Before heading to the grocery store, you make a list of the items you need. Eggs, milk, and bread. “I won’t buy anything other than these,” you tell yourself. As you wander down the store's aisles, a large sale sign catches your eye: there’s a shelf of all your favorite snacks, labeled 2 for $8. “What a steal!” you think to yourself, grabbing four packages without realizing that the original price was $3.99 per item. Once you get home, you realize you spent more than you intended and wonder how this happened yet again. Would you have bought the items if you realized you saved a mere $0.02?

This is an example of choice architecture, which means that our decisions are influenced by how choices are presented.¹ To this end, a choice architect is someone who creates the environment that influences decision-making. In the situation above, consumer decisions are influenced by directing their attention to a specific sale item and making it look like a good deal. 

Choice architecture is related to libertarian paternalism and nudge theory, which proposes that positive reinforcement and implicit suggestions can subtly influence behavior.² However, choice architecture is not always used to benefit those making decisions.¹ While it’s often leveraged to promote desirable behaviors, like choosing healthier food choices or improving public health outcomes, the effects of choice architecture can also make people do things that are not in their best interest, like buying things they don’t need. Understanding the impacts and issues with choice architecture is key to applying this behavioral science principle both effectively and ethically.

Choice Architecture: The presentation of choices in different ways to impact decision-making.

Choice Architect: Someone who frames information and designs the presentation of choices. Many people turn out to be choice architects without realizing it.

Libertarian Paternalism: The idea that it is both possible and legitimate for institutions to influence behavior while also respecting freedom of choice (i.e. without coercion).

Nudge: Any aspect of choice architecture that alters behavior in a predictable way without forbidding options or significantly changing economic incentives.

Reinforcement: The process of strengthening the likelihood of a future behavior. Positive reinforcement adds a desirable consequence, while negative reinforcement removes an undesirable consequence associated with the behavior.

The concept of choice architecture first emerged with Richard Thaler and Cass Sunstein’s 2008 book Nudge: Improving Decisions about Health, Wealth, and Happiness.² Thaler coined the term “choice architecture” to describe how insights from behavioral economics could be used to influence choices without changing their objective values. In the context of Nudges, choice architecture was said to minimize biases that result from bounded rationality. Limitations such as cognitive capabilities, the difficulty of the problem, and the time available to make a decision could be overcome if choice architects nudge humans toward beneficial choices.

Choice architecture has been used throughout history, long before Thaler and Sunstein named it and explored the concept in depth. For example, grocery stores have strategically placed products to influence shopping behavior for decades. When exploring and expanding on these examples of historic choice architecture, Thaler and Sunstein drew inspiration from cognitive scientist and design researcher Donald Norman’s 1990 book, The Design of Everyday Things.¹ The main lesson from Norman’s book was that designers must remember humans are confronted daily with many choices and cues, so products must be designed for ease of use. Thaler and Sunstein aimed to develop the same idea for choice architects: by detailing principles of effective choice architecture, the researchers wanted to help choice architects better design their environments to match how humans behave. To achieve this goal, Thaler and his colleagues highlighted six “tools” for choice architects: (1) understanding defaults; (2) expecting errors; (3) providing feedback; (4) understanding mappings; (5) structuring complex choices; and (6) incentivizing.

Defaults (point 1) are defined as the choices that people must take active steps to avoid, such as being enrolled in student health insurance.¹ To combat defaults, choice architects can require people to make their own choices, such as developing the student health insurance form to be empty, requiring students to either opt in or opt out by checking one of two boxes.

As for understanding mapping (point 4), this refers to exploring the different ways that the presentation of information can influence option comparison.¹ For example, mapping choices between ice cream flavors is much easier and less risky than mapping choices of health treatments. The researchers suggest that good choice architects should improve people’s ease of mapping choices, which can be done by making information about the various options more comprehensible.

Finally, choice architects should be cognizant of how they structure complex choices (point 5), since people are more likely to adopt simplifying strategies as the number of choices increase.¹ This is because facing a number of options increases feelings of cognitive overwhelm and contributes to decision fatigue, which makes us less likely to evaluate our choices carefully and more likely to resort to mental shortcuts. As options increase in number and become more complex, choice architects have more opportunities to present information in a way that will be more likely to influence behavior.

Starting out with the broader theory of nudges, Thaler and colleagues moved on to detail the six principles of good choice architecture. By outlining these principles and illustrating them with real-world examples, the goal of choice architecture was to help policymakers design physical environments that better align with human behavior, ultimately reducing cognitive biases and improving behavioral outcomes.

What are the three types of choice architecture?

The three main types of choice architecture are defaults, information restructuring, and information feedback.¹⁶ Defaults are the option that will be selected automatically if no one intervenes to change it. Information restructuring is all about presenting information in a format that is easy for people to process—this requires a solid understanding of what people want to achieve and the limitations that get in the way. Finally, information feedback refers to the linking of action and outcome to have people learn from their actions and change their behavior. Each of these tools is commonly applied through the use of nudges to help subtly construct people’s choices to influence their behavior.

What is the difference between a nudge and choice architecture?

The definition of choice architecture is the practice of designing how choices are presented to influence choice, while nudges are the actual interventions that steer people in a particular direction. More precisely, a nudge is a specific strategy within choice architecture. Think of choice architecture as the overall layout and design of a grocery store and nudges as the subtle tactics used within the store to influence choice—like putting certain products at eye level or making shoppers walk past tempting snacks to get to essentials like milk and eggs.¹⁷ One key feature of nudges is that they do not limit people’s freedom of choice but are simply intended to guide people’s decisions in one way or another. Other forms of choice architecture can include techniques that outright limit what people can choose, like banning plastic bags so people must purchase reusable bags or bring their own from home.

How can message framing influence consumer behavior?

Message framing is a common form of choice architecture that involves describing information in a way that influences consumer decisions. For example, messages can be framed to emphasize the consumer’s potential gains or losses, with gain-framed messages highlighting the benefits of buying a certain product and loss-framed messages typically highlighting the risks or drawbacks of not buying the product. Research suggests that the effects of message framing can have a powerful influence on consumer behavior. For example, a classic field experiment found that credit card holders were much more likely to use their cards after receiving messages that framed the card benefits in terms of the losses they could suffer from not using them.¹⁸ In fact, the people who received loss-framed messages charged more than twice as much to their credit cards as people who received gain-framed messages.

Richard Thaler

Richard Thaler has researched applied decision-making for over four decades.³ Thaler’s work has focused on how decisions made by both individuals and institutions alike are influenced by cognitive limitations and biases, refuting the assumption of economic theory that humans always act rationally and selfishly. He has identified many influences on behavior such as bounded rationality, social preferences, and lack of self-control, to name a few. Thaler was awarded the 2017 Nobel Prize for Economics for his contributions to nudge theory. Specifically, the award was presented because Thaler’s contributions have bridged economic and psychological decision-making analyses, synthesizing research that has been critical for the expansion of behavioral economics.

Cass Sunstein

Working alongside Thaler to develop nudges and expand on choice architecture, Cass Sunstein has not only worked on behavioral economics, but he is also an American legal scholar.⁴ Sunstein received his J.D. from Harvard Law School and has since worked on administrative, behavioral, constitutional, environmental, employment, and labor law. His impressive history landed him a position with the Obama administration from 2009 to 2012, working as the Administrator of the White House Office of Information and Regulatory Affairs. Sunstein has advised officials at the European Commission, the World Bank, and the United Nations regarding law and public policy. He is also an advisor to the UK’s Behavioural Insights Team, further applying his expertise in behavioral economics.

Supported by the 2017 Nobel Prize for Economics, nudge theory and choice architecture have had a large influence on behavioral economics research, as well as policy making.² Advocates of libertarian paternalism have supported choice architecture to nudge consumers toward desirable behaviors such as choosing healthier foods, registering as organ donors,⁵ signing up for health care, and even saving for retirement.⁶ Choice architecture also plays an important role in guiding prosocial behavior by subtly encouraging people to take actions that benefit others.

Choice architecture and COVID-19

During the COVID-19 pandemic, encouraging vaccine uptake, testing, social distancing, and self-isolation was incredibly important for managing the spread of disease and getting the virus under control. Choice architecture was implemented in several areas to guide people toward these behaviors.¹⁹ For example, the Singapore government enacted a policy on foreign travel requiring that people leaving the country cover their own medical expenses if they contracted COVID instead of issuing an outright travel ban. This effectively discouraged people from traveling without limiting their freedom of choice. 

Similarly, other countries applied defaults to encourage vaccinations during the pandemic. Poland, for example, pre-assigned vaccination dates, requiring that individuals actively cancel or change their appointments. This encouraged people to stick with their assigned schedules. Around the world, nudges were used to remind people about behaviors like mask-wearing and social distancing, often alongside social proof to position these behaviors as accepted norms. The widespread success of choice architecture during the pandemic, especially in countries with diverse cultural backgrounds, suggests that these behavior change methods can effectively drive large-scale public policy.

Nudges in the digital world

Choice architecture also has a significant role to play in influencing online decision-making. Unfortunately, these techniques aren't always employed to benefit the user. For example, recent research suggests that website hosts can alter the choice architecture of cookie banners to nudge visitors into sharing their data. One study found that manipulating cookie banners can increase consent by as much as 17%.²⁰ These manipulations can be subtle, like making the “accept” button large and green while the “decline” button remains small and grey. 

Many cookie banners also frame cookie consent positively, focusing on the benefits of sharing user data for improving the user experience rather than offering an honest explanation of what the company intends to do with their data. One of the main issues with these micro-environmental choice architecture interventions is that they subtly change the digital environment without significantly changing people’s options. This makes it difficult to call out companies for defying data protection laws—they are technically complying with regulations by asking users for consent before storing cookies on their browsers.

Choice architecture in policymaking

Choice architecture has also been impactful on significant policy-making groups: Sunstein was appointed to lead the White House’s Office of Information and Regulatory Affairs (OIRA) in 2009, while the Behavioural Insights Team was established under the UK’s Cabinet Office in 2010.⁷ Other countries soon followed in forming similar organizations with a focus on applying behavior science to improve public policy. For instance, international institutions such as the World Bank, the European Union, the Organisation for Economic Co-operation and Development, and UN agencies adopted behavioral insight units in the mid-2010s. These developments reflected a growing interest in using behavioral science to inform and improve public policy. To this day, teams at the federal level, such as Germany’s office of Federal Chancellor in Policy Planning Unit and the United States Office of Evaluation Sciences (OES), work directly with implementing agencies to diagnose, design, test, and evaluate interventions informed by choice architecture—though the changing political landscape in these respective countries may influence the future direction of these teams.

Since the implementation of nudges and choice architecture, research by behavioral economists, including Sunstein and Thaler, has shown that nudges can be more cost-effective than traditional policies in the domains of financial security, education, job training, and health, to name a few.⁸ The comparative advantages of choice architecture will normally be larger when policy objects are to change the daily behaviors of people who are making biased, rushed, and imperfect decisions. In this context, imperfection is determined by reference to the welfare of said people. The researchers acknowledge that nudges often work with other tools, such as simplifying procedures, but show that choice architecture can be beneficial nonetheless, regarding both cost and efficacy.

While Thaler and colleagues are credited with providing the first detailed account of choice architecture and outlining the six associated principles,¹ other behavioral scientists divided the “tools” into two categories: (1) those that structure the choice task; and (2) those that describe the choice options.⁹ The former addresses the issue of what information and options choice architects should present, and the latter focuses on how to present said options. When structuring the choice task, choice architects should consider the number of alternatives, decision aids such as technology, defaults, choice over time, and how task structure affects how people consider their options. As for describing choice options, choice architects should carefully divide options into groups and design the attributes associated with their options.

Despite the popularity of nudge theory and choice architecture, there is still some criticism. Since the original publication of Thaler and Sunstein’s book, debates over the effectiveness of nudges have suggested that they may not work the same across all situations. Recent literature reviews show that choice architecture can be more impactful in some areas than others—for example, food choices seem to be particularly responsive to choice architecture interventions.²² Researchers point to the importance of evidence-based implementations when applying choice architecture to influence behavior, especially when it comes to scaling nudge techniques beyond controlled experiments to real-world environments. Policymakers intent on using choice architecture also have to carefully consider the ethics of these implementations.

The ethics of choice architecture

Although libertarian paternalism is built on the foundation that behavior can be influenced without any coercion, the main concern is whether choice architecture is entirely ethical.¹⁰ Sunstein, an active advocate for nudge policies with significant influence over federal regulations as head of OIRA, faced many of these criticisms. In fact, he was famously labeled “the most dangerous man in America'' by radio talk-show host Glenn Beck back in 2010. Critics like Beck often interpreted Sunstein’s ideas as attempts to limit people’s personal freedoms.

While Sunstein’s unfortunate label was largely hyperbolic, criticisms of his ideas have some basis in reality. By designing options to overcome decision-making biases, there may be times when choice architects must impose costs on agency, such as limiting choices.¹¹ Although individuals are free to choose between the choices presented to them, the original design of choices may threaten autonomy by outright banning certain options entirely or influencing people to the extent that they act in ways that don’t align with their usual decision-making processes. Economist and philosopher Luc Bovens goes so far as to claim that choice architecture leaves people with “fragmented selves” that behave one way under nudges and another without such influences.¹⁰ The main point of criticisms around the ethics of choice architecture is that if people become used to being nudged, they may become passive and tolerable of more controlling tactics, similar to a “Big Brother” world.

The ethics of choice architecture continues to be an important point of concern as the world becomes increasingly digitized. As more decision-making occurs online, dark patterns are becoming more and more common. These misleading design techniques trick users into making unintended decisions, like accepting extra fees or signing up for subscriptions by accident. With this in mind, behavior change interventions must be careful to avoid deceiving people into behaving against their own will, especially when target behaviors prioritize corporate gain over individual benefit.

Decision rights under AI choice architecture

Decision rights have become a hot topic as organizations increasingly rely on AI to support decision-making processes. “Intelligent Choice Architecture” (ICA) is transforming enterprise decision-making, going beyond simply making recommendations by actually presenting options in a way that reduces the influence of human cognitive biases.²¹ While this has great potential to help leaders make better decisions, it also shifts the power from those who make decisions to those who design decision-making environments with the help of AI. This raises several important questions: who should have the authority to design ICAs? And who is responsible when ICAs lead to poor or harmful outcomes? This becomes an even bigger problem when we consider that AI-driven decision environments can pick up human biases from training data, meaning they are not the neutral, objective systems they are often made out to be.

Nudging as a rationality paradox

Nudging has been labelled by some as a “rationality paradox,” such that it represents an approach that emphasizes bounded rationality without reflecting on its own limits to rationality.¹² If individual decision-making is limited in rationality, then the policy groups composed of individuals are also limited. Although governments possess superior resource bases relative to those of individuals—in terms of knowledge, finances, organization, and authority—and have safeguards in place, at the heart of these organizations are people who are individually limited in their rationality. In other words, policymakers themselves can be biased. This means that bounded rationality cannot be dismissed based on having more resources, which lends to the question of whether such institutions are in a position to nudge others through policy.

Defaults in organ donation systems

The discrepancy between demand and supply of transplantable organs is a global health problem.¹³ Thus, policy debates in such countries focus on solving this shortage and mostly compare two systems: explicit consent versus presumed consent. In the former, the default is that nobody is a donor unless they explicitly register and opt-in. The main drawback with the explicit consent system is that donation rates tend to be low and less than reported willingness to donate. On the other hand, the presumed consent system registers every adult as a donor unless they explicitly opt-out. However, a drawback with presumed consent is that families of a deceased relative may object on their behalf and withdraw consent, which lowers donation rates, and governments can be seen as taking advantage of citizens’ consent.

To address these issues, researchers from the Netherlands considered a third option: a system of mandated choice.¹³ In this situation, there is no default, and citizens must actively choose. The researchers used hypothetical choices from a national survey in 2011 and chose to sample citizens of the Netherlands—which, at the time, had an explicit consent system for organ donations. 2069 individuals aged 16 years and older were divided into three groups representing explicit consent, presumed consent, and mandated choice systems. Participants were asked whether they would stay with the default or actively opt-in or out for explicit and presumed consent, respectively, and whether they would opt in or out for mandated choice. All participants were also given the choice to answer that they didn’t know.

The study results showed that—at least in the Netherlands—both systems of mandated choice and presumed consent were more effective than explicit consent in increasing the number of registered donors.¹³ 

The researcher in the Netherlands study suggested that people may be more inclined to register in the mandated choice system than the explicit consent system due to humans’ propensity to procrastinate when making decisions. Such an explanation would support the power of defaults and encourage choice architects to mandate choices, as suggested by Thaler and colleagues in the first principle of choice architecture.¹ While results from a study in the Netherlands do not guarantee equivalent results in other countries that are experiencing an organ shortage, the results suggest that a mandated choice system should be considered in policy debates and point to the power of defaults in which people are opted out. Since this study, the Netherlands has enacted an opt-out organ donation system, presuming that all adults are donors unless they actively register to opt-out. 

Save more tomorrow: easy and painless retirement

As average life expectancies increase, the time spent in retirement commonly increases as well, subsequently increasing the money required for security and comfort. The life cycle theory in economics holds that the general population will solve this problem like economists would: by calculating their required savings and distributing their finances to match this.¹⁴ However, this is not always the case for two reasons: it is difficult to make such calculations, and in the case that correct calculations are made, people may lack the required self-control to appropriately reduce their consumption rates.

Richard Thaler and behavioral economist Shlomo Benartzi recognized that people have a loss-aversion bias: when people are forced to cut their present spending to save for their pension, they perceive a loss instead of a future gain.¹⁴ This is where the problem of self-control ties in. Additionally, people tend to procrastinate when making decisions, especially for unpleasant or difficult tasks. While people may want to save for retirement, they keep putting it off. This leads to the status quo bias, allowing people to remain in their current situation regardless of what they know the rational optimal choice is.

To address these issues and to help people save for retirement, Thaler and Benartzi designed the Save More Tomorrow plan.¹⁴ The intervention implements choice architecture through seemingly simple but very influential components. The first aspect is automatically enrolling all employees in a savings plan right when they become eligible. This means that while employees can choose to opt out of the plan, their default choice is a commitment. The second aspect is that increases in savings rates are linked to future pay raises. This way, take-home pay always increases and deductions are taken from money that the employee was not used to having, helping overcome loss aversion.

Thaler and Benartzi’s choice architecture intervention has proven to nudge a higher savings rate for long-term gains and has appealed to people who were not otherwise interested in a savings plan.¹⁴ Designed in 2004, it was enforced as part of the Pension Protection Act of 2006, encouraging companies to adopt the core principles.¹⁵

Dancing to the tunes of a choice architect

If you’re interested in more examples of choice architecture at work, look no further! This article provides a great breakdown of how choice architects can influence behaviors through their design of choices, as well as how we can use our awareness of choice architecture to move forward.

Giving people the tools to nudge themselves

Moving past research on the effectiveness of nudges, some people wonder if implicitly influencing others’ behaviors is ethical. Who is in a position to make the decisions of a choice architect for someone else? An interview with Samuli Reijula, a researcher who has studied how nudging interventions can be used as an individual tool, provides some alternatives to consider. If nudging others is indeed unethical, perhaps we can nudge ourselves!

1. Thaler, R. H., Sunstein, C. R., & Balz, J. P. (2013). Choice architecture. The Behavioral Foundations of Public Policy, 428-439.
2. Thaler, R. H., & Sunstein, C. R. (2008). Nudge: Improving Decisions about Health, Wealth, and Happiness. Yale University Press.
3. Duigan, B. (2020, September 8). Richard Thaler. Encyclopedia Britannica. https://www.britannica.com/biography/Richard-Thaler
4. Harvard Law School. (n.d.). Cass R. Sunstein. https://hls.harvard.edu/faculty/directory/10871/Sunstein
5. Johnson, E. J., & Goldstein, D. (2003). Do defaults save lives? Science, 302(5649), 1338-1339.
6. Benartzi, S., Peleg, E., & Thaler, R. H. (2013). Choice architecture and retirement saving plans. The Behavioral Foundations of Public Policy, 245-263.
7. Afif, Z. (2017, October 25). “Nudge units” - Where they came from and what they can do. World Bank Blogs. https://blogs.worldbank.org/developmenttalk/nudge-units-where-they-came-and-what-they-can-do
8. Benartzi, S., Beshears, J., Milkman, K. K., Sunstein, C. R., Thaler, R. H., Shankar, M., ... & Galing, S. (2017). Should governments invest more in nudging? Psychological Science, 28(8), 1041-1055.
9. Johnson, E. J., Shu, S. B., Dellaert, B. G. C., Fox, C., Goldstein, D. G., Häubl, G., … & Weber, E. U. (2012). Beyond nudges: Tools of a choice architecture. Marketing Letters, 23(2), 487-504.
10. Selinger, E., & White, K. (2011). Is there a right way to nudge? The practice and ethics of choice architecture. Sociology Compass, 5(10), 923-935.
11. Mitchell, G. (2004). Libertarian paternalism is an oxymoron. Northwestern University Law Review, 99(3), 1245-1287.
12. Lodge, M., & Wegrich, K. (2016). The rationality paradox of nudge: Rational tools of government in a world of bounded rationality. Law & Policy, 38(3), 250-267.
13. Van Dalen, H. P., & Henkens, K. (2014). Comparing the effects of defaults in organ donation systems. Social science & Medicine, 106, 137-142.
14. Thaler, R. H. & Benartzi, S. (2004). Save More Tomorrow: Using behavioral economics to increase employee saving. Journal of Political Economy, 112(1), 164-187.
15. Save More Tomorrow. (2017). Shlomo Benartzi. http://www.shlomobenartzi.com/save-more-tomorrow
16. Camilleri, A., & Larrick, R. P. (2015). Choice architecture. Emerging Trends in the Social and Behavioral Sciences. Hoboken, NJ: John Wiley and Sons. DOI: 10.1002/9781118900772.etrds0039 
17. Thorndike, A. N. (2024, August 9). Small changes in food 'choice environments' in stores can make a healthy difference. STAT. https://www.statnews.com/2024/08/09/small-changes-food-choice-environments-healthy-food/ 
18. Ganzach, Y., & Karsahi, N. (1995). Message framing and buying behavior: A field experiment. Journal of Business Research, 32(1), 11-17. https://doi.org/10.1016/0148-2963(93)00038-3 
19. Czerwonka, M., & Siembida, J. (2022). Usage of behavioural innovation in the context of the COVID-19 crisis. Annales Universitatis Mariae Curie-Skłodowska, sectio H–Oeconomia, 56(3), 7-17. DOI:10.17951/h.2022.56.3.7-17 
20. Bauer, J. M., Bergstrøm, R., & Foss-Madsen, R. (2021). Are you sure, you want a cookie?–The effects of choice architecture on users' decisions about sharing private online data. Computers in Human behavior, 120, 106729. https://doi.org/10.1016/j.chb.2021.106729 
21. Schrage, M., & Kiron, D. (2025, January 28). The Great Power Shift: How Intelligent Choice Architectures Rewrite Decision Rights. MIT Sloan Management Review. https://sloanreview.mit.edu/article/the-great-power-shift-how-intelligent-choice-architectures-rewrite-decision-rights/ 
22. Mertens, S., Herberz, M., Hahnel, U. J., & Brosch, T. (2022). The effectiveness of nudging: A meta-analysis of choice architecture interventions across behavioral domains. Proceedings of the National Academy of Sciences, 119(1), e2107346118. https://doi.org/10.1073/pnas.2107346118