Virtuous Violence and the Psychology of Nuclear Escalation

DANIEL R. POST: We see every nuclear state modernizing its nuclear weapons, expanding their ideas of potential use cases for nuclear weapons, right?

So it's actually–all of the sort of trend is going the opposite direction, that these tools may be more usable and relied upon more for military purposes or security purposes, than even in the past. 

RYAN VEST: Welcome to Horns of a Dilemma, the podcast of the Texas National Security Review. I'm Ryan Vest, executive editor of TNSR, and I'm here with our editor-in-chief, Dr. Sheena Chestnut Greitens. We're pleased to have joining us today Paul Slovic, Rose McDermott, and Daniel Post, who along with Missy Peterson and Daniel Vastfal, wrote the article "When Our Minds Go Nuclear: Rethinking Nuclear Strategy Through the Psychology of Risk and Decision Making," which is featured in Volume 9, Issue 3 of TNSR.

Paul Slovic is a professor of psychology at the University of Oregon and a senior scientist at the Oregon Research Institute. He holds a BA from Stanford University, an MA and PhD from the University of Michigan. He studies human judgment, decision-making, and the psychology of risk. His recent work examines psychic numbing and the failure to respond adequately to mass human tragedies.

Rose McDermott is the David and Mariana Fisher University Professor of International Relations at Brown University and a fellow in the American Academy of Arts and Sciences. She works in the area of political psychology. She received her PhD in political science and MA in experimental social psychology from Stanford University and has also taught at Cornell and University of California, Santa Barbara.

Commander Daniel R. Post is a permanent military professor in the Strategy and Policy Department at the U.S. Naval War College. He received his BS in mathematics from the United States Naval Academy, an MA in national security and strategic studies from the U.S. Naval War College, and an MA in political science from Brown University, and a PhD in political science from Brown University. His research focuses on nuclear strategy, deterrence, escalation dynamics, limited nuclear war, and conflict termination. His views are his own and do not represent the Department of Defense, U.S. Navy, or U.S. Naval War College.

Welcome, all, to Horns of a Dilemma.

ROSE MCDERMOTT: Thanks for having us.

PAUL SLOVIC: Thank you.

DANIEL R. POST: Thank you.

SHEENA CHESTNUT GREITENS: Let me get us started with a question about how all of you came to write this article. You're a pretty diverse group that bridges psychology, international relations, and political science, and data analysis, and experimental methods. And so I just wanted to ask you a little bit about how you all connected, and what led you to develop this kind of interdisciplinary research.

ROSE MCDERMOTT: I think that we recognize that this topic of decision-making around nuclear weapons was something that had been sort of in a niche in international relations and political science, and that it really required a broader constellation of knowledge and background, especially from psychology.

And so, the inherent nature of it seemed to really call for interdisciplinary work, and I'd done a lot of interdisciplinary work. I, of course, was very familiar with Paul's work from when I was in graduate school. And, he was a collaborator of someone who trained me, Amos Tversky. And so, we met at a couple of different conferences and started talking about the fact that there was, you know, a lot of rich theoretical tradition around nuclear weapons and nuclear strategy, and political science and international relations.

But the knowledge that we had about decision-making and risk from psychology was not being properly incorporated into that. And so we started, you know, thinking about that and talking about that. And then Dan was my graduate student at Brown, and I knew of his extensive experience actually in the military working with these topics, and so that's how we sort of came together.

And I think the main thing was just our recognition that it was really important to incorporate the knowledge that existed from psychology into this domain of nuclear weapons, where it hadn't really been applied before.

RYAN VEST: So the foundations of this article are really built on psychological research on human judgment and particularly on Herbert Simon's theory of bounded rationality. To start off, why is bounded rationality especially important for understanding nuclear strategy?

PAUL SLOVIC: It's because the human mind is very central to the decisions about whether or not to use nuclear weapons, what kind of weapons to acquire and so forth. But we have had the feeling to the best of our knowledge, that the importance of the mind is not respected in the nuclear strategy. That there's an assumption based on 80 years that we've had these weapons and haven't used them, that it's because of the deterrence that says that, "You know, if you use these weapons against us, we will retaliate in kind," and that our leaders are wise enough to know that's not a thing to do, so we have a kind of a mutually assured destruction.

It's an assumption of rationality and stability that is well-documented by 80 years of success. But, as you mentioned, in the middle of last century, Herbert Simon, a very famous psychologist, operations research analyst and so forth, who won a Nobel Prize, developed a theory of bounded rationality, which said that decision-making needs to respect the limitations of human minds. That is, we have limited memory, limited attention span, limited ability to process the information that comes to our attention and analyze it in various ways, and that should be factored into decision-making.

And his views were very important in motivating the behavioral science community, primarily psychologists, to begin to take a hard look and do studies to assess the rationality of decision-making. And at that time, the leading theorists of decision-making were economists who said that "We don't need to study psychology because everyone is basically rational. If they weren't, market forces would punish them and shape them up. So long as people have proper information, they'll make a decision that's in their best interests, and who needs psychology?" 

So that, plus Simon's assertion that we do need psychology, started a movement within psychology to study behavioral aspects of decision-making. This was around 1960, and I 'm old enough to have been one of the early members of that. I was a graduate student at the time. And so with others, we began to study that.

And so we've had over 60 years now of studying the way the human mind processes information for decision-making. And we've learned a lot about the ways that we make mistakes in decision-making. And to the best of our knowledge, that work has not been respected and incorporated into nuclear strategy. Dan, who's more central to that world than we are, maybe, could comment on whether that's an accurate assessment or not.

So, when we began to see that there are some studies of decision-making, like the study by Scott Sagan and Ben Valentino that has motivated us to study in a hypothetical Iran war scenario. We became quite interested in that and pursued that in spades, and this article is quite an extensive elaboration of previous work in that way.

So our hope is to gain respect for the importance of the human mind in decision-making and management of nuclear weapons, and to use the knowledge that has been gained in 60 years of research, plus these recent experiments that we've been doing, which is represented by the article, to help guide policy, to help policymakers and the military make wiser decisions about nuclear weapons.

SHEENA CHESTNUT GREITENS: I want to come to the experimental setup and the scenario in just a minute because I do think it's timely and really relevant. 

But before we get to that, I wanted to loop back to something that you said a moment ago, which is that it has been more than 80 years since the last use of nuclear weapons in war. And so I do think there is a sort of belief, or a sort of optimism that maybe leaders are too rational, that nuclear weapons are too obviously destructive, and so this is going to make people at the highest levels of government and of statecraft cautious and constrained, and prevent those weapons from being used again.

But what you find is that in these survey experiments that are featured in the article, that human psychology and psychological mechanisms like numbing, and the affect heuristic, and comparative evaluation really can increase support for nuclear strikes among citizens, and your survey respondents, and that's differentially true among individuals with different characteristics, which we can get to in a minute.

But, it sounds like one of your central premises is that conventional wisdom and that assumption is not only sort of empirically and experimentally incorrect, but maybe dangerous, and that one of the things you talk about a bit in the article is that the history of war suggests that maybe, you know, there's actually more historical evidence for some of these mechanisms shaping wartime decision-making than we think.

If that's an accurate summation, can you tell us a little bit about why you think that assumption is dangerous, and where we can look to really say, "Okay, and here's how it's been dangerous in the history of war before."

DANIEL R. POST: Yeah, sure. I'll take that one. So, you're right. You know, we do just–we talk about this, the 80 years of non-use creates–we talk about as a dangerous illusion, right? That's because they've not–they haven't been used, it's evidence that they won't be used, right? They're being used to prevent war, and they will never be used in war, right? And we just disagree with that. And, as you mentioned, history proves otherwise.

But before even talking about the history, I mean, so long as they exist, they can be used, right? They're there, and many in the military and civilian leadership feel very much like if we need to use them, we'll use them, right? If they're tactically useful or if they're militarily useful, then why would we leave them on the shelf?

And we've seen evidence of that in history, in fact, today, we see the trends all going in the opposite direction, right? So we see every nuclear state modernizing its nuclear weapons, expanding their ideas of potential use cases for nuclear weapons, right? So it's actually all of the sort of trend is going in the opposite direction, that these tools may be more usable and relied upon more for military purposes or security purposes than even in the past.

So then you look back, and you can also see in history, you know, leaders deceive themselves, right? They exaggerate things, the potential for a quick victory or a winning–a war-winning shot of some kind, right? Comments like, "We'll be home by Christmas," and all these other things. 

So the leaders deceive themselves. They underestimate their adversaries. They underestimate and misunderstand potential escalation dynamics. They think, in fact, sometimes they think about escalation as de-escalation. You may have heard the phrase escalate to de-escalate. Well, that's sort of like a–the phrase itself is counterintuitive, right? Are you either escalating or are you de-escalating? And so you have these concepts where escalation is underestimated and also where they mistake, you know, we may be able to achieve an immediate tactical success or short-term success, but at the expense of the longer term consequences are ignored or not valued, and you see that in history as well.

So the sort of danger we're most worried about is that a nuclear strike might appear to solve a problem, an immediate problem, whether it's to save troops or avoid humiliation or defeat, or destroy a really difficult target or something. We even hear talk about these things today, restoring deterrence. But it also might create vastly greater long-term dangers or dynamics, or consequences that are hard to imagine because human beings have a difficulty with imagining things that are farther away, and abstract, and difficult to understand.

I think that's where the biggest risk comes from, is not really being able to fully understand and comprehend the long-term consequences. And if you can't do that, then you can't really make a rational decision, because if you can't understand the costs and benefits fully, how can you then decide between options, right? And so it makes it difficult to do that.

So the absence of nuclear war since 1945 is not, you know, we should be grateful for that, but not confident that it's going to continue. We should continue to do everything we can to keep that going, and not be complacent in any way. So I'll stop there if any of my colleagues want to weigh in. 

PAUL SLOVIC: A brief comment. When I give talks about this, I show a slide of kind of a cartoon of the guy who jumps off the skyscraper. As he plummets past the floors, he shouts, "Okay so far." And we have been okay so far, but, so one might ask, well, why are we so comfortable in expanding this kind of what's called the new arms race, the modernization which expands the arsenals? We're developing better ways to deliver the weapons with hypersonic missiles and so forth, maybe put them in space. All of this, and making different kinds of weapons that might be more usable. I mean, all of this is quite horrible to contemplate, and many observers of this are shocked by what's going on, but it's happening.

And the question is, well, why, is it? And I think it's because after 80 years of non-use and then the confidence in deterrence holding, people feel it's okay to develop all these, you know, modern nuclear weapons because they won't be used. You know it's just kind of a numbers game. If Russia is doing something or China, we have to do the same thing to have so they don't get ahead of us in the numbers game, and so forth. But it's okay because, you know, they'll never be used.

And we don't believe that. As we study this and reflect from a psychological standpoint, we think this is not very sane to do this because they likely will be used, and we need to recognize and manage weapons appropriately with that knowledge.

RYAN VEST: With some of this big picture stuff down, you know, kind of like Sheena mentioned, I'd like to dig in a little bit to talk about the scenario that your study is based on, which I think is really pertinent right now. Now, you obviously wrote this article many months ago and have been putting it together probably for more than a year at this point, but you chose the Sagan and Valentino Iran war scenario as a foundation for your discussion on these nuclear weapons, and that has become very relevant in the last six months.

But it's based on a public opinion experiment to understand nuclear crisis decision-making, and I'm wondering what are some of the limitations of using public opinion to represent policymakers, and do you think that those results might differ if this was done, actually done with policymakers?

PAUL SLOVIC: Well, we think that policymakers are human beings just like our respondents. I mean, we're studying in these experiments factors that are common across, you know, maybe all people in terms of–we're looking at fundamental ways that the mind processes information and makes decisions that are tied to Simon's fundamental psychological factors such as attention and processing capability.

And so we think that what we find is relevant to top-level political and military decision-makers, many of whom have not been trained in how to think about these factors, and avoid them. It's not, you know, the president isn't trained in these, the ways that the decisions go wrong. And I think the military, although they study decision-making, have not been keeping up with what we've learned about this.

So we think that our findings apply until demonstrated that they don't. I mean, if people think that we're exaggerating the risk, well, show us why the procedures that exist now are such that we don't have to worry about these things. And because the consequences are so great, I think we need to assume that they will matter, unless we can be shown that we don't have to worry about these. And I don't think that will happen. So, that's basically why we have confidence.

And also, the original studies by Sagan and Valentino, they were wise enough to point out that their results, which were quite sobering because they found a lot of use of nuclear weapons in their scenarios, much of which we've confirmed, that should be corroborated with what they called historical resonance. That is, you don't stop at just these findings. These findings just create ideas that then need to be checked out in other ways. And one is by looking at the history of war, and history of other kinds of decision-making, to see if you see the same kind of mistakes happening outside of the experimental situation, which we do.

So that's the kind of back and forth between the careful experimental study and what's going on in the world, to see if it matches. And there's plenty of evidence that it does.

SHEENA CHESTNUT GREITENS: So what I'm hearing is that you really think that the safe assumption here is to believe that leaders are subject to the same cognitive, moral, emotional mechanisms that you found in the survey experiment, and that people haven't been well-trained in how to account for those mechanisms in nuclear decision-making. 

So if that training hasn't been done, do you think that training could help with some of these mechanisms? And maybe to what extent would training of key decision-makers be helpful in a world in which we might be able to do it?

PAUL SLOVIC: Elite training could change the picture if it's done. I don't think there is training to, you know, to get people to understand, and counter, and change procedures to counter what we're talking about. And that's a hope that the article that we've written, if people read and take it seriously, that will lead to training and procedural change that will reduce the seriousness of what we've talked about.

Although we can also get into the fact that it won't completely eliminate the problem, even training and that these problems are so serious that they really argue for abolition of nuclear weapons, because even as we manage them more conservatively, there still will be weapons left, and likely to be used. So, we may want to discuss that more later.

ROSE MCDERMOTT: I would just like to add one thing to what Paul said, which is that, you know, the training can take different forms, and some of it, you know, we have all kinds of things in the established mechanisms within the Department of Defense to have these standard operating procedures.

And you can imagine having similar kinds of standard procedures where if somebody has to make a decision involving nuclear weapons, that they're required to go through particular steps, and that they practice those steps so that it's not like the first time it ever happened to you and you, you know, don't have a notion of the repertoire of possibilities, including not responding with nuclear weapons, right? That universe is available to you in a particular way. And those kinds of protocols can be very helpful.

And the example I use, the example that I think we have actually in the paper is that, when they implemented a series of checklists in American hospitals that are really simple and stupid. They are things like washing your hands, right? Like put a sticker on about which limb you're replacing or removing, or whatever. It saved 100,000 lives a year, right? And it's just because everybody had to hit certain checklists that seemed obvious that everybody should know, but you have to do it anyway.

And so if you establish a similar kind of protocol where everybody has to do a certain thing a certain way, and you practice it so that people have a sense of, you know, what's involved in actually thinking that through, I think it can be really helpful to prevent inadvertent escalation in particular, or escalation in response to something that's an accident, right?

So rather than having artificial intelligence tell you, "Okay, you have an incoming weapon, you have to respond," to say, "Huh, take a minute. Is that really an intentional use? Could it be an accident? Could it be a mistake?" and those kinds of things I think can be very helpful in the training process.

DANIEL R. POST: The policymakers and elites are making these decisions in environments that are known to make these biases and problems more likely. And they, you know, time pressure, fatigue, stress, all of those things, right? Those are known to make us more susceptible to the weaknesses of the human mind, right? You know, I'm too tired, I can't think straight, right?

So in a sense, leaders in some ways are more likely to struggle with these than the general public because they're going to be making these decisions under circumstances that put these things to the real test. And so I'd just stress you know, so that's another reason to think like, "Oh, we didn't study policymakers." Well, again, they're still human beings, and their minds and their bodies work the same as the rest of us.

PAUL SLOVIC: We're continuing to study these issues and learn. And, I realized after the article was published that we've neglected an important finding in the behavioral decision-making realm, which has to do with a term called the hot and cold empathy gap. And what that says is that, you know, when we are in a calm, cool, deliberative stage, we think of things in a certain way. But then the crisis occurs and we are now emotionally charged up, and scared, and under time pressure. The consequences are huge, and we are now in what is called a hot stage.

And there is a lot of evidence, not from this domain, but from other kinds of domains, that those decisions are different when you're in the midst of the crisis. And so as worrisome as our findings are, our respondents are basically in a cold state when they're projecting how they think we should act, you know, in a hot stage, but they're not in that stage. And it's likely that given circumstances where we are really being threatened, and we're losing a battle, and there's existential loss and the leader feels that they're in a quagmire, and about to, you know, lose their status and power and all that matters to them, that's in a hot stage. That things will be even worse than what we show in our studies.

And training should anticipate that and give our leaders, both military and civilian, that practice in the decision-making in as realistic a setting as possible.

ROSE MCDERMOTT: Yeah, if I could just add one more thing to both those comments, which is that people don't act alone. You know, in these kinds of decision-making communities, there's groups of people, even small groups of people, and they likely have a degree of emotional social contagion, right? Like, if one person's angry, one person is freaked out or scared, it can amplify in the group.

And so you get, you know, I know a lot of people are familiar with the term groupthink. I think it's a little more complicated than that, but it's very difficult to have somebody outside the situation come in and be like, "Hey, you guys are overreacting," right? And so there's a way in which the group can amplify the time pressure and the other things that Dan and Paul mentioned.

RYAN VEST: This is a great discussion on this. Thank you very much for going down that road. 

I'm going to pivot though a little bit and go back to the scenario. So for those that haven't read the article, or for those that may not be familiar with the Sagan and Valentino scenario, what this study did is looked at groups of people that compared the use of nuclear weapons in Iran versus different numbers of American troops killed, to achieve the same objectives.

And as you go through here, one of the pieces that really came out of this is the idea of virtuous violence, where people can see violence, whether that's nuclear or whether that's a more conventional violence, as being morally justified or necessary. I was wondering if you could talk a little bit about how that concept explains nuclear preferences better than traditional strategic models, and really, does the role of political identity suggest that ideology is less important than moral disposition? Is there a difference between the two, or are they difficult to separate? What do you think?

PAUL SLOVIC: Yeah in the initial study by Sagan and Valentino, they used a single question about an American domestic policy that punished people. Namely, do you support or not the death penalty for people convicted of murders? And they found a very strong correlation between people who endorse that, of having a much higher likelihood of supporting the use of nuclear weapons in their scenario, even though it has nothing to do with nuclear weapons.

And in our studies, we have extended that. We have added other questions that in various ways, represent an attitude towards punishing Americans for doing things that you find threatening or immoral, or whatever. Like, do you support punishing women seeking abortion, or criminalizing people in this country, immigrants who don't have extensive or adequate documentation, and so forth. These have nothing to do with nuclear war, but they are highly predictive of the support for the weapons. And the more of these items that people endorse, the higher the likelihood they would endorse using nuclear weapons.

There's a sense that it is appropriate and moral to punish bad people. And so in thinking about this, while we were writing this up, I came across a book, a 2014 book by anthropologist, Alan Fiske, and psychologist, Todd Ray, titled Virtuous Violence. And they were looking at nuclear weapons, but they were looking at a wide range of violence in our society, and arguing that perpetrators of violence most often feel that they are justified and virtuous in harming the people that they target.

And that was actually a surprise to me. I thought that people who commit violence kind of do it out of need and regret doing it, but that's apparently a small percentage of that. And we find that in our studies. And we're sensitized to that, we see that not only do people, who like to punish others more prone to nuclear weapons, but they feel good about it.

They feel that, if you ask them, "Is your decision to support using nuclear weapons to win this war and save the troops, is it virtuous?" A high percentage of them, you know, say it is virtuous. As high a percentage as those who don't use nuclear weapons because they think it's immoral and unconscionable to kill civilians. They also think their decision is virtuous, but you don't get any difference. Those who use nuclear weapons feel virtuous in doing that, and that is really scary. It means that, if and when weapons are used again, most likely the person who orders that to happen will feel virtuous, and that is frightening.

SHEENA CHESTNUT GREITENS: I think that is a really striking finding in the article, and one of the things that I liked about this article is how much it unpacked, how much individual characteristics can shape people's attitudes and preferences toward the use of nuclear weapons under different conditions, that each, you know, seem to varying degrees realistic at times when we think about geopolitics.

So, you know, individuals with these strong punitive attitudes are more willing to think of nuclear weapons use as morally right, and appropriate, and even protective. Individuals, you know, opposed to nuclear weapons use, who start out that way as troop fatalities rise get more willing to think about the idea. And people who are weighing the idea of a nuclear strike that causes a small number of casualties versus a conventional conflict that causes a large number of casualties weigh the use of nuclear weapons use differently.

So it's really an interesting exercise in the article, in unpacking these different nuances to the way people think about nuclear weapons use. 

I wanted to ask about one that I thought was interesting, and maybe a little bit surprising, which is that you found that under certain conditions women were more likely than men to support the use of nuclear weapons, if they saw it as protective of troops, and that was really interesting to me.

So I wondered if you could talk a little bit about how to interpret those results, maybe if possible without falling back on some stereotypes that we all have about gender and violence, because I thought this was both counterintuitive and really interesting, and potentially significant as we look at public attitudes.

ROSE MCDERMOTT: A quick and easy read of this, and also the Sagan and Valentino piece, is that women appear more hawkish than men. But in fact, it's much more our results, and I think it's fair to say the Sagan and Valentino ones, are actually more nuanced than that.

And part of it involves the unpacking that you just mentioned, but also is tied to the notion of virtuous violence that we just discussed, and the hot/cold empathy gap. In that, what's really happening is that when you mention troop losses, women do appear more supportive of the use of nuclear weapons than the men do on average.

But when you give people an opportunity to explain why they make the choices that they do. What you get are a lot of the women saying, "It's really important to protect our troops. It's really important to protect American soldiers." So, you know, perhaps they're thinking about their sons being at risk or, you know, their daughters at risk, or whatever it is.

And so we have a condition where there aren't any American soldiers who die, or that isn't mentioned. And in that condition, men are more supportive of the use of nuclear weapons than women are. So you have a situation where it's quite nuanced and context-dependent. When the loss of American life is made salient, then the switch becomes, "Okay, we need to protect our troops, our guys," and if that costs those guys, the other, you know, the outsiders, that's fine.

But when it isn't made salient, then you get what you would describe as the more stereotypic response, right? Where men are more supportive of these weapons than women are. And I guess the important point for me here is that it's not the stereotypic response, right? Where like, women are more hawkish, or men are more hawkish, but it's very dependent on context.

And so it's not just that women are more aggressive, but that they're actually defensive in support of in-groups, you know, of American troops, of their children, their friends, and so on. And I think that, it's important not to overstate the gender result, but also to think about how it contributes with things like punitiveness, with things like ideology, to explain the responses that people make.

And I guess the final point I'd make here is that, you know, compassion takes different forms, and it doesn't always mean peacefulness. Sometimes it means actually protecting our own families, our own community, our own country, and that it can be misread in different ways if we think of it too simplistically as more hawkish, or more dovish.

RYAN VEST: Just to keep moving on. 

The article underscores evaluability theory, or the idea that lives and costs are easier to judge when put in comparison.

How might the way options are framed during crisis times push leaders toward escalation?

PAUL SLOVIC: That relates to the finding, which is now quite well known, that when we contemplate loss of life, we become insensitive to larger losses of life. That is, a single life is very important to protect, and we do a lot to protect or rescue a single person, or a single animal in distress. But then if we look around us, we see that when large numbers of people are in distress, suffering from whether it's famine, or other kinds of social ills, or military violence, that we often do very little to protect them.

This became apparent to me when I was studying genocide, trying to understand why we, who value individual lives so greatly, turn a blind eye to genocide or other mass atrocities. And we do very little. And what we find is that it's because we are relying not on any kind of careful analysis of how serious the problem is, but we're relying on our feelings. And our feelings don't do arithmetic properly. We have a term for that. We say the more who die, the less we care. That is a non-rational form of arithmetic that our feelings do.

And so evaluability theory makes a contribution by showing that when you look at a particular situation or a particular number of things like loss of life by itself, you may not get a careful, clear picture of how serious it is. It's just a number. But when you have two or more numbers to compare, then you can make sharp comparisons about things that individually you don't comprehend.

And in our study, we did something that has never been done before. We gave people a chance to make a decision about using nuclear weapons twice. So for the first time, they made a decision about launching a single strike against Iranian civilians to end this war that might kill, in one case, 100,000 civilians, or in a second case, two million. But they only had one bomb, and the option was to continue with the ground war that we will win, but we'll lose troops.

And, you know, some people went in different directions. Some went nuclear, and some didn't. Then we brought them back at time two, and we took the bomb that they had not seen in the first time and put it in the choice set. So now they had three choices. Continue the ground war, choose a bomb that might kill 100,000 people, and choose one that might kill two million–and what you would do. And what we found was a lot of people who, if they had the choice between the 2 million fatality bomb and ground war at time one, they wouldn't go there because 2 million they felt was unconscionable or, you know, it just was not right. They would rather, as much as they hated to lose American lives, they couldn't do the 2 million.

Then you put the two–the 100,000 bomb in the choice, and here's another, you know, make the decision again. Many of those people who moved from the ground war to the 100,000 because they wanted to save the troops, and they clearly could see that the 100,000 fatalities is better than, far better than 2 million. And so they chose the 100,000 to save the troops, and they felt virtuous in doing that because they saved the troops with a better bomb.

And we realized that this is–first we call this the better bomb effect, then we realize, well it's really a form of a broader way of thinking that is universal in human cognition, which is the lesser evil. And I think it was a common comment by Ryan that asked about, you know, taking our findings, you know, historically, or looking at in a broader sense that led us to go in that route. And we found that the concept of the lesser evil goes all the way back to Aristotle, who recognized that we sometimes choose something that's a lesser evil. And he said that is good, that what we choose is good.

But he missed something that another person, a moral philosopher named Hannah Arendt, who was observing the Eichmann trial after World War II, and Eichmann's defense was that he, you know, he was just a train master. He didn't operate the gas chambers, so he wasn't as bad as the other people. She recognized that as the lesser evil as well. And she went further than Aristotle in recognizing that the mind confuses a lesser evil with goodness. And she made the statement, "Let us not forget that the lesser evil is still evil."

So what this shows is that we draw meaning from comparisons. That sometimes things are hard to evaluate on their own, but we can evaluate them. That's what evaluability theory has shown, is that we can make sharp and good decisions by comparing things. And so we see that very strongly.

And getting back to a comment that Sheena made earlier about the individual differences that come out from the study, and not only about gender, but other factors. We then created a typology for every person in our study. We had over, almost 3,000 people in this study. For every one, we could give them a genetic typing here based on three factors: gender–male or female, how punitive they were in terms of supporting domestic punitive policies. And we have three levels of that–low, medium, and high.

And then we also asked them, this is at time 2. "Did you think about comparing the two bombs? Is that as important in your decision?" we dichotomized that–yes or no. So these three factors, gender, recognizing a better bomb, and punitiveness, that created 12 different types of people. And then we see enormous differences among those people in the percentage of people within a type who would choose to use a nuclear weapon. And the most extreme is among women.

So you have women who are not punitive and don't pay attention to comparing. Only 2% of those women chose to use the 100,000 fatality bomb at time 2–2%. Then you go to the women who were punitive and recognize that one bomb was better than the other, that went to 80%. So you had a 40-fold increase in the likelihood of a person choosing to kill 100,000 people with a nuclear weapon just on the basis of those two factors there.

Again, we don't really know the probability of a leader choosing nuclear. What's important here is the relative finding of a 40-fold increase based on two factors that are–one is a character factor, the punitiveness, the other is a contextual factor of placing two bombs in the same choice set.

That's something to worry about, and to take seriously, and to think about how that should affect who we choose as our leaders and what kind of situations–whether we're creating a better bomb type feeling in people by the way we present choice sets.

DANIEL R. POST: Yeah. And that stuff really matters in terms of how options are considered or presented by the president and by civilian leadership, or even the military leadership, right? So everything Paul is talking about has a direct implication for how options are presented and how you can trigger these effects in leaders, where they might choose an option that on its own is strategically unwise, but when placed with another option, seems strategically wise or politically wise for that matter.

And so, our findings suggest the processes, procedures, and training really ought to factor that in. They need to understand that, right? That you might create a decision-making environment that makes evil options, evils, lesser evils look good. But they're not independently good, right? They're good in comparison to something else, but that doesn't make it wise, right? And so recognizing that is really crucial, I think.

Hopefully that's something that sticks out to people when they read this paper, because it's really important. It has direct implications for the decision-making process, and how options and decision-making procedures are constructed and executed.

So it's a really important discussion, Paul.

ROSE MCDERMOTT: I think it also introduces the possibility that advisors can use this to manipulate leaders to make a choice that they want them to make, that might not have been the leader's original choice.

So, for example, if an advisor knows that by adding a more extreme choice, the leader will be more inclined to make the middle choice, even though the middle choice otherwise would be too extreme without the really extreme choice, there can be a kind of implicit level of manipulation that the leaders may not be aware of.

And so it also introduces this possibility of manipulation that people may not be aware of, but it is very real and very powerful.

SHEENA CHESTNUT GREITENS: Yeah, I think this is really important, and I have sort of two thoughts right now. One is that these findings are so nuanced and so context-dependent. I really do encourage anybody who's listening to the podcast to go look at the article, to go through some of this, because there's just a lot to sit with here about how these tendencies shape individual preferences and views. And given the sort of moral stakes, the human consequences, just wanted to sort of throw in an encouragement for everybody to go actually look through, and spend some time sitting with the details, and the amount of information that's in the article.

I wanted to close by asking you a little bit about what we do with this information. So, you know, one thing the article suggests is, you know, we shouldn't put this solely in the hands of one decision-maker, and that there are some particular things to think about, particularly as AI becomes embedded in nuclear systems, and maybe in risk assessment and nuclear decision-making.

So the question is, you know, what would a more, I guess, cognitively resilient decision process that accounts for these kinds of things, the way we structure decisions, the way they're presented in brief to senior leaders, right? What would a good cognitively resilient decision process look like, and what safeguards should we absolutely prioritize to make sure that some of the risks associated with bad or immoral nuclear decision-making are actually reduced?

ROSE MCDERMOTT: I think that the most important thing here is that the decision to use nuclear weapons should never be in the hands of a single person. And certainly former Secretary of Defense, Bill Perry has been advocating for this for a very long time, as well as others. And I think the importance here is that it shouldn't be one person's decision based on, you know, whatever personal and contextual forces may be imposing on him or her at that time, that there is required to be, you know, multiple people involved in the decision.

I think that we've talked about some of the other issues related to having training, specific training around questions that need to be asked before the decision to use nuclear weapons is taken, that that has a particular protocol around it and a particular kind of training around it, and that those kinds of policies and procedures could make a big difference.

I think there's other factors. I'll let my colleagues add to that. But, for me, the big takeaway is it should never be in the hands of one person.

PAUL SLOVIC: Yeah, there are a lot of implications. So what we have identified both in this article, and also the broader community who studies decision-making, are ways that decisions go wrong. And with that knowledge, we can then say we need to correct those decisions. And some of it is through training, as we mentioned. The other is through changing procedures.

One of the ways that decisions like this go wrong is a focus on short-term, immediate benefits that we understand to a certain degree and carry a powerful emotional pull on us. And like winning a war, gaining political points against an adversary, or showing that, you know, our nation is tough and one shouldn't mess with us and so forth. These are very strong and understandable factors that capture our attention.

And because our attention is limited, and because these other things are complex, like what it does, it leads us to give too little attention to the longer-term consequences of crossing the nuclear red line, and using these weapons again for the first time in 80 years. It's like opening Pandora's box, that we don't know what's in it because we haven't had the experience to learn what happens when you have a nuclear war.

We have vague memories of what happened in Japan. But if in today's world, starting a nuclear war will cause these ripple effects, which will have far greater impact in a negative way, than the short-term benefits of that will come from using the weapons.

We know that's a bias. People will focus on the short term. So we need to have procedures, either people in the room who don't let the leaders forget about the long term, or we can have AI assistants that kind of warning, like warn us not to leave out important factors that we know people will leave out if they're not–there's not some procedural factor that captures their attention on the civilizational effect consequences, the humanitarian consequences. So that's the way that AI can be beneficial.

AI can also–is also going to be disruptive by flooding the information ecosystem with information of all kinds, much of which may be of dubious validity, they will overwhelm our ability to process it. So, AI will, you know, create time pressure to respond to things. And time pressure will exaggerate all the findings that we've observed in a calm, quiet state that our respondents are responding in. So AI will make anything we've found, that's dangerous even worse. Maybe we need an AI system that will help us deal with AI because otherwise we'll be overwhelmed with it.

We can't handle it. So things are very complicated. That's another way in which things are different now than they were in the previous 80 years.

DANIEL R. POST: My colleagues have covered it all, but you can kind of boil it down to sort of two areas, right? So like Paul started with, you know, you talk about cognitive resilience, right? So part of that is going to be the training, and education, and knowledge, that people have in these positions, right?

And there's improvements that can be made there. And then there's the process part of this. What can we do in the process and procedures, and standard operating rules, and all of those things? And there's some key themes in there.

You know, where and when can we build in time for deliberation, right? When can we relieve artificial pressure, or prevent artificial pressure that's being, you know, these situations are likely going to be pressure cookers to begin with. Are there ways we can reduce that or mitigate it somehow that we're not doing now, right?

The structured dissent is an idea that comes up. So, you know, –I think this is a concept mentioned in the book World War Z, if you ever read that book. You know, it's the 10th man idea. If there's 10 people in a room and nine of them agree, it should be one person's job to disagree with everybody and challenge every assumption that all the other nine people made, you know, just to give an opportunity for somebody to point out, "Did you think of this or that?"

And then the presentation of the options and the choice sets, and the way that is done should be informed by these psychological things that we know. That should all be built in. And so you got the training and education piece, you got the process piece, and I think if you do that, even the most ardent nuclear optimists and hawks who want their nuclear deterrence, and want their nuclear weapons, they can still have that with these procedures, right?

It'll mitigate risk, at least in that way. I would encourage the sort of practitioners listening that this is advice for how to make better decisions. That's what this is about.

PAUL SLOVIC: And don't respond immediately to provocation of the adversary. You know, force time for deliberation on these issues.

 

RYAN VEST: Thanks for joining us on Horns of a Dilemma from the Texas National Security Review. Our guests today have been Paul Slovic, Rose McDermott, and Daniel Post, authors of "When Our Minds Go Nuclear: Rethinking Nuclear Strategy through the Psychology of Risk and Decision Making," which as always can be accessed for free on our website. Paul, Rose, and Dan, thank you very much for joining us today.

DANIEL R. POST: Thank you.

ROSE MCDERMOTT: Thank you.

PAUL SLOVIC: Thank you.

RYAN VEST: If you enjoyed this episode, be sure to subscribe and leave a review wherever you listen, and you can always find more of our work at TNSR.org. Today's episode was produced by TNSR Digital and Technical manager Jordan Morning, and made possible by the University of Texas System. This is Ryan Vest and Sheena Chestnut Greitens.

Thanks for listening.