The Hubris of Choice: Can We Actually Diagnose Bad Decisions?

The Shaky Foundations of 'Bad Decisions'

For decades, behavioral economics has influenced public policy and product design, operating on the premise that experts can identify human cognitive biases and design environments to 'nudge' people toward better choices. However, groundbreaking new research challenges this very foundation, revealing that the primary methods used to diagnose a 'bad' decision fundamentally contradict each other.

Unmasking the Hubris of Choice Architects

The core idea behind 'nudging'—popularized by Richard Thaler and Cass Sunstein's book *Nudge* and implemented by government 'nudge units' worldwide—is that people make irrational choices due to cognitive biases (like procrastination, loss aversion, or present bias). Experts then design 'choice architecture' to gently steer individuals toward 'optimal' decisions. This approach is seen in auto-enrollment for 401(k)s, opt-out organ donation systems, and even diet-tracking apps or auto-renewing subscriptions, all justified by the belief that experts know what's best for the individual, sometimes even better than the individual themselves.

This implicit claim represents a significant leap of faith. But what if the foundational assumption—that we can actually *diagnose* a bad decision—is itself flawed? This is the central question tackled by B. Douglas Bernheim, Aldo Lucia, Kirby Nielsen, and Charles D. Sprenger in their NBER working paper, "When Are Decisions Improvable? An Evaluation of Diagnostic Methods," published in *Nature Human Behaviour*.

The Three Conflicting Diagnostic Methods

The paper identifies three main ways behavioral economists have historically tried to determine if someone has made an "improvable" decision:

1. Characterization Assessment Method

This method flags a choice as a mistake if the decision-maker holds specific, factual misconceptions about it. Researchers ask subjects to describe the parameters of their decision; if they cannot accurately calculate expected value, misunderstand probabilities, or fail a basic comprehension test, their decision is deemed 'improvable.' For example, an employee choosing a high-deductible health plan but unable to explain what a deductible is might be seen as having made a mistake. The critique here is that an inability to mathematically articulate a rational choice doesn't necessarily mean the choice itself is irrational; it might simply mean the individual isn't an economist.

2. Decision Confidence Method

Also known as measuring "Cognitive Uncertainty" (CU), this method gauges how confident a person is in the choice they just made. If a subject reports low confidence—say, 50%—their decision is flagged as suboptimal. The assumption is that if you're not confident, you probably made a mistake. An example would be a retail investor buying a volatile stock and admitting they're just guessing if it was the optimal financial move. Critics argue this relies heavily on self-reporting and assumes subjective doubt perfectly correlates with objective error, ignoring inherent uncertainties in the world or individual cautiousness.

3. Pattern Matching Method

This method identifies a behavior as 'improvable' if it mirrors a specific behavioral pattern known to be mathematically and objectively suboptimal in another domain, often violating Expected Utility Theory. For instance, if someone exhibits extreme risk aversion in a small-stakes game, or buys an extended warranty on a cheap item despite terrible expected value, this method would flag it as a mistake because the math doesn't add up, aligning with a known cognitive bias.

The Shocking Contradiction

The Bernheim paper's critical finding emerged when researchers applied all three diagnostic methods to the *same set of decisions* made by participants in an experiment involving binary lottery choices. The results were stark: the three methods completely contradicted each other, implying that entirely different choices were the 'improvable' ones. One method might deem a safe option a mistake due to misunderstanding, while another might flag a risky option due to low confidence, and a third might point to a completely different choice because it violated a known economic pattern.

This means that whether a human being's decision is labeled 'rational' or 'irrational' depends entirely on which academic tool a researcher chooses to use. The diagnostic tools, in essence, argue among themselves over what constitutes a 'wrong' decision, exposing a fatal flaw in distinguishing between a cognitive error and a valid, legitimate appetite for risk or preference.

The Academic Skirmish and Its Implications

The paper's findings sparked immediate academic debate. Weeks before its official publication, Benjamin Enke and Thomas Graeber published a rebuttal, specifically pushing back on the interpretation of the Decision Confidence method. They argued that subjects aren't necessarily conflating 'best decision' with 'best outcome,' but rather that the very presence of *outcome uncertainty* makes the choice subjectively difficult. Their 'interpretational middle ground' suggests that low confidence might reflect imperfect information or the inherent difficulty of decision-making in uncertain environments, rather than a diagnosable 'mistake.'

Regardless of the nuances in interpretation, the core conclusion remains: the field struggles to objectively identify genuine human error. Researchers, often trained to maximize Expected Value, may project their own risk aversion onto the public, labeling any deviation as an 'error,' even if it reflects a legitimate, different utility function (e.g., valuing peace of mind over maximizing returns).

This research calls for a reevaluation of decades of behavioral insights and the policies built upon them, as they may be founded on flawed diagnoses of human error.

Navigating a Nudged World: Key Takeaways for the Reader

This scientific reckoning has massive implications for individuals living in a 'nudged' world:

• Trust Your Preferences: What an economist labels as a 'cognitive bias' or 'error' might actually be your perfectly legitimate risk preference. If you prefer a safe, low-yield savings account because it helps you sleep at night, that's a valid choice, regardless of what 'Pattern Matching' might suggest.
• Question the Defaults: Recognize that every default option you face—from your 401(k) contribution rate to your data privacy settings—was designed by someone who assumed they knew what was best for you. This paper proves they often don't. These defaults are powerful, but they are not neutral or objectively superior.
• Demand Better Science: Policymakers need to clarify the assumptions underlying their behavioral interventions. Until the field can establish a unified, reliable way to objectively diagnose a 'bad' decision, institutions should default to maximizing freedom of choice rather than paternalistic nudging. The 'God Complex' of behavioral economics is being dismantled from the inside out, which is a positive development for individual autonomy.

This research should make us extremely skeptical of 'nudge-washing,' where corporations or policymakers use the veneer of behavioral science to justify paternalistic interventions that primarily serve their institutional bottom lines rather than the individual's true preferences. If the scientific justification for identifying a 'mistake' is built on such shaky ground, then claims of 'optimizing' individual behaviors through nudges warrant rigorous scrutiny.