The Architecture of Choice

It is such a pleasure to be here. They also promised my parents, who are actually not that proud.

As she mentioned, I used to be a violinist, and I sustained a really serious hand injury when I was 17 that doctors deemed to be a career-ending diagnosis. As you can imagine, I was pretty demoralized by the situation and also deeply concerned that I was never going to find something I loved as much as the violin.

Fortunately for me, one summer I was helping my parents clean their basement, as any good daughter does, and I stumbled across a book called The Language Instinct by a psychologist named Steven Pinker.

I was instantly mesmerized by the topics I was reading about. It articulated how something that had been so effortless for me and just about everybody else, namely our ability to learn, comprehend and speak language, was actually the result of extremely complicated, sophisticated cognitive architecture and machinery.

I remember thinking, “Wow, I never really even thought about my capacity for language.” It was something that came so effortlessly to me. It was so intuitive. Yet scientists over the years had discovered such fascinating and surprising findings about language acquisition.

This is what propelled me to study cognitive science over the last 15 years, a fascination with how the mind works. I ultimately landed in a particular sector of my field, decision science, and have been working to figure out applications of decision-making.

When I learned about the science of decision-making, one of the things I was excited to discover was that it was just as surprising as all the insights I had learned about language.

Again, decision-making feels so effortless and intuitive. I don’t really think about it, but I’m making tens of thousands of decisions each and every day.

What I’m hoping today is that I can share with you some of the surprising things from my field. You’re all experts in this area, and I’m hoping that together we can forge some intersections between behavioral science and the problems and challenges you’re all trying to solve.

I wanted to open with a story that can help articulate what I mean when I talk about applied behavioral science.

Back in 2015, I was working on veterans’ issues in the Obama White House. When veterans return from their time overseas, they can face a lot of challenges when it comes to finding work, getting back to school and so forth.

The federal government offers free job-assistance programs and educational counseling to help veterans get the support they need. But one thing we found was that not enough veterans were signing up. We weren’t quite sure why.

We had to conduct a behavioral analysis and devise a solution that kept us within the same budget we had initially been allocated. We didn’t have much breathing room.

Ultimately, we changed just one word in a marketing message to veterans about the program. Instead of telling veterans that they were eligible for the program, we simply reminded them that they had earned it through their years of service.

This one-word change led to a nine percent increase in access to the program, with the potential to positively affect millions of veterans around the country.

This example illustrates a principle called the endowment effect, which says that we value things more when we feel that we have earned or own them. This is one of many principles in behavioral science that could potentially be relevant to the kinds of decisions all of you make.

What actually is behavioral science?

It is the study of how and why we make the decisions we do. There is also research in this space that tells us how we form our attitudes and beliefs about the world.

It draws from research in psychology and economics and tries to blend the two fields to give us a more comprehensive understanding of our decision-making patterns.

The reason we should care about this body of research is that it tells us our decisions are actually pretty surprising. Our behaviors are very surprising, and they’re often influenced by factors we really ought not to be influenced by.

I’ll give you two concrete examples.

We like to think that when people go to the voting booth, they’re going to vote for the person they would most like to see elected into office. That’s pretty straightforward, right?

But research shows that the order in which candidates’ names appear on the ballot has a significant impact on voter behavior. In Texas, researchers found that when a candidate’s name appeared first on the ballot, that candidate received 10 additional percentage points in vote share, which, as you know, is more than enough to turn any election.

Another surprising factor affects how we feel about one another. There is research showing that when we’re holding a warm beverage in our hands, we actually feel more warmly toward the people we’re interacting with.

There is this really interesting misattribution going on, where we feel physical warmth and misinterpret it as interpersonal warmth.

At this point, I’d like to ask anyone with an iced beverage to kindly put it down for the remainder of the talk.

As you can imagine, our field is ripe with these really compelling, fascinating insights. After studying in academia for a long time, my goal was to figure out how we could forge intersections with the real world. How do we put these insights to work to make sure we’re solving real-world challenges faced by real people?

I brought this to the Obama White House, the glory days, back in better times.

Our goal there was to build a team of dedicated behavioral scientists who could systematically look at different policy challenges and engineer solutions based on what behavioral science research says about those challenges.

What I learned from our experience working with many different departments, including the Department of Education, the Department of Veterans Affairs, Health and Human Services and the Department of Defense, was how universal the application of these principles really is.

They are particularly relevant to how we think about end-of-life care and decision-making.

As you know, decisions are not made in a vacuum. The choice architecture we create for people, whether they are caregivers, providers, doctors or patients, really matters. It can have an outsized influence on the way people make decisions, how we plan for the end of life, how we create human-centered solutions to ease caregiving and support, and even how we try to minimize things like implicit bias.

Today, I’m going to talk about five behavioral science principles that I used time and again during my time at the White House. There are many more where these came from, but I think these five are particularly relevant to the conversations you’ve been having over the course of the day.

The first principle is called social identity priming.

What it says is that people tend to act in ways that are consistent with the identities they either currently associate with or aspire to associate with.

The Red Cross ran an experiment that leveraged this insight. They were sending letters to people who had previously donated and asking whether they would be willing to donate again.

They ran an A/B test and found that when they simply reminded the recipients that they had been previous donors, priming them for their identity as charitable, generous people with golden hearts, it led to a 30 percent increase in charitable contributions.

Impressively, priming people for their identity as donors also made them more generous. It increased the size of their charitable contributions compared with the previous year.

We used this inside government as well.

For example, when trying to support people receiving unemployment insurance as they returned to work and searched for jobs, it was really important that we not refer to this population as “claimants.” We referred to them as “job seekers.”

Similarly, when creating resources and handbooks for people leaving prison, it was important for us to call this population “members of the community,” people who were also searching for work.

These small changes in framing can have a huge impact.

In terms of the relevance to end-of-life care and decision-making, there are many roles and identities that people occupy within the health ecosystem. It is extremely important that we are thoughtful about how we label people, how we talk to them and how we talk about them.

It may seem like a subtle distinction, but it can transform people’s self-perception, their mental framework and the way they ultimately make decisions.

The second principle is about user agency and control.

People really like being in the driver’s seat. That’s the takeaway here.

There is an entire body of research on the importance of maintaining agency in people’s lives, but there are three particular principles I want to communicate to you.

The first is that when people actively choose something, they tend to like it more. We choose the things we like, and then, in turn, we like the things we have chosen even more.

That’s partly because we don’t like experiencing cognitive dissonance. We like to think of ourselves as really good decision-makers. When we decide something, we like to pat ourselves on the back and say, “Yeah, that’s right. I made a great decision.”

That is important to keep in mind because it shows that people have greater investment in the choices they make.

This leads me to my second point, which is that we feel more invested in experiences we have contributed to.

This is informally known as the IKEA effect. When you get the little box with 15 million pieces in it and build an unsteady stool, it has to sit in your house for a long time.

You actually like it a little more. You might feel a little more affection toward that stool than toward something that arrived preassembled.

The third principle is that we predict better outcomes when we’re in the driver’s seat compared with when a process is driven by others.

There is fascinating research showing that when people were allowed to tweak an algorithm, even slightly, they were more satisfied with its results. They actually chose to use that algorithm over other algorithms they knew performed better but had not involved their input.

It’s pretty irrational behavior, right? But it shows that there is some egocentricity here. It matters. We really value our own contributions.

I think this matters when it comes to end-of-life care and healthcare generally because it is crucial that we help people maintain a sense of agency and control over their own healthcare experiences.

We have to ask ourselves: Are doctors giving patients options? Are patients allowed to choose? In cases where options aren’t available, do patients feel empowered to question a doctor’s authority or give input into what they would like their treatment plan to look like?

The third principle is related to, but still different from, agency and choice. It’s called operational transparency.

Basically, what it says is that when we pull the curtain back, even just a little, on the thought process behind the decisions we’re making, why we reached a certain conclusion or what progress we’re making toward a particular goal, it can really help build trust with the people we’re communicating with.

The City of Boston took advantage of this insight.

People were complaining about all the problems in the city. “There are so many potholes. The street signs are all messed up. No one is fixing anything. The government is just one big black box.”

What the city decided to do was create a visual map online. It identified all the places where there were problems.

“There’s a pothole here. We’re going to mark it blue, and when we fix it, it’s going to turn green.”

Or, “There’s a broken sign here.”

Remarkably, even though the city did not increase the speed with which it was making progress on any of these issues, the map led to a 40 percent increase in trust in government. It also made people much more civically engaged than they had been before.

That’s really interesting because the outcome was the same. But the city pulled the curtain back, and suddenly people were more invested.

I think you can already see the application to the work you’re all doing.

Are we giving patients or caregivers insight into why the initial diagnosis was made? Why was treatment plan A chosen over treatment plan B? Are we making sure people understand how we arrived at the conclusions we reached?

Doing this can be a really effective way of building trust with the people we are communicating with.

In the case of patients, if they feel they have been privy to the thought process behind a treatment plan, they’re more likely to adhere to it. As we know, that can be a big challenge when it comes to encouraging well-being.

The fourth principle concerns the way we construct memories.

This one is really interesting. I think we intuitively assume that when we have an experience and think back on what it was like, every moment makes an equal contribution to our overall memory of it.

But research shows that we place disproportionate weight on the most intense moment of an experience, whether positive or negative, and on the end of the experience.

This is known as the peak-end rule.

A researcher who went on to win the Nobel Prize in Economics ran a fascinating experiment in which people’s hands were plunged into a freezing bucket of ice water for 60 seconds.

There was another condition in which people’s hands were plunged into ice water for 90 seconds. For 60 seconds, the water was as cold as it was in the first condition. Then, for the final 30 seconds, the temperature was increased by just one degree Celsius.

Let’s look at the two conditions again.

In condition A, your hand is in ice water for 60 seconds.

In condition B, your hand is in ice water for 90 seconds, but the final 30 seconds are slightly less unpleasant.

Remarkably, when people were asked which experience they preferred, they actually preferred the 90-second one.

That is totally wild because the overall amount and duration of discomfort are much greater. Yet the fact that the pain eased just a little during the last 30 seconds was disproportionately weighted in their construction of the memory.

The reason this matters is that when we think about how we design different procedures, many of which can be really unpleasant, it is important that we take these insights into account.

I don’t want to get into any of the details, but researchers have conducted these tests with procedures like colonoscopies, which can be extremely unpleasant. When you ease the experience toward the end, people’s memories of it are actually much better, and they are more likely to return for important follow-up visits and annual exams.

We want to be really thoughtful about accounting for this insight when designing procedures that are important for patients but are not necessarily pleasant.

Finally, this is one of my favorite principles: the power of the messenger.

This one is self-evident in terms of what it means, but what we find in behavioral research is that the messenger really, really matters.

There are a couple of examples of us testing this in the White House context.

At the time, and unfortunately still today, there was a lead-in-water crisis in Flint, Michigan. When this emergency broke, the federal government was tasked with creating fact sheets about water safety to make sure important information reached Flint residents.

People needed to know how to engage with their water properly so it would be safe for themselves and their families.

This was particularly important because misinformation was spreading. For example, some families were being told, “If you boil your water, then you’re fine.” In fact, that makes the situation much worse when it comes to lead.

We worked together on these fact sheets, making them comprehensive and clearly explaining where people could get resources.

But then there was the final-mile problem: Who delivers the message?

Typically, when the Environmental Protection Agency signs off on a fact sheet and puts its seal on something, it carries a great deal of legitimacy and weight. So we put that seal on there.

However, considering the context at the time, there had just been a huge breach of trust between Flint’s local government and its residents. It was likely that there would be a spillover effect involving the federal government and that residents would not necessarily trust us as the messenger.

Instead, the EPA mobilized canvassers from within the Flint community, including people from the Red Cross, the local YMCA and churches. These were people whom Flint residents already trusted.

They reviewed and endorsed the fact sheets, then knocked on doors and said, “Look, I really care about you as a member of my community. Here is a fact sheet that will help you and your family stay safe.”

We also found this principle useful when trying to motivate low-income students to fill out the Free Application for Federal Student Aid and seek financial aid for college.

These students lived in housing units assisted by the Department of Housing and Urban Development. We initially sent letters from the government agency.

Then we convinced Michelle Obama and her office that perhaps the postcards should come from her instead.

She shared a personal testimonial about feeling a little like the underdog and being unsure whether she had what it took to attend Princeton.

As you can imagine, anything Michelle Obama touches turns to gold. In this particular case, it was extremely effective in motivating students to complete the Free Application for Federal Student Aid.

In the context of the end of life, we need to be really thoughtful about who the messengers are.

I think we know the messenger matters, but we might not appreciate how much of an outsized influence the messenger can have on the decisions people make.

Most importantly, the messenger can influence whether people are willing to make a decision at all. When people don’t feel psychological safety, it can deter them from making a decision altogether.

I consider this one of the most important principles.

To summarize, behavioral science can help enhance policy design and shed light on how people make end-of-life decisions and how we think about end-of-life care.

Again, I’m no expert in those particular areas, but I do think behavioral science has a lot to offer.

It is especially important that we account for the insights scientists have been gathering over the years to make sure that when we design policies and programs, they are optimized with those insights in mind.

Thank you so much. I really appreciate it.