Cave team · Updated July 23, 2026 · 9 min read

How people actually make big decisions

TL;DR: People strongly favor whatever they're already doing — the status quo wins even when a neutral framing would flip the choice. People also overestimate how bad the road not taken will feel, because they forget how fast they adapt. And when 22,511 stuck people let a coin decide their dilemma, the ones it pushed into change were happier six months later. Lifetime regrets run 84% toward the changes never made.

In 2013, an economist put up a website inviting strangers to let a coin toss settle a question they couldn't settle themselves. Not trivia — the real ones. Should I quit my job? Should I end it? Should I move? You'd expect a handful of thrill-seekers. He got 22,511 usable flips, and "Should I quit my job?" alone drew 2,186 of them. Hold that number for a moment: two thousand people so evenly torn, for so long, that they were willing to hand the verdict to physics. We'll come back to what happened to them — but first, the machinery that got them stuck.

Why staying put usually wins

The first thing research shows about big decisions is uncomfortable: mostly, people don't make them. In a classic 1988 experiment, Samuelson and Zeckhauser gave people a set of investment options to choose between. When every option was presented fresh, 18% picked a particular portfolio. When the same portfolio was described as the one they already held, 52% kept it. Nothing about the option changed; being the incumbent nearly tripled its support. They named the effect status quo bias.

That was a questionnaire, so the same paper checked how people behave when the stakes are real. Harvard employees could switch their health plan every year: 97% didn't. Among 850,000 professors with TIAA-CREF retirement accounts, 72% had never changed their investment allocation — not once.

The extreme case is organ donation. In European countries where becoming a donor requires ticking a box, consent runs from 4% to 28%. In neighboring countries where donation is the default and the box is for opting out, consent runs from 86% to 99.98% (Johnson & Goldstein, Science, 2003). Similar people, opposite outcomes — the checkbox decides. And it isn't for lack of encouragement: the Netherlands, an opt-in country, once mailed more than 12 million letters asking residents to register. The rate barely moved.

So if a real decision has kept you stuck for months, here is the reframe: being stuck is what it feels like when the default machinery fails — when, for once, something in you refuses to let the incumbent win automatically. The agonizing means the decision is real.

The self you're deciding for doesn't exist yet

Here is the quietly disorienting finding. Across six studies of more than 19,000 people aged 18 to 68, people at every age reported having changed markedly over the past decade. And the same people predicted they would barely change over the next decade (Quoidbach, Gilbert & Wilson, Science, 2013). Twenty-year-olds looked back at a stranger and forward at themselves, permanent at last. So did sixty-year-olds. One of their measures put a price on the illusion: people would pay $129 to see their current favorite band play in ten years, but only $80 to see the band they loved ten years ago play today. The paper's closing line has the ring of an epitaph: "History, it seems, is always ending today."

The philosopher L. A. Paul pushed the point to its logical end. Some choices — parenthood is her canonical case, but a vocation change qualifies — are what she calls transformative: "Until you have had the experience, you cannot know what it is like," and having it rewrites the very preferences you'd use to evaluate it (Transformative Experience, 2014). Her thought experiment: would you become a vampire? Everyone who has says it's magnificent. But they're no longer the person who was asked. This is philosophy rather than data, and worth labeling as such — but it names the wall the spreadsheets keep hitting. The deepest decisions are identity problems dressed as information problems.

The futures you fear are overdrawn

Meanwhile, the simulations running in your head are wrong in a known direction. When researchers asked assistant professors to predict life after a failed tenure case — the career catastrophe of their world — forecasts landed at 3.42 on a 7-point happiness scale. Professors who had actually been denied within the past five years reported 4.71, significantly happier than the forecasts said they'd be (Gilbert et al., 1998). People who'd never been through a breakup predicted 3.89 two months after one; people actually two months past a breakup reported 5.42 — about the same as everyone else. The mechanism has a name, immune neglect: we each carry a psychological immune system that metabolizes catastrophe — rationalizing, refocusing, quietly repairing — and our forecasts forget to include it. As the authors put it, the system works "to repair one, not to improve one."

Daniel Kahneman added the other half: "Nothing in life is as important as you think it is, while you are thinking about it." While a decision is under deliberation, it fills the entire frame; lived, it becomes one thread in a fabric of hundreds. Both errors inflate the same thing — the imagined cost of the feared branch — which is precisely the weight holding the incumbent life in place.

What the coin said

Put the three findings together: people default when they can, decide on behalf of a self they won't remain, and overestimate the pain of the road not taken. So what happens when something forces the issue? That was the accidental elegance of the coin-flip site. The flips were random, and people obeyed them often enough — heads made a change 25 percentage points more likely — that the study works like a randomized trial run on real lives (Levitt, Review of Economic Studies, 2021).

Two findings. People told to change complied only half the time, while people told to stay complied three-quarters of the time — the status quo pulling even against an oracle they'd volunteered to obey. And six months later, the people the coin had pushed into change on the big questions were substantially happier than those it told to stay. Levitt's abstract says it with academic restraint: people "may be excessively cautious when facing life-changing choices." His participants were self-selected and already torn — this is a finding about people on the fence, not a universal summons to resign. But for the fence-sitters, the direction was unambiguous.

The pattern matches what people report at the end of life. Asked about the past week, we regret our actions and inactions about equally — 53 to 47. Asked about our whole lives, 84% of people name the thing they didn't do (Gilovich & Medvec, 1995). In the Terman study — 720 Americans tracked by psychologists since their 1920s childhoods — participants surveyed at an average age of 74 regretted inactions over actions 54% to 12%, and the single most common answer was an education not finished. The researchers' explanation deserves quoting whole: "The consequences of regrettable actions tend to be finite; the consequences of regrettable inaction tend to be psychologically infinite."

Regret flips over a lifetime: actions now, inactions forever Two panels. Asked about the past week, 53 percent regret an action more and 47 percent an inaction. Asked about their whole lives, 16 percent regret an action more and 84 percent an inaction. What people regret more: last week vs their whole life Which do people regret more? · Gilovich & Medvec, 1994/1995 Asked about the past week 53% an action 47% an inaction Asked about their whole lives 16% an action 84% an inaction "The consequences of regrettable actions tend to be finite; the consequences of regrettable inaction tend to be psychologically infinite."
Recent regrets split evenly between doing and not-doing; lifetime regrets don't. Gilovich & Medvec, Psychological Review, 1995.

So how does anyone decide?

Not, apparently, by finishing the spreadsheet. One intervention that reliably improves reasoning about one's own dilemma is embarrassingly literary: get outside the first person. People reason more wisely about a friend's crisis than their own identical one, and instructing them to view their own from a distanced, third-person perspective erases the gap (Grossmann & Kross, 2014) — a move as old as Caesar writing his own battles in the third person, and as available as talking the thing through out loud to anyone, or anything, that listens well.

And the research on people who actually made the change suggests it was never one big leap anyway. Career changers who succeeded, in Herminia Ibarra's studies, didn't introspect their way to certainty; they ran small experiments and let the evidence accumulate. "We learn who we have become — in practice, not in theory — by testing fantasy and reality." Which brings the problem back to L. A. Paul's reframe. If the biggest choices can't be simulated — if the self on the other side is genuinely unknowable — then the question was never which future scores higher. It's whether you want to find out who you'd become. The coin-flippers, the tenure-denied professors, the 74-year-olds naming the degree they never finished — they all point the same way. Go and see.

This essay comes from the team behind Cave, an AI companion built for decisions like these — not to compute the answer, but to talk it through with you, ask the questions only you can answer, and keep the thread going across the weeks a hard decision actually takes. The studies above are linked in full precisely so you don't have to take our word for anything.

Sources