Is Utilitarian Sacrifice Becoming More Morally Permissible? Ivar R

-- forthcoming in Cognition --

Is utilitarian sacrifice becoming more morally permissible? Ivar R. Hannikainen, Edouard Machery, & Fiery A. Cushman

A central tenet of contemporary moral psychology is that people typically reject active forms of utilitarian sacrifice. Yet, evidence for secularization and declining empathic concern in recent decades suggests the possibility of systematic change in this attitude. In the present study, we employ hypothetical dilemmas to investigate whether judgments of utilitarian sacrifice are becoming more permissive over time. In a cross-sectional design, age negatively predicted utilitarian moral judgment (Study 1). To examine whether this pattern reflected processes of maturation, we asked a panel to re-evaluate several moral dilemmas after an eight-year interval but observed no overall change (Study 2). In contrast, a more recent age-matched sample revealed greater endorsement of utilitarian sacrifice in a time-lag design (Study 3). Taken together, these results suggest that today’s younger cohorts increasingly endorse a utilitarian resolution of sacrificial moral dilemmas.

Keywords: moral dilemmas, cohort effect, aging, utilitarianism.

In 1967, the philosopher Philippa Foot large-scale cross-sectional studies (total N > published an essay on an obscure ethical 70,000) of the Interpersonal Reactivity principle, the doctrine of double effect. Her Index (Davis, 1980)—a multidimensional essay introduced now-famous cases like the measure of self-reported affect—younger trolley problem in order to crystalize the participants report lower scores on the competing mandates of deontology (never empathic concern subscale than either to use someone as a means to an end) middle-aged or older adults (O’Brien, versus utilitarianism (to promote the good Konrath, Grühn, & Hagen, 2013). of the many). Next, it argued that moral Meanwhile, a cross-temporal meta-analysis judgments regarding abortion and of 72 administrations of the IRI among euthanasia reflect this precise tension. United States college students revealed a In the decades since, public attitudes general decline in self-reported empathy toward euthanasia and especially abortion between 1979 and 2009 (Konrath, O’Brien have become substantially more permissive & Hsing, 2011). This generational trend (Inglehart, 1997; Norris & Inglehart, 2011). predicts a weakening prohibition on Might these developments reflect a deeper, utilitarian sacrifice, since dispositional more systematic shift in the relative balance empathy—as reported on the IRI of deontological versus utilitarian concerns? (Gleichgerrcht & Young, 2013; Patil & To answer this question, we assess evidence Silani, 2014)—is linked to deontological for historical change in the way that people reactions to the trolley problem. resolve the kinds of moral dilemmas posed Second, numerous Western cultures by Foot fifty years ago. have undergone processes of secularization Two lines of evidence motivate the (Norris & Inglehart, 2011), characterized by prediction that utilitarian moral values are religious disaffiliation and declines in on the rise. The first concerns cohort church attendance (Schwadel, 2010). In changes in trait empathy and their predicted turn, studies in moral psychology reveal consequences for moral psychology. In that religious believers are more likely to INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE oppose utilitarian sacrifice (Conway & Participants Gawronski, 2013; Piazza & Landy, 2013)— 1. 2007-08 wave. a pattern which may arise from their more Between October of 2007 and June of 2008, intuitive cognitive style (Shenhav, Rand & 4,134 volunteers (1662 women, 2472 men; Greene, 2012) and a corresponding age IQR: 20 – 37), took part in the present preference for the intrinsic moral evaluation study. Many participants were either of acts (Hannikainen, Miller & Cushman, college students (“Some college”: 1052 2017). Together these results provide [25%]) or graduates (“Bachelor’s degree”: 968 additional grounds to suspect that [23%]), and almost half of all participants utilitarian ethics may be proliferating, at were US nationals (1898 [46%]). Many least in secularizing societies. other participants came from Australia (119 Motivated by these existing lines of [3%]), Canada (162 [4%]), Germany (233 evidence, we examine the hypothesis that [6%]), Poland (284 [7%]), and the United utilitarian moral values are spreading over Kingdom (301 [7%]). Approximately half of time. Our methods are based on three the participants reported no religious complementary designs: In Study 1, we affiliation (“None”: 2051 [50%]), and many evaluate the relationship between age and others were of Christian denomination (“- moral judgment in a cross-sectional design, Catholic”: 512 [12%]; “-Orthodox”: 55 [1%]; i.e., comparing the moral judgment of “-Other”: 329 [8%]; “-Protestant”: 409 different age groups at a fixed point in time. [10%]). Next, in Study 2, we examine changes in 2. Longitudinal panel. moral judgment over the human life span in Between July 2016 and March 2017, we re- a fixed panel adopting a longitudinal design. contacted all 752 participants from the Finally, in Study 3, we employ a time-lag 2007-08 administration who voluntarily approach, comparing the moral judgment provided their e-mail address to take part in of similar age groups at different periods. future research: 166 (22%) e-mails bounced, Consistent effects in longitudinal 161 (21%) participants started the survey and cross-sectional designs (e.g., greater and, after excluding 21 incomplete condemnation of utilitarian sacrifice in participations, our re-test sample consisted older age) can be treated as indicative of a of 123 participants (73 men, 50 women; maturation effect—i.e., that processes of retest age IQR: 34 – 55) born between 1930 aging promote deontological views. If time- and 1994. Further demographic lag and cross-sectional analyses reveal information was retrieved from the first consistent findings, a predominant effect of phase: United States was the primary cohort may be assumed to be present (such nationality (53 [41%]), followed by as greater endorsement of utilitarian Germany (13 [10%]), United Kingdom (13 sacrifice among recent generations). [10%]), Poland (8 [6%]) and Canada (6 Finally, convergent effects in longitudinal [5%]). Many participants reported no and time-lag studies (e.g., more utilitarian religious affiliation (“None”: 80 [49%]), and judgment in recent test administrations) some were Christian (“-Catholic”: 15 [12%]; would suggest the prevailing influence of a or “-Protestant”: 12 [10%]). Many period effect—changes resulting from the participants held either a Master’s (36 passage of time that affect everyone alike. [28%]) or a Bachelor’s (31 [24%]) degree. General Methods Tests of attrition bias are reported in All reported studies were approved by the Supplementary Analysis 1. institutional review board at Harvard 3. 2015-17 wave. University.

2 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE

Between October of 2015 and March of revealed very good internal consistency in 2017, 9337 volunteers (4825 women, 4076 the present studies (Cronbach’s alpha ≥ .87) men, 130 other, 156 preferred not to and also test-retest reliability in Study 2 (r = specify; age IQR: 19 – 28) took part in the .67). present study. Most participants were either 2. Interpersonal Reactivity Index. high school graduates (“High school/GED”: Participants completed a widely-used 2018 [22%]), college students (“Some assessment of self-reported empathy (Davis, college”: 2481 [27%]) or graduates 1983). The IRI contains 28 items, organized (“Bachelor’s degree”: 1956 [21%]). Over half in four subscales: of all participants were US nationals (5314 a. perspective-taking, the tendency to [57%]). Many other participants came from evaluate situations from the point of Australia (292 [3%]), Canada (521 [6%]), view of others (e.g., “I try to look at Germany (214 [2%]), and the United everybody's side of a disagreement Kingdom (608 [7%]). More than half of the before I make a decision”); participants reported no religious affiliation b. fantasy, the capacity to transpose (“None”: 5397 [58%]), and many others oneself into the feelings and actions of were of Christian denomination (“- characters in fictional contexts (e.g., Catholic”: 976 [10%]; “-Orthodox”: 153 [2%]; “After seeing a play or movie, I have “-Other”: 790 [8%]; “-Protestant”: 707 [8%]). felt as though I were one of the Procedure characters”); Participants visited the Moral Sense Test c. empathic concern, the tendency to feel website (www.moralsensetest.com), either compassion and concern for others voluntarily (in Studies 1 and 3) or upon (e.g., “I often have tender, concerned receiving an e-mail request (Study 2). After feelings for people less fortunate than providing informed consent, participants me”); and completed at least the following three d. personal distress, own feelings of unease sections: and discomfort in reaction to the 1. Moral dilemmas. emotions of others (e.g., “Being in a Participants viewed a battery of thirteen, tense emotional situation scares me”). high-conflict personal dilemmas, previously 3. Demographic information. employed in numerous studies in moral Participants were asked to provide psychology (see Koenigs et al., 2007). Each information about: their gender; age (in hypothetical situation was narrated in the years); educational attainment (1: “Less than second person (placing the reader in the role high school” – 5 : “Graduate degree”); of actor), and presented a dilemma whether religious affiliation (“Buddhist”, “Christian – to personally sacrifice someone in order to Catholic”, “Christian – Orthodox”, “Christian save a larger number of lives. After each – Other”, “Christian – Protestant”, “Hindu”, dilemma, participants were asked to rate the “Jewish”, “Muslim”, “None”, “Other”); and permissibility of the utilitarian action on a religious self-identification (anchored at 1: seven-point scale, anchored at (1) “Not religious”, and 7: “Very religious”). “forbidden”, (4) “permissible”, and (7) Stimuli, data and scripts are available “obligatory”. We calculated a moral online at osf.io/ks3wz/. judgment index per participant, by Power analysis averaging permissibility ratings across all Given our large sample sizes in Studies 1 thirteen dilemmas, such that higher values and 3, our analyses were highly-powered to indicate more characteristically utilitarian detect small effects (r = .10, d = 0.20): i.e., α moral views. The moral judgment index < .001 (consistent with Benjamin et al.,

3 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE

2017), and 1 – β > .99, setting β/α ratio to judgments about the permissibility of 4:1. Our longitudinal study depended on a utilitarian sacrifice. If either maturation or more limited sample of 123 re-test cohort effects are present, we should observe participants. With α = .05, and 1 – β = .80, a correlation. In contrast, if only period our planned analysis (paired t-test) afforded effects are present, we should observe no us enough statistical power to detect effects differences in moral judgment by age. larger than or equal to Cohen’s d = 0.25. We also seek to replicate previously We adopt pairwise deletion reported relationships between empathic throughout this report: Each statistical concern and both age (O’Brien et al., 2013) analysis includes all participants for whom and moral judgment (Gleichgerrcht & the data are available, resulting in some Young, 2013). variation in sample sizes across analyses. Results Study 1: Cross-sectional age differences Summary statistics are displayed in Table 1 In Study 1, we examine the relationship below. between participants’ age and their

Table 1. 2007-08 wave: Summary statistics and correlation table. Mean SD (1) (2) (3) (4) (5) (6)

(1) Age 30.3 13.1 -- (2) Moral judgment 3.86 1.37 -.23 -- (3) Fantasy 3.43 0.83 -.14 -.00 # -- (4) Empathic concern 3.65 0.77 .18 -.29 .33 -- (5) Perspective-taking 3.54 0.76 .12 -.11 .21 .47 -- (6) Personal distress 2.59 0.79 -.14 -.08 .17 .09 -.13 -- (7) Religiosity 2.70 1.87 .08 -.25 .06 .20 .06 .02 # Note. # indicates non-significance. Otherwise p < .005.

Empathic concern and moral judgment. empathy (see Figure 1B). An F-test

Consistent with prior research, the indicated that the polynomial model (AIC2 condemnation of utilitarian sacrifice = 8770) provided better fit than the simple

(Cronbach’s α = .90) was associated with linear model (AIC1 = 8778), F(1, 3822) = greater self-reported empathy, r(3810) = -.29 10.7, p = .001. [-.32, -.26], p < .001. Controlling for the Age and moral judgment. remaining IRI subscales in a multiple linear Our primary analysis indicated that regression, the effect of empathic concern younger participants were more likely to remained significant, B = -0.58, t = -18.09, p endorse utilitarian sacrifice, r(3808) = -.23 [- 2 < .001, ηp = .087. .26, -.20], p < .001. Given that age revealed Empathic concern and age. a quadratic effect on empathic concern, we We also replicated past findings concerning tested a corresponding model with moral age differences: Age correlated positively judgment as the dependent measure: with empathic concern, r(3823) = .18 [.15, Indeed, age (centered) exhibited both linear, 2 .21], p < .001. In a polynomial model, age B = 0.03, t = -14.4, p < .001, ηp = .052, and (centered) demonstrated both linear, B = quadratic, B = -6.02 × 10-4, t = -5.77, p < 2 2 0.013, t = 10.29, p < .001, ηp = .031, and .001, ηp = .009, effects on moral judgment quadratic, B = -1.94 × 10-4, t = -3.27, p < (see Figure 1A). As with empathic concern, 2 .001, ηp = .003, effects on self-reported removing the quadratic term significantly

4 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE

reduced model fit (AIC2 = 12955, AIC1 = 12986), F(1, 3807) = 33.3, p < .001.

Figure 1. Age curve of (A) moral judgment and (B) empathic concern, with locally-weighted (loess) smoothing. The secondary x-axis displays corresponding birth years.

Study 2: Longitudinal analyses Overall, participants had aged an average of

Study 1 revealed differences in moral 8.4 years between phases (age2007-08 Mdn = judgment as a function of participants’ age: 32, Q1 = 26, Q3 = 47; age2016-17 = Mdn = 41,

Namely, younger participants reported Q1 = 34, Q3 = 55; CLES = 0.68), yielding a greater approval of utilitarian sacrifice than probability of .68 that randomly selecting did older participants. As noted, this result an individual from each phase would return may reflect differences between cohorts, the an older participant in the second phase. influence of maturation, or both. Maturation effects Longitudinal designs offer a IRI scores (FS α = .81; EC α = .81, PT α = window into maturation effects, by .80, PD α = .83) were highly correlated examining patterns of change within between phases (FS r = .80, EC r = .63, PT individuals. So, in Study 2, we re-contacted r = .66, PD r = .62, all ps < .001). No a panel of visitors to the Moral Sense Test significant changes in empathic concern, website and invited them to take part in an t(122) = -0.28, p = .78, d = 0.02, or identical retest, with at least an eight-year perspective-taking, t(122) = 0.99, p = .33, d interval between phases. = 0.09, were observed. However, we did If processes of aging magnify find a significant decrease in fantasy, t(122) attitudes of moral opposition toward = -3.66, p < .001, d = -0.33, and an increase utilitarian sacrifice, we should observe in personal distress, t(122) = 2.94, p = .004, greater moral condemnation (and perhaps d = 0.26 (see also Supplementary Table 1). also empathic concern) in the second phase Similarly, religiosity was highly of this study than in the first phase—a result correlated between phases, r(123) = .73 that would suggest that the age differences [.65, .81], p < .001. A paired t-test revealed in Study 1 are driven by maturation. a significant decline, (religiosity2007-08 = 2.37,

Results SD = 1.78; religiosity2016-17 = 2.00, SD = Manipulation check

5 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE

1.70), t(122) = -3.18, d = -0.29 [-0.46, - Critically, a paired t-test revealed no

0.11], p = .002. In Study 1, we saw a weak significant shift in moral judgment, (MJ2007- positive correlation between religiosity and 08 = 4.02, SD = 1.38; MJ2016-17 = 3.92, SD = age, suggesting that the present decline 1.18), t(122) = 1.11, d = -0.10 [-0.28, within individuals is attributable primarily 0.08], p > .250. From a Bayesian to the influence of period (and not of perspective, these results provide “positive” maturation; see Schwadel, 2010). or “substantial” support (BF01 = 5.71) for

Finally, moral judgment (MJ2016-17 α the absence of an aging effect, and the result = .87) also correlated strongly between was robust to different Cauchy prior widths phases, r(123) = .67 [.56, .76], p < .001. (maximum BF10 = 1).

Figure 2. Histogram and density plot of longitudinal change in moral judgment. A dashed line depicts the mean level of change.

Replicating Study 1, empathic than younger participants, but Study 2 concern predicted deontological judgment revealed a negligible and non-significant in both phases (2007-08: B = -0.69, t = - shift within participants over an eight-year 2 4.00, p < .001, ηp = .15; 2016-17: β = -0.50, interval. Furthermore, a Bayesian analysis 2 t = -3.04, p = .003; ηp = .08), controlling for lent substantial support for the absence of differences on other IRI subscales. Taken an aging effect. together, these results hint towards This pattern of results opens up the relatively stable cohort differences in possibility that cross-sectional age empathy and moral judgment, and provide differences in moral judgment reflect no support for the influence of aging changes in utilitarian proclivity across processes. However, we note that Study 2 generations. Thus, we should observe an cannot rule out the presence of a effect of cohort succession in a comparison maturation effect smaller than the between age-matched waves. minimum effect size we could confidently To test this prediction, in Study 3 detect. we adopt a two-point time-lag approach. A Study 3: Time-lag analyses new sample of online volunteers completed In Study 1, older participants tended to an identical study on empathy and morality make more deontological moral judgments between 2015 and 2017. We then apply an

6 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE exact matching (Iacus, King, Porro, & Katz, they antecede and complement statistical 2012) algorithm to create a 2015-17 inference in order to reduce dependence on subsample that strongly mirrors our 2007-08 parametric model assumptions and thereby sample on the primary pre-treatment improve estimates of treatment effects. covariates, thus eliminating differences in Our resulting matched sample age and other potentially confounding consisted of 3377 participants per wave demographic measures. In conjunction with (total N = 6754). The CEM algorithm the correlation reported in Study 1, a achieved perfect univariate balance on difference between waves would point educational attainment, religious affiliation, towards a cohort effect. and gender (all χ2s = 0, ps = 1), and age was

Results also adequately matched (age2007-08: M =

Imbalance correction 28.6, SD = 11.8; age2015-17: M = 28.7, SD = To evaluate whether our control (2007-08) 11.7), t(6752) = 0.36, p = .72, d = .01. and treatment (2015-17) samples differ in Descriptive statistics are reported in their composition, we entered every Supplementary Table 2. demographic predictor into a logistic Manipulation check regression model with wave (control vs. As a result of our matching procedure, treatment) as the dependent measure: participants in the second wave were born

Participants in the 2015-17 wave were an average of 8.3 years later (birth-year2007-08 overall younger, OR = 0.98 [0.97, 0.98], Mdn = 1983, Q1 = 1974, Q3 = 1988; birth- more likely to be women, OR = 1.40 [1.36, year 2015-17 Mdn = 1991, Q1 = 1983, Q3 = 1.57], and varied also in terms of 1996; CLES = .73), yielding a probability of educational attainment, χ2(4) = 257.70, .73 that a randomly-selected second wave nationality, χ2(24) = 939.83, and religious participant was born after a randomly- affiliation, χ2(9) = 47.25, ps < .001. selected first wave participant. Because visitors were not randomly Cohort effects assigned to one or the other wave, the Differences in interpersonal reactivity and imbalance in pre-treatment covariates may religiosity between waves were all reflect bias in selection. Thus, prior to negligible, albeit significant (EC d = -0.08, estimating our target effect, we sought to FS d = -0.12, PD d = -0.12, religiosity d = - balance the distribution of covariates 0.13, ps < .001, PT d = 0.05, p = .03). We between waves. To this end, numerous did not anticipate these results: Since age methods for pre-processing observational differences in self-reported empathy were data have been developed, whose general not related to processes of maturation, we aim is to compensate for non-random expected a larger difference between waves. assignment by either weighting or matching In the general discussion, we return to this observations as a function of their covariate issue. values. In our present case, given the large As predicted, a two-sample t-test sample size and substantial overlap between revealed a clear shift in moral judgment waves on the primary demographic (MJ2007-08 = 3.91, SD = 1.36; MJ2015-17 = covariates, we were able to seek exact 4.34, SD = 1.22), t(6752) = 13.6, d = 0.33 matches, using a coarsened exact matching [0.28, 0.37], p < .001, suggesting that recent algorithm (see the MatchIt R package cohorts in the treatment group were more introduced in Ho, Imai, King, & Stuart, likely to endorse utilitarian sacrifice than 2011; Iacus, King, Porro, & Katz, 2012). earlier cohorts in the control group. The Importantly, these techniques do not difference between waves was significant substitute for regression adjustment. Rather, across a range of regression models

7 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE incorporating different sets of covariates (see Table 2).

Table 2. Multiple regression models of moral judgment after coarsened exact matching of 2007-08 (control) and 2015-17 (treatment) waves. Model 1 Model 2 Model 3

Adj. r2 = .130 Adj. r2 = .161 Adj. r2 = .231 Wave 0.43 ** 0.43 ** 0.38 ** (0: 2007-08, 1: 2015-17) [0.37, 0.49] [0.37, 0.49] [0.33, 0.44] -0.03 ** -0.02 ** -0.02 ** Age [-0.03, -0.02] [-0.03, 0.02] [-0.02, -0.02] Gender -0.57 ** -0.53 ** -0.38 ** (0: male, 1: female) [-0.63, -0.51] [-0.59, -0.47] [-0.44, -0.32] Religious affiliation -0.06 0.26 * Buddhist [-0.28, 0.15] [0.04, 0.47] -0.45 ** -0.12 * -Catholic [-0.55, -0.37] [-0.22, 0.02] -0.65 ** -0.32 * -Orthodox [-0.90, -0.39] [-0.57, -0.07] -0.54 ** -0.13 * -Other [-0.65, -0.43] [-0.26, -0.01] -0.45 ** -0.06 -Protestant [-0.55, -0.35] [-0.18, 0.06] -0.46 * -0.06 Hindu [-0.77, -0.15] [-0.35, 0.25] -0.08 0.14 Jewish [-0.38, 0.22] [-0.15, 0.43] -0.53 ** -0.06 Muslim [-0.83, -0.22] [-0.37, 0.24] -0.11 0.14 * Other [-0.24, 0.01] [0.01, 0.27] Educational attainment -0.26 ** -0.20 ** Bachelor’s Degree [-0.37, -0.15] [-0.31, -0.10] -0.14 * -0.12 * High School Graduate [-0.26, -0.03] [-0.24, -0.01] -0.22 ** -0.18 ** Graduate Degree [-0.34, -0.11] [-0.29, -0.07] -0.20 ** -0.15 * Some College [-0.30, -0.10] [-0.25, -0.05] -0.36 ** Empathic concern [-0.40, -0.32] 0.10 ** Fantasy [0.06, 0.14] -0.02 Perspective-taking [-0.06, 0.02] -0.13 ** Personal distress [-0.17, -0.09] -0.11 ** Religiosity [-0.13, -0.09] *: p < .05, **: p < .001.

8 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE

Figure 3. Moral judgment by birth year and wave, with locally-weighted (loess) smoothing.

As seen in Figure 3, the cohort increase in Together, these results are most utilitarian judgment begins with individuals straightforwardly interpreted by appealing born approximately in the 1960s, and to a shift in the moral psychology of recent accelerates among birth cohorts after 1990. cohorts toward greater endorsement of Interestingly, among earlier generations— utilitarian sacrifice. those showing no cohort effect—an At the outset, we suspected that influence of period can be discerned (i.e., processes of secularization (Norris & more utilitarian views in the second wave Inglehart, 2011) and declining empathic relative to the first), perhaps because the concern (Konrath, O'Brien, & Hsing, 2011) factors shaping the formation of recent might be driving change, since the cohorts’ utilitarian views also somewhat opposition to welfare trade-offs has been influence the moral psychology of mature resoundingly attributed to religious and adults. affective prohibitions on intentional harm. General Discussion Although we replicated declines in self- In a cross-sectional design, younger reported empathy and religiosity, these participants were found to make more effects were smaller than changes in moral utilitarian judgments than older participants judgment. Below we summarize attempts to (Study 1). In order to understand whether interpret this unexpected result and our this relation was primarily the consequence recommendations for future research. of aging processes that reinforce First, there is some reason to deontological intuitions or a generational question the reliability of self-reported shift toward utilitarian judgment, we empathy. High trait empathy has a clear conducted longitudinal and time-lag positive connotation, so the IRI subscale studies. A longitudinal approach failed to may be contaminated by social desirability detect an effect of maturation (Study 2). (Watson & Morris, 1991). As a Meanwhile, a comparison between two age- consequence, the self-report of victim- matched samples separated by an eight-year oriented feelings does not always predict interval revealed increased utilitarian empathic behavior (Devlin, Zaki, Ong , & judgment in the recent wave (Study 3). Gruber, 2014; Marsh et al., 2014) and may

9 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE primarily serve to validate moral intuitions We must also take note of a few post hoc (Ditto & Liu, 2011). Instead, methodological limitations. A notorious complementary measures of affect, using complication, known as the identification neuroimaging (Lamm, Batson & Decety, problem, plagues the study of long-term 2007) and psychophysiological (Cushman, change. Because age = period – cohort, Gray, Gaffey & Mendes, 2012) methods, simultaneously estimating all three may provide a clearer window into the temporal effects on an outcome of interest affective processes shaping moral judgment. yields an infinite number of solutions. Second, from a dual-process Attempts to circumvent their collinearity perspective, utilitarian moral judgments are have long been viewed with suspicion often construed as the product of a (Glenn, 1976; Luo, 2013). Instead of cognitive suppression of affective reactions attempting to decompose these influences, (Conway & Gawronski, 2013; Cushman, our approach was to infer the most 2013; Greene, 2007; but see Bartels & plausible dominant mechanism from the Pizarro, 2011). So, growing endorsement of present pattern of results. utilitarian sacrifice may be due, not Independent evidence suggests that exclusively to a reduction in the intensity of most cultural change is driven by cohort affect, but also in its relative influence upon effects that withstand substantial variability moral judgment. Indeed, a diverse body of over periods (Vaisey & Lizardo, 2016; also neuroscientific (Greene et al., 2001; Davis, 1992). Vaisey and Lizardo (2016) Koenigs et al., 2007; Shenhav & Greene, demonstrate that historical change in moral 2014), physiological (Cushman et al., 2012; attitudes and values, in particular, stems Youssef et al., 2012) and behavioral from processes of cohort replacement more (Koenigs, Kruepke, Zeier, & Newman, so than from conversion due to 2012; Miller, Hannikainen, & Cushman, contemporaneous sociocultural factors, 2014; Patil, 2015) evidence now suggests which comports with our inference about that affective prohibitions of interpersonal increasing utilitarian judgment. Still, the harm are not triggered exclusively—perhaps reported findings could also emerge from not even primarily—by an emotional tie to the combination of a negative maturation the proximal victim, but also by an aversion effect and a positive period trend—an to the harmful act. Nonetheless, victim alternative we cannot decisively rule out. concern may be an important Discriminating between these candidate developmental precursor to harm aversion explanations requires future research (Cushman, 2013) and also play a decisive following different cohorts in new periods. role in motivating helping behavior (Decety In addition, most online volunteers & Yoder, 2016; Habashi, Graziano, & were nationals of North American and Hoover, 2016; Hu, Strang, & Weber, 2015). Western European countries. We therefore Third, scale ratings of IRI items remain agnostic as to whether cohort effects may suffer from recalibration over time emerge more generally across cultures. It (Golembiewski, Billingsley, & Yeager, may be, for instance, that religious and 1976). In contrast to our outcome measure, cultural traditions vary in their baseline anchored by terms like “obligation” and emphasis on deontological versus utilitarian “prohibition”, which may be expected to norms and that shifts in their relative maintain relatively stable meaning over prevalence coincide somewhat with widely- time, the self-assessment of dispositional studied processes of modernization and affect may be anchored by comparison to cultural change (see Inglehart, 1997). one’s peers, confounding the test of Finally, because our participants self- differences between administrations. selected, replicating these findings using

10 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE representative sampling methods is a the welfare-maximizing resolution of worthwhile task for future research. contemporary moral challenges. The evolution of moral norms holds Future research is clearly necessary inherent interest to philosophers and social to attest to these predicted consequences of scientists (Pinker, 2011; Singer, 1981). In shifting moral ideologies. To this end, the moral psychology, the received wisdom has present study underscores the value of been that the folk are fundamentally epidemiological tools for moral psychology deontologists—opposing harmful acts research, which enable us to detect long- carried out as a means toward the greater term fluctuation in basic moral values. good (Bartels, 2008; Cushman, Young, & Hauser, 2006; Royzman & Baron, 2002). References Our findings suggest, however, that this is Bartels, D. M. (2008). Principled moral not an immutable fact: Recent cohorts sentiment and the flexibility of moral (often referred to as Millennials) are judgment and decision making. Cognition, significantly more likely to support 108(2), 381-417. utilitarian sacrifice than their predecessors Bartels, D. M., & Pizarro, D. A. (2011). The mismeasure of morals: Antisocial personality (especially Baby Boomers, born before traits predict utilitarian responses to moral 1970)—a divide which may contribute to dilemmas. Cognition, 121(1), 154-161. the patent disagreement between younger Benjamin, D. J., Berger, J. O., Johannesson, and older adults in real-world debates about M., Nosek, B. A., Wagenmakers, E.--J., ethics and policy. Berk, R., Bollen, K. A., Brembs, B., Brown, A vibrant discussion among L., Camerer, C., Cesarini, D., Chambers, C. philosophers and cognitive scientists has D., Clyde, M., Cook, T. D., De Boeck, P., focused on distinguishing the virtues and Dienes, Z., Dreber, A., Easwaran, K., pitfalls of the human moral faculty (Bloom, Efferson, C., Fehr, E., Fidler, F., Field, A. 2017; Greene, 2014; Singer, 2005). On a P., Forster, M., George, E. I., Gonzalez, R., Goodman, S., Green, E., Green, D. P., pessimistic note, our results dovetail with Greenwald, A., Hadfield, J. D., Hedges, L. evidence about the socialization and V., Held, L., Ho, T.--H., Hoijtink, H., Jones, development of recent cohorts (e.g., J. H., Hruschka, D. J., Imai, K., Imbens, G., Shonkoff et al., 2012): Utilitarian judgment Ioannidis, J. P. A., Jeon, M., Kirchler, M., has been shown to correlate with Laibson, D., List, J., Little, R., Lupia, A., Machiavellian and psychopathic traits Machery, E., Maxwell, S. E., McCarthy, M., (Bartels & Pizarro, 2011), and also with the Moore, D., Morgan, S. L., Munafó, M., reduced capacity to distinguish felt Nakagawa, S., Nyhan, B., Parker, T. H., emotions (Patil & Silani, 2014). At the same Pericchi, L., Perugini, M., Rouder, J., time, leading theories credit highly- Rousseau, J., Savalei, V., Schönbrodt, F. D., Sellke, T., Sinclair, B., Tingley, D., Van acclaimed instances of moral progress to the Zandt, T., Vazire, S., Watts, D. J., Winship, exercise of rational scrutiny over prevailing C., Wolpert, R. L., Xie, Y., Young, C., moral norms (Greene, 2014; Singer, 2005), Zinman, J., & Johnson, V. E. (2017). and the persistence of parochialism and Redefine statistical significance. Nature prejudice to the unbridled command of Human Behavior. doi:10.1038/s41562-017- intuition (Bloom, 2017). From this 0189-z. perspective, greater disapproval of intuitive Bloom, P. (2017). Against empathy: The case for deontological principles among recent rational compassion. Random House. cohorts may stem from the documented rise Conway, P., & Gawronski, B. (2013). in cognitive abilities (i.e., the Flynn effect; Deontological and utilitarian inclinations in see Pietschnig & Voracek, 2015) and moral decision making: a process foreshadow an expanding commitment to

11 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE

dissociation approach. Journal of Personality Greene, J. D. (2014). Moral tribes: Emotion, and Social Psychology, 104(2), 216. reason and the gap between us and them. Cushman, F. (2013). Action, outcome, and Atlantic Books Ltd. value a dual-system framework for Greene, J. D., Sommerville, R. B., Nystrom, L. morality. Personality and Social Psychology E., Darley, J. M., & Cohen, J. D. (2001). An Review, 17(3), 273-292. fMRI investigation of emotional engagement Cushman, F., Gray, K., Gaffey, A., & Mendes, in moral judgment. Science, 293(5537), 2105- W. B. (2012). Simulating murder: the 2108. aversion to harmful action. Emotion, 12(1), 2. Habashi, M. M., Graziano, W. G., & Hoover, Cushman, F., Young, L., & Hauser, M. (2006). A. E. (2016). Searching for the prosocial The role of conscious reasoning and intuition personality: A Big Five approach to linking in moral judgment: Testing three principles personality and prosocial behavior. of harm. Psychological Science, 17(12), 1082- Personality and Social Psychology Bulletin, 1089. 42(9), 1177-1192. Davis, J. A. (1992). Changeable weather in a Hannikainen, I., Miller, R., & Cushman, F. cooling climate atop the liberal plateau: (2017). Act versus Impact: Conservatives Conversion and replacement in forty-two and liberals exhibit different structural General Social Survey items, 1972–1989. emphases in moral judgment. Ratio, advance Public Opinion Quarterly, 56(3), 261-306. online publication. doi:10.1111/rati.12162 Decety, J., & Yoder, K. J. (2016). Empathy and Ho, D. E., Imai, K., King, G., & Stuart, E. A. motivation for justice: Cognitive empathy (2007). Match It: Nonparametric and concern, but not emotional empathy, preprocessing for parametric causal predict sensitivity to injustice for others. inference. Journal of Statistical Software, 42(8), Social neuroscience, 11(1), 1-14. 1-28. Devlin, H. C., Zaki, J., Ong, D. C., & Gruber, Hu, Y., Strang, S., & Weber, B. (2015). Helping J. (2014). Not as good as you think? Trait or punishing strangers: neural correlates of positive emotion is associated with increased altruistic decisions as third-party and of its self-reported empathy but decreased relation to empathic concern. Frontiers in empathic performance. PloS one, 9(10), behavioral neuroscience, 9. e110470. Iacus, S. M., King, G., Porro, G., & Katz, J. N. Ditto, P. H., & Liu, B. (2011). Deontological (2012). Causal inference without balance dissonance and the consequentialist checking: Coarsened exact crutch. The social psychology of morality: matching. Political Analysis, 1-24. Exploring the causes of good and evil, 51-70. Inglehart, R. (1997). Modernization and Foot, P. (1967). The problem of abortion and postmodernization: Cultural, economic, and the doctrine of double effect. political change in 43 societies. Princeton Gleichgerrcht, E., & Young, L. (2013). Low University Press. levels of empathic concern predict utilitarian Koenigs, M., Young, L., Adolphs, R., Tranel, moral judgment. PloS one, 8(4), e60418. D., Cushman, F., Hauser, M., & Damasio, Glenn, N. D. (1976). Cohort analysts’ futile A. (2007). Damage to the prefrontal cortex quest: Statistical attempts to separate age, increases utilitarian moral judgements. period and cohort effects. American Nature 446:908–911. Sociological Review, 41(5), 900-904. Koenigs, M., Kruepke, M., Zeier, J., & Golembiewski, R. T., Billingsley, K., & Yeager, Newman, J. P. (2012). Utilitarian moral S. (1976). Measuring change and persistence judgment in psychopathy. Social Cognitive and in human affairs: Types of change generated Affective Neuroscience, 7(6), 708-714. by OD designs. The Journal of Applied Konrath, S. H., O'Brien, E. H., & Hsing, C. Behavioral Science, 12(2), 133-157. (2011). Changes in dispositional empathy in Greene, J. D. (2007). The secret joke of Kant’s American college students over time: A soul. Moral psychology: Historical and meta-analysis. Personality and Social contemporary readings, 359-372. Psychology Review, 15(2), 180-198.

12 INCREASING PERMISSIBILITY OF UTILITARIAN SACRIFICE

Lamm, C., Batson, C. D., & Decety, J. (2007). Royzman, E. B., & Baron, J. (2002). The The neural substrate of human empathy: preference for indirect harm. Social Justice effects of perspective-taking and cognitive Research, 15(2), 165-184. appraisal. Journal of Cognitive Neuroscience, Schwadel, P. (2010). Period and cohort effects 19(1), 42-58. on religious nonaffiliation and religious Luo, L. (2013). Assessing validity and disaffiliation: A research note. Journal for the application scope of the intrinsic estimator Scientific Study of Religion, 49(2), 311-319. approach to the age-period-cohort Shenhav, A., & Greene, J. D. (2014). Integrative problem. Demography, 50(6), 1945-1967. moral judgment: dissociating the roles of the Marsh, A. A., Stoycos, S. A., Brethel-Haurwitz, amygdala and ventromedial prefrontal K. M., Robinson, P., VanMeter, J. W., & cortex. Journal of Neuroscience, 34(13), 4741- Cardinale, E. M. (2014). Neural and 4749. cognitive characteristics of extraordinary Shenhav, A., Rand, D. G., & Greene, J. D. altruists. Proceedings of the National Academy of (2012). Divine intuition: cognitive style Sciences, 111(42), 15036-15041. influences belief in God. Journal of Miller, R., Hannikainen, I., & Cushman, F. Experimental Psychology: General, 141(3), 423. (2014). Bad actions or bad outcomes? Shonkoff, J. P., Garner, A. S., Siegel, B. S., Differentiating affective contributions to the Dobbins, M. I., Earls, M. F., McGuinn, L., moral condemnation of harm. ... & Committee on Early Childhood, Emotion, 14(3), 573. Adoption, and Dependent Care. (2012). The Norris, P., & Inglehart, R. (2011). Sacred and lifelong effects of early childhood adversity secular: Religion and politics worldwide. and toxic stress. Pediatrics, 129(1), e232-e246. Cambridge University Press. Singer, P. (1981). The expanding circle. Oxford: O’Brien, E., Konrath, S. H., Grühn, D., & Clarendon Press. Hagen, A. L. (2013). Empathic concern and Singer, P. (2005). Ethics and intuitions. The perspective taking: Linear and quadratic Journal of Ethics, 9(3-4), 331-352. effects of age across the adult life span. The Vaisey, S., & Lizardo, O. (2016). Cultural Journals of Gerontology Series B: Psychological fragmentation or acquired dispositions? A Sciences and Social Sciences, 68(2), 168-175. new approach to accounting for patterns of Patil, I. (2015). Trait psychopathy and utilitarian cultural change. Socius, 2, moral judgement: The mediating role of 2378023116669726. action aversion. Journal of Cognitive Watson, P. J., & Morris, R. J. (1991). Psychology, 27(3), 349-366. Narcissism, empathy and social desirability. Patil, I., & Silani, G. (2014). Reduced empathic Personality and Individual Differences, 12(6), concern leads to utilitarian moral judgments 575-579. in trait alexithymia. Frontiers in Psychology, 5, Youssef, F. F., Dookeeram, K., Basdeo, V., 501. Francis, E., Doman, M., Mamed, D., ... & Piazza, J., & Landy, J. F. (2013). “Lean not on Legall, G. (2012). Stress alters personal your own understanding”: Belief that moral decision making. morality is founded on divine authority and Psychoneuroendocrinology, 37(4), 491-498. non-utilitarian moral judgments. Judgment and Decision Making, 8(6), 639. Pietschnig, J., & Voracek, M. (2015). One century of global IQ gains: A formal meta- analysis of the Flynn effect (1909–2013). Perspectives on Psychological Science, 10(3), 282- 306. Pinker, S. (2011). The better angels of our nature: The decline of violence in history and its causes. Penguin UK.