The dish nobody orders

Why does a menu carry a dish priced so high that almost nobody orders it?

Near the top of the menu sits a dish at 68 when everything else is between 16 and 34. The kitchen sells perhaps one a week. A restaurant runs on thin margins and edits its menu constantly, so an item that does not sell should have been cut long ago. Unless selling is not its job. What could a dish do for a restaurant without ever leaving the kitchen?

Begin with a harder question than it looks: what is the right price for a plate of food? Sit with that. Not what you would pay — what it should cost. Most of us have no answer. We have no independent scale for the value of a piece of fish and some potatoes, and no way to audit the kitchen's costs. So when we judge 34 as expensive or reasonable, what are we comparing it to?

Almost always, to the other numbers on the same page. In the absence of an absolute standard, judgement becomes comparison, and the comparison set is whatever is in front of us. That is the foundation, and it is well supported: valuations of unfamiliar goods shift substantially with the reference points supplied alongside them.

Now put the 68 dish back. What does it do to the page? It stretches the range. Without it, 34 is the top of the menu — the extravagant choice, the one you feel slightly foolish ordering. With it, 34 is in the middle. Nothing about the dish at 34 changed, but its position did, and position is what we were judging by. That is the anchoring account, and it is the simplest thing that explains the observation.

There is a second, related tendency: people avoid extremes. Faced with low, medium and high, a substantial share pick the middle, and which item counts as the middle depends entirely on what else is listed. Add an expensive item at the top and yesterday's high option becomes today's compromise — and the item that gains usually carries a good margin, so the manoeuvre is worth more than the price difference alone.

The name most often attached to this is the decoy effect, or asymmetric dominance, from work by Huber, Payne and Puto in the early 1980s. Strictly it means something more specific than a high anchor: you add a third option clearly worse than one existing option on every dimension, but not clearly worse than the other. The decoy is not meant to be chosen; its function is to make the option that dominates it look obviously good, and share shifts that way. On a menu that would be a smaller portion of the same dish at nearly the same price, which makes the full portion the sensible buy.

Here is where honesty is required, because this effect is far more often asserted than demonstrated. The decoy effect is a textbook staple, but its replication record in recent work is mixed: it shows up reliably in certain abstract, numerically-described choice tasks and becomes weak or absent when people choose among real products with perceptual rather than numerical attributes. It has not vanished, but "widely believed and widely used" is a fairer description than "firmly established". And note that the menu case does not actually need it. The plain anchoring and extremeness-aversion account explains the expensive dish perfectly well, uses weaker assumptions, and does not depend on the contested result. When a simpler explanation covers the observation, reaching for the glamorous one is a habit worth resisting.

Duller explanations also deserve a hearing before we credit psychology with everything. The dish may be genuinely profitable at one a week, since a high price on an occasional order can beat a thin margin on many. It may signal what the kitchen can do. It may exist for the table that wants to celebrate. Any of those can be true alongside the anchoring story, and probably several are.

The refinement that makes this work is that we are not bad at judging value so much as we are comparative judgers by default. That is not a flaw to be embarrassed about; for most things there is no absolute scale available, and comparison is a reasonable procedure. It just means the answer depends on the comparison set, and the comparison set is chosen by someone else.

Two corrections. First, this is not a claim that people are being fooled into overspending on something they dislike — the diner who orders the 34 dish because it now reads as moderate may enjoy it entirely. The effect is on the framing of the price, not on the food. Second, resist the urge to say "the decoy effect proves it". Take the modest version: relative judgement is well established, extremeness aversion is well established, and the specific asymmetric-dominance mechanism is plausible, popular in practice, and less securely replicated than its fame suggests.

The practical implication follows directly from the mechanism. If your judgement of a price comes from the range on the page, then the way to have your own opinion is to fix a number before opening the menu, or to ask what the dish is worth to you rather than where it sits among its neighbours.

• Why removing currency symbols and aligned price columns changes what diners spend.
• How the same relative-judgement logic sets subscription tiers, where the top tier exists mainly to sell the middle one.

Every term the collection defines is gathered in the glossary.