Apollo gave Cassandra, a princess of Troy, the power to see what was coming, and when she turned him down he made sure no one would ever believe her. In Aeschylus's Agamemnon, she arrives at the palace of Argos as a captive, sees her own murder waiting inside, and walks in anyway. Greek myth makes her foresight a curse rather than a gift.
The psychologist Gerd Gigerenzer of the Max Planck Institute for Human Development in Berlin and Rocio Garcia-Retamero of the University of Granada named a 2017 paper after her: "Cassandra's Regret: The Psychology of Not Wanting to Know", in Psychological Review. Interviewers from the survey firm GfK sat down face to face with 1,016 adults in Germany and 1,002 in Spain, in samples drawn to be representative of each country. Asked whether they would want to know today when their partner will die, 90.2 percent of those who answered said no. Asked whether they would want to know in advance what they are getting for Christmas, 64.4 percent said no.
Most said no on nine of the ten questions
The researchers asked about five pieces of possible bad news and five events they class as positive, mostly matters of surprise and suspense. Here is the share who said they would not want to know, across both countries:
- When your partner will die: 90.2 percent
- What your partner will die of: 89.4 percent
- When you will die: 86.7 percent
- What you will die of: 85.8 percent
- Whether your marriage will end in divorce, imagining you are newly married: 86.2 percent
- How a recorded soccer match ended, from a friend who already watched it: 73.6 percent
- What you are getting for Christmas: 64.4 percent
- Whether a 2,000-euro sapphire bought on vacation is real, when a test costs 50 euros and the stone cannot be returned: 51.9 percent would skip the test
- Whether there is life after death: 50.4 percent
- The sex of your baby before birth: 37.4 percent
Only 1.1 percent of the Germans and 0.6 percent of the Spaniards consistently wanted to know.
The Christmas figure hides a large group of fence-sitters. In Germany, 59.6 percent said no, 33.5 percent were unsure, and just 6.8 percent said yes. Spaniards were more likely to say no, at 69.2 percent. On the baby's sex the two countries flipped: 40.3 percent of Germans wanted it kept a surprise, against 34.7 percent of Spaniards.
"All men by nature desire to know"
That is the opening line of Aristotle's Metaphysics in W. D. Ross's translation, and Gigerenzer and Garcia-Retamero quote it on their first page. Aristotle's evidence was ordinary pleasure. People delight in their senses, he wrote, sight above all, "even when we are not going to do anything," because seeing helps them know things. On his account, knowing is valued for its own sake.
Modern decision theory works from a practical cousin of that idea: if information is valid and the search is not too costly, seek it and use it. The paper traces this default through the philosopher Rudolf Carnap and Bayesian statistics, while noting that it has since been qualified. Borrowing the economist Gary Becker's model of forward-looking people, Gigerenzer and Garcia-Retamero even sketch what knowing a spouse's time of death could be good for: planning savings, deciding when to move a partner to hospice rather than continue toxic chemotherapy, making sure the family is there at the end.
The survey answers turn Aristotle around. His people prized knowing even when it served no purpose. These respondents turned it down even on death and divorce, where the authors make a case that knowing would be useful, and they trace the refusals to how the knowledge would feel.
Bad news lines up like a ladder
Answers about death hung together tightly, and divorce went along more loosely. People who did not want to know when a partner would die usually did not want to know the cause either; in Germany the correlation was .82.
The five bad-news questions also fell into an order: partner's time of death, partner's cause of death, own time, own cause, divorce. Statisticians call this a Guttman scale. With few exceptions, anyone who wanted to know an item near the top of that ladder also wanted to know everything below it (in this analysis, "unsure" counted as wanting to know). The fit was .95 in Germany and .94 in Spain, above the .90 that Louis Guttman set as the minimum.
The other five questions formed no such ladder. Wanting a surprise at Christmas said little about whether a person wanted to know if there is life after death; the correlation was .11 in both countries. Only the three everyday suspense questions, soccer, Christmas and the baby's sex, showed a small cluster. Across all ten items there were no meaningful negative correlations, the lowest being −.01, indistinguishable from zero. Not wanting to know one thing, the paper says, "never appears to lead to wanting to know about another one."
The regret explanation
Gigerenzer and Garcia-Retamero explain the pattern with a rule from decision theory, adapted from R. Duncan Luce and Howard Raiffa: avoid the option that could leave you with the most regret. Learning that a partner will die young would be crushing, and on this account that worst case outweighs the comfort of learning they will live long. At Christmas, the worst case of knowing is a spoiled surprise.
They derived four predictions from the idea, and the data mostly went their way, though the gaps were small. Each participant worked through two sets of choices between a lottery and a sure amount of money, one about winning and one about losing. People who took the sure amount even when the lottery had the better average outcome were a little more likely to say they would not want to know than people who gambled even when the sure amount was the better deal. The gap averaged 4.0 percentage points in the winning version, in the same direction on all ten questions, and 3.2 points in the losing version, on nine of the ten.
People who had bought optional insurance leaned the same way in 33 of 40 comparisons. And on the questions about death, the share saying no rose with age: 87.3 percent of 18- to 35-year-olds did not want to know when their partner would die, against 92.3 percent of those over 50.
Education seemed to matter at first, but its effect disappeared once age, risk attitude and the other background factors were taken into account. One further factor, religious practice, falls outside the regret theory. People who attended services were more likely to want to know, which runs against the stereotype of faith as not wanting to know. The exception was the baby's sex: religious respondents were 17 percentage points less likely to want to know it, an effect traced mainly to the Spanish sample.
What a survey answer can carry
All of this rests on what people said, not on what they did. The authors raise that objection themselves: people might answer differently if the choice were real. For death and divorce there is no way to check, because the technology for predicting either in an individual is, in their words, "still too unreliable for testing this conjecture." The clearest case where reliable technology already exists is the baby's sex, and there the evidence lines up. In earlier studies that asked expectant parents at an actual test, 31 percent of 210 Dutch women having amniocentesis and 42 percent of 1,340 women and partners at a Boston ultrasound clinic did not want to know, close to the 35 to 40 percent in this survey.
The ten questions were the researchers' own choice, and they call the results "an existence proof" for those events, not a measure of how often people avoid knowledge in general. Everyone was surveyed once, with age standing in for nearness to death, so the study cannot say whether people say no more often as an event approaches or simply as they grow older. The paper gives no interview dates. It was first submitted in December 2015, so the answers are at least a decade old.
"Nine in 10" is the pooled figure: 89.5 percent in Germany and 90.5 in Spain. The links to risk-taking and insurance, though consistent, amount to a few percentage points. The paper argues that not wanting to know need not be self-deception or moral weakness. It stops short of saying it is the better choice.
Cassandra had no say in what she saw. Early in the paper, after noting that researchers report biomarkers that help predict when a person will die, the authors spell out what has changed for everyone else: "Unlike Cassandra, who was both empowered and condemned to foresee the future, we increasingly often have a choice."