A Single Counterexample Leads to Moral Belief Revision
Abstract
What kind of evidence will lead people to revise their moral beliefs? Moral beliefs are often strongly held convictions, and existing research has shown that morality is rooted in emotion and socialization rather than deliberative reasoning. In addition, more general issues—such as confirmation bias—further impede coherent belief revision. Here, we explored a unique means for inducing belief revision. In two experiments, participants considered a moral dilemma in which an overwhelming majority of people judged that it was inappropriate to take action to maximize utility. Their judgments contradicted a utilitarian principle they otherwise strongly endorsed. Exposure to this scenario led participants to revise their belief in the utilitarian principle, and this revision persisted over several hours. This method provides a new avenue for inducing belief revision.
1 Introduction
Conventional wisdom advises one to steer clear of religion and politics in polite conversation. One reason for avoiding these topics is that they can quickly turn to disagreements about what is moral. Such disagreements are often fruitless: People's moral beliefs are highly resistant to revision (e.g., Lord, Ross, & Lepper, 1979; also see Skitka, 2010) as they are thought to be rooted in emotion, motivation, and socialization (e.g., Greene, 2008; Haidt, 2001; Prinz, 2007; Rai & Fiske, 2011). Some of these roots go deep: There is evidence that infants have moral beliefs as early as 19 months of age (e.g., Sloane, Baillargeon, & Premack, 2012; also see Hamlin, Wynn, & Bloom, 2007) and have developing moral frameworks by age 5 (e.g., Kohlberg & Hersh, 1977; Rochat et al., 2009; Turiel, 2006). Moral beliefs may be particularly steadfast because, as Skitka (2010) observes, “to support alternatives to what is right, moral and good is to be absolutely wrong, immoral, if not evil.” Yet people can and often do change their moral beliefs, so it seems that research on moral conviction and moral development bring a puzzling psychological picture into focus. Moral beliefs are at once stubbornly intransigent and resistant to revision, and yet their change and development throughout people's lifespan is almost inevitable.
Nonetheless, assimilating new evidence into one's moral framework is not a trivial task. The vast literature on confirmation bias has shown that people have an overwhelming tendency to seek out evidence that confirms their beliefs, and to interpret any new evidence (even evidence objectively contrary to their beliefs) as confirmatory (e.g., Klayman, 1995; Nickerson, 1998). This resistance to change has also been observed in the moral domain; for instance, people do not temper their credences about the death penalty in the face of compelling contradictory evidence. Rather, they often view evidence that contradicts their belief as confirmatory, leading to even more polarized attitudes (e.g., Lord et al., 1979).
There are, of course, instances where people do revise even their deeply held beliefs. For a particularly notable example, consider former U.S. Federal Reserve chair Alan Greenspan, who changed many of his views on U.S. government regulation in light of the economic crises of 2008 (Time, October 23, 2008). The overarching point of confirmation bias is not that people never change their minds—just that it takes more evidential support than seems (optimally) warranted, such as a global economic collapse. Even here, Greenspan altered his beliefs about economic policy, but there is reason to think his attitudes about the moral or normative aspects of economic policy did not change. In describing his new position on increasing market regulation, Greenspan prefaced by saying, “As much as I would prefer it otherwise, in this financial environment I see no choice …” (emphasis ours). All of this research raises the question, what kind of evidence leads people to revise their moral beliefs?
We propose that the method of cases, a common practice in philosophy (e.g., Nagel, 2012), provides an avenue for addressing this question. In ethical theory, moral dilemmas (a kind of case) are constructed with the aim of advancing or countering moral theories. Consider the Transplant dilemma, a case in which one must decide whether to harvest a single person's organs to save the lives of five dying patients. Many philosophers think it is intuitively wrong to kill the single person, even to save many lives (e.g., Kamm, 2007; Thomson, 1976) and, therefore, see their judgments about this dilemma as inconsistent with utilitarianism.1 Likewise, we hypothesized that people would view their judgments about the Transplant dilemma as being inconsistent with some of their moral beliefs (for instance, the belief that you should always save the most lives). If so, then exposure to this dilemma may lead to moral belief revision.
2 Coherence, the method of cases, and belief revision
An important epistemic standard for evaluating a belief is how it coheres with one's other beliefs. Coherence may be particularly important for ethics, where there is little hope of independent and objective verification of beliefs through, say, perception (e.g., Campbell & Kumar, 2012). Is there any evidence that people value coherence when evaluating their own or others' beliefs? First, research on cognitive dissonance suggests that incoherence between beliefs and actions can lead to psychological discomfort—motivating people to revise their beliefs (see Cooper, 2007 for a review). Second, people's distaste for hypocrisy also suggests a certain level of concern for coherence. For example, consider the U.S. congressman from Florida who was recently charged with cocaine possession. The public seemed less concerned with his crime (a misdemeanor) than with his prior vote in favor of a drug-testing requirement for welfare recipients (New York Times, January 27, 2014). More than his drug use, it was his inconsistency that was repugnant.
Still, maintaining coherence in one's system of moral beliefs is not always easy. People hold a multitude of moral beliefs, some portion of which are recruited when making judgments in any particular moral situation (e.g., Nichols, Kumar, & Lopez, Draft, unpublished data; Nichols & Mallon, 2006). It is possible that incompatibilities between stored moral beliefs could go largely unnoticed, as only a handful of beliefs may ever be actively represented together. Moral dilemmas pit opposing attitudes against one another, making salient the tensions between the stored beliefs one recruits when considering these dilemmas (e.g., Nichols, 2004; Prinz, 2007). This leads us to predict that making judgments about moral dilemmas will lead to moral belief revision by encouraging people to be coherent.
Confirmation bias often occurs when people are presented with external evidence (e.g., statistics about the efficacy of the death penalty for lowering homicide rates) that contradicts a belief they hold. Faced with this sort of external conflict, participants tend to either reject the presented evidence, or distort it until it conforms to their beliefs. However, when considering a case like the Transplant dilemma, tension arises due to an internal conflict among participants' own beliefs. In light of distinction between internal and external conflict, we predict that considering a moral dilemma will lead people to revise their moral beliefs.
Consistent with this prediction, one recent study suggests that people revise their beliefs about the impermissibility of killing after considering moral dilemmas. Horne, Powell, and Spino (2013) asked participants to make a judgment about a moral dilemma and then immediately after to rate their agreement with different moral beliefs.2 These researchers found that considering the Footbridge dilemma—a moral dilemma that tends to elicit deontological judgments—led people to lower their credence in the utilitarian belief, “In the context of life or death situations, always take whatever means necessary to save the most lives.”3 They concluded that people revise their moral beliefs after exposure to a single dilemma.
It is striking that these researchers did not have to coax a belief change from their participants. Rather, participants seemed to recognize that their deontological judgments were inconsistent with a utilitarian belief they otherwise strongly agreed with, and updated their agreement with the belief accordingly.
3 The present research: Does the method of cases actually lead to moral belief revision?
Although suggestive, Horne et al.'s (2013) findings leave open two important questions concerning (a) the nature of the changes in people's moral beliefs and (b) the mechanisms behind these changes.
First, it remains unclear whether the changes that Horne et al. (2013) observed in their participants' belief ratings are truly reflective of changes in their participants' underlying beliefs. For example, their participants may have changed their belief reports to appear consistent (e.g., in response to demand characteristics), without actually revising their beliefs. Another possibility is that Horne and colleagues' findings reflect simple emotional priming, or some other transient contextual factor caused by considering an emotionally evocative moral dilemma. If the changes in participants' belief reports reflect genuine changes in their moral beliefs, then they should remain relatively stable over time. In contrast, if demand characteristics or emotional priming can explain participants' reports, then any shift in belief ratings should be fleeting.
A second question concerns the mechanisms underlying the hypothesized belief revision process. Does moral belief revision occur spontaneously, as a side-effect of simultaneously representing two conflicting beliefs, or does it require effortful deliberation, for example, in response to explicit questions about one's beliefs? This latter possibility suggests that people will not revise their moral beliefs unless they are immediately prompted to do so. On the other hand, if participants revise their beliefs without immediate prompting, then this would suggest the process occurs more spontaneously, further cutting against the possibility that people only change their belief reports to appear consistent. Regardless of whether belief revision in response to moral dilemmas requires prompting or is spontaneous, demonstrating that people change their beliefs in response to such dilemmas would provide an avenue for overcoming confirmation bias. Although one's beliefs may be comparatively impervious to external, contradictory evidence, perhaps they are less impervious to one another.
We conducted two experiments to investigate moral belief revision after exposure to a single dilemma. In Experiment 1, we examined how considering the Transplant dilemma affected participants' agreement with a utilitarian belief both immediately following exposure to the dilemma and after a 6-hour delay. We found that exposure to a dilemma-induced changes in people's moral beliefs that remained stable over a delay, suggesting that the method of cases provides an avenue for approaching moral belief change. Then, in Experiment 2, we examined whether participants would spontaneously revise their beliefs in response to the Transplant dilemma by asking participants to rate their agreement with moral beliefs only after a delay.
4 General methods for Experiments 1 and 2
4.1 Belief selection
Much of the research in moral psychology has focused on utilitarian moral judgment (e.g., Bartels & Pizarro, 2011; Greene, Nystrom, Engell, Darley, & Cohen, 2004; Greene, Sommerville, Nystrom, Darley, & Cohen, 2001). We examined one operationalization of utilitarianism previously used in the literature: “In the context of life or death situations, always take whatever means necessary to save the most lives.”4 Existing research has found that the majority of people agree with this moral principle, and that their agreement ratings tend to be quite strong (Horne et al., 2013).
Although we assume that people's reports of their beliefs are reliable indicators of those beliefs, recent data also suggest that utilitarian principles guide people's judgments about moral dilemmas. Indeed, researchers have argued that people's judgments about these moral dilemmas do not owe purely to the idiosyncrasies of each case but are instead influenced by general moral rules, some of which are utilitarian in nature (e.g., Horne et al., 2013; Lombrozo, 2009; Nichols & Mallon, 2006; Nichols et al., unpublished data; Royzman, Goodwin, & Leeman, 2011). For instance, Lombrozo (2009) found that participants' agreement with utilitarian principles predicted their judgments about moral dilemmas. In line with this finding, several studies have shown that people tend to make utilitarian moral judgments—choosing to sacrifice the life of one person to save many—across a number of different moral dilemmas (e.g., Côté, Piff, & Willer, 2013; Cushman, Young, & Hauser, 2006; Greene et al., 2001, 2004; Hauser, 2006; Moore, Clark, & Kane, 2008). So it does not appear that participants naively assent to utilitarian principles without appreciating at least some of their negative implications. Of course, this is not to say that people wholeheartedly endorse the doctrine of utilitarianism, as some philosophers do; it is quite likely that they also hold deontological beliefs (e.g., Nichols et al., unpublished data).
4.2 Dilemma selection
Prior research suggests that people revise their moral beliefs after considering the Footbridge dilemma (Horne et al., 2013)—a moral dilemma that tends to elicit deontological moral judgments (e.g., Greene et al., 2001, 2004). However, as many people in our target population (Amazon Mechanical Turk workers) have been exposed to this dilemma in prior studies (e.g., Bauman, McGraw, Bartels, & Warren, 2014), we chose to use the Transplant dilemma, which also elicits high rates of deontological judgments (e.g., Greene, Morelli, Lowenberg, Nystrom, & Cohen, 2008), but it has been less commonly examined.
4.3 Power analysis
An a priori power analysis was conducted based on an expected effect size estimated from the analyses reported by Horne et al. (2013). A good point of comparison for the current experiments is provided by the contrast between their participants' utilitarian agreement ratings after reading the Footbridge and control dilemmas, in which they observed a large effect (d = 1.10; Cohen, 1988). A power analysis conducted in G*power (Faul, Erdfelder, Buchner, & Lang, 2009) revealed that at least 23 participants would be required in each experimental group to achieve 95% power to detect effects of this size (α = .05). Because it was difficult to predict dropout rates in our studies, we often exceeded the sample size required for 95% power.
5 Experiment 1
In Experiment 1 we sought to examine whether changes in participants' moral beliefs would remain stable over time. We hypothesized that people would revise their moral beliefs after considering the Transplant dilemma. As we have discussed, people almost universally agree that it is inappropriate to harvest a single patient's organs to save others, despite the fact that this action “maximizes the good” (Greene et al., 2008). Likewise, people also strongly endorse the utilitarian moral principle “In the context of life or death situations, always take whatever means necessary to save the most lives” (Horne et al., 2013). We predicted that participants in our study would revise their agreement with this principle after making a judgment about the Transplant dilemma, because they would view their judgments about the Transplant dilemma as inconsistent with this principle. If our participants genuinely change their agreement with this utilitarian belief, then their revised agreement ratings should remain stable even after a delay. In contrast, if participants' revisions are no longer present after a delay, then this would suggest that their reports changed due to demand characteristics, emotional priming, or other transient contextual factors. The first outcome would provide substantial evidence that moral dilemmas, and the method of cases more generally, can lead to belief revision. The second outcome may undermine this proposal.
5.1 Method
5.1.1 Participants
This experiment was conducted online with 195 participants (49% female; Mage = 34.9 years) recruited via the Mechanical Turk Work Distribution website between the times of 10:00 a.m. and 11:30 a.m. Central Standard Time. These participants were then encouraged to return later that same day between 5:00 p.m. and 6:30 p.m. to complete part two of the experiment. Of the 195 participants who initially completed part one, 67 returned and completed part two. Total compensation for completing both parts of the experiment was $1.50.
5.1.2 Design, materials, and procedure
This study utilized a 2 × 2 (vignette × delay) factorial design. We manipulated the moral vignette that participants read immediately before reporting on their moral beliefs: Participants were assigned to either the Control or the Transplant condition (vignette; between-subjects). Participants were also asked to rate their agreement with the utilitarian moral belief both immediately following their moral judgments about the Transplant dilemma and after a delay of approximately 6 h (delay; within-subjects). Fig. 1 provides a diagram summarizing the experimental procedures.
Part one of the study began with a judgment task. Participants were presented with a series of three moral and three non-moral vignettes and were asked to make judgments about these situations. In the Transplant condition, the judgment task began with five distractor vignettes and ended with a judgment about the Transplant case. In the Control condition, the judgment task ended with a non-moral control vignette, and the transplant case was replaced with a different moral vignette (a vignette about incest) to maintain the same number of moral and non-moral distractors. Participants made their moral judgments on a 6-point Likert scale with endpoints labeled “Completely Inappropriate” and “Completely Appropriate.”
After the judgment task, participants completed a belief task: Participants rated their agreement with six statements expressing moral and non-moral beliefs. They made these ratings using a sliding scale, the endpoints of which were labeled “Completely Disagree” and “Completely Agree.” To make it difficult for participants to remember their responses to these belief statements in the second part of the experiment, there were no hashmarks or numerical indicators of their response. The only visual feedback provided by the slider was the position of the marker relative to the endpoints. The position of the marker was coded by the computer into a response between 0 and 100. Our primary-dependent measure was participants' ratings of the utilitarian belief statement “In the context of life or death situations, you should always take whatever means necessary to save the most lives.” This statement was presented at the beginning of the belief task, immediately following the moral judgment task. Five other belief statements, two moral and three non-moral, were presented after the utilitarian belief statement. An example of a non-moral belief statement was, “I laugh out loud when someone tells me a joke that I think is funny,” and a moral belief statement was, “Incest is always morally wrong.” In addition, half of these belief statements corresponded to the stories that participants made judgments about earlier in the experiment. The moral and non-moral distractors and belief statements were intended to conceal the aim of the experiment, and further reduce the likelihood of demand characteristics. After rating the six belief statements, participants completed a series of comprehension questions to ensure they paid attention to the stories that they read and then were asked to participate in part two of the experiment.
Part two was identical for participants assigned to either condition. Participants first made judgments about four unrelated distractor vignettes (two novel, two old; two moral, two non-moral), none of which were the Transplant dilemma. After making judgments about these vignettes, participants again rated their agreement with five distractor belief statements and the utilitarian belief (our dependent measure). Three of the belief statements were originally presented in part one of the experiment and three were novel.
5.2 Results and discussion
Participants' agreement ratings for the utilitarian belief statement are shown in Fig. 2. Of the 195 participants who initially completed part one, 67 returned and completed part two. To rule out the possibility of self-selection effects, we compared the immediate utilitarian belief ratings of those participants who returned for part two and those who dropped out. Results of two independent samples t-tests revealed no reliable differences between these two groups in either the Control condition (t(99) < 1, p = .40) or the Transplant condition (t(94) = 1.48, p = .14).
We computed a two-way repeated measures anova to determine how participants' endorsements of the utilitarian belief were affected by reading the Transplant dilemma and by the delay. Participants in the Transplant condition gave lower ratings to the utilitarian belief than those in the Control condition, as indicated by a reliable main effect of Vignette (F(1, 134) = 11.32, p < .001). Moreover, and consistent with our prediction that people's beliefs would remain stable over time, participants' ratings were not affected by an approximately 6 h delay (F(1, 134) < 1, ns). Moreover, there was no reliable interaction between Vignette and Delay (F(1, 134) < 1, ns). Subsequent independent t-tests revealed that participants in the Transplant condition gave reliably lower agreement ratings for the utilitarian belief statement both immediately following the Transplant dilemma in part one, (t(67) = 3.26, p < .01, d = 0.79) and after a delay in part two (t(67) = 2.43, p < .05, d = 0.59). Both these effects are conventionally considered medium to large effects (Cohen, 1988). Finally, a paired sample t-test found no reliable difference in belief ratings in the Transplant condition between parts one and two of the experiment, t(38) = 0.92, p = .36.
5.2.1 Discussion
Consistent with our predictions, the results of Experiment 1 demonstrate that people revise their beliefs after exposure to a moral dilemma and that this revision persists for at least 6 h after the dilemma is considered. The stability of participants' revised beliefs suggests that this effect is not the product of emotional priming or other transient contextual factors. Rather, the results suggest that people consider their judgments about the Transplant dilemma as inconsistent with a utilitarian belief about sacrificing others to save lives, and revise their agreement with this utilitarian belief accordingly.
6 Experiment 2
Experiment 1 suggests that the method of cases may provide a means for inducing moral belief revision. Still, the mechanisms underlying moral belief revision remain unclear. One possibility is that people will only revise their beliefs when they are asked about them immediately after considering a dilemma. Experiment 2 examined whether participants would spontaneously revise their beliefs without prompting. This experiment was identical to Experiment 1, but with two additional conditions. In these new conditions, participants only rated their agreement with the utilitarian belief after the delay. If participants spontaneously revise their agreement with the utilitarian belief after considering the Transplant dilemma, then we should be able to observe the effects of this revision even when participants are not immediately prompted to rate their agreement with this belief.
6.1 Methods
6.1.1 Participants
This experiment was conducted online with 373 participants (52% female; Mage = 36 years) recruited via the Mechanical Turk Work Distribution website. These participants were then encouraged to return later that same day to complete part two of the experiment. Of the 373 participants who initially completed part one, 120 returned and completed part two. Three participants were removed for completing the study more than once, leaving a sample of 117 participants. Total compensation for completing both parts of the experiment was $1.50.
6.1.2 Materials and procedure
The stimuli and procedure of Experiment 2 were identical to those of Experiment 1, except for the inclusion of two additional conditions. In the Transplant Delay-Only and Control Delay-Only conditions, participants rated their agreement with the utilitarian belief only after the delay. In all other respects these conditions were identical to the Transplant and Control conditions in Experiment 1. Fig. 3 provides a summary diagram of the procedures in Experiment 2.
6.2 Results and discussion
The results of Experiment 2 are shown in Fig. 4. First, to rule out the possibility of self-selection effects, we again compared the immediate utilitarian belief ratings of those participants who returned for part two and those who dropped out. The results of two independent samples t-tests revealed no reliable differences in participants' immediate utilitarian belief ratings between these groups. This held for both the Control condition, t(96) < 1.66, p = .10, and the Transplant condition, t(94) = 1.69, p = .09.
The distribution of data in Experiment 2 was non-normal, rendering parametric statistics inappropriate for any subsequent hypothesis testing. As parametric effect sizes were also inappropriate, we calculated a non-parametric common language effect size (McGraw & Wong, 1992) CL, as recommended by Grissom and Kim (2012). This statistic represents the probability that a score randomly drawn from one population A will be greater than a score randomly drawn from another population B.
First, we computed a series of Mann–Whitney U tests to determine whether Experiment 2 replicated the results of Experiment 1. Our findings replicated all previously observed effects, demonstrating again that participants revised their beliefs after reading the Transplant dilemma, and that those revisions remained stable even after a significant delay. Agreement ratings for the utilitarian belief were higher in the Control condition than in the Transplant condition when tested immediately (U(27, 36) = 792, p < .001, CL = .815), and after a delay (U(27, 36) = 773.5, p < .001, CL = .796). Participants' immediate and delayed belief ratings were examined with a related-samples Wilcoxon Signed Rank test. The ratings did not reliably differ over time in the Transplant condition (W(26) = 430.5), p = .058), although this effect approached significance. However, a similar trend was observed in the Control condition (W(27) = 200, p = .059), suggesting that this trend was not associated with our experimental manipulation.
Of particular interest in Experiment 2 was whether people would spontaneously recruit and revise their beliefs after considering the Transplant dilemma—that is, whether participants' agreement ratings with the utilitarian belief would be lower in the Delay-Only Transplant condition than in the Delay-Only Control condition. This test revealed a significant effect, where participants in the Delay-Only Transplant condition gave lower agreement ratings for the utilitarian belief than participants in the Delay-Only Control condition, U(27, 26) = 471.5, p = .032, CL = .673. We also compared participants' delayed utilitarian agreement ratings from the Immediate-Delay Transplant condition and the Delay-Only Transplant condition. There were no reliable differences in their agreement ratings (U(36, 26) = 360, p = .12), even though participants in the Immediate-Delay Transplant condition also rated their agreement with the utilitarian belief immediately after considering the Transplant dilemma earlier that day. These results are particularly striking as they suggest that people spontaneously recruit and revise their beliefs when considering moral dilemmas.
7 General discussion
How is it that moral convictions are at once steadfast, and yet people can and do change their moral beliefs? The present research demonstrates that considering a moral dilemma can produce authentic change in people's moral beliefs. We found that making a judgment about the Transplant dilemma changed people's beliefs in the utilitarian principle, “In the context of life or death situations, always take whatever means necessary to save the most lives.” This was the case both immediately following exposure to the dilemma and 6 hours later. Experiment 2 demonstrated that this change occurs even when participants are not immediately queried about their agreement with the utilitarian principle. This latter result suggests that being exposed to moral dilemmas may lead people to spontaneously revise their beliefs, even when they are not prompted to do so.
It is unlikely that participants' belief reports owe simply to a desire to appear consistent, as our experiments included distractor vignettes and distractor belief statements that concealed our aims. In addition, research on analogical problem solving has shown that, in the absence of featural support for memory retrieval, it is rare for participants to recall a relevant problem when attempting to solve a new problem, even after a short delay (e.g., Gick & Holyoak, 1983; Wharton, Holyoak, & Lange, 1996). Accordingly, it is unlikely that participants in Experiment 2 recalled the Transplant dilemma when they rated their agreement with the utilitarian belief 6 h later. Instead, a more plausible interpretation of the results of Experiment 2 is that exposure to the Transplant dilemma changed participants' utilitarian beliefs when they first encountered the dilemma. These changes then persisted, affecting agreement ratings six hours later.
8 What are the mechanisms underlying moral belief revision?
Our results suggest that moral dilemmas can lead people to revise their beliefs by pitting inconsistent moral beliefs against one another. This marks a significant advance for understanding the psychological processes underlying moral conviction and confirmation bias. It seems that people possess many moral beliefs that can be mapped onto a specific moral situation (e.g., Nichols & Mallon, 2006; Prinz, 2007). Once such a mapping is computed, a belief may suggest a certain response to the moral situation under consideration. When thinking about a moral dilemma, people might successfully map multiple beliefs to the same situation. If different beliefs entail different responses to the same situation, then this suggests that the elicited beliefs are inconsistent. In this case, one or more beliefs must be revised to restore coherence. For example, our participants may have reduced their agreement with the utilitarian belief after considering the Transplant dilemma because they mapped a utilitarian belief to the dilemma but ultimately recognized that their judgment about the dilemma was inconsistent with this belief.
Of course, if convictions are held strongly enough, it is possible that people might respond to perceived inconsistency by rejecting the mapping between a case and their belief. This outcome seems especially likely for convictions that have been publicly expressed, such as positions on political issues, which have often been the focus of research on moral conviction (e.g., Skitka, 2010). Further research is necessary to more completely understand the connection between people's judgments about dilemmas and moral belief revision.
9 Conclusion
The present studies reveal a means for inducing moral belief revision. Rather than presenting people with data that directly contradict their existing moral beliefs—and which could, therefore, be ignored or discounted due to confirmation bias—moral dilemmas, and the method of cases more generally, may circumvent confirmation bias by highlighting existing (but perhaps heretofore unnoticed) inconsistencies in people's beliefs. While moral belief revision is a necessary first step toward moral progress, contemporary moral psychologists have often seen this step as insurmountable. This study gives reason for optimism, as we have provided a method for inducing moral belief revision.
