Choice Architecture and Nudging: Defaults, Friction, and Decision Environments

Last Updated June 21, 2026

Choice Architecture and Nudging appears here not as a collection of behavioral tricks, but as a serious field of decision-environment design. It asks how options are arranged, how defaults shape participation, how friction changes behavior, how salience directs attention, how simplification affects access, and how institutional design influences action before people experience themselves as making a deliberate choice.

The field matters because no choice environment is neutral. A retirement plan must decide whether enrollment is automatic or optional. A public agency must decide how forms are structured. A platform must decide which button is visible, which option is hidden, and how easy cancellation is. A school, hospital, city, workplace, marketplace, or climate program must decide how people encounter choices. Choice architecture makes those design decisions visible, analyzable, and ethically accountable.

GitHub Repository

The companion repository for this knowledge series should be created as choice-architecture-nudging-code under the Content-Catalyst-LLC GitHub organization. It should use article-level folders under articles/, shared reusable methods under _shared/, and documented synthetic-data workflows so the pillar can support conceptual articles, default-effect simulations, friction-cost models, digital interface audits, choice-environment diagrams, intervention-evaluation templates, and reproducible demonstrations of nudging and behavioral design.

Choice Architecture as a Core Behavioral Science

Choice architecture is a core behavioral-science concept because it makes the environment of choice visible. People do not choose from abstract option sets floating outside context. They choose from menus, forms, interfaces, defaults, deadlines, pathways, public programs, institutional rules, product pages, enrollment systems, recommendation feeds, signs, messages, and social environments. The arrangement of those environments shapes behavior.

Choice architecture does not claim that people lack agency. Rather, it shows that agency is exercised under conditions. A person may choose differently depending on whether an option is defaulted, whether the form is simple, whether the cost is immediate, whether the benefit is salient, whether the process is confusing, whether others appear to participate, and whether exit is easy. The chooser matters, but the architecture matters too.

This makes choice architecture a bridge between psychology and design. Cognitive psychology explains attention, memory, cognitive load, and framing. Behavioral economics explains bounded rationality, defaults, loss aversion, and present bias. Public policy explains institutional systems. Design explains interfaces, forms, pathways, and environments. Choice architecture asks how these layers combine in real decision settings.

Nudging as Behavioral Intervention

Nudging refers to behavioral intervention that changes the way choices are presented while preserving formal freedom of choice. A nudge may use a default, reminder, simplification, social norm message, timely prompt, feedback signal, or salience cue to make one action more likely. Nudges became influential because many behavioral problems arise not from lack of options, but from attention, inertia, complexity, timing, friction, and context.

Yet nudging should not be treated as automatically harmless or automatically effective. A nudge can help people enroll in beneficial programs, save for the future, attend appointments, conserve energy, complete forms, or understand risk. It can also steer people without transparency, exploit vulnerability, reduce meaningful consent, or serve institutional goals rather than human welfare. Effectiveness alone does not settle legitimacy.

For that reason, this series treats nudging as part of a broader field of behavioral design. It includes nudges, boosts, choice architecture, institutional design, friction reduction, behavioral public policy, digital interface design, and ethical intervention governance. The goal is not simply to make people choose differently. The goal is to understand when environmental design supports clearer, fairer, more accountable, and more agency-respecting choices.

Choice Architecture as Institutional and Design Practice

Choice architecture is not limited to psychology experiments or policy units. It is a practical reality of institutions. Every public agency, school, hospital, workplace, financial system, digital platform, marketplace, and civic process has to arrange choices. Some options are visible; others are hidden. Some actions are easy; others require effort. Some defaults apply automatically; others require enrollment. Some processes respect users; others create burden.

This means institutions are always behavioral designers, even when they do not use that language. A form can discourage access. A deadline can structure urgency. A default can increase uptake. A platform can make cancellation difficult. A public service can reduce friction or multiply it. A sustainability program can make low-carbon action easy or leave it aspirational. Choice architecture helps analyze these design choices.

The field is strongest when it treats design as both behavioral and ethical. A decision environment should be evaluated by what it makes likely, who benefits, who bears friction, who understands the choice, who can exit, who is vulnerable, and whether the design can be publicly justified. Choice architecture therefore belongs not only to behavioral science, but also to governance, ethics, public administration, technology design, and institutional accountability.

What Choice Architecture and Nudging Studies

Choice Architecture and Nudging studies how environments shape decisions. At the cognitive level, it examines attention, salience, framing, cognitive load, defaults, simplification, memory, effort, timing, and comprehension. At the motivational level, it studies incentives, loss framing, commitment, feedback, reminders, progress signals, and perceived difficulty.

At the institutional level, it studies forms, enrollment systems, administrative burden, benefit access, compliance, public-service design, program uptake, regulation, disclosure, and accountability. At the digital level, it studies interface design, personalization, recommendation systems, subscription flows, consent screens, notifications, dark patterns, algorithmic nudging, and platform governance.

The field also studies evaluation. Choice architecture can change behavior, but behavioral change must be measured. Did the default increase uptake? Did simplification improve access? Did the nudge work only for one group? Did it produce unintended consequences? Did people understand the choice? Did it increase autonomy or merely increase compliance? These are empirical and ethical questions together.

What This Pillar Covers

This pillar covers the major domains through which choice architecture and nudging are studied. It includes defaults, opt-ins, opt-outs, salience, simplification, cognitive load, friction, hassle costs, administrative burden, ordering effects, menu design, prompts, reminders, feedback, social norm nudges, disclosure, commitment devices, boosts, nudges, digital choice architecture, dark patterns, algorithmic nudging, public-policy applications, measurement, field trials, A/B testing, heterogeneity, spillovers, ethics, transparency, consent, autonomy, and governance.

These domains differ in scale. Some concern micro-design: a button, label, prompt, reminder, or default checkbox. Others concern institutional architecture: enrollment systems, public benefits, health programs, financial products, energy systems, digital platforms, and regulatory processes. Choice architecture connects these scales by asking how the design of options influences action.

The series also treats choice architecture as a field with ethical stakes. Since every decision environment shapes behavior, the issue is not whether influence exists. The issue is whether influence is visible, justified, reversible, proportionate, equitable, evidence-based, and aligned with human agency. That ethical question runs through the entire pillar.

Mathematics, Computation, and Modeling in Choice Architecture

Mathematics and computation help choice architecture make design assumptions explicit. A simplified choice model can represent the probability that a person selects an option as a function of value, default status, salience, effort cost, trust, social cues, and comprehension:

where:

In this simplified model, \(V_i\) represents perceived value, \(D_i\) default status, \(S_i\) salience, \(F_i\) friction, \(T_i\) trust, \(N_i\) social-norm signal, and \(K_i\) comprehension or clarity.

A default effect can be represented by comparing opt-in and opt-out regimes:

A friction-cost model can represent how completion probability declines as the number or difficulty of required steps increases:

These formulations do not reduce human choice to equations. They clarify central behavioral insights: options are not evaluated only by objective value. They are shaped by defaults, visibility, effort, timing, trust, norms, comprehension, and institutional design. Computation is especially useful when choice environments contain many users, heterogeneous responses, digital records, repeated decisions, and multiple intervention conditions.

R supports experiments, field trials, treatment-effect estimation, survey analysis, policy evaluation, and visualization. Python supports simulations, digital-interface analysis, behavioral analytics, A/B testing workflows, friction models, and platform-data pipelines. SQL supports treatment assignment, user flows, decision records, intervention metadata, consent states, and audit trails. Other languages can support typed records, reproducible tools, interface prototypes, formal constraints, and governance systems.

Major Domains of Choice Architecture and Nudging

Choice Architecture and Nudging includes several major domains. Foundational choice architecture studies decision environments, defaults, salience, simplification, cognitive load, ordering, friction, feedback, and the claim that no choice environment is neutral. Nudging studies small design changes that alter behavior while preserving formal choice.

Digital choice architecture studies interfaces, platforms, consent flows, recommender systems, personalization, notifications, subscription design, dark patterns, and algorithmic nudging. Institutional choice architecture studies public benefits, health systems, retirement savings, tax compliance, energy programs, education systems, and administrative burden.

Measurement and evaluation study whether interventions work, for whom, under what conditions, and with what unintended consequences. Ethics and governance study transparency, consent, autonomy, manipulation, vulnerability, equity, dark-pattern regulation, and public accountability for behavioral design.

Why Choice Architecture and Nudging Matters

Choice architecture matters because design changes behavior even when rules do not formally restrict choice. Automatic enrollment can increase participation. Simplified forms can improve access. Timely reminders can reduce missed appointments. Default settings can change savings, privacy, energy use, subscription behavior, and public-program uptake. Small design choices can produce large population effects.

Nudging matters because many real-world barriers are behavioral rather than purely informational. People may intend to act but procrastinate. They may want a benefit but fail to complete the process. They may support sustainability but remain shaped by convenience. They may choose the visible option because attention is limited. Choice architecture helps explain how these small environments shape large systems.

The field also matters because behavioral influence can be used well or badly. A public agency can reduce burden and increase access. A platform can make cancellation difficult. A health system can support adherence. A financial product can exploit inertia. A sustainability program can make lower-impact choices easier. Choice architecture therefore requires not only design skill, but ethical and institutional accountability.

Choice Architecture, Nudging, and Human Agency

Choice architecture changes how agency is understood. It shows that freedom of choice is not only about whether options formally exist. It is also about whether options are understandable, visible, accessible, reversible, comparable, and free from manipulative friction. A person can technically have a choice while being steered by complexity, confusion, asymmetry, or hidden design.

This does not mean that all influence is illegitimate. Good choice architecture can support agency by reducing burden, clarifying consequences, improving feedback, making beneficial options easier, and helping people act on commitments they already value. A reminder can help. A simpler form can respect people’s time. A default can protect people from costly inaction. A boost can increase capacity.

The ethical challenge is to distinguish support from manipulation. Choice architecture should be judged by whether people can understand the design, contest it, exit it, and endorse it under reflective conditions. It should support agency under real-world constraints rather than exploit those constraints.

Choice Architecture and Nudging Pillar Map

The map below organizes the Choice Architecture and Nudging knowledge series into conceptual domains, moving from foundations of decision-environment design toward nudging tools, digital choice architecture, public-policy applications, measurement, ethics, dark patterns, algorithmic influence, climate adaptation, public trust, inequality, digital rights, collective action, and future governance. The article titles below are intentionally left unlinked until their individual pages exist.

Measurement, Evaluation, and Behavioral Design Practice

Choice architecture requires measurement because plausible design changes do not always work. A reminder may be ignored. A default may increase uptake but reduce understanding. A social norm message may backfire if it normalizes the unwanted behavior. A simplification may help one group while failing another. A dark pattern may increase conversion while harming trust and welfare.

A serious behavioral design process should specify the target behavior, target population, mechanism, decision point, intervention feature, ethical constraints, expected effect, and evaluation method. It should measure not only immediate uptake, but also comprehension, persistence, distributional effects, unintended consequences, autonomy, and trust. Behavioral success should not be defined only as more clicks, enrollment, compliance, or conversion.

Evaluation should also consider whether the intervention helps people act on goals they would endorse under clearer conditions. Did the design reduce unnecessary friction? Did it improve access? Did it make consequences more visible? Did it preserve meaningful exit? Did it protect vulnerable users? Choice architecture is strongest when it treats behavioral design as both empirical and accountable.

Choice Architecture, Technology, and the Modern World

Technology has made choice architecture more powerful because digital environments can be personalized, tested, optimized, and changed at scale. Platforms can decide what users see, which options appear first, which defaults apply, how easy exit is, what notifications arrive, how consent is framed, what recommendations are made, and how friction is distributed. Digital systems are therefore among the most important choice environments in modern life.

This creates both opportunity and risk. Digital choice architecture can make complex decisions easier, improve access, provide timely feedback, reduce administrative burden, support learning, and help people act on long-term goals. It can also exploit attention, obscure costs, manipulate consent, personalize pressure, make cancellation difficult, and optimize institutional gain at the expense of user welfare.

A mature choice architecture of technology must therefore ask what is being optimized and for whom. Engagement is not the same as agency. Conversion is not the same as benefit. Retention is not the same as wellbeing. The future of choice architecture will increasingly depend on whether digital systems are designed, audited, and governed in ways that respect human autonomy.

Choice Architecture, Sustainability, and Public Systems

Choice architecture is central to sustainability because ecological behavior often depends on defaults, infrastructure, feedback, convenience, salience, and norms. Energy plans, transportation options, recycling systems, food choices, water use, appliance settings, building design, and consumption pathways all involve choice environments. Sustainable behavior becomes more likely when systems make it visible, easy, affordable, socially supported, and institutionally reinforced.

Yet sustainability nudges should not be treated as substitutes for structural change. A household energy comparison may help, but inefficient housing, unaffordable upgrades, poor transit, weak regulation, and unequal access still matter. Choice architecture can support sustainable behavior, but it cannot carry the full burden of sustainability policy by itself.

For Sustainable Catalyst, choice architecture is most valuable when it connects individual behavior to systems design. It asks how public systems, markets, technologies, communities, and institutions can make sustainable action more possible without blaming individuals for structural barriers. It belongs alongside regulation, infrastructure, education, public investment, and democratic governance.

R Section: Simulating Default Effects and Friction Costs

For analytical readers, R is useful for modeling default effects, friction costs, heterogeneous uptake, and intervention evaluation. The example below simulates a synthetic choice environment in which default status, salience, friction, trust, and social norms affect whether people select an option. It is not real behavioral data. It is a reproducible scaffold for thinking clearly about how choice architecture can influence action.

# Synthetic choice architecture model in R
# Educational example only.
# This script simulates default effects and friction costs in a decision environment.

# install.packages(c("tidyverse", "broom", "scales"))

library(tidyverse)
library(broom)
library(scales)

set.seed(6363)

n_people <- 2400

choice_data <- tibble(
  person_id = 1:n_people,
  perceived_value = runif(n_people, 0.20, 0.95),
  trust = runif(n_people, 0.15, 0.95),
  social_norm_signal = runif(n_people, 0.00, 0.80),
  comprehension = runif(n_people, 0.20, 0.95),

  # Randomized intervention conditions.
  default_status = rbinom(n_people, 1, 0.50),
  salience_boost = rbinom(n_people, 1, 0.50),
  simplified_process = rbinom(n_people, 1, 0.50),

  baseline_friction = runif(n_people, 0.10, 0.85)
) |>
  mutate(
    friction = pmax(0, baseline_friction - 0.25 * simplified_process),

    choice_score =
      -2.20 +
      1.40 * perceived_value +
      0.95 * default_status +
      0.70 * salience_boost -
      1.35 * friction +
      0.80 * trust +
      0.55 * social_norm_signal +
      0.65 * comprehension,

    probability_choose = 1 / (1 + exp(-choice_score)),
    chose_option = rbinom(n_people, 1, probability_choose),

    condition = case_when(
      default_status == 1 & simplified_process == 1 ~ "default_and_simplified",
      default_status == 1 & simplified_process == 0 ~ "default_only",
      default_status == 0 & simplified_process == 1 ~ "simplified_only",
      TRUE ~ "standard_choice_environment"
    )
  )

summary_by_condition <- choice_data |>
  group_by(condition) |>
  summarise(
    choice_rate = mean(chose_option),
    mean_probability = mean(probability_choose),
    mean_friction = mean(friction),
    mean_trust = mean(trust),
    .groups = "drop"
  )

print(summary_by_condition)

choice_model <- glm(
  chose_option ~ perceived_value + default_status + salience_boost +
    friction + trust + social_norm_signal + comprehension,
  data = choice_data,
  family = binomial(link = "logit")
)

print(tidy(choice_model, conf.int = TRUE, exponentiate = TRUE))

ggplot(summary_by_condition, aes(x = condition, y = choice_rate)) +
  geom_col() +
  labs(
    title = "Synthetic Choice Rate by Decision Environment",
    x = "Choice architecture condition",
    y = "Choice rate"
  ) +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 25, hjust = 1))

ggplot(choice_data, aes(x = friction, y = probability_choose)) +
  geom_point(alpha = 0.20) +
  geom_smooth(method = "loess", se = FALSE) +
  labs(
    title = "Synthetic Relationship Between Friction and Choice Probability",
    x = "Friction",
    y = "Predicted choice probability"
  ) +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal()

This workflow models a core choice-architecture intuition: a decision may depend not only on perceived value, but also on default status, salience, process simplicity, trust, social cues, comprehension, and friction. In real research, such models require careful study design, ethical review, and attention to external validity.

Python Section: Modeling Choice Environments and Nudging Regimes

Python is useful for simulating alternative choice environments. The example below compares standard, default-enhanced, simplified, and high-friction regimes. The goal is not to produce real predictions, but to show how small design features can change aggregate outcomes under different assumptions.

# Synthetic choice architecture simulation in Python
# Educational example only.
# This script compares choice outcomes across decision-environment regimes.

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

rng = np.random.default_rng(6363)

n_people = 3000

def simulate_regime(
    regime_name,
    default_effect,
    salience_effect,
    friction_level,
    simplification_effect,
    trust_shift
):
    perceived_value = rng.uniform(0.20, 0.95, n_people)
    trust = np.clip(rng.uniform(0.15, 0.95, n_people) + trust_shift, 0, 1)
    social_norm_signal = rng.uniform(0.00, 0.85, n_people)
    comprehension = rng.uniform(0.20, 0.95, n_people)

    friction = np.clip(
        rng.normal(friction_level, 0.12, n_people) - simplification_effect,
        0,
        1
    )

    z = (
        -2.15
        + 1.35 * perceived_value
        + default_effect
        + salience_effect
        - 1.45 * friction
        + 0.85 * trust
        + 0.55 * social_norm_signal
        + 0.65 * comprehension
    )

    probability_choose = 1 / (1 + np.exp(-z))
    chose_option = rng.binomial(1, probability_choose)

    return pd.DataFrame({
        "regime": regime_name,
        "perceived_value": perceived_value,
        "trust": trust,
        "social_norm_signal": social_norm_signal,
        "comprehension": comprehension,
        "friction": friction,
        "probability_choose": probability_choose,
        "chose_option": chose_option
    })

results = pd.concat([
    simulate_regime(
        "standard_choice_environment",
        default_effect=0.00,
        salience_effect=0.00,
        friction_level=0.55,
        simplification_effect=0.00,
        trust_shift=0.00
    ),
    simulate_regime(
        "default_enhanced_environment",
        default_effect=0.85,
        salience_effect=0.10,
        friction_level=0.50,
        simplification_effect=0.00,
        trust_shift=0.00
    ),
    simulate_regime(
        "simplified_supportive_environment",
        default_effect=0.25,
        salience_effect=0.40,
        friction_level=0.45,
        simplification_effect=0.25,
        trust_shift=0.10
    ),
    simulate_regime(
        "high_friction_environment",
        default_effect=0.00,
        salience_effect=0.00,
        friction_level=0.78,
        simplification_effect=0.00,
        trust_shift=-0.10
    )
], ignore_index=True)

summary = results.groupby("regime").agg(
    choice_rate=("chose_option", "mean"),
    mean_probability=("probability_choose", "mean"),
    mean_friction=("friction", "mean"),
    mean_trust=("trust", "mean"),
    mean_comprehension=("comprehension", "mean")
).reset_index()

print(summary.sort_values("choice_rate", ascending=False))

plt.figure(figsize=(10, 6))
plt.bar(summary["regime"], summary["choice_rate"])
plt.xticks(rotation=25, ha="right")
plt.ylabel("Choice rate")
plt.title("Synthetic Choice Rate Across Decision Environments")
plt.tight_layout()
plt.show()

plt.figure(figsize=(10, 6))
for regime in results["regime"].unique():
    subset = results[results["regime"] == regime]
    plt.hist(
        subset["probability_choose"],
        bins=35,
        alpha=0.35,
        label=regime
    )

plt.xlabel("Predicted choice probability")
plt.ylabel("Number of people")
plt.title("Synthetic Choice Probability Distributions")
plt.legend()
plt.tight_layout()
plt.show()

results.to_csv("choice_architecture_nudging_simulation.csv", index=False)

For analysts, designers, and policymakers, the key lesson is that choice architecture does not replace value, preference, or agency. It shows how those factors are filtered through defaults, salience, friction, trust, comprehension, and social context. A decision environment can support action, obstruct it, or manipulate it.

Interpretive Limits and Choice Architecture Cautions

Choice architecture is powerful, but it can be misused. A nudge is not automatically ethical because it preserves formal choice. A default is not neutral because users can technically opt out. A design is not legitimate merely because it increases uptake, compliance, conversion, or engagement. Behavioral influence must be evaluated by transparency, consent, autonomy, equity, accountability, and public justification.

Choice architecture can also distract from structural problems. A reminder may help people complete a form, but the form may still be unnecessarily complex. A social norm message may encourage energy conservation, but inefficient housing and unequal access still matter. A default may increase enrollment, but people may not understand the program. Nudges are often useful, but they are not substitutes for justice, infrastructure, regulation, or democratic governance.

The field is strongest when it combines behavioral realism with ethical humility. People are affected by defaults, friction, salience, and timing. That does not make them objects to be steered. Good choice architecture should reduce unnecessary burden, support meaningful choice, and make environments more accountable. Poor choice architecture exploits behavioral limits for institutional gain.

Choice Architecture and Nudging in a Wider Intellectual Context

Choice Architecture and Nudging belongs to a long intellectual history of inquiry into judgment, agency, freedom, persuasion, paternalism, institutions, and design. Philosophers have debated autonomy, welfare, consent, manipulation, and paternalism. Psychologists have studied attention, bias, framing, and cognitive load. Economists have studied incentives, utility, and market behavior. Designers and public administrators have shaped forms, interfaces, procedures, and institutional pathways.

Modern choice architecture brings these traditions together around a practical question: how do environments shape choice? The answer matters for public policy, digital platforms, health systems, education, sustainability, finance, consumer protection, and governance. Every system that arranges options also arranges behavior.

For Sustainable Catalyst, this field is essential because it connects knowledge architecture with behavioral architecture. It asks not only what people know, but what environments make possible. It provides one of the clearest bridges between psychology, design, ethics, technology, institutions, and sustainable action.

Related Reading

• Behavioral Science & Behavioral Psychology
• Behavioral Economics
• Behavior Change and Habit Formation
• Motivation, Reinforcement, and Learning
• Social Norms and Behavioral Influence
• Behavioral Public Policy
• Behavioral Research Methods
• Ethics of Behavioral Intervention
• Psychology
• Decision Science
• Content Frameworks
• Technology & Systems Intelligence
• Institutions & Governance
• Sustainable Development

Further Reading

• Behavioural Insights Team (2014) EAST: Four Simple Ways to Apply Behavioural Insights. London: Behavioural Insights Team.
• Halpern, D. (2015) Inside the Nudge Unit: How Small Changes Can Make a Big Difference. London: WH Allen.
• Hansen, P.G. and Jespersen, A.M. (2013) ‘Nudge and the manipulation of choice: A framework for the responsible use of the nudge approach to behaviour change in public policy’, European Journal of Risk Regulation, 4(1), pp. 3–28.
• Johnson, E.J. and Goldstein, D. (2003) ‘Do defaults save lives?’, Science, 302(5649), pp. 1338–1339.
• Kahneman, D. (2011) Thinking, Fast and Slow. New York: Farrar, Straus and Giroux.
• OECD (2019) Tools and Ethics for Applied Behavioural Insights: The BASIC Toolkit. Paris: OECD Publishing.
• Sunstein, C.R. (2014) Why Nudge? The Politics of Libertarian Paternalism. New Haven, CT: Yale University Press.
• Sunstein, C.R. (2016) The Ethics of Influence: Government in the Age of Behavioral Science. Cambridge: Cambridge University Press.
• Thaler, R.H. and Sunstein, C.R. (2021) Nudge: The Final Edition. New Haven, CT: Yale University Press.
• World Bank (2015) World Development Report 2015: Mind, Society, and Behavior. Washington, DC: World Bank.

References

• Behavioural Insights Team (2014) EAST: Four Simple Ways to Apply Behavioural Insights. London: Behavioural Insights Team. Available at: https://www.bi.team/publications/east-four-simple-ways-to-apply-behavioural-insights/ (Accessed: 21 June 2026).
• Dolan, P. et al. (2012) ‘Influencing behaviour: The mindspace way’, Journal of Economic Psychology, 33(1), pp. 264–277. Available at: https://doi.org/10.1016/j.joep.2011.10.009 (Accessed: 21 June 2026).
• Halpern, D. (2015) Inside the Nudge Unit: How Small Changes Can Make a Big Difference. London: WH Allen.
• Hansen, P.G. and Jespersen, A.M. (2013) ‘Nudge and the manipulation of choice: A framework for the responsible use of the nudge approach to behaviour change in public policy’, European Journal of Risk Regulation, 4(1), pp. 3–28. Available at: https://doi.org/10.1017/S1867299X00002762 (Accessed: 21 June 2026).
• Johnson, E.J. and Goldstein, D. (2003) ‘Do defaults save lives?’, Science, 302(5649), pp. 1338–1339. Available at: https://doi.org/10.1126/science.1091721 (Accessed: 21 June 2026).
• Kahneman, D. (2011) Thinking, Fast and Slow. New York: Farrar, Straus and Giroux.
• Kahneman, D. and Tversky, A. (1979) ‘Prospect theory: An analysis of decision under risk’, Econometrica, 47(2), pp. 263–291. Available at: https://www.jstor.org/stable/1914185 (Accessed: 21 June 2026).
• OECD (2019) Tools and Ethics for Applied Behavioural Insights: The BASIC Toolkit. Paris: OECD Publishing. Available at: https://www.oecd.org/gov/regulatory-policy/tools-and-ethics-for-applied-behavioural-insights-the-basic-toolkit-9ea76a8f-en.htm (Accessed: 21 June 2026).
• Sunstein, C.R. (2014) Why Nudge? The Politics of Libertarian Paternalism. New Haven, CT: Yale University Press.
• Sunstein, C.R. (2016) The Ethics of Influence: Government in the Age of Behavioral Science. Cambridge: Cambridge University Press.
• Thaler, R.H. and Sunstein, C.R. (2021) Nudge: The Final Edition. New Haven, CT: Yale University Press.
• World Bank (2015) World Development Report 2015: Mind, Society, and Behavior. Washington, DC: World Bank. Available at: https://www.worldbank.org/en/publication/wdr2015 (Accessed: 21 June 2026).

Back to top ↑