Cognitive Psychology
Decision Framing Laboratory
Four decisions about seedlings, servers, grain and bicycles. Each one offers a safe option and a gamble with exactly the same expected outcome. The only thing that varies between people is whether the consequences were described as things saved or things lost.
Fictional scenarios with no people in them — chosen so nothing turns on how you feel about the stakes
Learning objective
By the end you should be able to state a framing effect in terms of a reference point, and distinguish a robust aggregate pattern from a claim about any individual.
About 25 minutes. Nothing is timed. Nothing you do here is saved or sent anywhere.
Before you start — what a framing effect is not
It is not proof that people are irrational. The effect is a group-level shift in the proportion choosing each option, it varies enormously with how the problem is written, and a preference that depends on a reference point is not obviously a mistake — reference points are how almost every evaluation people make actually works.
Your own four decisions cannot show the effect. You see one frame per scenario, so there is nothing to compare them with. The reveal shows you the other framing of the same arithmetic, and a simulated class dataset supplies the aggregate pattern one person cannot produce.
- Answer the prediction below — it unlocks the laboratory.
- Set how certain the safe option is, and how the stock is described.
- Make four decisions. Take as long as you like; nothing is timed.
- Read the reveal: the same arithmetic, written the other way round.
- Compare with the simulated class data, then rewrite a scenario in the challenge.
- Prefer not to decide anything? Load the worked example first.
First, a prediction
In each scenario something is at risk. One option produces a fixed, known result. The other is a gamble: a small chance that everything comes through, and a larger chance that nothing does. The two have the same expected outcome, to the unit.
The laboratory
Four scenarios, one at a time. Two of them will be written as gains and two as losses, assigned at random, and you are not told which is which until afterwards.
Key terms
- Frame
- How an outcome is put into words: as a gain from where you stand, or as a loss from somewhere else.
- Reference point
- The state outcomes are read as departures from. It is not fixed by the numbers alone.
- Expected value
- The average outcome of a gamble: each possible result weighted by how likely it is.
- Risk-averse and risk-seeking
- Preferring the certain option, or preferring the gamble, when the two carry the same expected value.
Four decisions
Make all four. Nothing is revealed until you have.
Progress Answer the question above to unlock this.
Answer the question above to unlock the laboratory.
The same arithmetic, twice
Your four decisions, with both framings
| Scenario | Frame you saw | Your choice | Safe option | Gamble | Expected outcome of each |
|---|
| Frame | Safe option was | Decisions | Safe option chosen |
|---|
Challenge — rewrite it so the framing matters less
Here is one of the scenarios in its original gain-framed form. Below it are six possible edits. Select the ones you think would reduce the difference between how people respond to the gain-framed and loss-framed versions. Your selected edits are then applied, so you can read what you have built.
What this demonstrates
A framing effect is a shift in a proportion
Nobody's arithmetic changes. What changes is how many people in a group choose the gamble, and the shift is typically large enough to reverse which option is in the majority. That is a real and repeatedly replicated finding about aggregates. It is not a statement about any individual, and it is not a claim that any particular person made an error.
Reference dependence is the mechanism, and it is not a bug
Outcomes are evaluated as changes from a reference point rather than as final states. "Two hundred saved" is a gain from a reference point of nothing; "four hundred lost" is a loss from a reference point of everything. Losses loom larger than gains of the same size, so a gamble that might avoid a loss altogether becomes attractive in a way that a gamble which might increase a gain does not. Notice that reference dependence is how almost every everyday evaluation works — is this a good price, is this a warm day — and a system without it would be strange rather than wise.
Certainty is doing part of the work
Some of the pull of the safe option in the gain frame comes not from its being safe but from its being certain. Take away the certainty — a very high probability instead of a guarantee, with the expected outcome unchanged — and the safe option loses a good deal of its appeal. That is why the certainty setting exists here: it separates a preference for the sure thing from a preference for low variance.
From a demonstration to a claim about rationality
The distance is considerable. Framing effects shrink when problems are stated so that both complements are visible, when people are asked to justify their choice, when the sample is expert in the domain, and when the description makes the reference point explicit. They also vary in size across replications. All of that is compatible with a real effect and incompatible with "people are irrational" as a conclusion. What survives is narrower and more useful: descriptions carry information about reference points, and whoever writes the description has some influence over the decision.
For teaching elsewhere: take this activity as one self-contained block of HTML, on the clipboard or as a file. Either way it is styled so that it will not disturb the page you put it into.