When do we need civilizational glue to bind us closer together? When do we need to proactively promote and protect cognitive diversity? When I want to ask such questions, I pose them to Cass R. Sunstein. This present conversation focuses on Sunstein’s book Conformity: The Power of Social Influences. Sunstein is the Robert Walmsley University Professor at Harvard Law School, where he founded the Program on Behavioral Economics and Public Policy. From 2009 to 2012, Sunstein served as the Administrator of the White House Office of Information and Regulatory Affairs. His recent books include Can It Happen Here?: Authoritarianism in America, The Cost-Benefit Revolution, How Change Happens, and On Freedom.
ANDY FITCH: People who have attended American schools and absorbed American popular culture might bring negative connotations to the topic of conformity (itself perhaps a conformist response). So could you first flesh out a basic human impulse to conform — not only in the most regrettable cases of assaults on perceived rival groups and persecution of nonconforming individuals, but also in the more benign sense of an everyday civilizational glue allowing us to assemble some constructive, collaborative conception of what we believe true and what we believe right?
CASS R. SUNSTEIN: When people want to know what they should do (like run from a threat, or pack an umbrella, or smile or scowl), they look at what others do. Conformity has two underlying justifications. First, what other people do provides relevant and helpful information. When you see people running out of the open fields, you can infer that a big thunderstorm is coming. And second, conformity helps us get along together. If you do what other people do, all else being equal, this will probably benefit you, by keeping you in their good graces (not always, but often). So if everybody plans to wear formal attire to the party, it benefits your own self-interest to know this in advance, and probably to do the same.
Here your book frames itself as offering a description, rather than an evaluation, of conformity. Nonetheless, it does point to foundational problems brought about by conformist thought and behavior. In terms of informational dynamics, conformity can distract us from and deprive us of the information that we most need (both as individuals and as groups) in order to make the best decision. In terms of reputational dynamics, conformity can coax us to free-ride, to come across as cooperative “good sports” rather than selfish standouts — even if maintaining such status requires us to withhold some valuable contribution that could have helped the group achieve better outcomes. So before we get to any specific social, political, legal circumstances, could you speak more generally to why a collectivity might want to support certain types of nonconformity: not solely to protect individual rights per se, but to promote the group’s best interests?
Right this book, on balance, does offer a plea for allowing room (a lot of room) for nonconformists. It acknowledges many good reasons for conformity, but ultimately concludes that societies suffer if they don’t create significant protections for those who go in different directions. Many types of groups (a company, a church, a labor union, a White House administration) need the kinds of information that nonconformists provide. A giant Silicon Valley social-media firm needs to allow its people to speak up when they sense a certain policy leading the company down the wrong path. That type of dissent benefits the company. It gives the company a sense of what it might not know.
I worked in President Obama’s White House. President Obama insisted on allowing and even encouraging the expression of many competing views. That helped us avoid a lot of mistakes — because sometimes the minority position actually turns out to have it right. Any well-functioning organization establishes norms of openness precisely in order to correct its own leaders’ or its majority’s mistakes. A great commentator on World War Two, Luther Gulick, observed that while many people expected dictatorships to outfight democracies, given the efficiency of top-down chains of command, the reverse ended up happening. Democracies (even within the military) felt freer to correct their mistakes, which provided a big safeguard. And what I personally observed in the White House, and what analysts have observed on a very large scale about World War Two, holds true even for little organizations. If you create a culture of conformity you’ll miss much information that could have prompted midcourse corrections or prevented disasters.
In your book’s basic terms, conformity helps to breed stability, and social cascades bring forth change. Yet cascades themselves exemplify certain conformist tendencies. Could you sketch, for instance, the self-amplifying processes by which an information cascade prompts participants to stop relying on their private assessment, and instead to treat other people’s public signals as a reflection of deliberative judgments worth emulating (and, in doing so, to send their own equivalent public signaling — prompting yet more conformists to follow suit)?
For one glaring example, the rise of Nazism came about in part through an informational cascade, at a moment when a bunch of Germans didn’t know what to think. Prominent voices told them to support Hitler and to persecute Jews. Many people believed in this cause. But many also went along primarily because signaling from fellow citizens gave a lot of weight to these policies — just by virtue of the numbers. As more supporters then continued to join these groups, the groups just sucked in more and more people. That’s not an example we should applaud.
But within the US Civil Rights Movement during the 1960s, Martin Luther King had equivalent successes as a norm entrepreneur harnessing a social cascade that brought in millions of people (though still not as many as one might hope). That movement gathered and distributed important information about injustice happening in the US. And as increasing numbers of Americans joined this movement, some of their fellow citizens thought: Well, so many people have joined with King. I probably should do the same. And once they did, you started to have a large swath of not only African American but also Hispanic and white supporters, and it became that much harder to resist the push for the Civil Rights Act of 1964, and the Voting Rights Act of 1965 — with the passage of both coming from cascades.
More recently, we’ve found ourselves within an information cascade on climate change. The vast majority of us have no expertise in the science. We don’t really grasp many of the details. We have to rely on trusted specialists. But our sense of this crisis and its severity keeps getting multiplied by the increasing numbers of people who seem to agree — sometimes culminating in significant policy change, including in the US under President Obama, or certainly in Germany under Chancellor Merkel, who has made addressing climate change a top priority. To some extent, even China has done so (again, as in the US, not as much as you’d like to see). And for each individual person, the actual increase of knowledge on a given topic might be modest. But at this point, within a matter of days, new information can reach tens of thousands or hundreds of thousands of concerned citizens.
With #MeToo as well, we’ve seen an accelerating social cascade, with people joining a movement that many previously wouldn’t have elected to join, or wouldn’t have felt comfortable joining — all based on informational signaling they see coming from others.
Among conformity’s basic mechanisms, belief polarization also stands out, particularly in our present age of demographically inflected partisan divides. But I find most intriguing in your account of belief polarization the implicit intra-group homogenizing pressures, rather than the explicit inter-group rivalry. So could we take the fact that self-consciously unified social groups tend towards convergent beliefs, that individuals within such groups tend towards increasingly extreme and unflinching beliefs, and could you outline certain ways in which the incredulity, the outrage, the fear that one belief-polarized group feels might come about through its own internal dynamics — as much as through its opponents’ provocations?
I’d describe it as even more dynamic than homogenization because, as you said, like-minded people typically push each other towards more extreme perspectives than their pre-deliberation tendencies would have suggested. When you bring together a group of people who think President Trump probably should be removed from office, and these individuals start talking with one another, they’ll likely leave that discussion convinced we should have removed President Trump for sure — yesterday. Or a group of individuals who suspect we probably shouldn’t remove President Trump from office might talk each other into the uniform position that this whole impeachment process is a hoax and a dangerous plot devised by terrible people.
In either case, when individuals predisposed to certain beliefs come together, they will typically exchange information that further supports those inclinations. At the same time, they won’t hear many perspectives pushing in opposing directions. They’ll listen to each other, and they’ll end up thinking a more extreme version of what they originally thought. They’ll make the group as a whole more extreme, and they’ll also (even when not starting off as a close-knit group) keep each other in line. They’ll want to stay in each other’s good graces. They’ll shift a bit as individuals to uphold this dominant tendency. You might at first have some right-of-center people who worry a lot about climate change, or who welcome immigration. But they’ll silence themselves or even shift their own beliefs to avoid becoming an outlier or “traitor.”
In reference now to how our Constitution might proactively provide for and preserve cognitive diversity, why might 21st-century Americans want to think of our republican (not purely democratic) traditions, our elected officials’ representative deliberation (with its reflective attempt to find the best solution for all, rather than some rigid commitment to the expressed will of a given district’s constituents), our federalism and separation of powers and bicamerality, not just as insulating elite interests from the capricious “mob,” but as protecting all of us from some of the more problematic social influences that this book traces? And with this emphasis on diffused decision-making power taking on less positive connotations (amid, say, present-day concerns of a dysfunctional American vetocracy, in which polarized parties and vested interest-groups have sufficiently gamed the system to ensure that nothing much gets changed), how might you persuade skeptical citizens to maintain faith that democratic decision-making still can offer something closer to an Aristotelian mean than to some schemer’s bait-and-switch?
Whether you like President Trump or President Obama or President Bush or President Clinton, our constitutional system historically has safeguarded against developments that many of our citizens would have loathed. Congress needs to pass laws or at least agree to allow certain executive initiatives. The House and Senate have many opportunities to check any especially threatening political movement.
Again, it’s no accident that some of Hitler’s first acts as German Chancellor involved eliminating checks and balances. We likewise have a system where our House checks our Senate, and our Senate checks our House, and our president checks Congress, and Congress checks our president. We have courts which can and often do strike down laws and executive policies, both under Republican and Democratic presidents. And if the White House or the Washington apparatus moves in an undesirable direction, the states likewise can counter that. California has taken many significant steps lately on climate policy and environmental protections, providing an important check against White House rollbacks or Washington inaction. You might not agree that climate change deserves the attention California has given it. But you nonetheless might affirm that in a free, diverse nation like ours, Mississippi or Massachusetts or Ohio or California sometimes has to go in its own direction, to safeguard against mistakes from the center.
Returning then to how to structure group dynamics to produce the best decisions, how does rewarding group success (rather than rewarding comparative individual advantage) incentivize full disclosure — providing the most complete and accurate and actionable picture? And by contrast, how might rewarding individual conformity produce the opposite?
Suppose you have a group deciding whether to create a new product, or to launch a startup, or to go in one policy direction rather than the other. Suppose this group has eight people. If life plays out as it often does, every one of these eight people will have something to contribute. Each individual will possess slightly different information, and have a distinct angle on the overall situation. Though then this group’s leader or most powerful or most credentialed person will indicate a view — and, as a result, the information elicited from all other group members will take on less weight. People will feel a bit frightened of the leader. Or people might assume the leader knows best, even if they should have good reason to think otherwise.
I mean, when I worked in the White House, the day after the US Senate confirmed my nomination, I noticed my staff, with so many talented people, acting like I’d suddenly become the wisest person in the room. Obviously, I had not. But when I’d offer even just an initial inclination towards one view, my excellent staff responded with disproportionate enthusiasm to this insufficiently informed provisional statement of conviction. This signal from the group led to my sense that, as the leader, I’d better shut up and let them talk first.
So I had to learn that lesson in real time, but a bunch of empirical research also confirms leaders have this power. As a result, influential people in the group often do best by staying as quiet as possible, and creating a culture of receptivity — not just because they want everybody to feel happy and respected (though that helps too), but because they want the group to reach the right decision. To make the best decisions, a group often has to overcome certain conformity pressures. I’ve witnessed this in companies large and small, as well as in governments all across the world. If you can create a culture of receptivity, and can weaken conformity pressures, you tend to get much more valuable input from individual group members. Now, you get a lot of nonsense also. But one actual advantage of conformity pressures comes from how, in a reasonable discussion, foolish or uninformed people will tend to self-silence. That saves everybody time, and prevents distracting tangents. Though it also comes with its own high cost, since even that person who seems foolish once in a while has the freshest and best perspective.
If we still have close to 40,000 deaths on US highways, we need to change that. Various people in the Department of Transportation will have their own ideas — some of these quite inconsistent with the views of the department’s leadership. But precisely for this reason, department leadership needs to hear out these qualified but dissenting voices. And to get those voices into this conversation, you need to prevent conformity pressures from stifling the group.
Here for one quick indication of conformity’s pervasive impact even on the most insulated decision-making bodies, could you articulate your call for a reasonable diversity of viewpoints on our courts — to ensure that lucid argument, rather than lockstep group membership, wins the day? And why, even among disciplined professionals operating from relatively explicit codified principles, must we remain wary about courts which proceed from preestablished unanimity failing to hold themselves accountable to the law?
The striking findings from research I did with a bunch of students, and with some colleagues, show that, on our federal courts, if you put three Democratic appointees together on a three-judge panel, the likelihood that they will produce a stereotypically left-of-center ruling gets crazy high — compared to if you have two Democratic appointees and one Republican appointee (and by the way, we observed exactly the same patterns play out with Republican majorities). Now, what makes this finding so wild is that you of course only need two judges on such a panel to establish the majority opinion. So why would three Democratic appointees show such strongly left-of-center voting patterns, compared to the more moderate patterns of courts with a two-one split?
Or you could just ask: how should we go about interpreting the Civil Rights Act of 1964? How should we interpret the Clean Air Act? Well, to start with, the contingent fact of which political party appointed your fellow judges shouldn’t make much difference. Though this turns out to make a massive difference in what courts decide — apparently because two Republican appointees who find themselves seated with a Democratic appointee might hear that Democratic appointee say: “This woman accusing her employer of sexual harassment seems to have made a pretty reasonable argument, and let me tell you why.” But without that kind of dissenting view, you again see, even in our courts, another just amazing case of like-minded people moving to extreme positions when they only talk to each other.
We truly didn’t expect to find this with federal judges, who tend to be pretty independent-minded. They’re not young. They have strong and well-tested convictions. And yet even with this impressive group of people, you can better predict their decisions, even on quite ideologically inflected cases (such as civil-rights or criminal cases), based on who’s sitting next to them, rather than which president appointed them.
Then for a more benign mode of harnessing such informational and reputational dynamics to socially constructive ends, could you sketch some ways that a society might best deploy the expressive function of law — perhaps less to proclaim the moral superiority of a certain set of values, than to promote the social perception that others obey these laws, and believe them right? Could you give an example of how such norm-management might prove particularly efficient when norm-breaking stands out, and when private enforcement gets legitimated and activated? And could you situate law’s optimized expressive place in society: somewhat ahead of prevailing public views, but not too far ahead as to require substantial reinforcement?
For one quick example: during my time in government, President Obama issued an executive order that said federal employees could not text while driving. This kind of safety-first executive order is never easy to enforce. If someone in the State Department texts while driving, this executive order does not specify what will happen to the employee. Instead, this order signals more generally to the federal sector: “We consider it dangerous to text while driving.” And this expressive executive order had equivalent impacts beyond the federal sector. It encouraged and reinforced state-issued laws forbidding texting while driving. The UN Secretary-General soon prohibited employees from texting while driving at an official ceremony. We saw these ripple effects in the US and abroad. When the public trusts the relevant official or institution making such an expressive signal, and when things go your way, this type of gesture can significantly reshape a basic social norm. You still might need more than that. Too many people still text while driving. But ideally, when the commander in chief says “I don’t think you should text while driving,” a lot of people think: Well, I probably shouldn’t do that.
For an earlier example, think about norms with buckling your seatbelt. Where we have laws requiring seatbelt buckling, we’ve seen much better results. This doesn’t necessarily mean that police often ticket people who do not buckle. Instead, these laws convey the impression that an expressed norm has taken hold among the public. The norm then becomes increasingly easy to justify. Why should you buckle? Well, you and your loved ones (or even your liked ones, or even your disliked ones) deserve the added safety that comes from buckling. Click It or Ticket might linger as a threat. But the norm probably ends up shaping your behavior more.
Similarly, for more conspicuous norm-breaking, say with public-smoking bans, even without any police presence, ordinary citizens themselves now might feel inclined to say: “You know, you shouldn’t smoke here.” Or when the public gets galvanized, as it has with the #MeToo movement, you might see sexual-harassment prohibitions in our civil-rights laws get much more consistently emphasized (though still not yet enough). #MeToo has helped both to shift a norm and to put violations under bright lights. And that second part becomes especially crucial when you factor in all the types of sexual harassment that stop short of physical assault, but that nonetheless prevent people from receiving their fair share of equality and of dignity. Here again the expressive power of civil-rights laws can help draw attention to these unjust situations.
So, given these tensions between revelatory political movements and polarizing partisan distortions, to what extent should we assume, regardless of which “side” we find ourselves on today, that its most vigorously professed views derive from dubious intra-group dynamics — as much as from any lucid reasoning? When should we distrust our most committed fellow travelers more than we distrust our sober-argument-asserting rivals? When should all the reasonable people from all sides come together to form their own group? Or why might it better serve society if they distribute themselves as evenly as possible among whatever new tribalisms emerge?
Great questions. I wouldn’t expect that if you support the Republican Party and consider the Democratic Party just terrible, you can somehow force yourself to think: I probably should distrust strong voices on my own side more than the Democrats. That sounds psychologically hard to do, and we probably don’t have enough reason to try. But it does help to recognize that you likely support your own team’s policy preferences 10- to 40-percent more than reason alone would recommend, and that this does at times mean you probably end up making the wrong decisions.
I definitely would assume that when you restrict yourself to associating with like-minded people, you become your most highly fallible. And I do believe that nonconformists should sometimes band together and build a critical mass. Jonathan Haidt, a social theorist I greatly admire, has helped to found the Heterodox Academy, which questions intellectual orthodoxies in educational settings. I certainly consider that an important project. We need to preserve institutional space to challenge conventional wisdom really on any topic. This doesn’t mean we need to identify as chest-beating nonconformists. It probably works better to think: I will take every issue as it comes, and not try to put it through some filter of affiliation.
We already do this well for so many things. When you want to know whether some food preservative is dangerous, you probably don’t just ask your friends for their opinion. You probably think: Well, I’ll look it up on the Internet and see what qualified people have to say. Or when you have some health concern you probably go to a doctor or at least consult credible professional websites, rather than just rely on some implicit group affiliation. And with topics like climate change, or immigration visas, or gun control, that same approach would probably work best. When instead you find yourself just acting in lockstep, you should consider that a warning sign.