Commentaries and Rejoinder on

Commentary on Žeželj and Jokić (2014)

Han Gong and Douglas L. Medin

Department of Psychology, Northwestern University, Evanston, IL, USA

We appreciate the replication project conducted by Žeželj and Jokić (2014). Our beliefs about the reliability and generalizability of the effects of mental construals and psychological distance on moral judgment have been updated considerably from their results, and we appreciate the opportunity to continue the discussion. In our commentary, we revisit the consistent findings of how construal levels induced by How/Why manipulation affect moral judgment. We then speculate about the puzzling patterns of results regarding the divergent role of temporal distance in determining moral evaluations. And we conclude by discussing the complexity of morality and propose that it may be important to incorporate cultural systems into the study of moral cognition.

A Clearer Picture of How Moral Judgment Is Influenced by the How/Why Manipulation

Over the past few years, there has been a growing interest in replication within psychology (e.g., Pashler & Wagenmakers, 2012; Spellman, 2013). Žeželj and Jokić (2014; Ž&J) have conducted a series of studies, aiming to replicate the Eyal, Liberman, and Trope (2008; EL&T) and Gong and Medin (2012; G&M), which examined the effects of psychological distance and construal levels on moral judgment and observed apparently conflicting findings. The original study in G&M was designed as an extension, as well as a conceptual replication, of EL&T; unexpectedly, however, G&M found that moral judgments became more extreme after participants answered a series of How questions rather than Why questions. The new studies by Žeželj and Jokić, included a direct replication of G&M (Study 1) that mirrored the results from G&M, as did the just reported study by Eyal, Liberman, and Trope (2014). Given that the same result has been consistently and readily obtained by three independent research groups in three different locations, the effects of how versus why primes on moral judgment appears to be fairly robust.

We emphasize robustness because, given the increasing recognition of the role of culture in cognition, it is far from obvious that a cross-national study should yield the same pattern of results. Indeed, Haidt, Koller, and Dias (1993) reported both cultural and social class differences in moral judgments for scenarios involving disgust (including those used in the studies under discussion). So there is reason to think that cultural norms and culturally based emotions will influence moral judgments. Therefore, the effects of construal level might well depend on the (cultural) salience of emotions or attention to mitigating factors (such as whether they are seen as mitigating versus “poor excuses”).

The Divergent Effects of Temporal Distance on Moral Judgment

According to Construal Level Theory (CLT), a major determinant of which construal level is activated is psychological distance. Specifically, distancing a target on any psychological dimension (i.e., time, space, social distance, and hypotheticality) leads to greater activation of high-level construals rather than low-level construals (Trope & Liberman, 2010). Thus, we should expect similar effects of psychological distance on moral judgment in light of the robust findings concerning construal levels.

Surprisingly, neither G&M nor Ž&J managed to replicate the effect observed by EL&T that moral judgment becomes more extreme with increased temporal distance. In particular, G&M obtained opposite results while Ž&J didn’t observe any reliable differences between near versus distant future. There are other relevant studies as well. Consistent with EL&T, Agerstrom and Bjorklund (2009) found that people made harsher evaluations of other individuals who failed to act altruistically when this was highly desirable in a number of different situations that were temporally distant as compared to temporally close. Lammers (2012, Experiments 2 and 4) manipulated temporal distance and temporal mindset and found that judgments of one’s own morality became less harsh with distance but that judgments of other actors became harsher with distance. We agree with Eyal et al. (2014) that we need to know more about the specific kinds of features that are activated in different moral scenarios and not just whether they are abstract or concrete.

It could be the case that the effects of temporal distance may not perfectly correspond to those of construal levels. Williams, Stein, and Galguera-Garcia (2012), for instance, have suggested that the emotional consequences of psychological distance for judgments are distinct from those of abstract thinking. Indeed, not a single study on temporal distance mentioned above included manipulation checks on whether the manipulation succeeded in inducing high-level construals. Given that research has documented cross-cultural differences in how people perceive the future (e.g., Boroditsky, 2001; Ji, Nisbett, & Su, 2001), it is possible that temporal distance manipulation results in different effects for participants recruited at these different sites (i.e., Israel, US, Sweden, the Netherlands, and Serbia), therefore exerting differing influences on moral judgments. As Stroebe and Strack (2014) argue, replicating a specific operationalization of a construct with a different population might not reflect the same construct that the same procedure operationalized in the original study, especially when the variables are culturally mediated. One important contribution of Ž&J is to offer data from a different culture that may help to interpret the differences between EL&T and G&M.

The Complexity of Morality

Most scenarios in all three papers are adopted from previous work by Haidt et al. (1993) and sample one small slice of human morality (e.g., Haidt & Joseph, 2004). The scope of research on mental representation and moral judgment should be enlarged to include both the less obvious situations and the long-lasting conflicts in everyday morality, which would allow a more comprehensive understanding the influence of distance and construals on moral judgment. Finally, much cultural research on morality suggests that people with various cultural backgrounds may have a different understanding and conceptualization of what morality entails (Sachdeva, Singh, & Medin, 2011). The current apparently inconsistent results of psychological distance may reflect cultural or socioeconomic variation in moral reasoning.

Thinking of Why a Transgression Occurred May Draw Attention to Extenuating Circumstances

Manipulating “Why” Versus “How” Mindsets

We conducted the research in the original paper within the framework of Construal Level Theory (CLT), suggesting that moral values tend to be abstract and general whereas contextual circumstances are often specific and concrete, and therefore the former should receive more weight relative to the latter as psychological distance increases. We adopted scenarios by Haidt and colleagues, which present immoral acts that are rendered harmless by mitigating contextual circumstances. We theorized that those acts would be judged harsher, as being more immoral, with increasing distance. Extending this logic, G&M predicted that a “why” mindset versus a “how” mindset (manipulated via priming) would produce harsher judgments. They, as well as Ž&J, found the reverse.

We would like to raise the possibility that in the present context, the “why” versus “how” priming achieved the opposite of what it was theorized to achieve. Specifically, it is possible that focusing on “why” aspects, relative to “how” aspects, prompted seeking explanations of the transgressions which in turn made the mitigating factors more salient relatively to the violated moral principle. Thus, although extensive findings and theorizing within CLT suggest that “why” aspects comprise a higher-level of construal than “how” aspects (e.g., Fujita, Trope, Liberman, & Levin-Sagi, 2006; Gilead, Liberman, & Maril, 2013; Liberman & Trope, 1998), this simply might not be true in the present context. For example, consider the incest scenario: “A friend of yours and his sister are home alone and decide to make love just once. The sister is regularly taking birth control pills and the brother uses a condom. They enjoy it immensely and decide not to do it again or tell anyone about it.” We suggest that it is quite likely that thinking why the siblings behaved the way they did would bring to mind answers like “because they made sure that they were alone at home” or “because they loved each other” rather than “because they were immoral and sinful.”

To test this possibility, we conducted a study (N = 83)¹ with undergraduate students from Ben-Gurion University. Participants read the three vignettes from our previous studies with minor changes (for complete method and results see https://osf.io/u63k4/files/). We instructed participants to either consider why the actor performed the behavior, how the actor performed it, or, in the control condition, gave no instructions. Participants then rated the extent to which they thought about the violated moral value, the circumstances that lead to the behavior, and the specific activities the actor did, each on a scale ranging from 1 (not at all) to 4 (a lot). Participants also rated the wrongness of the action on a scale ranging from −5 (very wrong) to 5 (completely ok) and answered seven items from each of the following questionnaires: the disgust scale (Haidt, McCauley, & Rozin, 1994) and the right-wing authoritarianism (RWA) questionnaire (Altemeyer, 1981). Replicating G&M and Ž&J, we found that participants judged the acts more harshly when considering how compared to why the acts were done, F(1, 80) = 3.19, p = .079, although this effect was marginally significant. Interestingly, and consistent with our prediction, participants who thought about why the action was done, compared to those who thought how it was done, reported thinking less about the violated moral value, F(1, 80) = 3.19, p = .043, and reported thinking more about the circumstances leading to the behavior, F(1, 80) = 6.84, p = .011 (for all means and SD’s see Table 1 ). As suspected, considering why the transgression occurred led to thoughts about (mitigating) specific factors rather than about the violated high-level value. On a more general level, these findings suggest that the applicability of even well-established and widely used manipulations needs to be examined anew in the context of each study.

Table 1. Means of considerations and wrongness ratings as a function of condition. Standard deviations are in parentheses

	Flag	Incest	Dog	Total
How (N = 28)
Consideration
Violated moral principle	3.39 (0.88)	3.82 (0.61)	3.61 (0.69)	3.61 (0.35)
Circumstances	2.54 (1.17)	1.89 (0.92)	2.54 (1.20)	2.32 (0.73)
Concrete actions	2.50 (1.14)	2.04 (1.04)	2.21 (1.17)	2.25 (0.84)
Wrongness of action	−2.50 (3.00)	−4.36 (2.26)	−3.75 (2.08)	−3.54 (1.43)
Why (N = 31)
Consideration
Violated moral principle	3.10 (0.87)	3.81 (0.60)	3.23 (0.96)	3.38 (0.48)
Circumstances	2.94 (1.09)	2.16 (1.10)	3.35 (0.92)	2.82 (0.75)
Concrete actions	2.52 (1.09)	2.19 (1.11)	2.55 (1.09)	2.42 (0.76)
Wrongness of action	−1.94 (3.10)	−3.74 (2.41)	−2.32 (2.77)	−2.67 (1.96)
Control (N = 24)
Consideration
Violated moral principle	3.67 (0.57)	3.71 (0.75)	3.50 (0.72)	3.63 (0.44)
Circumstances	2.29 (1.08)	1.58 (0.93)	3.13 (1.04)	2.33 (0.69)
Concrete actions	3.04 (0.96)	2.33 (1.05)	2.62 (1.01)	2.67 (0.67)
Wrongness of action	−2.46 (3.08)	−3.88 (2.61)	−2.08 (3.22)	−2.81 (2.16)

Table 1. Means of considerations and wrongness ratings as a function of condition. Standard deviations are in parentheses

View as image HTML

Manipulating Psychological Distance

It is challenging to consider why Ž&J failed to replicate our findings with temporal distance, while at the same time replicated our findings with social distance. It is possible that the transgressions in the scenarios violated group-based, conservative values (purity, authority) but not against values that emphasize individuals' rights (harm, fairness), and that social distance, more than temporal distance, enhances the salience of this type of values.

It would be interesting to examine, in future studies, which values people think are relevant for each scenario. Possibly, some participants would name two values that conflict with each other (purity vs. no-harm), while others would find it difficult to name a single value. In both cases, focusing on values would not necessarily promote harsher moral judgments, and correspondingly the effect of psychological distance becomes difficult to predict. Based on our experience with running the studies, however, we agree with Haidt (2001) and G&M that people tend to have an initial overall negative affective and/or moral reaction to the scenarios. We thus predict that if this initial global reaction is highlighted (e.g., by making an overall affective evaluation of the behaviors), distancing would yield harsher moral judgments.

Acknowledgments

We thank Ronnie Weinbach and Maayan Taube for their help with data collection. Designed research: E. T., N. L.; Performed research: E. T.; Analyzed data: E. T.; Wrote paper: E. T., N. L., Y. T. We report all data exclusions, manipulations, and measures, and how we determined our sample sizes. Complete method, materials, and data are available on the project page on the Open Science Framework: https://osf.io/4fb7d/. The authors declare no conflict-of-interest with the content of this article. The study reported in this article earned Open Data and Open Materials badges: https://osf.io.yiupc.

A Rejoinder to Comments by Eyal, Liberman, & Trope and Gong & Medin

Iris L. Žeželj and Biljana R. Jokić

Belgrade University, Serbia

Abstract. In two replication attempts, procedural priming of a high construal mindset (“why” condition) unexpectedly led to less harsh judgment of moral transgressions compared to priming of a low construal mindset (“how” condition). Eyal, Liberman, and Trope (Eyal, Liberman and Trope, 2014) proposed the mechanism that explained these findings, and obtained some supporting data. We expand on it by testing the mechanism on virtuous acts instead of on moral transgressions. We conclude by discussing the need to re-evaluate the procedures in the context of specific studies.

Keywords: construal level, moral reasoning, pro-social acts, mindset priming

Operating within the framework of Construal Level Theory (CLT; Eyal, Liberman, & Trope, 2008; Trope & Liberman, 2010), two independent research laboratories tried to replicate the established impact of procedural priming of construal level on moral judgment (Gong & Medin, 2012; Žeželj & Jokić, 2014). They both obtained the results contradicting the initial assumptions. Namely, the respondents primed with a high construal mindset (introduced via series of “why” questions on an unrelated material) were expected to focus on values and therefore judge the moral transgressions more harshly in comparison to the respondents primed with a low construal mindset (introduced via series of “how” questions). However, the priming procedure seemed to have worked in the opposite way: a series of “why” questions led to less harsh judgments compared to a series of “how” questions. In an attempt to explain these findings, Eyal, Liberman, and Trope (2014) conducted an additional experiment in which they directly prompted the respondents to think about either how or why a moral transgression was performed. They assumed that focusing on “why” paradoxically led participants to think more about the mitigating circumstances than about the violated moral rule. Their data supported the newer assumption. More generally, procedural priming had the consequences opposite to the ones usually obtained in the domains other than moral reasoning (e.g., Fujita, Trope, Liberman, & Levin-Sagi, 2006; Wakslak & Trope, 2009).

The Effects of Direct “How/Why” Manipulation on Virtuous Acts

It can be hypothesized that the effect would be even more pronounced in an evaluation of the pro-social acts, as the research shows that it is easier to revise a positive judgment in the presence of compromising circumstances than to revise a negative in the presence of mitigating circumstances (Reeder & Coovert, 1986; Ybarra, 2001). Therefore, the respondents in the “why” condition would judge the behavior more harshly than the respondents in the “how” condition, as they would focus on the burdening circumstances instead of on the virtuousness of the act.

To test this possibility, we conducted an online study in which students from the Belgrade University (N = 107) assessed the morality of three pro-social acts. We used the same vignettes as in EL&T (2008), and G&M (2012). As in EL&T (2014), we instructed the participants to either consider why the actor performed the behavior or how the actor performed it; we gave them no instructions in the control group. On a scale anchored with 1 (= not at all) to 4 (= a lot) the respondents assessed the extent they thought about (a) the moral value that was advanced by the act, (b) the circumstances that led to the act, and (c) the specific actions that constituted the behavior. Finally, they assessed the virtuousness of the act on a scale ranging from 1 (= not virtuous at all) to 7 (= very virtuous). Materials, data, and the pre-registered proposal are available on the project page on the Open Science Framework: https://osf.io/yiupc/.

As expected, we obtained the main effect of condition, F(2, 98) = 3.10, p = .05, η² = .061; planned contrast analysis revealed that the respondents in the “why” condition assessed the behavior as less virtuous (M = 4.47, SD = 1.05) than the respondents in the “how” condition (M = 4.87, SD = 0.81), the effect was marginally significant, t(1, 96) = 1.72, p = .089.

Although the participants in the “why” group reported to focus on moral value less, and on the circumstances more than the participants in the “how” group, the differences were not statistically significant (means and SDs are detailed in Table 1 ).

Table 1. Means of considerations and virtuousness ratings as a function of condition. Standard deviations are in parentheses

	Inheritance	Adoption	Campaign	Total
Note. 1Based on the power analysis performed by G*Power for repeated measures design with three groups, we aimed for 30 respondents per group. As the test was put online, we ended up with larger groups.
Consideration	How (N = 381)
Principles	2.92 (0.95)	2.79 (0.84)	2.53 (0.76)	2.76 (0.48)
Circumstances	3.30 (0.66)	3.26 (0.76)	3.05 (0.99)	3.22 (0.52)
Action	2.76 (1.01)	2.95 (1.04)	3.05 (0.90)	2.92 (0.81)
Moral judgment	5.27 (1.33)	5.21 (1.32)	4.05 (1.51)	4.87 (0.81)
Consideration	Why (N = 34)
Principles	2.88 (0.87)	2.52 (0.91)	2.16 (1.04)	2.51 (0.64)
Circumstances	3.37 (0.75)	3.27 (0.76)	3.39 (0.80)	3.34 (0.54)
Action	2.87 (1.13)	3.06 (0.83)	2.94 (0.89)	3.01 (0.73)
Moral judgment	5.19 (1.40)	5.00 (1.44)	3.29 (1.55)	4.47 (1.05)
Consideration	Control (N = 35)
Principles	2.77 (0.97)	3.03 (0.82)	2.42 (1.00)	2.73 (0.60)
Circumstances	3.43 (0.74)	3.17 (0.79)	2.97 (0.81)	3.17 (0.61)
Action	3.23 (0.73)	3.09 (0.85)	3.36 (0.65)	3.22 (0.54)
Moral judgment	5.49 (1.10)	5.43 (1.58)	4.27 (1.53)	5.06 (0.99)

Table 1. Means of considerations and virtuousness ratings as a function of condition. Standard deviations are in parentheses

View as image HTML

Distinctiveness of Moral Judgment

Taken together, the results of G&M (2012), Ž&J (2014), and EL&T (2014) show that the rules applying to general judgments cannot be a priori applied to moral judgments. In their comment to our replication, G&M specifically raised this point stating that moral judgments could be sensitive to cultural norms, leading to different persuasiveness of “mitigating” circumstances, as well as that high and low construal mindset could produce unexpected consequences in different moral scenarios.

In addition, we would also like to present the possibility that the “why” question, expected to lead respondents to focus on a goal of a certain behavior, in the domain of moral reasoning might actually lead them to focus on a cause of a behavior. To take this point further, as the scenarios included conflicting causes, both moral principles and circumstances could be viewed as causes of a behavior.

Finally, for a stricter test of the proposed mechanism, instead of directly asking the participants to focus on how or why a behavior was performed, it would be useful to start with how/why procedural priming on an unrelated material, follow by a set of control questions assessing whether the respondents focused on moral value or circumstances, and end with their judgment of moral acceptability of the behavior.

We appreciate the opportunity to engage in this discussion and view it as beneficial to the advancement of our field. One of the important outcomes of replication projects could be raising awareness about the fact that even routinely used, well established procedures need to be re-evaluated in the context of every planned experimental design.

Acknowledgments

Designed research: I. Ž., B. J.,. Performed research: B. J., I. Ž.; Analyzed data: B. J., I. Ž.; Wrote paper: I. Ž., B. J. We report all data exclusions, manipulations, and measures, and how we determined our sample sizes. Complete method, materials, and data are available on the project page on the Open Science Framework: https://osf.io/yiupc/. The authors declare no conflict-of-interest with the content of this article. The study reported in this article earned Open Data and Open Materials badges: https://osf.io./yiupc.