Featured Video Play Icon

The Dark Side of Morality: Group Polarization and Moral-Belief Formation

Marcus Arvan
University of Tampa


Most of us are accustomed to thinking of morality in a positive light. Morality, we say, is a matter of “doing good” and treating ourselves and each other “rightly.” However, moral beliefs and discourse also plausibly play a role in group polarization, the tendency of social groups to divide into progressively more extreme factions, each of which regards other groups to be “wrong.”[1] Group polarization often occurs along moral lines[2], and is known to have many disturbing effects, increasing racial prejudice among the already moderately prejudiced[3], leading group decisions to be more selfish, competitive, less trusting, and less altruistic than individual decisions[4], eroding public trust[5], leading juries to impose more severe punishments in trials[6], generating more extreme political decisions[7], and contributing to war, genocide, and other violent behavior.[8]

This paper argues that three empirically-supported theories of group polarization predict that polarization is likely caused in substantial part by a conception of morality that I call the Discovery Model—a model which holds moral truths exist to be discovered through moral intuition, moral reasoning, or some other process.


Target Presentation in PDF

Target Presentation by Marcus Arvan (Tampa)

§1 of this paper clarifies the Discovery Model, showing how it is ubiquitous in everyday life and moral philosophy, cohering as well with empirical research on how people ordinarily form moral beliefs. §2 then argues that three dominant empirical theories of group polarization—Social Comparison Theory[9], Informational Influence Theory[10], and Self-Categorization Theory[11]—all predict that the Discovery Model likely plays a significant role in causing polarization. Finally, §3 argues that there are converse theoretical reasons to believe that an alternative Negotiation Model of morality—one according to which most moral truths are instead created by interpersonal negotiation—would likely mitigate polarization and perhaps even foster its opposite.

As a point of clarification, this paper’s aims are modest.[12] First, it aims neither to establish that the Discovery Model causes polarization, nor that the Negotiation Model is psychologically realistic or would reduce polarization. Because these are complex issues—and different possible explanations of polarization exist—this paper instead aims to provide strong theoretical grounds for investigating these matters further in future research. Second, this paper does not aim to settle a variety of important philosophical questions, such as whether group polarization is morally undesirable, or whether the Negotiation Model should be favored over the Discovery Model on meta- or normative-ethical grounds. Although I have defended Negotiation Model elsewhere on both grounds[13], this paper cannot settle these wide-ranging issues. Instead, it once again aims to provide new theoretical reasons—grounded in the psychology of group polarization—to investigate these and other related questions in future research.


1 The Discovery Model of Morality

Some philosophers[14] and everyday laypeople purport to be moral skeptics, alleging that there are no moral facts. Nevertheless, just about everyone plausibly forms moral beliefs in the course of everyday life. We regularly speak of people doing “right”, “wrong”, “good”, and “bad.” We also tend to do so in accordance with a particular model of morality: a Discovery Model according to which moral truths exist to be discovered through intuition, moral reasoning, or some other cognitive or affective process. The Discovery Model, as I propose we understand it, does not hold that we come to believe moral propositions passively or unreflectively. It is instead the conjunction of the following meta-ethical and psychological claims:

  • The discovery model of meta-ethics: there are preexisting truths about moral issues (e.g. truths about right, wrong, good, bad) that can in principle be ascertained by individuals “unilaterally”, via their own use of intuition, philosophical argument, or some other cognitive or affective process.
  • The discovery model of moral-belief formation: individuals who tacitly or explicitly endorse the discovery model of meta-ethics will tend to form moral beliefs though intuition, argument, or other such process, typically believing in at least some cases that they have discovered moral truths that other people should believe as well.

We can see just how ubiquitous the Discovery Model is by examining everyday discourse, the history and present of moral philosophy, and finally, social-psychological research.

Consider first everyday moral practice. When it comes to applied moral issues—such as whether abortion is morally permissible—people commonly presuppose that there are moral facts to be discovered through intuition, reasoning, or some other cognitive or affective process, typically forming beliefs about their “moral discoveries.” For example, individuals who believe that abortion is morally wrong often claim to base this “discovery” on the intuition or some argument that human beings have a moral right to life—arguing that since fetuses are human beings, we should all believe that abortion is wrong.[15] Conversely, people who believe abortion is morally permissible typically base their belief on different intuitive or argumentative “discoveries”, such as that fetuses do not have a moral right to life at certain stages of development[16], or alternatively, that a fetal right to life is not a right to depend on a mother’s body.[17] Similarly, consider the Israeli-Palestinian conflict. Here too, both sides appear to treat moral truths as discoverable through intuition or argument. Whereas pro-Palestinian voices defend the moral “discovery” Israel has wrongly occupied Palestinian lands, pro-Israelis typically defend the opposite “discovery”—that Israelis have rightly occupied Palestine, claiming that settlers are simply “living on land that Israel has liberated.”[18] In each of these cases, we see the Discovery Model’s meta-ethical and psychological components clearly exemplified. And these are not isolated cases. Indeed, the Discovery Model appears presupposed by every major world religion—with Judaism holding that we can discover moral truths through the Ten Commandments; Christianity that we can discover moral truths via Christ; Islam that moral truth is to be discovered through the Quran and/or Shari’a Law; Buddhism that moral truth is be found in the Noble Eightfold Path; and so on.

The Discovery Model similarly pervades the history and present of moral philosophy. In contemporary meta-ethics, the notion that moral facts exist to be discovered is central to many (though not all) theories of moral semantics and moral epistemology. Indeed, although there are non-cognitivist interpretations of moral language, the dominant semantic view of moral language is cognitivist: the view that sentences of the form, “X is morally wrong”, are true just in case it is a fact that X is morally wrong.[19] In addition, many moral realists argue that cognitivism is not only true, but that moral facts are mind-independent features of the world discoverable by us.[20] Further, consider a few dominant theories of moral epistemology: moral intuitionism, reflective equilibrium, and constitutivism. Intuitionists hold that moral truths can be discovered through a kind of moral perception analogous to sense-perception.[21] In contrast, reflective equilibrium treats moral facts as things we can discover by drawing our moral beliefs into greater coherence, holding that we should form new moral beliefs as a result of reflective argumentation.[22] Finally, constitutivists argue that moral truths can be discovered—and moral beliefs formed—by reference to constitutive features of agency.[23]

Now consider normative ethical theory. Here too we see the ubiquity of the Discovery Model. Act-utilitarians hold that an action is right if and only if the act maximizes utility—facts that can in principle be discovered. Kantians hold that an action is permissible if and only if its maxim can be willed as a universal law[24], respecting the humanity of oneself and others[25]—facts that once again can be discovered.[26] Aristotelian virtue ethicists hold moral virtues are beneficial character traits which we can discover to be necessary for living well.[27] And so on. Many other influential approaches to normative ethics—contractualism, Rossian pluralism, moral particularism, etc.—similarly hold that moral truths can be discovered by intuition, argument, or some other cognitive or affective process.[28]

The same is true in applied ethics. In the applied ethical literature on abortion, some argue that abortion can be discovered to be wrong because it violates the moral rights of the fetus[29]; others argue that abortion can be discovered to be morally permissible at certain stages of fetal development due to the fetus not being a person[30]; others still argue that abortion is permissible because a fetal right to life is not a right to depend on a mother’s body.[31] And so on. Once again, this is not an isolated case. The applied ethics literature is replete with works arguing that we can discover truths about applied ethical issues—about the ethics of torture, gun control, warfare, treatment of animals, etc.—through intuition, argument, or other cognitive or affective processes.

Finally, the Discovery Model’s psychological account of moral-belief formation coheres with the dominant empirical model of moral-belief formation in social psychology: the Social Intuition Model (SIM) which holds that moral beliefs are the result of sudden flashes of affectively laden intuitions, with moral reasoning largely serving a subservient role to justify one’s beliefs ex post facto.[32]


2 The Discovery Model and Group Polarization

Group polarization, once again, is the tendency of social groups to divide into progressively more extreme factions, each of which regards other groups to be “wrong.” There are two leading empirical theories of the causes of polarization: Social Comparison Theory and Informational Influence Theory. According to Social Comparison Theory, polarization results from people seeking to “fit in” with those around them, aiming to impress members of their group by endorsing progressively extreme views.[33] Informational Influence Theory holds that, in addition, polarization results from people hearing new arguments and information in support of their position—processes that make group members more receptive to progressively more extreme views.[34] Importantly, these two theories are not mutually exclusive, nor are they “mere theories.” Because both theories have significant empirical support[35], it is more correct to say that they have identified two primary causal mechanisms of polarization. Finally, a third theory, Self-Categorization Theory, also has some empirical support.[36] This theory holds that polarization results from individuals identifying with the prototypical view of their “in-group”—a group which then casts views of “out-groups” as threatening, causing the views of the in-group to shift even further away from those of the outgroup as a kind of defense mechanism.[37] Allow me now to briefly explain each of these three theories in more detail.

According to Social Comparison Theory, individuals in groups have a psychological tendency to want to gain acceptance and be perceived favorably by other members of their group. This desire for acceptance causes individuals in the group to adopt ever-so-slightly more extreme views than those already typical in the group, so as to “impress” and “prove” themselves to other members of the group[38]—something which often takes the form of moral grandstanding.[39] Group polarization then occurs when many individuals in the group do more or less the same thing, progressively adopting more extreme views to impress each other, causing the representative beliefs of the entire group to become progressively more extreme. Finally, and importantly, studies indicate that this phenomenon is even more likely to occur with respect to “judgmental issues”, such as moral or political matters.[40] For instance, a recent study on Twitter regarding the shooting of an abortion doctor indicated that like-minded individuals on both sides of the issue tend to group together, reinforcing and progressively polarizing pro-life and pro-choice views.[41]

Informational Influence Theory supplements this account with a complementary mechanism: the tendency of people to group together with likeminded individuals to present novel arguments and information in favor of their preferred views, leading individuals in the group to become more easily convinced of even more extreme views.[42] For example, members of different political parties tend to frequent different news sources and social media networks.[43] These differential sources of information tend to provide members of each group with new information and arguments supporting their members’ preexisting views, ignoring or delegitimizing countervailing information and arguments, thereby making individuals in each groups more likely to develop more polarized views.[44] Further, research indicates this mechanism is especially strong for “intellectant” issues—or issues involving intellectual debate, including moral issues.[45] Informational Influence Theory also coheres with a well-established individual bias: confirmation bias, the tendency of people to selectively seek and privilege information confirming their preexisting beliefs, while ignoring or minimizing contrary information.[46]

Finally, Self-Categorization offers a third mechanism for polarization. When individuals are confronted with a risky or threatening outgroup, there is a pronounced human tendency to coalesce around the views of one’s self-identified in-group as a kind of protection-mechanism of solidarity against the threatening out-group.[47] This general mechanism is familiar from everyday life and history—as when Adolf Hitler used perceived threats (by Jews and others) to rally the German people behind his extreme Nazi ideology, and in polarized debates over moral issues today. For instance, anti-abortionists may be cast proponents of abortion as complicit in “genocide”[48] whereas proponents of abortion cast anti-abortionists as “waging a war on women.”[49] Finally, this mechanism appears to strengthen the more threatening the outgroup is perceived to be.[50]

We can now provide several theoretical arguments that Discovery Model likely plays a significant role in group polarization. First, the Discovery Model appears to play directly into the phenomena described by Social Comparison Theory. Once again, Social Comparison Theory shows that people tend to seek approval of those they interact with, adopting progressively more extreme views to impress those in their group. As such, Social Comparison Theory predicts that if people cluster around opposing moral “discoveries”—if, for instance, some believe they have discovered abortion is wrong, whereas others believe they have discovered abortion is permissible—there will be a progressive tendency for each group’s members to adopt more extreme positions in order to impress members of their own group. Which, of course, is basically what we do see across a wide variety of moral issues. We see people cluster together in opposing moral groups—pro-abortion and anti-abortion groups, gun-control and gun-rights groups, pro-Israel groups and pro-Palestinian groups, etc.—with members on each side often “ramping up” their moral claims to impress fellow group-members.[51] Social Comparison Theory thus not only predicts that the Discovery Model likely plays a significant causal role in polarization. Social Comparison Theory and the Discovery Model together provide a potential explanation for why polarization is so pronounced on “judgmental” issues (specifically, moral issues).[52] Because moral beliefs involve or are related to reactive attitudes such as anger and blame[53], when members of opposing moral groups believe they have made opposing “moral discoveries” (viz. the Discovery Model), members of each group may adopt more extreme views in order to appeal to the reactive attitudes of members of their group (playing to their group’s anger, resentment, etc.).[54] Social Comparison Theory thus predicts not only that the Discovery Model likely plays a causal role in polarization, but that likely plays a prominent role.

Now turn to Informational Influence Theory, the theory which holds that polarization occurs by people in groups providing each other novel arguments and information that confirm their pre-existing beliefs, making them more amenable to even more extreme views. Here again, the Discovery Model appears to play directly into these phenomena.  Informational Influence Theory predicts that that if people cluster around opposing moral “discoveries”—if, for instance, some believe they have discovered abortion is wrong, whereas others believe they have discovered abortion is permissible—people will tend to provide new arguments and information to members of their own favored moral group, making members of each group progressively more amenable to more extreme beliefs. Yet this too is broadly what we see in everyday life. In the abortion debate, for instance, we see many novel philosophical arguments generated on each side of the debate, with each side tending to emphasize the novel arguments for their own moral beliefs while ignoring or dismissively discounting arguments for the other side’s beliefs.[55] This is clearly not an isolated case, as people are known to cluster in political groups around different moral issues—groups which tend to expose their members to different information and arguments.[56] Further, as we saw earlier, Information Influence Theory predicts that polarization tends to be particularly pronounced for “intellectant” issues.[57] Because people who subscribe to the Discovery Model commonly treat moral matters as issues of intellectual debate—things to debate at dinner tables, on television, in university seminars, etc.—Informational Influence Theory thus predicts that the Discovery Model likely plays a prominent role in polarization.

Finally, the Discovery Model also appears to play directly into the phenomena described by Self-Categorization Theory. Self-Categorization Theory predicts that group polarization tends to occur when an in-group is confronted by a threatening out-group. It is plain from everyday experience, however, that in-groups and “threatening outgroups” are often defined precisely in moral terms—in terms of “moral truths” people believe members of their group to have discovered. For instance, both sides of the abortion debate clearly find the other group threatening[58]—and the same is clearly true across a variety of moral issues, including the Israeli-Palestinian conflict, gun-control, and so on. Insofar as human beings often do identify as members of moral groups—people who cluster around similar moral “discoveries”, viewing opposing groups as threatening out-groups—Social Categorization Theory also predicts that the Discovery Model plays a causal role in polarization.


3 An Anti-Polarizing Alternative? The Negotiation Model

I have argued elsewhere[59] on meta- and normative ethical grounds that although a few moral ideals (of coercion-minimization, mutual assistance, and fair bargaining) can be discovered through rational argumentation, all other moral truths—including how the above ideals should applied to applied ethical topics—should be thought of not as discoverable through intuition, argument, or any other cognitive or affective process, but instead as created by interpersonal negotiation. Let us call this the Negotiation Model of morality.

The basic idea behind the Negotiation Model is straightforward. Consider again the issue of abortion. The Discovery Model holds that there are preexisting facts about the morality of abortion (viz. its rightness, wrongness, etc.) that we can discover through intuition, argument, or some other cognitive or affective process. In contrast, the Negotiation Model holds that the moral status of abortion is literally indeterminate unless and until a social compromise has been arrived at, after which point abortion’s moral status should be seen as defined by norms negotiated, in essence settling abortion’s moral status via negotiated compromise (as in: “We have now negotiated a compromise that abortion is permissible in conditions A, B, and C, but impermissible in conditions X, Y, and Z”). On the Negotiation Model, as such, it is a meta- and normative-ethical mistake to form moral beliefs prior to social negotiation. People should instead withhold judgment on controversial moral issues, not forming moral beliefs on those issues until after clear public norms have been negotiated, after which point they should form beliefs in line with the negotiated norms (as in: “I now believe that abortion is morally permissible in conditions A, B, and C, but impermissible in conditions X, Y, and Z, because these are the standards that have been publicly negotiated as a compromise”). Importantly, on the Negotiation Model, these publicly negotiated norms—and the moral beliefs they prescribe—are not mere “maxims” or rules to follow for some further moral aim (such as, say, utility-maximization). The norms instead express genuine moral propositions about the issue in question (viz. the moral permissibility or impermissibility of abortion) that individuals should believe.[60]

The Negotiation Model obviously raises many empirical and philosophical questions. First, is the model psychologically realistic? Can people really believe (for instance) that abortion’s moral status is indeterminate prior to public negotiation, and then come to believe that public negotiation settles its moral status? Second, is the model meta-ethically and normatively justifiable? Can negotiation truly settle the moral status of abortion? Although I have argued elsewhere that the Negotiation Model is indeed justifiable meta-ethically, normatively, and psychologically[61], these are broad issues that we cannot settle here. Instead, let us examine the Negotiation Model’s theoretical relationship to this paper’s topic: group polarization.

On my preferred version of the Negotiation Model, moral truths are created by the outcome of negotiated agreements between all agents plausibly motivated by moral ideals of coercion-minimization, mutual assistance, and equal bargaining power.[62] However, because my favored account is controversial[63], let us define the Negotiation Model here more broadly, in terms of the following meta-ethical and psychological claims:

  • The negotiation model of meta-ethics: aside from perhaps some moral ideals which may be discovered by rational argument (more on this shortly), moral truths do not exist to be discovered by intuition, argument, or any other cognitive or affective process, but are instead created by interpersonal processes of moral negotiation that, at the very least, involve all those with interests in the issue in question seeking to arrive at a compromise agreement on moral norms for that issue.
  • The negotiation model of moral-belief formation: individuals who endorse the negotiation model of meta-ethics should tend to seek common ground with others who share relevant moral ideals; believe that morality requires moral negotiation as such; and form moral beliefs only after negotiating compromises, in line with whatever norms have resulted from compromise agreement.

Because these claims are complex, allow me to spell out their components a bit more.[64] “Moral negotiation”, as I wish to understand it, is a term of art intended to cover any and all forms of interpersonal human behavior (such as conversation, voting, and so on) that involve those with interests in a given moral issue (abortion, etc.) obeying certain discursive rules—rules that at the very least include a commitment to (A) certain regulative moral ideals as background beliefs and motivations, (B) suspending moral judgment on the issue in question before interpersonal agreement is reached, and (C) seeking a compromise agreement on moral norms for the issue in question (e.g. abortion) with others who plausibly satisfy conditions (A) and (B). Allow me to illustrate using my preferred version of the Negotiation Model—a version which, again, I do not want to presuppose the truth of here, but merely use for illustrative purposes.

As mentioned earlier, I have argued elsewhere that moral ideals of coercion-minimization, mutual assistance, and equal bargaining power can be established (i.e. discovered) through rational argument.[65] However, I argue that these are only regulative ideals—ideals that people can have legitimate interests in weighing against one another. Accordingly, on my favored version of the Negotiation Model, moral truths on controversial moral issues (e.g. abortion) should be understood in terms of the outcome of a compromise agreement between all those who plausibly share the above regulative ideals (of coercion-minimization, mutual assistance, etc.), but who may have different priorities on the issue in question. The example of abortion is, I think, instructive here. On my account, a major reason why the morality of abortion remains such a divisive issue is because individuals on both sides of the debate plausibly share relevant regulative moral ideals. Anti-abortion advocates, for example, typically claim that fetuses have a “right to life”—a claim clearly intended to help fetuses, protecting them from having their lives coercively ended. Pro-abortion advocates, on the other hand, claim women have a “right to choose”—a claim clearly intended to help women, protecting their reproductive choices from being coercively reduced. Both sides are thus plausibly motivated by regulative ideals of coercion-minimization and assisting others. The primary difference between the two sides—on my version of the Negotiation Model, at least—occurs at the level of moral priorities: anti-abortion advocates currently think the rights of the fetus should “trump” (or outweigh) the rights of women, whereas pro-abortion advocates think the rights of women should take priority. On my favored version of the Negotiation Model, because both sides share relevant regulative ideals but have different priorities, they have a duty to negotiate a compromise: it is wrong for both sides to think that they can “discover” moral correctness of their own preferred view (e.g. “Abortion is wrong!”) via intuition, argument, or any other process of discovery. Instead, the Negotiation Model requires both sides to both recognize the legitimacy of the other side’s concerns (since both sides plausibly share relevant regulative ideals), and then demonstrate a willingness to forge a compromise agreement.

What kind of negotiated compromise might emerge from such a process, say in the case of abortion? One obvious possibility—but not the only possible one—is this: because prevailing scientific knowledge indicates fetuses first become sentient between 18 to 25 weeks of gestation[66], both sides of the abortion debate could  (if they were willing to conform to the Negotiation Model) arrive at a compromise agreement that (i) early abortion prior to fetal sentience is morally permissible, (ii) abortion after fetal sentience is normally impermissible, except perhaps in cases of rape, incest, or danger to the mother’s life, and finally (iii) members of society share a duty to devote ample social resources to provide sexually-active women with ready access to family-planning resources (including access to affordable early-term abortion) to prevent abortion after fetal sentience.

Such a compromise would almost certainly not fully satisfy many parties to the abortion debate—in part, I think, because the Discovery Model is so deeply entrenched in how people think about moral issues (viz. “But abortion is murder!”), but also because of the very nature of compromise (which requires “give and take”). Still, as uncomfortable as compromise may be, if the Negotiation Model is correct (as I have argued elsewhere it is), we should nevertheless accept that it is what morality requires. Further, and importantly, because people might not be wholly satisfied with a given compromise, compromises following the Negotiation Model would plausibly leave many matters open to renegotiation. If, for instance, the above compromise on abortion had serious negative effects on women’s lives (e.g. by requiring single women to bear children if the father dies late in pregnancy), individuals on the “pro-choice” side of the debate could bring that new information to bear publicly in the aim of renegotiating abortion norms (as in: “I know we have currently agreed that abortion is permissible in cases X, Y, and Z. However, because this standard of permissibility is having negative effects on women, I would like us to consider a new compromise on different standards of permissibility”). In this way, the Negotiation Model entails—plausibly, I believe, albeit provocatively—that moral truths on controversial issues can literally evolve as people lobby for and negotiate new compromises.

It might seem hard to imagine many people ever accepting the Negotiation Model, as it would require us to give up many moral convictions (about abortion, etc.) that we may believe very deeply. Indeed, the Negotiation Model’s psychological model of moral-belief formation might turn out to be difficult or even impossible for people to reliably conform to, if Haidt’s Social Intuition Model (SIM) of moral-belief formation is correct.[67] Further, some may worry that the Negotiation Model would have undesirable moral consequences, such as requiring gays and lesbians to negotiate on same-sex marriage—compromises that might set back the gay liberation movement.[68] All of these are important questions worth investigating in more detail in future research. However, because we cannot settle them here, let us instead examine the Negotiation Model’s theoretical relationship to our primary topic of inquiry: group polarization.

…[Abridged for conference presentation]…

Let us begin with Social Comparison Theory: the theory which holds that group polarization results from people adopting more extreme views to impress fellow group members. The Negotiation Model promises to undercut this polarizing mechanism in at least two ways: by (A) discouraging people from adopting first-order moral beliefs (e.g. “Abortion is wrong!”) prior to negotiating, and (B) encouraging people to see answers to controversial moral questions as created through negotiated compromise. These two discursive elements of the Negotiation Model promise to defuse polarizing mechanisms of social comparison at “step 1”: it would lead people have no settled moral beliefs for groups to cluster around (e.g. “Abortion is wrong”) in the first place. On the contrary, it would plausibly give people an anti-polarizing ideal to cluster around despite their differences: the ideal of negotiating compromises with people with different priorities who share relevant moral ideals. Importantly, existing research already indicates that group organization around such a cooperative ideal does indeed mitigate polarization and promote cooperation.[69] As such, the Negotiation Model theoretically promises to harness the forces that Social Comparison Theory identifies as responsible for group polarization to produce the very opposite: progressive convergence on a cooperative norm of negotiated compromise.

Now consider Informational Influence Theory, the theory which holds that group polarization is generated precisely by people seeking out and attending selectively to information that confirms their preexisting belief. The Negotiation Model promises to undermine the social psychological forces responsible for group polarization here as well. When people have preexisting first-order moral beliefs on a given side of an issue, as on the Discovery Model—such as the preexisting belief that abortion is wrong, or alternatively, that women have a right to abortion—the phenomenon of informational influence leads each side to attend to information in support of their preexisting moral beliefs, making each side more amenable to more extreme beliefs. The Negotiation Model, on the other hand, holds that people should not have opposing first-order moral beliefs in the first place, but should instead conceive answers to moral questions (such as the morality of abortion) as created through negotiated compromise. But now if people increasingly held this kind of first-order moral belief—the belief that answers to controversial moral questions must be negotiated—then Informational Influence Theory predicts that people would become progressively more inclined to seek out and attend selectively to their belief that moral answers must be created cooperatively through negotiation. Consequently, the Negotiation Model theoretically promises to harness the phenomena described by Informational Influence Theory to prevent polarization and promote cooperation.

Finally, consider Social-Categorization Theory, the theory which holds that group polarization results from the development of in-groups which then treat out-groups as threatening. As we have seen, in-groups and out-groups often form around divisive moral “discoveries.” For example, whereas anti-abortionists often cast defenders of abortion as a threatening out-group (“They are baby-killers!”), defenders of abortion often cast anti-abortionists the very same way (“They want to take away women’s rights!”). The Negotiation Model once again promises to halt this polarizing force at “step 1.” Insofar as it (A) holds that people should not have settled moral beliefs on issues prior negotiation, and (B) should be willing to negotiate compromises with those with different priorities, the Negotiation Model would theoretically prevent the formation of divisive in-groups and out-groups, instead promoting the development of a cooperative in-group: people who have different priorities on controversial moral issues (abortion, gun control, etc.), but who are nevertheless unified around anti-polarizing ideals of negotiation and compromise (something which, again, has indeed been found to promote cooperation[70]). Finally, Social-Categorization Theory does plausibly predicts that the Negotiation Model would generate certain types of polarization—namely, polarization between those who accept the Negotiation Model and those who accept the Discovery Model (who might indeed regard each other as threatening “enemies”), as well as polarization with those who reject relevant regulative ideals (e.g. racists, sexists, etc.). However, while Social-Categorization Theory plausibly predicts that there would be polarization between these groups—with each potentially treating the others as threatening out-groups—these would not obviously be bad forms of polarization according to the Negotiation Model, as the model itself suggests that we should not tolerate the Discovery Model or false regulative ideals.

In sum, all three empirical theories of group polarization predict that the Negotiation Model is likely to substantially reduce polarization relative to the Discovery Model, and perhaps even harness social-psychological forces to generate the opposite: a progressive willingness to cooperate and compromise.



Laypeople and philosophers tend to treat moral truths as discoverable through intuition, argument, or other cognitive or affective process. However, we have seen that there are strong theoretical reasons—based on three empirically-supported theories of group polarization—to believe this Discovery Model of morality is a likely cause of polarization: a social-psychological phenomenon known to have a wide variety of disturbing social effects. We then saw that there are complementary theoretical reasons to believe that an alternative, Negotiation Model of morality might not only mitigate polarization but actually foster its opposite: an increasing willingness for to work together to arrive at compromises on moral controversies. While this paper does not prove the existence of the hypothesized relationships between the Discovery Model, Negotiation Model, and polarization, it demonstrates that there are ample theoretical reasons to believe that such relationships are likely and worthy of further empirical and philosophical research.



Abrams, D.; M. Wetherell; S. Cochrane; M.A. Hogg; J.C. Turner (1990). ‘Knowing what to think by knowing who you are: Self-categorization and the nature of norm formation, conformity and group polarization’, British Journal of Social Psychology 29 (2): 97–119.

Andrews, B., Levintova, H., Liss-Schultz, N. & P.R. Lockhart (2017). ‘Trump Brought the War on Women Mainsteam in His First 100 Days’, Mother Jones, http://www.motherjones.com/politics/2017/04/first-100-days-reproductive-rights-abortion-trump, accessed on 6/5/2017.

Arthur, J. (2001). ‘Personhood: Is a Fetus a Human Being’, Pro Choice Action Network, http://www.prochoiceactionnetwork-canada.org/articles/fetusperson.shtml, accessed 3/20/16.

Aristotle, Nichomachean Ethics, T. Irwin (trans.) (1999). Aristotle: Nicomachean Ethics, 2nd Ed. (Indianapolis: Hackett).

Audi, R. (2015). ‘Intuition and Its Place in Ethics’, Journal of the American Philosophical Association 1(1): 57-77.

Baldassarri, D., & Gelman, A. (2008). ‘Partisans without constraint: Political polarization and trends in American public opinion’, AJS: American journal of sociology114(2), 408.

Barcalow, E. (2007). Moral Philosophy: Theories and Issues, 4th ed. (Belmont, CA: Thomson Wadsworth).

Black, P.; B. Wedeman, & J. Mullen (2015). ‘Israeli Palestinian Conflict: Two Viewpoints; One Tragic Outcome’, CNN, http://www.cnn.com/2015/10/21/middleeast/israeli-palestinian-root-of-violence, accessed 6/20/16.

Bray, R. M. & A. M. Noble (1978). ‘Authoritarianism and decisions of mock juries: Evidence of jury bias and group polarization’, Journal of Personality and Social Psychology 36 (12): 1424–1430

Brewer, M. B. (1996). ‘When contact is not enough: Social identity and intergroup cooperation’, International journal of Intercultural relations20(3-4), 291-303.

Brink, D. (1989). Moral Realism and the Foundations of Ethics (Cambridge: Cambridge University Press).

Cunningham, G. (2009). ‘Is Abortion Genocide?’, Canadian Center for Bio-Ethical Reform, https://www.endthekilling.ca/training/classroom/genocide, accessed 6/2/2017.

Dancy, J. (2013). ‘Moral Particularism’, in E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Fall 2013 Edition), http://plato.stanford.edu/archives/fall2013/entries/moral-particularism/.

—– (1986). ‘Two Conceptions of Moral Realism’, Proceedings of the Aristotelian Society, Supplementary Volume 60: 167-87.

Daniels, N. (2013). ‘Reflective Equilibrium’, in E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2013 Edition), http://plato.stanford.edu/archives/win2013/entries/reflective-equilibrium/, accessed 6/20/16.

Durkheim, E. [1893]. The Division of Labor in Society, G. Simpson (trans.) (New York: the Free Press).

Finlay, S. (2007). ‘Four Faces of Moral Realism’, Philosophy Compass 2: 820-49.

Haidt, J. (2012). The righteous mind: why good people are divided by politics and religion (New York: Vintage Books).

—- (2001). ‘The emotional dog and its rational tail: a social intuitionist approach to moral judgment’, Psychological review108(4), 814.

Hinsz, V.B.; J.H. Davis (1984). ‘Persuasive Arguments Theory, Group Polarization, and Choice Shifts’, Personality and Social Psychology Bulletin10: 260–268.

Hogg, M.A.; Turner, J.C.; Davidson, B. (1990). ‘Polarized norms and social frames of reference: A test of the self-categorization theory of group polarization’, Basic and Applied Social Psychology 11: 77–100.

Huckfeldt, R., Mendez, J.M., & Osborn, T. (2004). ‘Disagreement, ambivalence, and engagement: The political consequences of heterogeneous networks’, Political Psychology25(1), 65-95.

Hursthouse, R. (1999). On Virtue Ethics (Oxford: Oxford University Press).

Isenberg, D.J. (1986). ‘Group Polarization: A Critical Review and Meta-Analysis’, Journal of Personality and Social Psychology 50 (6): 1141–1151.

Iyengar, S. & Hahn, K.S. (2009). ‘Red Media, Blue Media: Evidence of Ideological Selectivity in Media Use’, Journal of Communication 59(1): 19-39.

Johnston, L.C. & C.N. Macrae. ‘Changing social stereotypes: The case of the information seeker’, European Journal of Social Psychology 24.5 (1994): 581-592.

Joyce, R. (2015). ‘Moral Anti-Realism’, in E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Summer 2015 Edition), http://plato.stanford.edu/archives/sum2015/entries/moral-anti-realism/.

—– (2007). The Myth of Morality (Cambridge, UK: Cambridge University Press).

Kant, I. (1797). The Metaphysics of Morals, in M.J. Gregor (ed.), The Cambridge Edition of the Works of Immanuel Kant: Practical Philosophy (Cambridge: Cambridge University Press, 1996): 363-603.

—-(1785). Groundwork of the Metaphysics of Morals, in M.J. Gregor (ed.), The Cambridge Edition of the Works of Immanuel Kant: Practical Philosophy (Cambridge: Cambridge University Press, 1996): 38-108.

Kaplan, M.F. (1977). ‘Discussion Polarization Effects in a Modified Jury Decision Paradigm: Informational Influences’, Sociometry 3 40(3): 262–271.

Katsafanas, P. (2011). ‘Deriving Ethics from Action: A Nietzschean Version of Constitutivism’, Philosophy and Phenomenological Research 83(3): 620-60.

Koch, C. (2009). ‘When Does Consciousness Arise in Human Babies?’, Scientific American, https://www.scientificamerican.com/article/when-does-consciousness-arise/, accessed 6/1/2017.

Korsgaard, C.M. (2009). Self-Constitution: Agency, Identity, and Integrity (Oxford: Oxford University Press).

—– (2008). The Constitution of Agency (Oxford: Oxford University Press).

—– (2000). Creating the Kingdom of End (New York: Cambridge University Press).

Liberty Women’s Health (2016), ‘Freedom to Choose’, http://www.libertywomenshealth.com/rtc.php, accessed 6/20/16.

Luhan, Wolfgang J.; Martin G. Kocher; Matthias Sutter (2009). ‘Group Polarization in the Team Dictator Game Reconsidered’, Experimental Economics 12 (1): 26–41

Mackie, D.M. (1986). ‘Social Identification Effects in Group Polarization’, Journal of Personality and Social Psychology 50: 720–728.

Mackie, J.L. (1977). Ethics: Inventing Right and Wrong (London: Penguin UK).

Marquis, D. (2007). ‘An Argument Abortion is Wrong’, in Russ Shafer-Landau (ed.), Ethical Theory: An Anthology (Blackwell Publishing): 439-50.

McGarty, Craig; John C. Turner, Michael A., Barbara David; et al. (1992). ‘Group polarization as conformity to the prototypical group member’, British Journal of Social Psychology 31: 1–19

Morris, J.S. (2007). ‘Slanted Objectivity? Perceived Media Bias, Cable News Exposure, and Political Attitudes’, Social Science Quarterly 88(3): 707-28.

Myers, D.G.; S.J. Arenson (1972). ‘Enhancement of dominant risk tendencies in group discussion’, Psychological Reports 30: 615–623.

Myers, David G.; G.D. Bishop (1970). ‘Discussion effects on racial attitude’, Science 169 (3947): 778–779

Newman, L.S. (2002). Understanding Genocide: The Social Psychology of the Holocaust. New York, NY: Oxford University Press. pp. 143–161

NRLC (2016). ‘Mission Statement’, www.nrlc.org/about/mission, accessed 6/20/16.

Parfit, D. (2011). On What Matters, Vols. 1&2 (Oxford: Oxford University Press).

Pew Research Center (2016). ‘A Wider Ideological Gap Between More and Less Educated Adults’, http://www.people-press.org/2016/04/26/a-wider-ideological-gap-between-more-and-less-educated-adults/, accessed on June 6, 2017.

Plous, Scott (1993). The Psychology of Judgment and Decision Making (McGraw Hill).

PoliticsUSA (2017). ‘Proof of the GOP War on Women’, http://www.politicususa.com/proof-war-women-2, accessed 6/5/2017.

Pojman, L.P. (1998). ‘Abortion: A Defense of the Personhood Argument’, The Abortion Controversy 25 Years After Roe v. Wade: A Reader (Cenage).

Pro-Life Perspective (2016). ‘About the show’, Pro-Life Perspective, www.prolifeperspective.com, accessed 6/20/16.

Pruitt, D. (1971).’’Choice shifts in group discussion: An introductory review’, Journal of Personality and Social Psychology 20 (3): 339–360.

Rapp, C. (2016). ‘Moral opinion polarization and the erosion of trust’, Social science research58, 34-45.

Ross, W.D. (1930). The Right and the Good (Oxford: Oxford University Press, 2002).

Rostenberg, G. (2014). ‘Why Abortion is Genocide’, Live Action News, https://www.liveaction.org/news/why-abortion-is-genocide/, accessed 6/2/2017.

Scanlon, T.M. (1998). What We Owe to Each Other (Cambridge, MA: Harvard University Press).

Shafer-Landau, R. (2003). Moral Realism: A Defense (Oxford: Oxford University Press).

Shoemaker, D.W. (2011). ‘Psychopathy, Responsibility, and the Moral/Conventional Distinction’, Southern Journal of Philosophy, 49(s1): 99-124

Stoner, J.A. (1961). ‘A comparison of individual and group decision involving risk’, Unpublished master’s thesis, Massachusetts Institute of Technology.

Stratton-Lake, P. (2014). ‘Intuitionism in Ethics’, in E.N. Zalta (ed), The Stanford Encyclopedia of Philosophy (Winter 2014 Edition), http://plato.stanford.edu/archives/win2014/entries/intuitionism-ethics/, accessed 6/16/2016.

Strawson, P.F. (1962). ‘Freedom and resentment’, Proceedings of the British Academy 48:1-25.

Stroud, N.J. (2010). ‘Polarization and partisan selective exposure’, Journal of Communication60(3), 556-576.

Sunstein, C.R. (2002). ‘The Law of Group Polarization’, The Journal of Political Philosophy 10 (2): 175–195

Tawia, S. (1992). ‘When is the capacity for sentience acquired during human fetal development?’, Journal of Maternal-Fetal Medicine1(3), 153-165.

Thomson, J.J. (1976). ‘A Defense of Abortion’, in Biomedical Ethics and the Law (Springer).

Timmons, M. (2007). Disputed Moral Issues: A Reader (New York: Oxford University Press).

Tosi, J. & Warmke, B. (2016). ‘Moral Grandstanding’, Philosophy and Public Affairs 44(3): 197-217.

Van Roojen, M. (2015). ‘Moral Cognitivism vs. Non-Cognitivism’, in E.N. Zalta (ed.) The Stanford Encyclopedia of Philosophy (Fall 2015 Edition), http://plato.stanford.edu/archives/fall2015/entries/moral-cognitivism, accessed 6/20/16.

Vaughn, L. (2009). Bioethics: Principles, Issues, and Cases (Oxford: Oxford University Press).

Vinokur, A. & Burnstein, E. (1974). ‘Effects of partially shared persuasive arguments on group induced shifts: A group problem-solving approach’, Journal of Personality and Social Psychology 29: 305–315.

Walker, T.G. & Main, E.C. (1973). ‘Choice shifts and extreme behavior: Judicial review in the federal courts’, The Journal of Social Psychology. 2 91 (2): 215–221.

Warren, M.A. (1973). ‘On the Moral and Legal Status of Abortion’, The Monist: 43-61.

Yardi, S. & Boyd D. (2010). ‘Dynamic Debates: An analysis of group polarization over time on Twitter’, Bulletin of Science, Technology and Society 30 (5): 316–27.




[1] See Isenberg (1986) and Pruit (1971).

[2] See Haidt (2012).

[3] Myers and Bishop (1970).

[4] Luhan, Kocher, and Sutter (2009).

[5] Rapp (2016).

[6] Bray and Noble (1978)

[7] Walker and Main (1973)

[8] Newman (2002), Sunstein (2002).

[9] Bray and Noble (1978), Mackie (1986).

[10] Stoner (1961); Myers and Arenson (1972); Hinsz and Davis (1984).

[11] Abrams et al (1990).

[12] I thank two anonymous reviewers for inviting me to rethink and foreground my argument’s scope.

[13] In [book title redacted], I argue that seven meta-ethical principles of theory-selection support a new normative ethical theory in line with the Negotiation Model.

[14] See e.g. Joyce (2007, 2015) and Mackie (1977).

[15] See e.g. NRLC (2016), Pro-Life Perspective (2016).

[16] See e.g. Arthur (2001).

[17] See e.g. Liberty Women’s Health (2016)

[18] See Black, Wedeman, and Mullen (2015) for a brief overview.

[19] See van Roojen (2015) for an overview.

[20] See e.g. Shafer-Landau (2003), Brink (1989), Dancy (1986), Finlay (2007).

[21] See e.g. Audi (2015) as well as Stratton-Lake (2014): introduction, for an overview of intuitionism’s resurgence.

[22] See Daniels (2013), Timmons (2007): 27-31, Vaughn (2009): 46-7, and Barcalow (2007): 14-5 for brief summaries of how pervasively reflective equilibrium is used to evaluate moral theories and arguments.

[23] See e.g. Kant (1785, 1797), Korsgaard (2008, 2009), and Katsafanas (2011).

[24] Kant (1785): 4:421.

[25] Ibid: 4:429.

[26] Ibid: 4:422-3, 4:429-31.

[27] See Nichomachean Ethics. Also see Hursthouse (1999): ch. 1.

[28] See e.g. Scanlon (1998): 4, 191; Ross (1930); Dancy (2013); and Parfit (2011): §49.

[29] See e.g. Pojman (1998) and Marquis (2007).

[30] See e.g. Warren (1973).

[31] Thomson (1976).

[32] Haidt (2001).

[33] Bray and Noble (1978), Mackie (1986).

[34] Stoner (1961); Myers and Arenson (1972); Hinsz and Davis (1984).

[35] For major confirming evidence of Social Comparison Theory, see Bray and Noble (1978), Myers and Bishop (1970), and Luhan et al (2009). For major confirming evidence of Informational Influence Theory, see Stoner (1961), Myers and Arenson (1972), Kaplan (1977), and Hinsz and Davis (1984).

[36] See Hogg et al (1990) and McGarty et al (1992).

[37] Abrams et al (1990).

[38] Bray and Noble (1978), Myers and Bishop (1970), and Luhan et al (2009).

[39] See Tosi & Warnke (2016).

[40] Isenberg (1986).

[41] Yardi and Boyd (2010).

[42] See Vinokur and Burnstein (1974)

[43] See e.g. Iyengar & Hahn (2009).

[44] See e.g. Morris (2007).

[45] Isenberg (1986).

[46] Plous (1993):  233.

[47] Hogg et al (1990).

[48] Cunningham (2009).

[49] Andrews et al. (2017).

[50] McGarty (1992).

[51] Tosi & Warmke (2016).

[52] See e.g. Haidt (2012) and Isenberg (1986).

[53] Strawson (1963).

[54] See Tosi & Warmke (2016): §2. For examples, see e.g. Cunningham (2009), Rostenberg (2014), and PoliticsUSA (2017).

[55] For instance, whereas anti-abortion websites such as Arthur (2001) and Pro-Life Perspective (2016) tend to emphasize philosophical arguments defending the moral status of human fetuses (viz. Marquis 2007 and Pojman 1998), pro-choice websites such as NLRC (2016) tend to emphasize arguments defending women’s rights to their bodies (viz. Thomson 1976).

[56] See e.g. Iyengar & Hahn (2009).

[57] Isenberg (1986).

[58] Again, see Andrews et al. (2017) and Cunningham (2009).

[59] [Reference redacted to preserve anonymized review].

[60] I thank an anonymous reviewer for suggesting this clarification.

[61] [Reference redacted for anonymized review].

[62] [Reference redacted for anonymized review].

[63] [Reference redacted for anonymized review].

[64] I thank an anonymous reviewer for inviting me to explicate the nature of negotiation and the model’s discursive rules in more detail.

[65] [Reference redacted for anonymized review].

[66] See Tawia (1992) and Koch (2009).

[67] See Haidt (2012). It is important to note here that empirical data supporting the SIM model have been collected under prevailing social conditions—in which most people appear to tacitly or explicitly accept the Discovery Model. Consequently, it is an open question whether a “paradigm shift” in the direction of the Negotiation Model might substantially change how individuals form moral beliefs.

[68] I thank an anonymous reviewer for raising these concerns.

[69] Brewer (1996).

[70] Brewer (1996).

Invited Comments from Michael Bishop (Florida State)

Comments on Marcus Arvan’s “The Dark Side of Morality: Group Polarization and Moral Belief Formation”

Michael Bishop
Florida State University


Marcus Arvan’s interesting and ambitious paper argues that the mechanisms responsible for group polarization would be undone – or at least seriously weakened – by the negotiation model of morality. Let’s start with three basic ideas.

  • Group polarization: The tendency of social groups to adopt ever more extreme attitudes or positions (typically toward moral or value-laden propositions).
  • Discovery Model of Morality: Individuals can and sometimes do discover moral truths without essential appeal to other people (e.g., via reasoning or intuition).
  • Negotiation Model of Morality: The truths of morality are the products of a fair negotiation.

We can understand the Negotiation Model in a number of different ways. Here are two.

  • Idealized Negotiation Model: The truths of morality are the products of an idealized (and hence non-actual) fair negotiation. At any particular time, there’s a fact of the matter about how this idealized negotiation would go – even if that fact is epistemically remote or inaccessible. And so there’s always a fact of the matter about (say) the morality of abortion – even if it’s epistemically remote or inaccessible.
  • Actual Negotiation Model: The truths of morality are the products of real, fair negotiations. Prior to such negotiations, there’s no fact of the matter about a moral issue. Moral facts are the products (creations, constructions) of such negotiations. One ought, therefore, to withhold judgment about non-negotiated moral propositions.

There are various bells and whistles one might add to the Actual Negotiation Model. But essential to it is that there’s no truth of the matter about not-yet-negotiated issues; and so one ought to withhold judgment about them.

Here’s the core of Arvan’s argument: The mechanisms of polarization work by amplifying disagreement. They turn small disagreements into big ones. Remove small disagreements and the mechanisms of polarization would stall. That’s what the Actual Negotiation Model, if fully implemented, would do. Take any potentially polarized issue. Prior to negotiations, there wouldn’t be any disagreement, since everyone would “withhold judgment” (13). And after fair negotiations, everyone would congeal around the view that was produced by those negotiations – not merely as an imperfect compromise, but as the truth about the issue. So again, with unanimity, there would be no disagreements for the mechanisms of polarization to amplify.


1. A Preliminary Worry: How Does the Actual Negotiation Model Work?

The Actual Negotiation Model creates consensus on potentially polarized issues. But then what’s to negotiate? People who don’t disagree don’t need to negotiate. In fact, it’s not clear that a substantive negotiation (as opposed to a “negotiation” that ends in immediate agreement) is even possible among people who don’t disagree.

Arvan describes a situation in which negotiators start with agreement about what counts as a fair negotiation but have “different priorities on the issue in question” (16). For example, on abortion, negotiators would start by placing different priorities on the value of protecting life and protecting autonomy. And then they’d negotiate by withholding judgment on the concrete issue (abortion) and negotiating over the more general issue (weighing the relevant values against each other). But this more general issue is a moral issue. Apply the Actual Negotiation Model to this general issue, and it would eliminate diversity by making everyone withhold judgment on it prior to a fair negotiation. So here’s the worry: Apply the model to every contested moral issue and it’s not clear how negotiations ever get started.

I’m going to assume there’s a straightforward way to dodge this worry. I can think of a few. They’re in the following parenthetical remark. But if I were you, I’d skip it so we can press on to more substantive concerns.

(Here are two potential solutions: (a) Negotiators start by withholding belief about (say) abortion but they have different hypotheses about the right way the issue should be resolved. And so they negotiate about that. (b) The Actual Negotiation Model might allow a small class of moral beliefs to conflict, such as perhaps general ones, e.g., how to weigh different values against each other.)


2. Is the Actual Negotiation Model Too Conservative?

On the Actual Negotiation Model, before fair negotiations everyone withholds judgment; after fair negotiations everyone takes the outcome of those negotiations to be not merely an imperfect compromise but the truth of the matter. The model undercuts mechanisms that amplify moral diversity (mechanisms of polarization) by eliminating moral diversity. But by eliminating moral diversity, the model also seems to undercut some pretty important mechanisms of moral change and progress.

I should hasten to note that the Actual Negotiation Model allows for certain types of moral progress. If a negotiation were only apparently fair and not genuinely fair, this would be grounds for changing a culture’s moral norms and practices. Or if a fair negotiation were based on descriptive, empirical falsehoods, this would also be grounds for changing a culture’s moral norms and practices. But the Actual Negotiation Model doesn’t allow for moral progress that’s driven by the knowledge that there are deep injustices in a culture’s norms and practices.

Here’s a way to appreciate this point: Would a culture where everyone embraced the Actual Negotiation Model allow for the existence of a Frederick Douglass, a Mary Wollstonecraft, or a Stonewall uprising? It seems that essential to being a moral beacon is being fired by conviction in the rightness of your cause. You can’t be a Douglass, a Wollstonecraft, or a person who resisted at Stonewall if you withhold judgment about the rightness of your cause or if you embrace as true compromised principles negotiated with your opponents, no matter how fairly arrived at.

A culture where everyone embraced the Actual Negotiation Model would be a profoundly conservative, change-averse culture. Take a culture that has (from our perspective) misguided norms and practices with respect to a group of people. Prior to fair negotiations, no one would have any beliefs about whether their norms are good ones or whether they can be improved. So it’s not clear what grounds anyone would have for demanding change. And after fair negotiations, once the norms and practices agreed to in those negotiations have been implemented, everyone would agree that these are the morally best norms and practices. Again, it’s not clear what grounds anyone would have for demanding change. The Actual Negotiation Model would produce a culture that lacked the potential for value-driven moral change.

The Actual Negotiation Model faces a dilemma. Does it make room for moral beacons, people who take strong action on principles that contradict a culture’s norms and practices? If not, that’s a problem that calls on us to consider alternative ways to handle polarization. If the Actual Negotiation Model does allow for the existence of moral beacons, then it must allow for the existence of moral diversity. And once you have moral diversity, what’s to prevent the mechanisms of polarization from doing their (often dirty) work?


3. Is There A Better Way to Handle Group Polarization?

The problem with group polarization isn’t the mere fact of disagreement. Some ways of achieving agreement are unwelcome. Political polarization in the US would disappear if everyone just agreed with the President’s tweets (were they consistent). And some disagreement is to be welcomed, such as when it helps enact needed social change. Group polarization tends to be harmful when people are seriously overconfident in their beliefs: when their level of justification doesn’t approach their level of confidence. So perhaps a better way to reduce harmful polarization is to eliminate the most egregious sources of overconfidence, such as:

  • Poor critical thinking about empirical claims: For example, people often oppose changing norms of practices on the basis of poor reasoning about consequences (e.g., permitting gay marriage will lead to bestiality).
  • Failing to think rationally about costs and benefits: For example, many people supported the war in Iraq because they thought it had net expected benefits. Even if this were the right conclusion to draw at the time (and I doubt it), just because something has a net expected benefit doesn’t mean it’s the right thing to do. There might be other obvious options with far better net expected benefits.
  • Poor moral reasoning: Deriving an ‘ought’ from an ‘is’ is always (or typically, depending on your philosophical predilections) a mistake.
  • Failures of epistemic virtue: The failure to be open-minded, curious, intellectually honest and courageous, etc.

I’m sure the reader can identify other sources of unjustified confidence (e.g., information bubbles). The suggestion here is that you can undo damaging group polarization by going after sources of overconfidence. The Overconfidence Model has three advantages over the Actual Negotiation Model.

  1. More Unified: Polarization arises for normative and descriptive claims (e.g., climate change, heliocentrism, evolution). The Overconfidence Model can explain both types of polarization in the same way: They’re the result of unjustified confidence. Reduce overconfidence and you’ll reduce unhealthy polarization. Unless we’re to adopt a form of the Actual Negotiation Model for descriptive claims (e.g., there’s no truth of the matter about climate change until after a fair negotiation about the issue), the Overconfidence Model provides a more unified and powerful explanation for unhealthy polarization.
  2. Healthy Diversity: If there is more than one way to reason rationally (and I think there is), then the Overconfidence Model can reduce harmful polarization while retaining some healthy diversity of opinion. But (for the reasons specified above) it seems that the Actual Negotiation Model, if fully implemented, can’t retain healthy moral diversity.
  3. More Feasible: It’s really difficult to convince someone who’s confident that abortion is morally wrong (or morally permissible) to withhold judgment on purely metaethical grounds. For example, I’m more confident that my views on abortion are roughly true than I am that my own metaethical views are roughly true. Getting me to be more confident in the Actual Negotiation Model than in my views about abortion seems a really difficult task. (And matters get even tougher if you replace abortion with the racial views at issue recently in Charlottesville.) There’s more hope for the Optimism Model. It’s not difficult to convince someone about sources of overconfidence: Just show them how those sources work on people they disagree with. The trick is to convince them that they’ve been influenced by those same sources of overconfidence. That’s a big job, to be sure. But it’s a job many of us have had some success with. It’s part of what we do when we teach philosophy.

Invited Comments from Hrishikesh Joshi (Michigan, Ann Arbor)

Comments on Marcus Arvan’s “The Dark Side of Morality: Group Polarization and Moral-Belief Formation”

Hrishikesh Joshi
University of Michigan – Ann Arbor


Marcus Arvan’s novel paper draws a connection between the way we think of moral truths and our tendency towards group polarization. Arvan argues that how we ordinarily think of morality – as consisting of facts to be discovered – could explain the tendency of groups to polarize around moral issues. Arvan contrasts this Discovery Model of morality with his preferred view, the Negotiation Model. On the Negotiation Model, moral facts consist of a compromise among parties who accept certain regulative ideals. That is, ordinary moral truths obtain in virtue of a compromise among people who are committed to, on Arvan’s substantive view, the ideals of coercion minimization, mutual assistance, and assuring equal bargaining power.

Arvan offers some reasons to think that the Negotiation Model, if adopted generally, would reduce the tendency towards group polarization. His strategy is to consider the three main drivers of polarization discussed in the politics and moral psychology literatures, and then argue both that i) the Discovery Model plausibly exacerbates polarization with all three frameworks, and ii) the Negotiation Model reduces polarization within the frameworks. Arvan doesn’t aim to produce a knockdown argument for this, nor a detailed empirical case – his goal is to rather make i) and ii) plausible, thus creating a stepping-stone for future research. While I think there is much to be said for the Negotiation Model in this regard, I want to offer two main points in response within these comments.

First, it would be interesting to see how this strategy maps on to group polarization with regards to non-moral, empirical claims. The example Arvan considers most closely is that of the morality of abortion. In this case, while there is little broad disagreement on the empirical facts of the matter – i.e. the various properties of the fetus, the economic and other costs of bearing children, etc. – there is strong disagreement on what morality demands. (Even this requires caveats; some of the parties in question might disagree on whether an unborn fetus has a soul, for instance. This disagreement would then be a non-moral one, though not quite empirical.)

However, not all issues on which there is polarized disagreement are like this. Consider the question of climate change, one of the most polarizing issues in modern politics (see PEW, 2016). While the issue of climate change does raise moral questions of intergenerational justice and population ethics, the chief disagreement in the modern context, I take it, is empirical. The disagreement is about the extent to which human activities contribute to global warming, and about the extent to which global warming will have severe consequences on humans. Both these points of disagreement are empirical. Some of the most well studied cases of social influence on individual judgments also involve empirical propositions. The famous Solomon Asch experiments (see Sunstein, 2002) revealed subjects’ tendencies to conform to group opinion; but the task at hand was to determine lengths of different lines – an empirical matter.

Now, the Discovery Model seems to be the only game in town in the case of such empirical facts – the facts about global warming themselves are not determined by negotiation (except in an indirect, causal way). This highlights the fact that Arvan’s strategy is limited to scenarios where there’s widespread agreement on non-moral facts, while there is disagreement on the moral import of those facts. However, it seems, a large portion of the issues on which there is partisan disagreement, involve parties not agreeing on the non-moral facts in question. As an additional example, consider the issue of minimum wage. Fiscal conservatives tend to believe that raising minimum wage will increase unemployment and thus will be bad for the poor. Fiscal liberals tend to disagree. This is, at heart, an empirical disagreement.

This raises a question for Arvan – when the different parties are negotiating as to whether some policy is the right thing to do – say the policy of setting a $15/hour minimum wage – are they supposed to agree on the empirical facts first? If so, the Negotiation Model will have very limited applicability in the context of modern political polarization. Even on an issue like the Israeli-Palestinian conflict, it will turn out that people on different sides of the issue consume different news sources, and thus are working from a different stock of empirical assumptions. It’s not obvious then whether negotiation in the purely moral domain can proceed. Further, it may turn out that there is no fundamental moral disagreement at all – the parties rather disagree on first order moral claims because they disagree on the empirical (historical, sociological, economic, etc.) facts of the matter.

Another issue that the defender of the Negotiation Model, qua way to decrease polarization, has to grapple with is that of determining how much of the polarizing tendencies are driven by deep features of our moral psychology – deep enough that the tendencies would survive attempts to move away from the Discovery Model. In the paper, Arvan lays out some of the research suggesting that polarization can be driven by people’s desires to impress their social group by, inter alia, “moral grandstanding,” (see Tosi and Warmke, 2016). Another driver of polarization is the tendency to disparage the out-group. Insofar as these propensities reflect deep psychological features, they can arise even if the Negotiation Model is accepted. For, in order to begin negotiation, one must think of the other party as being a reasonable interlocutor, committed to certain regulative ideals – for example: coercion minimization, mutual assistance, and equal bargaining power. However, claiming that the relevant out-group lacks these regulative ideals can be a way to impress one’s in-group, and to demonize the out-group. A proponent of the $15/hour minimum wage could thus claim that the opposing side does not care about the poor, and that a desire to have low (or zero) minimum wages reveals the lack of a commitment to assuring equal bargaining power. I want to remain neutral as to whether this is true, but want to flag that this kind of strategy will be available on almost any polarizing issue. That is, some of the group polarization dynamics can be maintained under the Negotiation Model by narrowing the group of parties who are considered to be reasonable interlocutors, having the appropriate regulative ideals. This is especially plausible against a backdrop of increasingly strong cross-partisan hostility in the U.S. and Europe (see Westwood et. al., 2017).

A different way to put the worry is this. Suppose that Haidt (2001) is actually correct in describing our moral psychology. That is, our moral judgments are formed through quick, intuitive processes, and then we rationalize our judgments through reason and argumentation. If this really is a feature of our psychology, produced as a result of millions of years of evolution, it seems that adopting the Negotiation Model just pushes the bump in the rug. Someone who believes in moral proposition P and accepts the Discovery Model will be disposed to give first-order reasons without referencing claims about social agreement (of the sort Arvan thinks are central) when challenged about P or asked to justify P. Someone who accepts the Negotiation Model, on the other hand, might defend P with different strategies – for example, by arguing that those who do not accept P lack the appropriate regulative ideals. In the end, a person’s accepting the Negotiation Model might just end up getting them to offer different sorts of reasons while making similar moral judgments.

The above considerations support some skepticism about whether the adoption of a meta- and normative ethical theory, such as the Negotiation Model, can in fact temper group polarization. That is, in the end there may be no way around the hard work required to generate institutional mechanisms for polarization reduction – such as promoting ideological diversity – and individual attempts at epistemic betterment – such as diversifying one’s information gathering processes, exposing oneself to the best counterarguments to one’s moral and political views, and so on.



Haidt, J. (2012). The righteous mind: why good people are divided by politics and religion (New York: Vintage Books).

—- (2001). ‘The emotional dog and its rational tail: a social intuitionist approach to moral judgment’, Psychological review108(4), 814.

PEW Research. (2016). “The Politics of Climate.” http://www.pewinternet.org/2016/10/04/public-views-on-climate-change-and-climate-scientists/ accessed 09/01/2017.

Sunstein, C.R. (2002). ‘The Law of Group Polarization’, The Journal of Political Philosophy 10 (2): 175–195

Tosi, J. & Warmke, B. (2016). ‘Moral Grandstanding’, Philosophy and Public Affairs 44(3): 197-217.

Westwood, S., Iyengar, S. Walgrave, S., Leonisio, R., Miller, L., & Strijbis, O. (2017). “The tie that divides: Cross-national evidence of the primacy of partyism.” European Journal of Political Research. DOI: 10.1111/1475-6765.12228

5 thoughts on “The Dark Side of Morality: Group Polarization and Moral-Belief Formation”

  1. This is just a quick note that I am currently unable to post responses to my two commentaries, as power is currently out at my hotel in Georgia (where I had to evacuate to because of hurricane Irma). I want to thank my commentators for their excellent sets of comments, and will post my replies as soon as possible!

  2. Reply to Mike Bishop’s commentary

    I’d like to begin by extending my thoughts and prayers to everyone affected by Hurricane Irma. As a resident of Tampa who had to evacuate to avoid the storm (I am writing these replies in a hotel in Georgia), it has been a real ordeal. Fortunately, Tampa appears to have been mostly spared from severe damage. However, not everyone was so lucky, especially those in South Florida and islands in the Caribbean. Please consider doing what you can to help those in need.

    I also want to apologize in advance if it takes me some time to respond to reader comments. Unfortunately, the internet at my hotel is somewhat unreliable (the power has been out all day), and in the next day or two my family and I will be making a long drive back to our home in Florida. I will do my best to respond to comments in a timely manner, and thank you all for your patience.

    I want to thank the Minds Online organizers for putting this conference together, and my two commentators for excellent sets of comments. Both sets of comments raise many good questions that I hope will serve at stepping stones to a good discussion. I will begin with my reply to Mike Bishop’s comments, and then discuss Rishi Joshi’s commentary in another comment to follow.

    In his commentary, Mike Bishop raises three major questions. Allow me to work through each one in order.

    Bishop’s first concern is about how the Actual Negotiation Model works. The way he glosses the model, because people are supposed to withhold judgment on (most) controversial moral issues until after a compromise is negotiated, there’s nothing to negotiate—as people don’t really disagree. More exactly, her worries the model “would eliminate moral diversity” by making everyone withhold judgment prior to negotiation.

    As I mention in my video, I had to abridge the paper to fit the conference’s length requirements—and unfortunately, this required cutting out long sections where I describe exactly how the Negotiation Model works. If anyone is interested, I’ve linked to an unabridged draft of the paper here. Here, though, is the short answer to Bishop’s first concern: the Negotiation Model doesn’t diminish moral diversity; it preserves it, just in a reconceived manner that promises to prevent polarization. On the Negotiation Model as I defend it in Rightness as Fairness, parties to actual negotiation must share regulatory ideals of coercion-minimization and mutual assistance, as well as a general commitment to fair bargaining (viz. the approximation of equal bargaining power – more on this below). Because the above ideals are merely regulatory ideals, all preferences consistent with them are to be considered legitimate to appeal to in bargaining. However, people do not need to share the same preferences, either before or after negotiating. For example, in the case of abortion, the Negotiation Model allows that some people may legitimately care more about the lives of fetuses than others, whereas others may care more about preserving women’s reproductive autonomy. The Negotiation Model at no point requires people to give up their different preferences. What it does require them to do is seek a common ground agreement with those who have different preferences, and then identify moral rightness with the outcome of that compromise agreement unless and until a new compromise is negotiated.

    This final point is important. Although the model does require people to identify moral rightness with the outcome of negotiated compromises, because negotiations tend not to fully satisfy people (due to differences in moral priorities), the model allows for future renegotiation. So, for example, suppose a public compromise on abortion is reached (say, norms permitting only abortion prior to fetal sentience). If as a Catholic this compromise does not satisfy me (thinking they have a soul at conception), the model allows me to try to renegotiate: I can try to convince others that fetuses have souls at conception, and so on. What it doesn’t allow me to do is form or appeal to beliefs about abortion being wrong on those grounds—as, again, the model holds that this is fundamentally the wrong way to think about morality: moral truths on the Negotiation Model cannot in principle be settled by discovering whether fetuses have souls, etc., because there is on the view no negotiation-independent fact of the matter of how conflicting moral priorities should be weighed against one another.

    In short, the Actual Negotiation Model doesn’t undermine moral diversity or render negotiation moot: it simply holds that we should reconceive the very nature of moral diversity and disagreement. Instead of seeing these things as disagreements over moral facts that can be discovered (viz. the rightness or wrongness of abortion), the model holds that we should understand moral diversity and disagreement in terms of recognizing that people can have morally legitimate differences in preference both prior to and after negotiation.

    Now let me turn to Bishop’s second worry, which is that the model may be too conservative. In brief, Bishop’s worry is that the model doesn’t allow for moral progress driven by the knowledge that there are deep injustices in a culture’s norms and practices. As Bishop puts it, “Would a culture where everyone embraced the Actual Negotiation Model allow for the existence of a Frederick Douglass, a Mary Wollstonecraft, or a Stonewall uprising?”

    Here again, it is somewhat unfortunate that I had to abridge the paper. For it is a central part of the Negotiation Model that it doesn’t permit negotiating with views or preferences directly antithetical to ideals of coercion-minimization, mutual assistance, and fair bargaining. The reasons why the model doesn’t permit negotiating with these perspectives is complex, but in a nutshell in Rightness as Fairness I argue that morality reduces to a form of diachronic rationality—specifically, a recursive set of interests we have to justify our actions to all of our possible future selves, so as to “safeguard” ourselves in future against outcomes (even unlikely ones) that could undermine the future pursuit of our ends. I then argue that because views and preferences that aim to dominate people (e.g. slavery, racism, sexism, etc.) by their very nature aim to undercut the ability of some people to obtain their ends (people whose ends one’s possible future selves can identify with), such views and preferences cannot by their very nature be the subject of a consensus agreement that can satisfy all of one’s future selves. Such views and preferences are therefore not to be negotiated with, but instead fought against on the Negotiation Model—just as people like Frederick Douglass, Wollstonecraft, the Stonewall uprising, and so on, all did.

    Indeed, it is worth noting here that I have previously published work examining how the Negotiation Model should be applied to injustice. In Rightness as Fairness I understand moral rightness in terms of actual or hypothetical fair negotiation. In my 2014 article “First Steps Toward a Nonideal Theory of Justice and my 2008 dissertation A Non-Ideal Theory of Justice, I show how injustice should be addressed in the latter way: through a hypothetical model—a “nonideal original position”—that models a fair procedure for addressing injustice. In these earlier works, I argue that this hypothetical procedure justifies principles of nonideal justice that require identifying just social progress with the aims and collective decisions of open and inclusive social reform groups that aim to give preferential bargaining weight to classes of people in proportion to the severity of the injustices they suffer. As I argue in “First Steps…” and my dissertation, these implications fare well in reflective equilibrium, both (A) fitting broadly with how many of us already intuitively understand the justice of social progress, including (I think) the views of people like Douglass, Wollstonecraft, and those who took part in Stonewall and other social uprisings, while also (B) providing substantial and important normative directives for social movements (such as requiring openness and inclusivity rather than exclusivity).

    Bishop’s final worry is that there may be other, better ways to prevent polarization. Specifically, he suggests an Overconfidence Model which he thinks might be more unified, better respect moral diversity, and be more feasible.

    It is of course possible that the Overconfidence Model would be superior in these ways—and it is ultimately an empirical question. However, I think there are reasons for skepticism. One interesting (and well-known) finding is that people have a pronounced tendency to think they are morally superior to others. My hunch is that a big part of why people think they are morally superior—why they are morally overconfident—is that they believe they have discovered moral truths. After all, if you have the “right” view, then everyone who disagrees is thereby wrong. There is also empirical evidence that people reason poorly in the ways Bishop mentions (viz. poor critical thinking about empirical claims, failures of epistemic virtue, etc.) because of their moral and political views, engaging in what is known as motivated reasoning (a very pervasive and well-established phenomenon).

    In other words, I suspect Bishop’s Overconfidence Model has things exactly the wrong way around. It is not that people arrive at bad moral views because they are overconfident. Rather, they are overconfident (and engage in bad critical thinking, motivated reasoning, etc.) because they think they have discovered the “right” moral and political views. Consequently, the Negotiation Model may not only be the best solution to polarization; it may also, by extension, be a unified solution to Bishop’s forms of overconfidence—leading people to engage in fewer polarizing behaviors by undermining the kind of overconfidence people tend to develop when they believe they have made a “moral discovery.” The Negotiation Model might, in other words, be the best means of preventing overconfidence and polarization in one fell swoop.

    Now, Bishop is almost certainly right that it may be very difficult to get people to reconceptualize morality—and change their moral psychology—to fall in line with the Negotiation Model. As Bishop notes autobiographically, “I’m more confident that my views on abortion are roughly true than I am that my own metaethical views are true.” However, I’ll lay my cards on the table here: I don’t think significant changes in moral thought or behavior are supposed to be easy. For thousands of years, many people basically took it for granted that morality comes from God (viz. Divine Command Theory). If history is any indication, it wasn’t easy at all, socially or psychologically, to get people to give up this view, and to this very day many religiously-inclined view still subscribe to it. Still, for all that, it’s a meta- and normative-ethically bad view—and, I think most of us know, a harmful one (insofar as it is predictably vulnerable to dogmatism, extremism, etc.). By a similar token, however natural the Discovery Model may be—and however sure people may be of their moral beliefs (on abortion and the like)—if my arguments are correct the view is, for all that, diachronically irrational and harmful, and must be unlearned (however difficult that may be to do). For let us reflect briefly on what plausibly underlies both (1) morally abominable views (e.g. racism, sexism, etc.) as well as (2) just about every miserable, violent conflict (ranging from the Israeli-Palestinian conflict to the “war on terror”): people thinking that they, and they alone, have “discovered the moral truth.” If I am right, the Discovery Model may seem attractive, it may be seductive, and it may be difficult to unlearn—but unlearn it we should. The Discovery Model may be natural, but if I am right it makes us into enemies, in ways that undermine our own ability to achieve our own goals over time.

    Finally, there are a few reasons to think the Negotiation Model is feasible. First, in Rightness as Fairness I derive the model from claims about moral motivation and cognition that I argue have substantial empirical support (for a similar take on the empirical picture, see Eleonora Vigano’s recent paper “Adam Smith’s Theory of Prudence Updated with Neuroscientific and Behavioral Evidence”). Although I argue it takes detailed philosophical argument to see precisely why the Negotiation Model is favored by our best moral psychology (and instrumental rationality), I argue that the model is wholly grounded in actual human motivation. Thus, it should be possible for people to conform to the model, at least with some thought and effort. Second, on that note, although my own experience putting the model to practice is largely anecdotal—comprised by the model’s effects on the behavior of myself and my spouse (who I’ve largely convinced of the model)—my experience is that the Negotiation Model is not only surprisingly easy to conform to once one accepts the view, but can also dramatically transform one’s perspective in ways that lead (as the present paper suggests) to better, more mutually-beneficial interactions with other people. Before accepting the model, I too had strong views about all kinds of moral issues: abortion, gun control, and so on. Now that I accept the model, my views have basically wholly changed: I sincerely think all of these issues must be settled through negotiation. I’ve also been largely able to convince my spouse (a PhD candidate in psychology) of the Negotiation Model, and the effects on our behavior toward each other has been noted by both of us. Although we still struggle sometimes, we’ve both noted that we now spend a whole lot less time arguing over things—on everything ranging from what is “right” with respect to my career, or our financial dealings, etc.—and a whole lot more time trying to arrive at mutually acceptable compromises. Again we still struggle sometimes, falling back on our own “moral discoveries”, thinking we have the “right” answer and the other person the “wrong” answer. But, for all that, the more we’ve internalized the Negotiation model (and we’ve both worked at it pretty consciously) the more we’ve been willing to forge compromises on matters of disagreement that we both regard as fair both to ourselves and the other person, given our different priorities.

    In any case, I thank Bishop for raising good concerns and proposing an alternative hypothesis (the Overconfidence Model). I think it would indeed be interesting to systematically test which hypothesis—the Negotiation Model or the Overconfidence Model—would best prevent polarization and overconfidence, and I hope these and other broader questions will begin to be seen as important research programs to pursue further.

    Thanks again, Mike, for your excellent commentary. I look forward to hearing your thoughts, and to a good discussion!

  3. Reply to Rishi Joshi commentary

    I’d like to thank Rishi Joshi for his excellent set of comments, which raise a number of important empirical questions about the Negotiation Model. Let me address each of his questions in turn.

    Joshi’s first question concerns how the Negotiation Model maps onto group polarization with respect to non-moral, empirical claims. As Joshi points on, on many moral issues (including climate change) there is broad, polarized disagreement on basic empirical facts—as people on different sides of the issue visit different news sources, are presented with different stocks of empirical assumptions, and so on. Joshi notes that when it comes to empirical facts, the Discovery Model is the only game in town—as empirical facts are of course things to be discovered. Yet here is Joshi’s concern: when people on different sides of a moral issue disagree on basic empirical facts, it is not clear how proper moral negotiation can proceed—and indeed, it may turn out that there is no fundamental moral disagreement at all, but rather merely disagreement over first-order empirical facts.

    This is a good concern. However, I believe the Negotiation Model has resources to address it, thanks to what we know about the causal bases of empirical polarization. As I mentioned in my response to Michael Bishop’s commentary, there is ample empirical evidence that people reason poorly about empirical facts—paying selective attention to information (on things like climate change, etc.)—because of their moral and political views. Notice that this is entirely intuitive. If I believe I have discovered the moral truth that the economy matters more than protecting the environment, then I may be apt to engage in motivated reasoning when collecting non-moral, empirical facts: I will be on the lookout for “information” that supports the moral or political view I hold, “bending the facts” to that view instead of paying proper attention to information that does not support my view. For these reasons, I think that although when it comes to empirical facts themselves, the Discovery Model is the only game in town (empirical facts must be discovered), the Negotiation Model of morality promises to help people reason better about empirical facts. For, insofar as the model requires people to remain agnostic about controversial moral issues prior to negotiation—such as whether jobs and the economy should matter more than protecting the environments—it should lead people to engage in the motivated reasoning described above less. If one does not have a settled moral view at all (about, say, the importance of protecting the environment), there may be less of a temptation to “bend the facts” to fit one’s own view—and one may be more open to different sources of information.

    Now, of course, this is an empirical question. Will the Negotiation Model, in preventing moral polarization, thereby also substantially lessen empirical polarization? I have just given some reasons to think that it might—and I take it we can all agree it would be a good thing if it turns out to do so. Still, the hypothesis must be tested, and for that reason I think Joshi’s very good question further indicates how important it is to pursue these types of questions further in future empirical work.

    Joshi’s second concern is that the Negotiation Model needs to grapple with whether and to what extent polarization is driven by deep features of the moral psychology—features that would survive attempts to move away from the Discovery Model. As Joshi notes, to begin legitimate negotiation on my view, the parties to negotiation must think of the other side as being reasonable interlocutors committed to regulative ideals of coercion-minimization, mutual assistance, and equal bargaining power. Alas, insofar as claiming the other side doesn’t have these ideals is a plausible way to impress one’s in-group and demonize the out-group, there may be a tendency—even on the Negotiation Model—for polarizing behaviors to still arise, just in a different guise than before. Joshi appends this concern by citing Haidt’s theory that our moral judgments are formed through quick, intuitive processes which we then subsequently rationalize. Given that these features of our moral psychology plausibly evolved through millions of years of evolution, Joshi worries our polarizing tendencies may be unavoidable.

    These are excellent worries, and because empirical matters here are unsettled I do not have anything like a knock-down reply available. However, I do think there may be some reasons for optimism.

    First, while there may always be some temptation or tendency to regard one’s moral opponents as unreasonable—as either not committed to regulative ideals of coercion-minimization, mutual assistance, and fair bargaining—we can (for reasons I examine in more detail in Rightness as Fairness) establish something like a standing public presumption that a moral or political view is consistent with these regulative ideals, a presumption which we should drop only in cases where there are no plausible arguments that the view is consistent with them. For example, consider Joshi’s example: disagreement over the moral desirability of a $15/hour minimum wage. Standard arguments against raising the minimum wage are that it would further coerce employers to pay more than they wish or can afford (especially small businesses with limited resources) and result in higher unemployment (thereby failing to help people in need of jobs and income). These arguments are prima facie consistent with ideals of coercion-minimization and mutual assistance, and there is no immediate reason to think sincere proponents of the above claims seek to deny bargaining power to workers (as the claim is that there is no feasible way to improve the bargaining power of some workers by raising the minimum wage without dramatically lessening the bargaining power of others). Thus, I think that as long as we develop strong norms of presumptively negotiating—unless and until it is very clear that the other side’s preferences are motivated by forms of animus that cannot be plausibly defended as consistent with coercion-minimization, mutual assistance, or fair bargaining (such as racism, sexism, etc.)—the Negotiation Model may be put into practice in a manner whereby we can consciously work to overcome the polarizing elements of our moral psychology to a substantial extent. Finally, and importantly, there is empirical work showing that something like this is possible. As I note in the paper, Brewer (1996) and others have found that groups committed to ideals of negotiation develop less polarization and a greater willingness to cooperate productively.

    Finally, on that note, I want to say a few things about Haidt’s (2001) Social Intuitionist Model of moral psychology—as I have received a lot of questions about it from other readers. Haidt’s theory suggests that people tend to moral beliefs through quick, automatic processes—and that we then tend to rationalize those beliefs after the fact through moral reasoning. There are, however, several open questions here. First, it is an open question whether our moral-belief forming processes really are as automatic as Haidt makes them out to be. Haidt doesn’t claim that we always form moral beliefs intuitively, just that we normally do. But of course I recognize that as well. I think people tend to form moral beliefs intuitively because we tend to think moral truths are the sort of thing we can discover. So, two further questions immediately arise. The first is whether the belief-forming mechanisms Haidt describes are really “deep” or hard-wired. Could they perhaps instead be at least partly an artifact of social learning (which might therefore be undone or preventable through socialization to the Negotiation Model)? Second, even if our initial moral-belief forming mechanisms turned out to be deep, could it be that we could override those mechanisms through conscious effort, unforming moral-beliefs after we initially form them automatically out of a conscious attempt to conform to the Negotiation Model?

    I think there are two related reasons for optimism here. The first reason is merely anecdotal, but (I think) relevant. As I mentioned in my response to Mike Bishiop, I used to have very strong moral beliefs about all kinds of things: about abortion, gun-control, and so on. At the time, I thought these beliefs were well-justified by moral reasoning—and, if Haidt is right, they were probably the result of automatic, intuitive processes. But what of it? As I mentioned in my response to Bishop, I now reject many of those moral beliefs I once held. I now sincerely believe that abortion and gun-control norms fundamentally need to be negotiated by people with different moral priorities. And why? It appears that, in coming to accept the Negotiation Model, I was able to undo the moral beliefs I had previously formed automatically. Thus, even if Haidt is right that we have a tendency to form moral beliefs intuitively, it seems plausible—and is an open empirical question—whether we can overcome these tendencies with conscious effort.

    Second, although Haidt’s Social Intuitionist Model is based on empirical research, so too is the Negotiation Model. In Rightness as Fairness, I derive the Negotiation Model from actual human motivations—specifically, from prospective concerns about our own possible futures (including capacities for transformative empathy) that an increasing body of empirical evidence indicates (1) normal, morally responsible human subjects share, (2) are systematically implicated in moral cognition and behavior, and with which (3) subjects with compromised capacities for moral responsibility (including psychopaths, children, and teenagers) show substantial deficits. In other words, if my arguments in Rightness as Fairness are correct, a set of actual human motivations—a substantial part of our moral psychology not examined by Haidt—makes the Negotiation Model diachronically rational. Accordingly, while it may take conscious effort to act rationally vis-à-vis these motivations—overcoming whatever automatic moral intuitions we may tend to have, and conforming to the Negotiation Model instead—I see no reason to think that human beings are incapable of it. While these are admittedly unsettled empirical questions, I think they once again highlight the importance of pursuing this research program further.

    In conclusion, I thank Joshi once again for his excellent comments. He has given me a lot to think about, and I think he pushes back against my view at precisely the right points. As I explain above, I think a lot of the issues here are empirical ones—and I hope that further empirical work will shed greater light on them moving forward.

  4. Thanks for the papers, comments and discussion. Two quick thoughts, one on the target and one on Joshi’s comment.
    (1) I was very sympathetic to the initial claims, and the psychological literature motivations for adopting them. My question is how to put this into the existing literature. In particular, to what extent is this Negotiation Model a specific formulation of first-order or metaethical constructivism? As far as I can tell, the Negotiation Model is a particular way of formulating a non-ideal constructivist method for normative claims.
    (2) To RJ, with apologies for not being a theoretical comment. Responding to the following: “While the issue of climate change does raise moral questions of intergenerational justice and population ethics, the chief disagreement in the modern context, I take it, is empirical. The disagreement is about the extent to which human activities contribute to global warming…” As someone who works on climate change, this seems like an odd point to make. While for some of the public, the disagreement might be empirical, the scientific view is straightforward (actually, this exact point is covered in ENV200 at Princeton!): more than 100% of the observed warming over the past few decades is due to anthropogenic forcing. The reason is that the vast majority (>90%) of the excess energy is absorbed by the oceans, so our observed warming is far less than the actual capacity for warming. In other words, the anthropogenic forcing that affects observed surface-level or ground-level warming is only a small fraction of the actual warming potential of energy produced anthropogenically. So yes, I don’t doubt that people disagree about this empirical fact, but that is not because it isn’t known.

  5. Hi Kian: Thanks for your comment! Yes, I think that’s probably the right way to fit the view into the literature. Standard constructivist views (Kantian, Humean, etc.) conform to the Discovery Model, holding that we can discover moral truths or justify moral beliefs by thinking through some procedure of moral construction (viz. deriving moral truths from the procedure). The Negotiation Model agrees that moral truths are constructed by a procedure, but holds that we cannot arrive at moral truths by merely thinking through the procedure (actual compromises must be negotiated instead). In short, it is a form of constructivism, but a unique form of it that should reorient moral-belief formation.

Comments are closed.