See the pattern

Moral Cases Need Decision Theory Without Becoming Decision Theory

Applied ethics brings theory to cases; decision research shows the framing and trade-offs through which cases are encountered. A model of choice is not yet a reason why one outcome ought to be chosen.

One system

applied ethics

The other

decision making

Applied ethical case reasoning and expected-utility violations meet where a choice must make several incommensurable considerations explicit without mistaking its psychology for its norm.

Difficult cases expose the structure of a decision and the insufficiency of treating that structure as its justification.

Karen Ann Quinlan entered a persistent vegetative state in 1975.

Her parents asked to remove her ventilator. Their lawsuit reached the New Jersey Supreme Court, which ruled for them the following spring.

The machine came off. Quinlan breathed independently, sustained by a feeding tube, for another nine years.

This is what applied ethics looks like before it becomes a seminar diagram. A family, a patient who cannot speak, a machine doing some of the speaking, and a decision that refuses to arrive with its reasons alphabetized.

Consequentialism asks about the prognosis, the burdens, and the benefits.

A Kantian account asks whether withdrawing treatment is allowing an underlying condition to proceed rather than causing death, and brings double effect and substituted judgment to the bedside.

Virtue ethics asks what compassion and practical wisdom require.

The three routes converge on withdrawal. They have not thereby become the same road.

That difference is the first useful thing a case does.

It makes a decision visible as a structure: outcomes, acts, duties, persons, risks, unknowns.

It also makes visible the otherwise easy-to-miss fact that a structure is not a verdict.

Laying out a decision is not justifying it.

A spreadsheet may have excellent posture and still possess no moral philosophy.

Decision theory enters this room with unusually clean stationery.

Von Neumann and Morgenstern proved that preferences over lotteries can be represented by expected utility if four axioms hold: completeness, transitivity, continuity, and independence. Savage extended the approach to subjective probabilities.

The result is formidable, and modest in the way a theorem is modest.

It says what follows if the preferences have that form. It does not report that human beings carry the four axioms around in their coats, next to their keys and bad receipts.

Maurice Allais put independence under precisely that pressure.

One choice offers a million for certain. Another offers a very high chance of a million, a small chance of five million, and a sliver of nothing.

A second pair strips the same common component out of both options. The ranking should survive. The familiar reversal, certainty first and the larger prize second, does not.

Daniel Ellsberg found a neighbouring trouble.

His urn holds ninety balls. Thirty are known to be red. The remaining sixty are black or yellow in an unknown ratio.

Betting on red has a known probability. Betting on black does not.

The recurring preference for the known chance, even when complementary bets are offered, is ambiguity aversion. Risk is one thing. Not knowing what the risk is is another, which proves irritating to any theory that has already put every uncertainty into a single number.

Prospect theory does not repair ethics with this material. It describes choice differently.

Value is measured from a reference point rather than final wealth, concave for gains and convex for losses. Probability weighting is nonlinear near the endpoints, so the approach to certainty receives an outsized role.

Its cumulative version weights ranked outcomes, which repairs the original formulation’s occasional violation of stochastic dominance.

So the machine looks ready to be pointed at a moral problem.

A moral case is a choice under uncertainty. Prospect theory describes choices under uncertainty. Therefore prospect theory can tell the moral agent what to do.

The syllogism has the smooth mechanical hum of a vending machine, which is not the same thing as an argument.

The Asian disease study supplies the seduction in miniature.

An outbreak is expected to kill six hundred people. In a lives-saved frame, a large majority chooses the programme that saves two hundred for certain over a one-in-three chance of saving everyone.

In the deaths frame, a similar majority chooses the corresponding gamble.

The outcomes are logically identical. The descriptions are not.

Prospect theory calls the shift risk aversion for gains and risk seeking for losses, the reflection effect made politely public.

The result tells an ethicist something important. A case does not arrive naked. Its reference point is already holding a clipboard.

But the finding cannot choose between the programmes.

It can expose that certainty, loss, and wording have rearranged a preference. It cannot say whether a certain rescue, a gamble, a right, a duty, or a character-based response ought to govern.

The is-ought gap has not vanished. It has merely acquired very good experimental design.

There is a further embarrassment for anyone hoping to make a vignette into an x-ray of conscience.

That a response to one reveals a settled moral commitment is speculation. The direction of the framing effect reproduces reliably across large multi-site efforts, and its magnitude is heterogeneous across samples and settings.

The famous percentages belong to that wording and that population, not to humanity’s permanent moral thermostat.

Three claims like to wear the same trench coat here, and they should be separated.

A choice pattern occurs. An account explains it. The pattern supplies a moral reason.

The first can be robust while the second remains contested and the third remains a category mistake.

What holds is not that applied ethics secretly becomes decision theory, or that decision research has caught moral philosophy in an old-fashioned suit.

It is that difficult choices require their moving parts to be made explicit, and then refuse to let the inventory stand in for justification.

Ethics asks why the choice should be made. Decision research asks how a choice comes to look like a choice at all.

Between those questions is the case. Full of reasons, probabilities, descriptions, and a silence no formula gets to inherit.

The tempting reading, refused

Prospect theory tells a moral agent what to do.

A descriptive account of choice does not cross the is-ought gap by itself.

What you can take forward

Applied ethics runs consequentialist, deontological, and virtue-ethical reasons through a case, where they can converge on an action without sharing its justification.
The von Neumann-Morgenstern expected-utility theorem represents lottery preferences only when completeness, transitivity, continuity, and independence hold.
The Allais paradox isolates violations of independence, while the Ellsberg paradox distinguishes aversion to ambiguity from aversion to known risk.
Prospect theory evaluates gains and losses from a reference point and uses nonlinear probability weighting; cumulative prospect theory applies weights to ranked cumulative outcomes.
The Asian disease study shows a gain-loss framing reversal, but its reported percentage split belongs to its wording and population rather than a universal constant.
A descriptive model of framing or risk preference does not supply the normative reason required to choose an outcome in applied ethics.
Robust descriptive direction and a settled mechanism or moral interpretation are different claims, and replication evidence must keep them separate.

Go deeper

What does applied ethics add to a hard case?: Applied ethics brings a moral theory to a concrete case and asks what follows from its reasons. Consequentialism weighs outcomes, deontology asks about duties and constraints, and virtue ethics asks what practical wisdom and character recommend. The theories can endorse the same act while reaching it by different routes. That difference matters because agreement on an outcome does not establish agreement about why it is justified.
Why did ethical theories support withdrawing Karen Ann Quinlan's ventilator?: Karen Ann Quinlan entered a persistent vegetative state in April 1975, and the New Jersey Supreme Court ruled on March 31, 1976 that her family could authorize removal of her ventilator. Consequentialist reasoning weighs prognosis and burdens. Deontological reasoning treats withdrawal as allowing an underlying condition to proceed, using the doctrine of double effect and substituted judgment. Virtue ethics invokes compassion and practical wisdom.
What does expected utility assume about a choice?: John von Neumann and Oskar Morgenstern's 1944 theorem represents preferences over lotteries with an expected-utility function when four axioms hold: completeness, transitivity, continuity, and independence. Completeness requires comparable options, transitivity requires consistent ranking, continuity rules out certain discontinuities, and independence requires a shared change to two lotteries not to reverse their ordering. The theorem establishes a consistency representation; it does not claim that ordinary people always satisfy it.
Why is the Allais paradox a problem for expected utility?: Maurice Allais's 1953 choices pair a certain $1 million with an 89% chance of $1 million, a 10% chance of $5 million, and a 1% chance of nothing. A second pair offers an 11% chance of $1 million against a 10% chance of $5 million. The common-outcome substitution is the same, yet the usual preference reversal violates independence and cannot be represented by expected utility.
How is ambiguity different from risk in the Ellsberg paradox?: Daniel Ellsberg's 1961 urn contains 90 balls: 30 known red and 60 black or yellow in an unknown ratio. A bet on red has a known probability of 1/3, while a bet on black has an unknown probability. Preference for the known option, including in complementary bets, shows ambiguity aversion. Risk concerns known probabilities; ambiguity concerns uncertainty that has not been reduced to one probability.
What does prospect theory change about expected utility?: Kahneman and Tversky's 1979 prospect theory evaluates outcomes as gains and losses relative to a reference point rather than as final wealth. Its value function is concave for gains and convex for losses, and its probability weighting is nonlinear near the endpoints. Kahneman and Tversky's 1992 cumulative prospect theory applies decision weights to ranked cumulative outcomes, avoiding a stochastic-dominance problem in the original formulation.
What happened in the Asian disease framing study?: Tversky and Kahneman's 1981 study presents an outbreak expected to kill 600 people. In a lives-saved frame, 72% choose a program that saves 200 people for certain over a gamble with a 1/3 chance of saving all 600. In a deaths frame, 78% choose the corresponding gamble. The outcomes are logically identical, but gain and loss descriptions reverse the majority preference.
Does a framing effect tell a moral agent which option is right?: No. A framing effect describes how the presentation of logically identical outcomes changes choices. It can reveal a reference point, a risk preference, or an inconsistency relevant to deliberation, but it does not establish whether saving 200 people for certain or taking a gamble is morally required. A moral conclusion still needs reasons about outcomes, duties, rights, character, or another normative standard.
Are the 72% and 78% results a fixed law of moral judgment?: No. The 72% and 78% figures describe the wording and population of Tversky and Kahneman's 1981 study. Risky-choice framing reliably reproduces in direction across large multi-site replications, including the Many Labs series, but its magnitude varies across samples and settings. The evidence supports a robust gain-loss reversal, not a universal percentage split or a fixed measure of anyone's moral commitments.
Does a response to a moral vignette reveal a settled moral commitment?: Not necessarily. A vignette response records a choice under a particular description, reference point, and set of options. It may reflect a settled moral commitment, but that stronger interpretation is speculation unless independent evidence connects the response to stable judgment and conduct. Framing research establishes that descriptions can shift choices even when underlying outcomes remain logically identical, which makes that inference especially precarious.
Why does replication matter when a result is used in ethics?: Replication matters because a descriptive finding can be sturdy in one respect and unsettled in another. Henrich, Heine, and Norenzayan's 2010 WEIRD paper raises sampling concerns; Simmons, Nelson, and Simonsohn's 2011 "False-Positive Psychology" raises analytic-flexibility concerns; and the Open Science Collaboration's 2015 project prompts a dispute over what its estimate means. A moral interpretation cannot be stronger than the evidence it borrows.
Why does a model of choice not become a moral theory?: A model of choice identifies how options are represented, compared, weighted, or selected. It can show that certainty, ambiguity, or framing changes a decision, and it can reveal assumptions hidden in a case. A moral theory supplies a different kind of claim: why one outcome ought to be chosen. Describing a decision's psychological structure therefore does not by itself derive its normative justification.