The Hedonimeter

If the whole difficulty is that utility is forever inferred and never seen — read off the choosing, because the choosing is all that shows — then the boldest reply is also the most obvious. Build the instrument. Lay it against the person and read the want directly, and the inference, with all its slippage, is no longer needed. The idea sounds like a caricature of scientific ambition, the economist with his calipers measuring joy. But it was proposed in earnest, and early, by one of the most mathematically serious minds the subject has produced. In 1881 Edgeworth imagined what he called a hedonimeter: an ideal instrument that would register, continuously and from instant to instant, the intensity of pleasure a person was feeling, tracing it as a curve through time. Integrate beneath the curve and you would have the quantity Bentham had only named — the total happiness of an hour, a day, a life — read at last from a dial rather than guessed from behavior.¹

For a hundred years this was filed as a charming impossibility, and the discipline congratulated itself on having outgrown it: revealed preference (the fourth plate) had shown how to do economics with no pleasure-meter at all, and the very wish for one came to seem a category mistake, like wanting a thermometer for justice. Then, at the end of the twentieth century, the impossibility was reopened — not by economists but at the border where economics meets psychology, and chiefly by Kahneman, who proposed that the question had been badly put. The error, he argued, lay in running together two different things under the one word utility. There is the weight an outcome carries in a decision — call it decision utility, and it is exactly what revealed preference measures, the utility inferred from what people choose. And there is the quality of an experience as it is actually lived through, moment by moment — call it experienced utility, and it is Edgeworth's quantity, the thing the hedonimeter would read. The discipline had assumed the two were one. Kahneman set out to measure the second on its own terms and to ask whether it matched the first.²

They did not match. The experiments are simple and their result is a knife at the root of the whole behavioral theory of value. Subjects underwent unpleasant episodes — a hand held in painfully cold water; a medical examination of some duration — while reporting their discomfort moment by moment, so that the lived total could be computed as Edgeworth prescribed, by summing the moments. Afterward they were asked how bad the whole thing had been, and, in one design, which of two such episodes they would choose to repeat. Their verdicts ignored the sum almost entirely. What governed the remembered badness of an episode was the average of its worst moment and its final moment; how long the suffering lasted scarcely entered the judgment at all. So an episode that was worse by the only measure that adds up the living of it — longer, with more total pain — could be the one people remembered as milder and chose to repeat, simply because it happened to end less badly.³ The moment-by-moment self and the remembering, choosing self gave different answers; and it is the remembering self that does the choosing, and so the remembering self whose verdict revealed preference records. The implication is severe. Choice does not track the goodness of experience. It tracks a memory of experience that systematically misrepresents it. The instrument revealed preference reads is not broken, but it is not reading what economics took it to read.

If choice cannot be trusted to register experience, then experience must be measured in its own right, and the back half of the program built the means. The Day Reconstruction Method has people reconstruct a day as a sequence of episodes and rate the feeling in each, yielding something close to Edgeworth's curve assembled from memory's better-behaved fragments; from such data one can compute a blunt but usable index — the share of waking time a person spends in an unpleasant state — and compare lives and circumstances by it.⁴ This is not the hedonimeter. It is testimony, refined and time-stamped, not a reading taken from the person without his word. But it is the nearest the social sciences have come to measuring lived feeling directly, and it already does what the construct was always meant to do: it lets two situations be ranked by the quality of the experience they produce, rather than by the choices people make about them.

The literal instrument, the dial that needs no testimony, is now sought where Edgeworth could not have thought to look: inside the skull. Neuroeconomics asks whether the brain keeps something like a common register of value, a single quantity in which the worth of a sip of juice and the worth of a sum of money are written in the same units, so that the brain can compare unlike goods and choose. There are real findings to point to. A signal carried by the neurotransmitter dopamine behaves, with uncanny fidelity, like an error term — the gap between the reward an animal receives and the reward it had learned to expect — which is exactly the quantity a system would need in order to learn what is worth wanting.⁵ Activity in particular regions of the prefrontal cortex tracks the subjective value a person places on an option closely enough that, from the activity, a rough ranking of the options can sometimes be recovered. The hedonimeter, in this telling, is not a fantasy but a research program with a budget.⁶

Suppose the program succeeds completely. Suppose that one day there is a true hedonimeter — a reading, taken from the brain without a word spoken, of the pleasure a person is feeling, accurate and continuous and beyond dispute. It is worth being precise about what such a triumph would, and would not, settle, because the precision is the whole lesson of this plate. It would settle less than its makers hope, and for three separate reasons, each of which the earlier plates have already prepared.

The first is the one the experiments forced on us, and it cuts the wrong way for economics. A perfected hedonimeter would measure experienced utility — the lived feeling. But the quantity a science of markets needs is decision utility, the thing that governs what people choose, and the two come apart, reliably, in the manner of the peak and the end. So the instrument would not predict behavior; it would contradict it. It would not vindicate revealed preference by grounding it in feeling at last; it would stand as a permanent witness that our choices and our experiences diverge — telling us not what we will do but that what we do is, by the measure of lived pleasure, often a mistake. The dream was an instrument that would underwrite the theory of choice. The instrument, realized, would indict it.

The second reason is the scandal of the eighth plate, and no instrument touches it. Grant a perfect reading of my pleasure and a perfect reading of yours. They are two readings on two instruments, each registering the state of one brain. Nothing in the two readings fixes a common zero or a common unit between them; nothing tells us whether the number my scanner reports as a seven means more felt pleasure, or less, than the seven yours reports. The comparison of one person's want with another's was shown, in its place, to be impossible not for want of measurement but as a matter of what the quantities are — and a quantity measured in the brain is still a quantity belonging to one person. No microscope, however fine, sees across the gap between two minds, because the gap is not a distance that finer resolution could close. The hedonimeter would let us read each person's want with perfect clarity and leave us exactly as unable as before to add one person's to another's.⁷

The third reason is the oldest, and it questions the prize itself. Even were pleasure perfectly measurable and perfectly comparable, there remains the question whether pleasure is the thing that ought to be measured — whether a life goes well to the precise extent that it feels good. A famous thought-experiment asks whether one would consent to be wired, for the rest of one's days, into a machine that produced any experience one wished, indistinguishable from the real thing, at the cost that none of it was real. Most people refuse, and the refusal reveals that they want something beyond the experience of a good life: they want the life. If that is so, then a reading of felt pleasure, however exact, measures one ingredient of a good life and not its sum, and the leap from how much pleasure to how well it goes is a leap the instrument cannot make for us.⁸ It is exactly this leap that the eleventh plate refuses to attempt — proposing that we abandon want and pleasure as the measure of advantage altogether, and ask instead what a person is actually able to do and to be.

So the hedonimeter, were it ever built, would be a triumph and a disappointment in the same instant. It would accomplish the thing economics had declared forever out of reach: it would let us see the ghost, read the want plainly off the matter of the brain, end the long inference at last. And in the same moment it would show that seeing the ghost was never the whole of the problem. We would at last be able to measure the want — and find that the want, made visible, still could not predict our choices, still could not be summed across our separate minds, and still could not tell us, by itself, what a good life amounts to. Edgeworth's curve would be drawn on the page, exact and continuous and real. And the questions that sent us looking for it would be standing, undisturbed, beside the instrument that found it.