Imagine you are offered a bet: a coin flip that pays you $1.50 if you win and costs you $1.00 if you lose. It is obviously worth taking — the expected value is positive — but most people refuse it, because a loss of a dollar hurts more than the pleasure of gaining a dollar fifty. That asymmetry is loss aversion, and it is one of the most reliable findings in the study of human decisions.

Now imagine the same bet described to you in a language you learned in a classroom. Same coin, same amounts, same odds. According to a body of research that began in 2012, people take it far more often.

This is the foreign language effect: the discovery that the language you are thinking in changes what you decide, how strongly you feel about it, and even which moral rules come to mind. It is one of the more useful findings for any adult learner, not because it makes you smarter — the evidence is clear that it does not — but because it shows something concrete about how a second language sits in your memory: it arrives with more distance and less heat attached than the language you grew up in.

Here is the original experiment, the cumulative numbers from the meta-analyses, the argument about why it happens, the situations where the effect disappears entirely, and what a learner should do with it.

The Experiment That Started It: Six People and a Disease

In 2012, Boaz Keysar, Sayuri Hayakawa and Sun Gyu An at the University of Chicago published a study in Psychological Science with a provocative title: "The Foreign-Language Effect: Thinking in a Foreign Tongue Reduces Decision Biases." They ran six experiments, and they leaned on a problem first designed by Kahneman and Tversky, the famous Asian disease problem.

The setup is this. A disease is expected to kill 600 people. You must choose between two programmes, which are described in one of two ways:

  • Gain frame: Programme A will save 200 people. Programme B has a one-third chance of saving all 600 and a two-thirds chance of saving nobody.
  • Loss frame: Programme A will result in 400 deaths. Programme B has a one-third chance that nobody dies and a two-thirds chance that all 600 die.

The two versions are mathematically identical. Yet in a native language, people flip: they prefer the certain option when it is framed as lives saved, and the gamble when the identical outcome is framed as lives lost. That flip is the framing effect, and in Keysar's experiments it simply disappeared when the problem was presented in a foreign language. Participants were no longer risk-averse for gains and risk-seeking for losses. They responded to the numbers instead of the wording.

Two further experiments in the same paper went after loss aversion directly. When choices were made in a foreign language, participants accepted more gambles with positive expected value — and the manipulation held not only for hypothetical bets but for real ones with real money on the line. The paper's conclusion was blunt:

"Using a foreign language reduces decision-making biases. … We propose that these effects arise because a foreign language provides greater cognitive and emotional distance than a native tongue does." — Keysar, Hayakawa & An (2012), Psychological Science 23(6), 661–668

What the Cumulative Evidence Says

A single striking study is not evidence. Fortunately, the field has now been meta-analysed.

Circi, Gatti, Russo and colleagues pooled the literature in Psychonomic Bulletin & Review in 2021. They found 17 studies containing 47 experiments, split into two domains: 38 experiments on moral decision-making and 9 on risk aversion. The results were consistent in both:

  • Moral decisions: a small-to-medium effect, g = 0.22 (95% CI [0.14, 0.30]), meaning people using a foreign language were more willing to accept harm if it maximised the overall outcome.
  • Risk decisions: g = 0.28 (95% CI [0.17, 0.39]), meaning a foreign language reduced risk aversion.

Two details in that meta-analysis matter more than the headline. First, for the risk domain the heterogeneity was essentially zero (I² = 0%), which means the studies agreed with each other closely. For the moral domain it was moderate (I² = 64.7%), so something was pushing results around — and the analysis found that the similarity between the two languages was a significant moderator, explaining about 37% of that spread. Second, proficiency in the foreign language did not moderate the effect in that dataset, which is the sort of result that should make you suspicious of tidy explanations.

That suspicion is warranted, because other work points the other way. A 2023 meta-analysis of the moral foreign language effect found that self-reported reading proficiency did moderate it: more proficient readers were less likely to make the harsh utilitarian choice. And a study comparing German learners put the effect mainly in lower-proficiency speakers. Meanwhile, studies of bilinguals who genuinely live in their second language — the ones who read, argue, and get shouted at in it — keep finding the distance shrinking.

The honest summary is this: the effect is real, it is small to medium, it is larger for people who learned the language later and use it less, and it fades as the language becomes part of your emotional life rather than a school subject.

The Moral Dilemma Debate, and the Explanation That Fell Apart

The most famous test bed for the effect has been the trolley family of dilemmas, and the details are worth following carefully, because the story of how the explanation changed is a good lesson in reading science.

Albert Costa and colleagues published "Your Morals Depend on Language" in PLOS ONE in 2014, testing participants across multiple language pairs. In the footbridge dilemma — push a large man off a bridge to stop a runaway trolley and save five people — using a foreign language increased the rate at which people accepted the sacrificial push. In the classic switch version of the trolley problem, where you merely divert the trolley with a lever, there was no difference. The effect also scaled roughly with the price of getting the decision wrong: for participants who scored worst on the Cognitive Reflection Test, using a foreign language increased utilitarian choices by 27 percentage points, almost exactly the overall 26-point effect.

Janet Geipel, Constantinos Hadjichristidis and Luca Surian then dug into the mechanism in PLOS ONE in 2015, and reported a number you can hold onto: in the footbridge dilemma, 22.2% of foreign-language participants endorsed the push versus 9.7% in the native language — but again, no effect at all on the trolley dilemma, and no effect on a non-moral control item.

The elegant explanation everybody had been repeating was that a foreign language numbs emotion, so the emotional veto weakens and cold reasoning takes over. Geipel's team put that explanation to the test — and it failed. Emotion ratings were lower in the foreign language for both dilemmas, but that reduction did not mediate the change in judgement. What did predict it was whether the action violated a social or moral norm. Pushing a person is prohibited; flipping a switch is not. Their conclusion was that a foreign language works mainly by making norms less mentally accessible.

Then in 2017, Hayakawa and colleagues published a process-dissociation study in Psychological Science that pulled the two moral inclinations apart — six experiments, one aggregate analysis — and found that a foreign language reduced the deontological response (the "do not do it" pull, d = 0.24) without increasing the utilitarian one. The paper's summary is the most quotable correction in the field:

"People are more utilitarian when using a foreign language not because they think more, but because they feel less." — Hayakawa, Tannenbaum, Costa, Corey & Keysar (2017), Psychological Science

Where the Effect Disappears

A common misreading is that a foreign language makes you a better thinker. It does not. The same team that ran the framing study also tested the Cognitive Reflection Test — the bat-and-ball problems that catch people using a fast intuitive answer — and found no benefit from using a foreign language. The Moses illusion, outcome bias and the representativeness heuristic have all been tested, mostly with null or inconsistent results.

The pattern that survives across the literature is narrower and more interesting. The foreign language effect shows up on problems that carry emotional charge or a norm. It shows up on:

  • Loss aversion and framing (Keysar et al. 2012; replicated by Costa and colleagues in 2014).
  • The sunk cost effect — three studies in 2023 found that thinking in a foreign language made people less likely to keep pouring money into a failing project, with emotion mediating the shift.
  • The endowment effect — sellers valued an object less highly in a foreign language, because their sense of psychological ownership weakened.
  • The truth bias, the self-serving bias, the hot-hand fallacy, the causality bias — a cluster of judgement biases that lean on intuition, summarised in the 2021 meta-analytic review.

It does not show up on cold logical puzzles. That distinction is the whole point: a second language does not raise your IQ, and it does not make your reasoning more rigorous. It changes what you feel about the problem in front of you.

Two Mechanisms: Distance You Can Measure, and Slower Processing

The body reacts differently

The emotional-distance claim is not metaphorical. It has been measured in sweat glands.

Catherine Caldwell-Harris and Ayşe Ayçiçeği-Dinn, working with Turkish-English bilinguals, recorded skin conductance responses — the tiny changes in how well your skin conducts electricity when you are emotionally aroused — while people heard or read words in each language. The categories included neutral words, positive words, negative words, taboo words, and one category that turns out to be the most telling of all: childhood reprimands, phrases like "shame on you" or "go to your room."

Skin conductance was higher in the first language, and the greatest gap between the two languages was on the reprimands. Being scolded in a language you learned at twelve produced almost none of the physiological flinch it produced in the language of your childhood. Later work with Mandarin-English and Spanish-English bilinguals refined the picture into the emotional contexts of learning theory: what predicts emotional resonance is not which language is "first" but where and when you learned it. Bilinguals who acquired a language in emotional situations, with real people, in childhood, showed similar physiological reactions in both languages. Bilinguals who learned a language in a classroom showed the gap.

The effort changes your processing style

The second mechanism is unglamorous: using a foreign language is hard, and effort changes how you think. Psychologists have long known that disrupting cognitive fluency — making text harder to read, for instance — pushes people toward more deliberate, less intuitive processing. Several researchers have argued the same thing happens with a second language: the difficulty itself acts as a brake.

But the puzzle is that a foreign language should increase cognitive load and therefore interfere with effortful reasoning, not help it. That contradiction is one reason the fluency-only explanation is not accepted as a full account. The best current position is that both factors operate, and that the emotional one dominates on emotionally loaded problems.

What This Means for You as a Learner

1. Use the distance deliberately — especially for embarrassing things

If a second language blunts emotional reaction, it is a good tool for the conversations that feel too hot to hold. Clinicians have documented this for decades: bilingual clients often switch into their second language to discuss painful material precisely because it gives them emotional reserve. You can use the same trick deliberately. The first conversation about a mistake you made, the negotiation, the apology, the medical question you are nervous about — doing those in a second language often feels easier than doing them in your first. That is not a bug in your learning; it is a property of it.

2. Do not mistake distance for fluency

Here is the trap. The distance is comfortable, so learners quietly build a life in the safe zone: reading, listening, typing, never having a real argument or a real friendship in the language. The distance then becomes permanent, and permanent distance is exactly what keeps a language feeling like an exam instead of a life. If you can only discuss a topic calmly, you have not finished learning it.

The emotional-contexts-of-learning research says what closes the gap: the same words learned in emotional situations behave like first-language words. That is a study-session instruction.

3. Stop making cards for single words

If emotional resonance attaches to contexts, then a card with frustrated = frustrado teaches you a label and nothing else. It gives you no scene to feel from. Replace most of those cards with three kinds that carry a situation:

  • A first-person line with a feeling in it. I can't believe I said that., I'm sorry I kept you waiting., This is driving me crazy. Cards that sound like something you would actually say in a bad moment.
  • A reaction card. The cue is a description of an event, the answer is the phrase a native speaker would actually produce — congratulating, apologising, complaining, sympathising.
  • A taboo or blunt card, privately kept. Every language has a register you will meet whether you study it or not — the word for angry that is not the dictionary word, the insult you need to recognise and never use. Learn to decode it. This is the vocabulary that carries the loudest emotional charge and the longest distance.

4. Learn your numbers and your money words cold

The decision experiments only work when comprehension is genuinely effortless. If you have to decode the numbers, you lose the effect and gain only confusion. Numbers, prices, probabilities, and the vocabulary of odds and risk — one in three, on average, worst case, it's worth it — are the highest-value flashcard content in the language if you ever intend to negotiate, invest, or argue anything where the stakes are real.

5. Borrow the second-language mind as a decision check

The practical trick is cheap and it is supported by the effect sizes above. When a decision feels urgent, or when the wording is doing the persuading, restate the choice in your second language. The framing effect, the sunk cost pull, and the endowment effect are all weaker there. Effect sizes of g ≈ 0.22–0.28 are not magic — they will not rescue a bad decision — but they are free, and the alternative is deciding while the wording does your thinking for you.

6. Expect the distance to keep shrinking, and stop worrying

Everything in this article describes a stage, not a destination. The effect is largest for late learners with low proficiency, it is smaller for people who read and argue in the language, and it is much smaller for words and phrases learned inside emotional situations. If you have ever worried that the flat feeling of your second language is permanent, the research disagrees: what changes it is not more time, but more emotional context — arguments, friendships, embarrassments, apologies, jokes at your own expense.

The foreign language effect is, in the end, a measurement of a gap you are trying to close. It explains why you can be rational about a problem in a language you do not love yet, and why that is not the same as fluency. The goal is not to keep the distance. It is to earn back the ability to feel something in a second language — and then to keep using it after the feeling arrives.