In a democracy, elected officials are accountable to the public they govern. Exactly how that accountability works is complicated, with checks and balances, partisanship, the media, and good old-fashioned bureaucracy all playing roles. Of late, another wrinkle has emerged: The president-elect, Donald Trump, has a habit of attacking the free press, one of the institutions that makes that accountability possible.

But accountability itself is a tricky thing. People react to being held accountable in different ways in different situations. One would think that having to answer to someone for your actions or your decisions would make you think more carefully about whether you’re doing the right thing. And according to research done by Philip Tetlock, a professor of psychology and political science at the University of Pennsylvania, it sometimes does make people engage in more “exploratory thought” before making a decision. But sometimes it doesn’t. Sometimes it just inspires people to defend more stridently what they already believe. Sometimes it motivates people to conform with the views of the audience to which they are accountable.

I spoke with Tetlock about when being held accountable leads people to make more careful decisions and when it doesn’t, and what lessons this suggests for the public’s ability to hold government accountable. A lightly edited and condensed transcript of our conversation is below.

Julie Beck: What are the conditions where being held accountable makes people consider other points of view, and when does it not really change how people make decisions?

Philip Tetlock: It virtually always influences how people make decisions, but it’s not always good. Accountability is a multidimensional concept. It refers to who must answer to whom for what under what ground rules. It’s not a single independent variable, to use the terminology of social science. So it’s not all that surprising that it has many different effects. In our work we found that only quite specialized, carefully crafted forms of accountability induce people to think more self-critically and make them more resistant to cognitive biases. Most forms of accountability do not have beneficial effects on judgment and choice.

Engineering forms of accountability that help requires a degree of patience. People need to believe that they’re going to be answerable to an audience whom they respect, and they need to feel accountable to an audience with unknown views. They also need to feel that they’re not under pressure to justify what they’ve previously said or done. That’s a form of accountability that induces a relatively unnatural cognitive act, which is preemptive self criticism—to think of possible ways in which you might be wrong before other people do it for you.

Beck: When can holding people accountable can have bad effects?

Tetlock: One obvious one is when people cope with accountability simply by subordinating their judgment to others. Conformity and obedience, Solomon Asch and Stanley Milgram and all that sort of thing. We call that strategic attitude shifting. That makes it sound a little more conscious than it sometimes is—sometimes people just move their attitudes totally unconsciously toward what they think is the social center of gravity. But sometimes they move quite deliberately. What we call elastic-band shifts are when people move to where the audience is, but when the audience disappears they snap back.

Perhaps the coping response to accountability that people regard as more perverse is actually rooted in the old cognitive-dissonance literature. It’s the idea that when you’re accountable for something you can’t change—you’ve said something or done something, and you can’t undo it, and you’re being held accountable for it—most of your cognitive effort goes into generating reasons why you’re right and your critics are wrong. That’s sometimes called post-decisional bolstering. And it makes people more extreme and more rigid.

It’s almost the opposite of what I call preemptive self-criticism. Preemptive self-criticism is where you de-center, you become less egocentric in your thinking, you try to anticipate grounds reasonable people might have for objecting. Whereas post-decisional bolstering is where you really hunker down and raise the defenses.

Beck: In the U.S. government, there are checks and balances, but I'm interested in how government is held accountable to the public. One would assume that elected officials are always accountable to the public, so that first condition where you need to know you’re being held accountable hopefully should be taken care of. But how can we think about this when it comes to government?

Tetlock: Well, in a democracy there is a flip. Usually, in most organizations, the people at the bottom are accountable to the people at the top, whereas in a democracy the people at the top are ultimately accountable to the people at the bottom. Of course it’s not that simple because the people at the top have a lot of resources—cognitive and financial—for manipulating the views of the people at the bottom. And the people at the bottom don’t have nearly as many resources for coping with the influence tactics. In the case of the U.S., where you have a winner-take-all system at the presidential level, you have incentives to compete for the median voter—post-primary season of course.

Beck: How does your research translate? Do you think the public should be creating a sort of accountability that would get leaders to have this preemptive self-criticism? Would that be the goal?

Tetlock: The U.S. government, obviously, is a very complex entity with many checks and balances. The executive branch is supposed to be continuously second guessed by both the congressional and the judicial, so those are forms of accountability that should, in principle, trigger more complex forms of thought. In fact, accountability can become so pluralistic that it can induce analysis paralysis, though that might not be the dominant concern at the moment.

[In the laboratory], you can control who must answer to whom under what ground rules. Whereas when you move to the real world, especially in a complex pluralistic constitutional checks and balances system, you’re really dealing with accountability to multiple audiences. They’re juggling multiple things.

I did a study a long time ago, in 1981, one of the first studies I ever published. It was called “Pre- to Postelection Shifts in Presidential Rhetoric.” It was looking at whether there’s a shift in how integratively complex presidential statements become immediately after they come into power. Before they've had an opportunity to learn very much about the intricacies of the job, is there an almost immediate shift? A president is coded as more integratively complex if the rhetoric is laden with qualifiers, like “however,” “although,” and “but,” as opposed to linguistic amplifiers like “moreover.” Integratively complex rhetoric is one sign that a political leader is reasonably comfortable living in a pluralistic accountability universe.

That naturally raises the question—to what extent has Donald Trump’s integrative complexity shifted since he was elected?

Beck: And what would you say?

Tetlock: You know, I haven’t been tracking it. I sense there’s been a bit of a shift, but I'm not sure.

Beck: For the other condition you were talking about, where the audience’s views are known or unknown, how do you think partisanship would play into that? Would officials only consider themselves accountable to voters from their party? Or might they assume that they already know the audience’s views already?

Tetlock: I think there it’s become really quite a refined science. It’s quite routine for political leaders to spend significant amounts of time with their pollsters, whose job is essentially to read the public mood. And that proves to be a nontrivial challenge because so much hinges on exactly how you phrase the question. So it is interesting that in a highly competitive pluralistic democratic system like the U.S. where the technology of surveys has become as refined as it has, that politicians have felt it is in their interest to spend as much time as they do with pollsters.

Beck: The review of yours I was reading said one of the reactions to being held accountable is just trying to conform to the views of the audience.

Tetlock: Strategic attitude shifting would be a version of that. I think part of what politicians try to do—to use the terminology of my field again—is stay within the latitude of acceptance of public opinion. They don’t want to be caught making positions that are way outside the mainstream of acceptability. So Donald Trump inadvertently blundered into saying women should be punished for getting abortions. And I think very few conservative Republicans were endorsing that position. He apparently hadn’t done his homework.

Beck: The other condition that I wanted to talk about, this is something I saw in your review, is how when you believe the audience is well-informed and interested in accuracy, that is more likely to make you do this exploratory thought. Obviously, Donald Trump has lied time and time again—lying about Hillary Clinton’s policies during the campaign, lying about voter fraud after, saying he didn’t mock a disabled reporter when he did. Not to say that his audience isn’t interested in accuracy, but Trump has continued to be successful in spite of these lies. So do you think he would even think that being held accountable for lies would hurt him at this point? How is that a factor?

Tetlock: I guess that’s the scenario that Cass Sunstein raised when he wrote about ideological echo chambers and the possible emergence of a post-truth society in which each side has its own facts.

How one sorts through those things is disturbing. I’m not as pessimistic as some people are about ideological echo chambers and people retreating into their own truth worlds. But I think there certainly has been a disturbing trend in that direction.

Beck: If that is the case, does that undermine the conditions for good accountability?

Tetlock: Yes. It’s a serious problem, if there’s not a shared reality. In a two-party, winner-take-all presidential system, the two parties have an incentive to converge on the median voter. They want to figure out what the median voter wants so they can form a minimal winning coalition, so they tailor their messages toward the middle. It would seem that if large factions of the electorate were living in different truth universes that would certainly complicate that process.

You’re looking at things that are way more sophisticated than anything you see in the lab. Politicians are very good at rhetoric. They’re very good at impression management. They’re very good at spinning things so that what might have looked rash looks bold, what might have looked timid looks cautious. They’re quite skilled at turning questions into opportunities to answer the question they want to answer. Regular people aren’t nearly as good at that. So what you're looking at are two elite teams of highly professional impression managers, the Democrats and the Republicans, each of whom is doing their thing, and they’re definitely doing it well enough to keep their community of co-believers in place. And sometimes one side really scores, and the other side reels back. There's an old expression that American politics is a game that's played between the 40 yard lines.

Beck: There was one finding from your review that seemed really relevant to the current political situation. You wrote, “Decision makers who sense that an illegitimate audience wants to influence their beliefs may react in a variety of counterproductive ways. They may respond by asserting their own views even more vigorously or by disengaging from the task.” This immediately made me think of how Trump consistently tries to paint the press as “dishonest,” which indicates that he will see any attempts by the press to hold him accountable as illegitimate. What does that mean for the state of accountability going forward?

Tetlock: Well, I think that’s quite clearly true and I think that when you look at the pattern of press endorsement in 2016, the number of Republican newspapers that refused to endorse Trump was quite staggering. He’s in a deep hole with the people who cover politics most closely. So it makes perfect political sense that he would try to delegitimize an institution that he considers to be irretrievably opposed to him.

I don’t think he’s hoping to win the press over, I think he’s hoping to control and manage the press. I’m sure he would prefer to have a better relationship with the press if he could. But I think at this point that’s going to be extremely difficult. It would require a major pivot. And even then, I think the relationship would be very sour.

Beck: The legitimacy factor, that also seems to play into partisanship—is it possible that Democrats would see any Republican attempts to hold them accountable as illegitimate and vice versa?

Tetlock: They always do that to some degree. It’s a question of how intensely polarized the political system becomes, the degree to which people start to deny the legitimacy of the claims of the other side across the board. That’s what happens when countries start to drift towards civil war.

Beck: It does seem to be a pretty polarized situation right now. Is it a situation where people would only do this careful pre-decisional consideration if the accountability was coming from their own side?

Tetlock: It’s a dance, in which each side is trying to find the upper and lower boundaries of the acceptable. If the nominee for attorney general Jeff Sessions was a KKK sympathizer at some point, does that automatically disqualify him from being the attorney general of the United States? If he said X, Y, or Z, does that automatically disqualify him? There’s going to be this vigorous back and forth of defining what the boundaries of the acceptable are, and each side is really good at this. Each side is really good at taking extreme instances of behavior, either good or bad, and trying to magnify them and portray them as typical.

Beck: Is there any sort of broader takeaway about how accountability’s going to play out going forward?

Tetlock: I would just say there is this big gap between what goes on in the lab and what goes on in the world. And what goes on in the world is very two-directional. Each side is making very active efforts to influence each other and influence the audiences to which each is accountable, and the influence is ebbing back and forth in a very dynamic, messy way. It's hard to draw simple extrapolations.

There comes a point where administrations get stuck with certain policies that they are responsible for and they face a choice and the natural psychological thing for them to do is to dig in and bolster. That famously happened with Lyndon Johnson during the Vietnam War. But other times they make tactical retreats and try to cut their losses.

In the lab, we can really specify with precision whether this is conformity, this is bolstering, this is preemptive self criticism. In the real world, these things become much blurrier and it depends what kind of a spin you want to put on it. With strategic attitude shifting, you could say it’s cowardice or you could say you’re being responsive to the will of the people. Post-decisional bolstering—you could say it’s rigid or you could say it’s principled. Pre-emptive self criticism—you can say it’s thoughtful and statesmanlike, or you can say you’re a flip-flopper the way George W. Bush successfully characterized John Kerry in the 2004 election. You’ve got this potential to flip and spin the meanings of the coping strategies that’s not there in the lab and you’ve got people who are good at it. They're spinmeisters. It's their job, it's their profession.

Beck: So it becomes hard, then, to know if you’re holding people accountable and that’s why they're doing things or if it's just part of the plan?

Tetlock: That’s right. It eventually becomes a hall of mirrors.