The $20,000 Question
You’re sitting across from a hiring manager. The offer is $140,000. Excellent money by any objective measure. You feel elated, ready to accept on the spot. Then, over lunch with your old college classmate, you learn that her younger sister, who graduated last year, just got an offer of similar to you.
Suddenly, your great offer doesn’t feel so great anymore.
Same salary. Same purchasing power. Same lifestyle. But a completely different emotional response. What changed?
Nothing about the money changed. Everything about your reference point did.
For decades, economists built elegant models assuming humans were rational calculators—Homo economicus weighing probabilities, maximizing utility, making optimal choices. But if we were truly rational, your salary would feel exactly the same whether your friend’s sister made $160,000 or $120,000. The absolute value wouldn’t change.
Yet here we are, making seemingly irrational decisions every day. We buy both life insurance and lottery tickets, two mathematically opposed bets. We drive blocks to avoid a $25 parking ticket but won’t travel the same distance for a $25 discount on groceries. We feel poorer even as our wealth increases, simply because the perception of those around us are getting wealthier faster.
The question isn’t why people are irrational. It’s how those “irrationalities” follow predictable patterns—patterns we can measure, model, and design around.
The Experiment That Changed Everything
In the late 1970s, two Israeli psychologists—Daniel Kahneman and Amos Tversky—were conducting experiments that would eventually earn Kahneman a Nobel Prize and reshape how we understand human decision-making.
The setup was deceptively simple. They presented people with hypothetical gambles and asked them to choose.
Problem 1:
Which would you prefer?
A: A sure gain of $450
B: A 50% chance to gain $1,000 (and 50% chance to gain nothing)
Most people chose A—the sure thing. Even though both options had similar expected values (mathematically), certainty felt better than risk when gains were involved.
Then Kahneman and Tversky flipped the frame:
Problem 2:
Which would you prefer?
A: A sure loss of $450
B: A 50% chance to lose $1,000 (and 50% chance to lose nothing)
Now most people chose B—the gamble. Suddenly, when facing losses, people became risk-seekers, hoping to avoid any loss at all.
Mathematically, these problems are identical—just framed differently. But psychologically, they couldn’t be more different. Flip gains to losses, and risk preferences flip too.
This finding, published in Econometrica in 1979 under the title “Prospect Theory: An Analysis of Decision under Risk,” fundamentally challenged classical economic theory (Kahneman & Tversky, 1979). More importantly, it revealed three psychological laws that govern how humans actually make decisions:
The Three Laws of Human Decision-Making
Law 1: Reference Dependence
We don’t evaluate outcomes in absolute terms. We judge them relative to a mental baseline—a reference point.
Think about temperature. Is 70°F warm or cold? It depends. If you just came in from 95° heat, it feels cool. If you just left a 50° room, it feels warm. The absolute temperature hasn’t changed, but your reference point has.
The same logic applies to everything: salaries, grades, product quality, relationship satisfaction. We’re constantly comparing what we have to what we expected, what we had before, and what others around us seem to have.
Law 2: Loss Aversion
Losses hurt roughly twice as much as equivalent gains feel good.
Losing $100 ruins your day more than finding $100 improves it. Missing your flight feels worse than catching it feels good. A negative performance review stings more than a positive one uplifts.
This isn’t just a feeling—it’s measurable. Across hundreds of studies, the pain of loss typically outweighs the pleasure of gain by a factor of about 2:1 (though this ratio varies by person and context). Your brain literally treats losses like physical threats, activating the amygdala—the same brain region that fires when you’re in danger (De Martino et al., 2010).
Law 3: Probability Distortion
We treat probabilities emotionally, not mathematically.
A 1% chance feels much more real than it should. We overweight small chances of disaster (hence buying insurance) and small chances of windfall (hence buying lottery tickets). Meanwhile, we underweight large probabilities—a 99% success rate doesn’t feel certain enough, and a 50% risk feels more like 60%.
This explains why the same person can simultaneously insure against unlikely risks and gamble on unlikely rewards—behaviors that seemed contradictory under traditional economics but are perfectly consistent under Prospect Theory.
The Problem With Static Models
Prospect Theory was revolutionary, earning Kahneman the Nobel Prize in 2002 (Tversky had passed away in 1996 and couldn’t share the honor). A 2020 global replication study across 19 countries confirmed that its findings hold across cultures: the empirical foundations for Prospect Theory “replicate beyond any reasonable thresholds” (Ruggeri et al., 2020).
But the original theory had a limitation: it treated your reference point as fixed—a snapshot frozen in time.
In reality, your baseline never stops moving.
The Moving Baseline: When Today’s Luxury Becomes Tomorrow’s Necessity
Here’s where the story gets interesting—and more predictive.
In 2006, economists Botond Kőszegi and Matthew Rabin published a groundbreaking extension of Prospect Theory in the Quarterly Journal of Economics. They proposed that reference points aren’t static—they’re shaped by three dynamic forces:
Past outcomes: Got a bonus last year? This year’s bonus needs to be bigger to feel like a “win.” Yesterday’s exceptional coffee makes today’s taste disappointing. Last quarter’s delightful product feature is now just expected baseline functionality.
Expectations: If you expect a promotion and don’t get it, it feels like a loss—even though objectively, nothing changed. Your expected reality becomes your reference point. Users who expect 1-second page loads will abandon sites at 2 seconds, even though 2 seconds is objectively fast.
Social comparison: Your salary feels different when you know what peers earn. Your test score of 76 feels great if the class average is 64, but disappointing if it’s 85. Startup founders judge success not by absolute metrics but by their position relative to other startups in their cohort.
This is the hedonic treadmill—and it never stops running. What was once a gain becomes the new normal. And when you fail to meet that new standard, it feels like a loss, even if objectively your situation hasn’t worsened.
The implications are profound. Traditional Prospect Theory could explain a single decision in isolation. Dynamic reference-dependent preferences (as the Kőszegi-Rabin model is formally known) can predict behavior trajectories—how satisfaction erodes over time, why “get-even” gambling persists, why employee motivation decays after raises, and why A/B test wins often fade as users adapt.
Beyond Psychology: Culture, Biology, and Environment
But understanding reference points is only part of the puzzle. To truly predict human behavior in the real world, we need to integrate insights from neuroscience, cultural psychology, and environmental design.
The Cultural Layer: Risk Means Different Things in Different Places
An international survey published in Management Science (later republished in Theory and Decision in 2017) tested Prospect Theory across 53 countries in 13 languages—the largest behavioral economics study ever conducted (Rieger et al., 2017).
The results? Prospect Theory’s core patterns held everywhere: people showed risk aversion for gains and risk seeking for losses in all cultures. But the degree of these effects varied dramatically.
Chinese respondents were significantly less risk-averse than Americans when pricing financial options (Weber & Hsee, 1998). But here’s the twist: this wasn’t because Chinese people were fundamentally more comfortable with risk. It was because they perceived the same options as less risky. When you controlled for risk perception, attitudes toward perceived risk were remarkably similar across cultures.
Cultural dimensions like individualism and uncertainty avoidance systematically influenced how people set reference points and evaluated risk (Wang et al., 2017). Western gamblers prefer solitary games—slots, scratch cards, sports betting. In many Asian cultures, gambling is social—mahjong, dice games—where luck feels communal, shared, even ritualistic. The math is identical; the meaning is completely different.
The Biological Layer: Your Brain on Loss
Our decisions aren’t just cognitive—they’re chemical.
Dopamine prediction errors fire when outcomes surprise us, teaching our brains to update expectations. This is why near-misses in gambling deliver the same neurological reward as actual wins, keeping players engaged despite negative expected value (Kuhnen & Knutson, 2005).
Amygdala activation treats potential losses like physical threats. Neuroscientists using fMRI have shown that losing money activates the same brain circuits as physical pain (De Martino et al., 2010). When patients with amygdala damage were tested on Prospect Theory-style gambles, their loss aversion disappeared entirely—they became rational expected-value maximizers, uninfluenced by framing.
Stress hormones alter risk preferences. Under chronic stress, cortisol levels change how we evaluate risk—some people become hyper-cautious, others swing to compulsive risk-taking (Sokol-Hessner et al., 2009). The same person makes radically different choices when well-rested versus exhausted, calm versus anxious.
This biological reality means rationality isn’t stable—it’s state-dependent. Your reference point, your loss aversion, and your probability weighting all shift with your neurobiological state.
The Social Layer: Who Bears the Consequences Matters
We also don’t decide in a vacuum. Our choices change dramatically based on who will experience the outcomes.
Research consistently shows that when we’re making decisions for close others—family members, close friends—we become more conservative, especially with potential losses (Pahlke et al., 2015). Accountability and empathy amplify the perceived weight of downside risk.
But flip the context to anonymous clients or distant stakeholders, and the pattern reverses. Fund managers handling “other people’s money” often take bigger risks than they would with their own wealth, because incentives replace empathy—and because losses feel less visceral when someone else bears them.
Even the direction of social comparison matters. Research in China found that when people compared upward (to wealthier neighbors), their reference points shifted dramatically, making objectively good outcomes feel like losses (Brumagim & Wu, 2005).
The Environmental Layer: Context Is Everything
Finally, there’s the environment where decisions actually happen.
Most real-world decisions aren’t repeated experiments with clear feedback loops. They’re one-off, high-stakes moments:
- Accepting a job offer
- Choosing a medical treatment
- Launching a startup
- Pulling out of an investment during a market crash
These decisions happen in complex environments with multiple interacting factors:
- Time pressure
- Incomplete information
- Emotional state
- Social context
- Recent experiences
- Cultural norms
- Power dynamics
The architecture of these environments—how choices are presented, what defaults are set, what information is salient—shapes behavior in predictable ways.
The Power of Choice Architecture
This brings us to one of the most actionable insights from modern behavioral economics: there is no such thing as a neutral presentation of options.
Richard Thaler and Cass Sunstein formalized this idea in their 2008 book Nudge, coining the term “choice architecture” (Thaler & Sunstein, 2008). Every decision happens in a context. The order of menu items. The default settings on a form. The framing of outcomes as gains or losses. The social cues embedded in the interface.
That context—the architecture of choice—shapes behavior by influencing:
- Which information is salient
- What cognitive shortcuts get activated
- Whether fast, automatic thinking or slow, deliberate analysis engages
- What reference point gets established
A meta-analysis of over 200 studies found that choice architecture interventions produce a small-to-medium effect size (Cohen’s d = 0.43), and that effectiveness varies significantly by technique and behavioral domain (Mertens & Wulff, 2021).
Designing Environments for Better Decisions
If behavior emerges from the interaction of dynamic reference points, neurobiological states, social context, and environmental cues, then to predict or change behavior, we must design decision environments themselves.
Here are evidence-based principles:
1. Reduce Cognitive Load
Our deliberate thinking system can’t engage when overwhelmed. Simplify complex choices. Break decisions into manageable steps.
At NRG Energy, we increased online enrollment by 7% simply by catching form errors in real-time rather than at submission—lowering cognitive load and preventing frustration-based abandonment.
2. Leverage Smart Defaults
Countries that switched to opt-out (rather than opt-in) organ donation saw participation rates jump from 15% to 99% (Johnson & Goldstein, 2003). The default exploits our mental efficiency while preserving choice.
3. Frame Carefully—But Anticipate Adaptation
“Avoid losing $20/month” often outperforms “Save $20/month” because loss framing creates urgency. But remember: framing effects can decay as reference points adapt. What works today may need refreshing tomorrow.
4. Make Social Context Visible
We decide differently when aware our choices affect others. Show how decisions impact teammates, clients, or loved ones. Make fairness and equity considerations explicit.
Research shows people become significantly more risk-averse when deciding for close others, but more risk-seeking when deciding for distant, anonymous beneficiaries (Pahlke et al., 2015).
5. Build Feedback Loops
Show outcomes of past decisions. Make consequences visible and measurable. Enable iteration and reflection. This shifts from single-loop optimization (doing things better) to double-loop learning (questioning assumptions).
Working Hypotheses From the Field
As someone who’s run thousands of experiments across millions of users at companies like NRG Energy and Silicon Valley Bank, here are patterns I’ve observed that deserve deeper investigation:
Hypothesis 1: The Distance-Risk Gradient
The farther we are from whoever bears the consequences, the more incentives—not empathy—drive behavior. Test this: run experiments where users decide for themselves versus advising a friend versus choosing for an anonymous stranger. I predict risk preferences will shift systematically with social distance.
Hypothesis 2: Astronomical Probability Blindness
When jackpots hit hundreds of millions, our mathematical brain turns off completely. Stories beat math. Vivid imagery beats logic. Traditional expected-value calculations fail to predict behavior in these extreme contexts—we need different models for “impossible probabilities.”
Hypothesis 3: The Adaptation Acceleration Curve
Users adapt faster to positive changes than negative ones. A delight decays to baseline faster than a pain point becomes tolerable. If true, this suggests asymmetric strategies: iterate positive features rapidly before adaptation, but fix critical pain points even if users are “getting used to it.”
Hypothesis 4: Environmental Complexity Amplifies Automatic Processing
The more factors present in a decision context, the more heavily we rely on fast, automatic thinking—even for supposedly “important” choices. Corollary: single-variable lab experiments may dramatically underestimate the role of automatic processing in real-world decisions.
Hypothesis 5: Social Reference Points Update Faster Than Personal Ones
Comparison-based baselines shift more rapidly than expectation-based ones in networked environments like social media. You can get used to your new salary over time, but seeing a peer’s success instantly resets your reference point.
Bringing It All Together: A Unified Framework
Traditional economics said: humans maximize expected utility through rational calculation.
Prospect Theory (1979) said: humans judge outcomes relative to reference points, with loss aversion and probability distortion.
Dynamic reference-dependent preferences (2006) said: those reference points continuously move with expectations, experiences, and social comparisons.
The unified view emerging today: Human behavior is predictable when you understand the interaction between dynamic reference points, neurobiological state, social context, cultural norms, and environmental design.
To predict real behavior in the wild:
- Map the current reference point (what does the person expect? What happened recently? Who are they comparing themselves to?)
- Account for state (are they stressed, tired, emotionally activated? These shift the parameters)
- Understand cultural context (individualism vs. collectivism, uncertainty avoidance, social norms around risk)
- Analyze the environment (defaults, framing, social cues, cognitive complexity, reversibility of the decision)
- Model adaptation trajectories (how quickly will reference points shift after this decision?)
- Consider social distance (are they deciding for themselves, for loved ones, or for anonymous others?)
This isn’t about eliminating irrationality. It’s about recognizing that “irrational” behavior follows systematic patterns shaped by evolution, culture, learning, and context.
Questions Worth Exploring
On well-being: If expectations always reset, can satisfaction ever be sustained—or only managed? What does this mean for how we structure rewards, measure happiness, or define success?
On ethics: Where’s the line between helpful environmental design and manipulative engineering of consent? How should nudges evolve over repeated exposures?
On measurement: Can we build “reference point monitors” that track baseline shifts in real-time? What would that enable in product design, policy, or personal decision support?
On social dynamics: How fast do peer comparisons cascade through networks? Can we model collective baseline shifts—and their potential to create bubbles, panics, or social movements?
On individual differences: Which aspects of risk preference are stable personality traits versus context-dependent states? How much do the parameters vary across people, cultures, and situations?
A Final Thought
Take a moment right now and reflect:
What decision are you facing where your reference point—not the absolute facts—is driving your feelings?
Maybe it’s a salary negotiation where comparison to peers matters more than the actual number. A product launch where user expectations are outpacing your delivery. A career choice driven more by social comparison than personal values. An experiment that’s “failing” only because your baseline for success was set too high.
The algorithms running your behavior are invisible until you look for them. But once you see the patterns, you can’t unsee them.
Every emotional reaction to gains and losses. Every comparison to peers. Every mental shortcut under pressure. Every adaptation to new baselines.
These aren’t bugs—they’re features of human cognition, shaped by millions of years of evolution, learned through daily experience, and triggered by the environments we inhabit.
The question isn’t whether to work with these systems. You already are, whether you realize it or not.
The question is: Will you design environments that amplify our better instincts, or exploit our vulnerabilities?
Atticus Li is a Growth & Experimentation Leader who has helped driven over $1B in client acquisitions and millions in recurring revenue through experimentation, behavioral economics principles, and data science. He writes about the psychology of decision-making, the science of experimentation, and what drives human behavior. Learn more at experimentationcareer.com.
References & Further Reading
Brumagim, A. L., & Wu, S. (2005). An examination of cross-cultural differences in attitudes towards risk: Testing prospect theory in the People’s Republic of China. Multinational Business Review, 13(3), 67-85.
De Martino, B., Camerer, C. F., & Adolphs, R. (2010). Amygdala damage eliminates monetary loss aversion. Proceedings of the National Academy of Sciences, 107(8), 3788-3792. https://doi.org/10.1073/pnas.0910230107
Johnson, E. J., & Goldstein, D. (2003). Do defaults save lives? Science, 302(5649), 1338-1339.
Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263-291. https://doi.org/10.2307/1914185
Kőszegi, B., & Rabin, M. (2006). A model of reference-dependent preferences. Quarterly Journal of Economics, 121(4), 1133-1165. https://doi.org/10.1162/qjec.121.4.1133
Kőszegi, B., & Rabin, M. (2007). Reference-dependent risk attitudes. American Economic Review, 97(4), 1047-1073. https://doi.org/10.1257/aer.97.4.1047
Kuhnen, C. M., & Knutson, B. (2005). The neural basis of financial risk taking. Neuron, 47(5), 763-770.
Mertens, S., & Wulff, A. (2021). The effectiveness of nudging: A meta-analysis of choice architecture interventions across behavioral domains. Proceedings of the National Academy of Sciences, 118(51), e2107346118. https://doi.org/10.1073/pnas.2107346118
Pahlke, J., Strasser, S., & Vieider, F. M. (2015). Responsibility effects in decision making under risk. Journal of Risk and Uncertainty, 51(2), 125-146.
Rieger, M. O., Wang, M., & Hens, T. (2017). Estimating cumulative prospect theory parameters from an international survey. Theory and Decision, 82(4), 567-596. https://doi.org/10.1007/s11238-016-9582-8
Ruggeri, K., Alí, S., Berge, M. L., et al. (2020). Replicating patterns of prospect theory for decision under risk. Nature Human Behaviour, 4(6), 622-633. https://doi.org/10.1038/s41562-020-0886-x
Sokol-Hessner, P., Hsu, M., Curley, N. G., Delgado, M. R., Camerer, C. F., & Phelps, E. A. (2009). Thinking like a trader selectively reduces individuals’ loss aversion. Proceedings of the National Academy of Sciences, 106(13), 5035-5040.
Thaler, R. H., & Sunstein, C. R. (2008). Nudge: Improving decisions about health, wealth, and happiness. Yale University Press.
Wang, M., Rieger, M. O., & Hens, T. (2017). The impact of culture on loss aversion. Journal of Behavioral Decision Making, 30(2), 270-281. https://doi.org/10.1002/bdm.1941
Weber, E. U., & Hsee, C. K. (1998). Cross-cultural differences in risk perception, but cross-cultural similarities in attitudes towards perceived risk. Management Science, 44(9), 1205-1217. https://doi.org/10.1287/mnsc.44.9.1205
You must be logged in to post a comment.