1.28), and a community group with problematic involvement in deviancy including 53 participants (M_age = 17.72 years, SD = 1.23). For the sake of clarity, hereafter we refer to these groups as “community control group” and “community deviant group,” respectively, while we refer to the group including youth in prison as “incarcerated group.”

Results

Comparison Between Incarcerated and Community Groups

A MANOVA was used to investigate group differences (community control group vs. community deviant group vs. incarcerated group) with respect to the study variables, age and SES. There was a significant effect of group, Wilks’s λ = .85; F(10, 562) = 4.80, p < .001; η² = .08. Univariate post hoc analysis (Table 1) showed that both community control group and incarcerated group were higher than community deviant group in EC, F(2, 285) = 10.99, p < .001; η² = .07. Moreover, the incarcerated group showed higher scores of UJ, F(2, 285) = 4.55, p = .021; η² = .03 and lower scores of PT, F(2, 285) = 4.50, p = .010; η² = .03 with respect to the community control group. No other significant differences emerged.

OPEN IN VIEWER

Table 1:

Comparison Between Incarcerated and Community Groups With or Without Problematic Involvement in Deviancy on All the Study Variables, Age, and SES

Measures	Community control group (n = 186)		Community deviant group (n = 53)		Incarcerated group (n = 67)		F (10, 562)	η2
	M	SD	M	SD	M	SD
Perspective taking	2.16 a	0.74	1.94	0.74	1.84 b	0.74	4.50*	.03
Empathic concern	2.75 a	0.65	2.27 b	0.73	2.82 a	0.81	10.99***	.07
UJ	3.67 a	1.44	3.95	1.51	4.36 b	1.57	4.55*	.03
Age	17.23	1.28	17.72	1.23	17.55	1.64	2.26	.02
SES	−0.09	0.84	0.16	1.16	−0.05	0.97	1.45	.01

Note. Univariate post hoc analysis. Different superscript letters indicate statistically significant differences between the groups (Bonferroni Post Hoc Test). UJ = utilitarian judgment; SES = socioeconomic status.

*

p < .05. ***p < .001.

Relationships Between Empathy, UJ, and Delinquency

We then tested our model of the association of delinquent behavior with empathy and UJ, using structural equation modeling implemented in MPLUS 7.2 (Muthén & Muthén, 1998-2012). More specifically, to ascertain whether the morality constructs we considered were related to delinquency and not to incarceration, we tested three models using as a dependent variable alternatively, (a) the probability of belonging to the community control (=0) versus incarcerated group (=1); (b) the probability of belonging to the community deviant (=0) versus incarcerated group (=1); and (c) the probability of belonging to the community control (=0) versus community deviant group (=1).

Missing data were handled using full-information maximum likelihood estimation, which has been shown to produce unbiased parameter estimates and standard errors (provided the missing values are at least missing at random). Our data set Little’s test for Missing Completely at Random (MCAR) data showed that missing values were completely at random, Little’s MCAR test; χ²(318) = 248.94, p = .99. This enabled us to use all observations in the data set.

In view of the sample size, complexity of the model and the number of items used to assess PT and EC, we used a partial disaggregation approach (Bentler & Wu, 1995; Dabholkar, Thorpe, & Rentz, 1996) for these variables in the measurement part of the model. This approach involves the creation of two or more composite variables for a construct by aggregating items that relate to it, using different strategies. In the present study, we built two parcels for each variable, balancing factor loadings. On the contrary, given to the measure of UJ consisted of only a small number of items, we used a total disaggregation approach for this variable, with the individual items as indicators of the latent variable.

In the structural part of the model, we posited that PT and EC were intercorrelated and directly associated with UJ and belonging to the group, and that UJ was directly associated with belonging to the group. Finally, we tested the mediating role of UJ. As in all three models our dependent variable was a dichotomous variable, the weighted least squares mean and variance–adjusted method for parameter estimation was used.

The first model—the one in which the dependent variable was the probability of belonging to the community control (=0) versus incarcerated (=1) group—also included age as a covariate, because of significant correlations with EC (r = .14 p = .026) and group (r = .17 p = .008). Results showed a good fit—χ²(21) = 18.70, p = .60; comparative fit index (CFI) = 1.00; Tucker–Lewis index (TLI) = 1.00; root mean square error approximation (RMSEA) = .00; 90% confidence interval (CI) = [.00, .047]; pclose = .96—and explained a reasonable percentage of variance in the probability of belonging to the incarcerated group (13.4%). The standardized path coefficients for the model are presented in Figure 2. The results showed that both EC and PT had no direct association with belonging to the group. Only PT had a direct, negative association with UJ, which in turn had a direct, positive association with the probability of belonging to the incarcerated group. Analyses of the indirect associations showed that UJ mediated the relationship (β = –.06, p = .035).

OPEN IN VIEWER

Figure 2:

Structural Equation Model Analysis of the Relations Between Empathy, Utilitarian Judgment (UJ), and Belonging to Group (Incarcerated vs. Community Control)

In the second model (see Supplemental Material, available with the online version of this article), the belonging to the community deviant (=0) versus incarcerated (=1) group was the dependent variable, including SES as a covariate, because of significant correlations with EC (r = –.21 p = .032) and UJ (r = –.20 p = .045). Results showed a good fit—χ²(21) = 25.29, p = .23; CFI = .94; TLI = .90; RMSEA = .04; 90% CI = [.00, .09]; pclose = .52—and explained a reasonable percentage of variance in the probability of belonging to the incarcerated group (28.6%). The results showed that the only significant effect was the positive, direct association between EC and the higher probability of belonging to the incarcerated group. No indirect association emerged.

In the third model (see Supplemental Material), the belonging to the community control (=0) versus community deviant (=1) group was the dependent variable including age as a covariate, because of significant correlations with group (r = .16 p = .014). Results showed a good fit—χ²(21) = 14.84, p = .83; CFI = 1.00; TLI = 1.00; RMSEA = .00; 90% CI = [.00, .033]; pclose = .99—and explained a reasonable percentage of variance in the probability of belonging to the community deviant group (27.2%). The results showed that the only significant effect was the negative, direct association between EC and the higher probability of belonging to the community deviant group. The association between PT and UJ was not significant. No indirect effect emerged.

Discussion

In this study, we investigated the association between two morality-related variables (empathy and UJ) and serious delinquent behavior, comparing incarcerated and community youth. Because, as we said above, the literature showed that individuals in the community may also be involved in delinquent behavior (Farrington et al., 2014; Loeber & Farrington, 2012; Moffitt et al., 2002), we distinguished among community youth, a community control group (including persons evaluating themselves as not involved in deviant behaviors), and a community deviant group (including persons evaluating themselves as highly involved in deviant behavior).

Our first aim was to determine whether the incarcerated group differed from the community control and community deviant groups with respect to empathy and UJ. Our second aim was to test an explanatory model of serious delinquent behavior in adolescence by evaluating the role of the two above-mentioned variables, representing two different dimensions of morality.

With respect to our first aim, we found that the incarcerated group did not differ from the community control group in affective empathy (EC) and cognitive empathy (PT), whereas the community deviant group exhibited lower levels of EC than the other two groups. Although this finding was not predicted, it is not an uncommon result, as research showed inconsistent results regarding the association between empathy and delinquent behavior. Studies found that empathy is not a stable personality disposition during adolescence, though it becomes increasingly consistent with age (Cauffman & Steinberg, 2000; Roberts & DelVecchio, 2000), given the developmental changes occurring during adolescence that can influence its consolidation. Moreover, when objective measures of delinquency are used, as in this study, it is not infrequent that the association between empathy and delinquency lacks significance (Lardén et al., 2006; Vachon et al., 2014). However, further research is needed to better understand why EC is lower in the community deviant group but not in the incarcerated group.

Moreover, we found significant differences between the incarcerated group and the community control group with respect to the tendency to make UJs. More specifically, incarcerated youth were more likely than community control participants to find it acceptable to kill one person to save five others, that is, they tend to be more likely to adopt a utilitarian perspective.

In fulfillment of our second aim, we were able to provide a more comprehensive account of the mechanisms underlying the relationships between morality-related variables and delinquency. We hypothesized that belonging to the group was directly associated with PT, EC, and UJ, and that PT and EC were directly associated with UJ. We also investigated the mediating role of UJ. Even if our primary aim was to evaluate the association between our morality-related variables and the probability of belonging to the incarcerated or community control group, we decided to take into account also the community deviant group, allowing us to ascertain whether the pattern of relations we found was related to a general deviant attitude (the construct measured by Rule-Breaking scale) or just to serious adolescent offending operationalized through incarceration. Thus, we tested three models using as a dependent variable, alternatively, the belonging to: incarcerated group versus community control group; incarcerated group versus community deviant group; and, finally, community control group versus community deviant group.

Our findings partially confirmed the hypothesized models. In the first model (community control vs. incarcerated group), only the higher tendency to make UJ was directly associated to a higher probability of belonging to the incarcerated group, while PT showed only an indirect association, mediated by UJ. More specifically, a lower PT was associated with a higher tendency to make UJ, which in turn was related to a higher probability of belonging to the incarcerated group. Finally, EC did not show any significant relation (except covariance with PT). The remaining two models did not show any significant effect of UJ. After summarizing the results, we try to give a more comprehensive explanation.

Overall, the role of empathy in affecting delinquency was weaker than expected, although in line with the most recent meta-analyses (Vachon et al., 2014; van Langen et al., 2014) and with evidence from the psychopathy literature. Research in this field has even questioned the predictive utility of this construct for juvenile offending, primarily because of the still developing personality and psychosocial changes typical of normative adolescent development (Piquero et al., 2012; Seagrave & Grisso, 2002). However, a more relevant role of cognitive empathy has emerged. This result is in accordance with studies that found cognitive empathy more strongly associated with delinquency than affective empathy (Lauterbach & Hosser, 2007; van Langen et al., 2014), particularly in groups similar to ours in age and gender (i.e., in younger people compared with adults), as well as for men/boys compared with women/girls (van Langen et al., 2014).

More intriguing, perhaps, is the role of UJ. Our results appear to be in line with previous studies showing a positive association between the tendency to make UJs and antisocial traits such as psychopathy (Bartels & Pizarro, 2011; Gao & Tang, 2013; Koenigs et al., 2012; Landon & Delmas, 2012), often related to delinquent behavior. However, our study is the first to use an objective measure of delinquency (incarceration). It is also important to keep in mind also other important issues related to the use of sacrificial moral dilemmas, discussed in an ongoing debate among researchers in this field.

Above all, we should consider a possible role of the specific scenarios we proposed in our moral dilemmas, in which the actor has to necessarily kill one individual to save the five others. The idea of killing someone, usually arousing aversive feelings in respondents—even if justified by a good purpose—could be less disturbing for delinquent youth, who have committed and have witnessed serious crimes. Because of these detrimental experiences, delinquency could have become more normative for incarcerated youth, so that the emotional response to the idea of killing someone in our moral dilemmas could have been weaker for these persons, making them more likely to give utilitarian responses. As this is the first study that compares incarcerated and community youth with respect to the tendency to make utilitarian choices using sacrificial moral dilemmas, further research and careful theoretical reflection are needed. According to Greene’s theoretical model, utilitarian choice results from reduced empathic activation. Consistent with this model, we found a significant negative association of PT and UJ. Unexpectedly, however, we did not find any relation between EC and UJ. Nonetheless, as we said in the Introduction, research investigating the relation between utilitarian choices and different components of empathy has reached inconclusive results, evidencing alternatively the role of affective and cognitive empathy.

Our findings are in line with Gleichgerrcht and colleagues (2013), who found more utilitarian choices in individuals with a selective impairment of cognitive empathy. Thus, with respect to the debate about whether utilitarian responses are more grounded on deliberation and cost-benefit analysis of the different outcomes or more a function of a decreased aversion to harming others (Duke & Bègue, 2015), our study seems to support this second hypothesis showing that a lack of PT encourages UJs. This seems also in accordance with Hoffman’s (2000) theoretical model, according to which one of the main functions of empathy is to emotionally load the individual’s moral principles. Individuals with low levels of empathy, therefore, will be more likely to show only “cold cognitions,” which could provide the basis for UJs, implying a relative lack of concern for proximal others’ welfare. This would be consistent with the interesting research field questioning whether UJs in response to the trolley dilemma are the result of genuinely moral concerns (see Bazerman & Greene, 2010; Bennis et al., 2010; Conway & Gawronski, 2013). Our findings, evidencing no association of affective empathy (i.e., EC) with the tendency to make UJs, give support to the research revealing that UJ does not actually reflect a real concern for others or a desire to maximize welfare (see, for example, Kahane et al., 2017; Kahane et al., 2015). Rather, it reflects a reduced concern over causing harm, due at least in part, according to our result, to lower levels of PT. It is also important to note that, although sacrificial dilemmas are by far the most dominant and reliable experimental paradigm to measure utilitarian decision-making, there was a weak relation between judgments in sacrificial dilemmas and a genuine utilitarian approach to ethics (Kahane et al., 2015).

Moreover, questions were raised regarding the adequacy of these specific dilemmas to distinguish between the two components of utilitarian decision-making (Kahane et al., 2017): the “positive,” impartial concern for the greater good and the “negative,” permissive attitude toward instrumental harm. Specifically, it was evidenced that sacrificial dilemmas enable measuring only the negative dimension of utilitarianism (Kahane et al., 2017). This would explain the association between utilitarian responses to sacrificial dilemmas and psychopathy and, more generally, aggressive and antisocial tendencies. As Kahane et al. (2015) also noted, “it seems rather implausible that individuals with antisocial traits or lower levels of empathy are especially morally committed to promoting the greater good, or harbor a special concern for humanity as a whole” (p. 194).

However, it is important to emphasize that the pattern of relations involving UJ was not significant in the second and third model, revealing further details regarding the role-played by this variable. A higher tendency to make UJs is associated with a higher probability to belong to the incarcerated group rather than the community control but not the community deviant group. This suggests that the variable is related to deviancy and that incarceration does not play any role. However, a higher tendency to make UJs is not significantly associated with a higher probability of belonging to the community deviant group rather than the community control group, suggesting that the seriousness of behaviors could play a role. As this is the first study investigating UJ in incarcerated and community adolescents, further research is needed to confirm our results and clarify this issue.

In conclusion, our study evidenced the role of UJ, providing initial evidence of the involvement of this variable in juvenile delinquency. Although more research is needed to confirm our results and investigate in more detail the patterns of relation that emerged, this seems to be a promising new perspective on juvenile delinquency. More generally, our results seem to suggest that the Greene’s dual process model, primarily directed to explain moral decision-making, could also be useful in explaining such behaviors.

Supplementary Material