You’re Probably Wrong: Group Polarization and Going to Extremes

In Going to Extremes, Cass Sunstein paints a grim picture. Men are prone to a variety of serious failings, and those failings only intensify through the processes of groups. Bolstering his argument, Sunstein points to numerous studies that have revealed humanity’s flaws. Stanley Milgram, for example, demonstrated that a majority of recruited participants could be convinced to administer what they thought were dangerously high shocks to an actor who responded with increasingly dramatic expressions of pain. “What Milgram revealed,” Sunstein writes, “is that the heuristic – in favor of obedience of apparently trusted authorities – does not always work well. In real-world cases, it leads to terrible moral errors” (Sunstein, 2009).

Philip Zimbardo’s Stanford Prison Experiment similarly seems to reveal humanity’s darkness. Healthy, average participants were randomly assigned to serve as ‘prison guards’ or ‘inmates.’ Within days, the guards displayed “growing cruelty, aggression, and dehumanization,” while the prisoners – after an initial attempt at uprising – were crushed; becoming “subdued and ‘zombie-like’” (Sunstein, 2009). The simulation had grown so dire and grotesque that the experiment had to be ended early. After just 5 days. Sunstein summarizes the lessons learned from this dark look into human nature:

In pointing to the apparent normality of those involved in Nazi war crimes, Zimbardo gives a social science twist to Hannah Arendt’s claims about the ‘banality of evil.’ And in explaining what makes atrocities possible, Zimbardo places a large emphasis on deindividualization – a process by which both perpetrators and victims become essentially anonymous and are thereby transformed into a type or a role. (Sunstein, 2009)

For Sunstein, these studies highlight a deeper challenge. Human beings are embedded in a social context, and that context serves as a significant driver of individual actions and opinions. Hearing friends express a view makes a person socially inclined to express the same view. Deliberating groups tend towards extremism in the direction of the pre-deliberation median because nobody wants to take the social risk of expressing an unpopular view. College students playing prison guards give each other permission – or may even encourage each other – to act in increasingly horrific ways. We each take our cues from the social context we’re embedded in; a problematic heuristic because the signals we receive are so often morally or factually wrong.

This presents a potentially paralyzing conundrum: if your own perceptions and opinions are merely a product of your social environment, how can you ever know what is truly good or right? Sunstein offers a small prescription of hope, arguing that “many human beings are able to resist situational pressures and to engage in forms of heroism. Even when group polarization is under way, some people, some of the time, will hold fast to their convictions and stay where they are, especially if group members go in destructive or violent directions” (Sunstein, 2009). Sunstein further argues that the policy prescription of ‘checks and balances’ serves as a bulwark against polarization. Indeed, “the institutions of our Constitution reflect an implicit fear of polarization, creating a range of checks on potentially ill-considered judgments.” For example, the constitution explicitly denies the president’s power to declare war, thus ensuring that a single person can not “do so without sufficient deliberation and debate among diverse people” (Sunstein, 2009).

It is reasonable to think that checks and balances provides some protection against polarization, yet the idea of deliberation “among diverse people” is laughable coming from an era when ‘diversity’ consisted entirely of the diversity between straight, white, property-owning men in cities and straight, white, property-owning men in rural areas. This narrow notion of diversity points to a significant oversight in Sunstein’s work: he puts a lot of attention on political diversity while giving very little thought to other forms of diversity. For example, Sunstein notes that “confident people are more prone to polarization” but he gives little attention to the constant social admonishment women receive for not being confident enough (Kay & Shipman, 2014). Perhaps, then, it should not be surprising that groups with higher numbers of women perform better at a range of tasks than equal or male-dominated groups (Woolley, Chabris, Pentland, Hashmi, & Malone, 2010). This isn’t because women are smarter or better, but because women are more likely to be socialized for group problem-solving. Indeed, Woolley et al. find their result to be “largely mediated by social sensitivity,” a skill which the women in their study displayed more strongly than men. Woolley et al. further argue that “groups, like individuals, do have characteristic levels of intelligence.” People socialized for group processes, then, tend to make groups smarter while individuals socialized with the destructive features of toxic masculinity – such as over confidence in their individual perspective and brash confrontation with any form of dissent – make for less productive groups. We don’t need individualistic heroes who “hold fast to their convictions” no matter what; we need thoughtful collaborators ensure that a variety of voices are heard.

Rather than reveal the risks of deliberation, these anecdotes highlight exactly why group deliberation is needed. Sunstein argues that the Stanford Prison Experiment exposes the ‘banality of evil,’ but Arendt didn’t mean this phrase the way Sunstein interprets it: that humanity’s evil is mundane. Rather, as Arendt writes, evil “possesses neither depth nor any demonic dimension. It can overgrow and lay waste the whole world precisely because it spreads like a fungus on the surface. It is ‘thought-defying,’ as I said, because thought tries to reach some depth, to go to roots, and the moment it concerns itself with evil, it is frustrated because there is nothing. That is its ‘banality’ (Arendt, 1963).” Evil, in its empty banality, cannot survive the rigors of reasoned thought and debate; it thrives when these habits are suppressed, when there is mere lip service hollowly lauding diversity. This is exactly why Bernard Manin argues for the normalization of debate as an essential feature of deliberation (Manin, 2005). As a political ideal, debate, isn’t about finding out whether you and I agree; it is about considering all possible reasons, all possible perspectives, and then co-creating solutions which none of us could have accomplished on our own. That is, as Dewey calls it (Dewey & Rogers, 2012), the essence of democracy as a way of living.


Thanks to Joshua Miller for the Arendt quote
Arendt, H. (July 24, 1963). [Letter to Gershom Scholem].

Dewey, J., & Rogers, M. L. (2012). The public and its problems: An essay in political inquiry: Penn State Press.

Kay, K., & Shipman, C. (2014). The confidence gap. The Atlantic, 14, 1-18.

Manin, B. (2005). Democratic Deliberation: Why We Should Promote Debate Rather Than Discussion. Paper presented at the Program in Ethics and Public Affairs Seminar, Princeton University.

Sunstein, C. R. (2009). Going to extremes: How like minds unite and divide: Oxford University Press.

Woolley, A. W., Chabris, C. F., Pentland, A., Hashmi, N., & Malone, T. W. (2010). Evidence for a Collective Intelligence Factor in the Performance of Human Groups. Science, 330(6004), 686-688. doi:10.1126/science.1193147



One thought on “You’re Probably Wrong: Group Polarization and Going to Extremes

  1. Kevin Dye

    You write: “Deliberating groups tend towards extremism in the direction of the pre-deliberation median because nobody wants to take the social risk of expressing an unpopular view.” This seems like a paradox – going towards an extreme would not seem to be going towards a median. My understanding of this effect from the literature is that unstructured deliberation tends towards an extreme, while structured deliberation tends towards an average.


Leave a Reply

Your email address will not be published. Required fields are marked *