Why Are We So Bad at Predicting What We Want?

I recently came across a Hidden Brain episode about a speed-dating experiment that got me thinking about how well we actually know what we want.

In a 2017 study, psychologists Samantha Joel, Paul Eastwick, and Eli Finkel collected an enormous amount of information about people's personalities, preferences, and characteristics before sending them on a series of four-minute speed dates. They then used machine learning to see whether they could predict who would be attracted to whom.

Although the models could predict certain general patterns, like who tended to attract more interest, they couldn't reliably predict which particular pairs of people would click.

It made me think about how often this happens in other areas of life. We have a pretty clear idea of the kind of partner we'd like, the job we'd enjoy, or the decisions we'd make in certain situations. And yet, when we're actually faced with those choices, we sometimes end up doing something completely different.

Does that mean we're not particularly good at understanding our own preferences? Or is there something about making a real decision that we can't fully anticipate when we're just imagining it?

I started looking into the research, and there's quite a lot going on.

Do We Actually Know What We Want?

We tend to think of preferences as things we already have. We like certain foods, prefer certain types of people, and value certain qualities in a job. When someone asks us what we want, we assume the answer is already somewhere in our heads, waiting to be retrieved.

Sometimes it is. We have plenty of preferences that remain relatively stable over time. But when decisions involve unfamiliar situations or difficult trade-offs, things become more complicated.

Our Preferences Form as We Make Decisions

Psychologists Sarah Lichtenstein and Paul Slovic have spent decades studying a concept called preference construction. Their research suggests that our preferences aren't always fully formed before we make a decision. In some cases, we actually develop them as we're evaluating our options.

Imagine you're choosing between two jobs. One offers a significantly higher salary but comes with longer hours and more responsibility. The other pays less but gives you greater flexibility and more interesting work. You might genuinely value both financial security and freedom, but until you're faced with those specific offers, you may not know exactly how much of one you're willing to sacrifice for the other.

There's a 2004 experiment that illustrates this particularly well. Researchers Dan Simon, Daniel Krawczyk, and Keith Holyoak asked participants to evaluate the characteristics of different job offers before, during, and after making a choice. As participants considered their options, their evaluations began to change. The characteristics of the offer they were leaning toward became more attractive, while those of the alternative became less appealing. Interestingly, these shifts happened before participants had actually committed to a decision. In other words, the decision-making process itself was influencing how they evaluated the options.

Other experiments have found that people can even reverse their preferences depending on whether they evaluate options individually or side by side. I find this particularly interesting because we often approach decisions as if the goal were to discover what we really want, when sometimes the process of comparing our options is part of how we figure that out.

It also raises questions about how useful it is to ask people about their preferences before they've encountered the actual situation. Their answers might be completely sincere, but that doesn't necessarily mean those preferences will remain the same when they're faced with a real choice.

We Underestimate How Much Context Shapes Our Choices

I've caught myself imagining everything I could accomplish if I didn't have to spend so much time working. I'd exercise more, read more, cook more, and spend more time on personal projects. And it seems perfectly reasonable to assume that if I can already manage these things while working long hours, I'd be able to do considerably more with all that extra time.

Except there's something missing from that prediction. I'm imagining myself with more free time, but I'm also assuming that I'd have the same motivation, routines, priorities, and energy that I have today. Would that actually be the case? I'm not so sure.

We Assume Our Future Selves Will Want What We Want Today

There's a concept in behavioral economics called projection bias, which describes our tendency to assume that our future preferences will be more similar to our current preferences than they actually turn out to be. In a 2003 paper, George Loewenstein, Ted O'Donoghue, and Matthew Rabin examined how we use our current preferences to predict what we'll want in the future, often underestimating the effect that changing circumstances might have on us.

Related research on the hot–cold empathy gap explores something similar. When we're in a relatively calm or comfortable state, we can struggle to anticipate how strongly emotions or physical states such as fear, anger, hunger, or pain will influence our behavior later. It's relatively easy to imagine how we'd handle a difficult conversation when we're not emotionally involved. But being in the middle of that conversation, feeling defensive or threatened, is a very different experience.

Wanting Something Isn't the Same as Choosing It

Our predictions aren't necessarily unreasonable. They're just based on information and feelings that may not reflect what we'll experience when the moment arrives. However, I think there's another distinction that's important here: just because someone chooses something different from what they said they wanted doesn't mean they misunderstood their own desires.

Someone might genuinely want to change careers but ultimately decide to stay in their current job because they're worried about financial security. They haven't necessarily changed their mind about wanting a different career. They're simply making a decision in which several needs compete with one another.

This connects to the intention–behavior gap, or the difference between what we intend to do and what we actually end up doing. A 2006 meta-analysis of 47 experimental tests found that even when interventions successfully changed people's intentions, the resulting changes in behavior were considerably smaller.

There are plenty of reasons why that might happen. Sometimes we lack the resources or opportunities to act on our intentions. Sometimes our habits get in the way, or the consequences of a decision feel different once they're real. And sometimes another priority becomes more important.

The point is that what someone chooses doesn't necessarily tell us everything about what they want. Their behavior also reflects their circumstances, their constraints, and the trade-offs they're willing or able to make. Which makes predicting future choices considerably more complicated than simply asking someone what they would like to do.

We Don't Always Understand What Influenced Our Decisions

All this research makes me wonder how accurately we get to understand the reasons behind our decisions.

In 2005, researchers led by Petter Johansson conducted an experiment in which participants were shown pairs of photographs and asked to choose which face they found more attractive. Once they'd made their choice, the researchers secretly switched some of the photographs so that participants were presented with the face they hadn't selected. They were then asked to explain why they'd made that choice. Many participants didn't notice the switch. Instead, they went on to explain why they'd chosen the photograph in front of them, sometimes giving detailed explanations for a decision they hadn't actually made. The researchers called this phenomenon choice blindness.

What I find so interesting about this experiment is that the participants weren't simply making random guesses. They were offering explanations that made sense to them, even though those explanations couldn't accurately describe the choice they'd originally made. Something to think about there in terms of introspection and how much access we really have to the processes behind our decisions.

We generally assume that when someone explains why they chose something, they're reporting the reasons that influenced them. And often that may be true. But research like this suggests that, under certain circumstances, we can also construct plausible explanations after the fact.

Of course, a controlled experiment involving photographs doesn't prove that we constantly invent reasons for our everyday decisions. But it does suggest we should be careful about treating people's explanations as a complete record of what actually influenced them, which is particularly relevant when we're trying to understand not only what someone prefers, but why they prefer it.

Why Are We So Bad at Guessing What Other People Want?

So far, we've looked at how difficult it can be to understand and predict our own choices. But when we're trying to predict what someone else wants, there's another complication: we don't have direct access to their perspective, so we often rely on our own. Psychologists call one version of this the false consensus effect, a tendency to overestimate how widely other people share our choices, opinions, and beliefs.

In a series of experiments published in 1977, Lee Ross and colleagues found that participants tended to see their own responses to different situations as more common than participants who had chosen differently did. It's easy to recognize this in everyday conversations: we sometimes assume that our interpretation of a situation is the obvious one, and when someone sees things differently, we struggle to understand how they've arrived at such a different conclusion.

We Tend to Assume Other People Share Our Preferences

And there's another experiment that I think illustrates the problem well. In five studies published in 2000, researchers investigated how people value objects depending on whether they own them or are considering buying them. There's already an established phenomenon called the endowment effect, which describes how people can value something more highly simply because they own it.

What these researchers wanted to understand was whether buyers and owners could accurately predict how the other group would value the same object. They found that both groups overestimated how similarly the other group would value it.

Owners expected buyers to value the object more highly than they did, while buyers underestimated how much ownership affected the owners' valuations. And perhaps the most interesting part was that people also struggled to predict how their own valuations would change if they were in the opposite position. So the problem wasn't simply that they didn't understand other people. They also had difficulty imagining how their own preferences would change under different circumstances.

I think this has implications for how we communicate and interact with others, particularly when we're trying to persuade someone or reach an agreement. Imagine presenting an idea to a colleague and being genuinely surprised that they don't see the benefits as clearly as you do. You've explained your reasoning, the evidence makes sense to you, and you feel like you've addressed the most important concerns, but what matters most to them might be completely different. Perhaps they're worried about something you haven't considered, or they're evaluating the decision based on experiences and priorities you don't share.

Understanding their perspective requires more than explaining your own reasoning clearly. You need to recognize that the same information can be evaluated very differently depending on who's making the decision.

Asking People What They Want Isn't Enough to Understand Their Choices

After looking through all this research (oof, if you made it this far, thanks for sharing this level of curiosity with me), I don't think the conclusion is that we should stop asking people what they want or assume that nobody understands their own preferences. In fact, there's evidence that stated preferences can predict behavior, even in the original area that got me thinking about this question.

Stated Preferences Can Still Help Predict Attraction

A speed-dating study published online in 2024, involving 1,145 participants, found that stated and observed preferences corresponded for four of nine traits. The researchers also showed through simulations that these relationships can be difficult to detect statistically when attraction depends on several characteristics at once.

There are important differences between this study and the 2017 experiment, including the fact that participants reported their ideal preferences after the dates rather than before them, so the results aren't directly comparable. But the findings offer a useful reminder that failing to predict someone's choices doesn't automatically mean their stated preferences are meaningless.

I think a more useful conclusion is that understanding someone's preferences is only part of understanding their decisions.

We also need to understand the circumstances in which they're choosing, the alternatives available to them, the consequences they're anticipating, and the other priorities that might influence their behavior. This is particularly relevant in marketing research, where we often ask people to describe what they'd like, what they'd buy, or how much they'd be willing to pay for something.

Hypothetical Answers Don't Always Match Real Behavior

The problem with hypothetical questions is that people aren't necessarily experiencing the same conditions they would face in real life. A meta-analysis of 77 studies examined the difference between hypothetical and actual willingness to pay for consumer goods and found that the two often differed, with hypothetical valuations generally exceeding real ones. That doesn't mean we should ignore what people tell us. It means we should be careful about treating hypothetical answers as reliable predictions of actual purchases.

The same applies to something as simple as asking readers what kind of content they prefer. Someone might say they want detailed, technical articles but consistently click on shorter pieces. We could conclude that they don't really know what they want, but we could also be missing something about the context. Perhaps they prefer detailed content when researching an important purchase and shorter articles when casually browsing. Perhaps their interests have changed, or perhaps the topics and presentation matter more than the length itself.

Looking at their behavior gives us useful information, but it doesn't necessarily explain their motivations. And asking about their preferences gives us another piece of the picture, but not the entire explanation.

I think this is why it's worth combining different approaches when we're trying to understand people's decisions. We can ask about their preferences, investigate their past experiences, observe what they actually do, and pay attention to the circumstances that might explain any differences. And when we're trying to predict our own behavior, it helps to recognize that we're making those predictions from our current perspective, without knowing exactly what the future situation will feel like.

Understanding What We Want Is Only Part of the Picture

What started as a question about romantic attraction turned into a much broader exploration of how we make decisions, and I think the most interesting thing I've taken away is that there isn't one explanation for why our predictions sometimes fail.

Sometimes our preferences are still being formed as we consider our options. Sometimes we underestimate how much our circumstances will change. Sometimes we know exactly what we want but choose something else because other needs matter more. And sometimes we may not fully understand the reasons behind our choices in the first place.

These are different psychological processes, and they don't all apply to every decision. But together, they challenge the idea that understanding what someone wants is as simple as asking them. We can be self-aware, thoughtful people and still make choices that surprise us. We can genuinely understand our current preferences and struggle to predict how we'll feel in a different situation. And we can make a sincere effort to understand another person's perspective without fully appreciating the circumstances that shape their decisions.

I think there's something useful in recognizing those limitations, both when we're trying to understand ourselves and when we're making assumptions about other people.

Rather than assuming that someone's behavior reveals what they really wanted all along, perhaps we should be more curious about what influenced their decision. What were their options? What mattered to them at that moment? What were they worried about, and what might have changed if the circumstances had been different?

Those questions won't always give us a definitive answer, but they might get us a little closer to understanding why people make the choices they do.

Frequently Asked Questions

What Is Preference Construction?

Preference construction is the idea that we don't always know what we prefer before making a decision. Sometimes our preferences develop or change as we compare options, consider trade-offs, and work out what matters most to us.

What Is Projection Bias?

Projection bias is our tendency to assume that our future preferences will be more similar to our current ones than they actually turn out to be. We underestimate how much changing circumstances might affect what we want.

What Is the Hot–Cold Empathy Gap?

The hot–cold empathy gap describes how difficult it can be to predict our behavior in a different emotional or physical state. For example, when we're calm, we might underestimate how strongly anger or fear could influence our decisions.

What Is the Intention–Behavior Gap?

The intention–behavior gap is the difference between what we plan to do and what we actually do. Someone might genuinely intend to exercise regularly but struggle to follow through because of competing priorities, habits, or practical constraints.

What Is Choice Blindness?

Choice blindness happens when people fail to notice that the outcome of a decision has been changed and may even explain why they supposedly chose that outcome. It shows that we don't always have an accurate understanding of our own decisions.

What Is the False Consensus Effect?

The false consensus effect is our tendency to overestimate how many other people share our opinions, preferences, or choices. We often assume that what seems obvious or reasonable to us will seem equally obvious to others.

What Is the Endowment Effect?

The endowment effect describes our tendency to value something more highly simply because we own it. This can help explain why sellers and buyers sometimes disagree about what an object is worth.

What Are Stated and Revealed Preferences?

Stated preferences are what people say they want or prefer. Revealed preferences are inferred from their actual choices or behavior. The two don't always match, although observed choices can also reflect constraints and competing priorities.

What Is Hypothetical Bias?

Hypothetical bias occurs when people's answers to imagined decisions differ from the choices they make in real situations. For example, someone might say they'd pay a certain amount for a product but be unwilling to spend that money when actually buying it.

References

1. Joel, S., Eastwick, P. W., & Finkel, E. J. (2017). Is romantic desire predictable? Machine learning applied to initial romantic attraction. Psychological Science.

2. Lichtenstein, S., & Slovic, P. (2006). The construction of preference: An overview. In The Construction of Preference. Cambridge University Press.

3. Simon, D., Krawczyk, D. C., & Holyoak, K. J. (2004). Construction of preferences by constraint satisfaction. Psychological Science.

4. Hsee, C. K. (1996). The evaluability hypothesis: An explanation for preference reversals between joint and separate evaluations of alternatives. Organizational Behavior and Human Decision Processes.

5. Loewenstein, G., O'Donoghue, T., & Rabin, M. (2003). Projection bias in predicting future utility. The Quarterly Journal of Economics.

6. Webb, T. L., & Sheeran, P. (2006). Does changing behavioral intentions engender behavior change? A meta-analysis of the experimental evidence. Psychological Bulletin.

7. Johansson, P., Hall, L., Sikström, S., & Olsson, A. (2005). Failure to detect mismatches between intention and outcome in a simple decision task. Science.

8. Ross, L., Greene, D., & House, P. (1977). The false consensus effect: An egocentric bias in social perception and attribution processes. Journal of Experimental Social Psychology.

9. Van Boven, L., Dunning, D., & Loewenstein, G. (2000). Egocentric empathy gaps between owners and buyers: Misperceptions of the endowment effect. Journal of Personality and Social Psychology.

10. Zhao, A. A. Z., Sidari, M. J., Lee, A. J., & Zietsch, B. P. (2025; first published online in 2024). Speed-dating and simulation data explain the discrepancy between stated and revealed mate preferences. European Journal of Personality.

11. Schmidt, J., & Bijmolt, T. H. A. (2020; first published online in 2019). Accurately measuring willingness to pay for consumer goods: A meta-analysis of the hypothetical bias. Journal of the Academy of Marketing Science.