The Miles-Per-Gallon Illusion
People treat fuel consumed as though it fell in a straight line as miles per gallon rises, and it does not. Fuel used per fixed distance is the reciprocal of miles per gallon, so the savings from improving an inefficient vehicle are far larger than the savings from improving an efficient one, and almost nobody sees it. The finding is Richard Larrick and Jack Soll's, published in Science in 2008, and it is one of the cleanest cases of a judgment result changing a public instrument.
What it is
The worked comparison that makes it obvious: replacing a vehicle that does ten miles per gallon with one that does eleven saves more fuel over the same distance than replacing one that does thirty-five with one that does fifty. Read off the miles-per-gallon numbers, the second upgrade looks enormously better. Measured in fuel actually burned, it is not.
The paper's recommendation followed directly. Express efficiency as fuel per fixed distance rather than distance per unit of fuel. The gallons-per-hundred-miles figure on the United States fuel economy label comes from this work.
In effect
The generalisation is bigger than cars. Any quantity expressed as a rate of the form output per unit of input will mislead in the same way when what the person cares about is input per unit of output. Kilometres per litre, words per minute, cases handled per hour, cost per click. The fix is always the same: invert the ratio so that the number moves in proportion to the thing being decided about.
The concept also arrives in this vault attached to an attribution failure worth recording. The Art of Clear Thinking uses the illusion at pages 48 to 49, and the substance is right: the illusion is described accurately and the arithmetic is correct. It is presented as the author's own worked example, introduced with a remark that decades of research in cognitive psychology show that our brains struggle, and there is no Larrick, no Soll and no Science entry anywhere in the book's forty-three item bibliography. Cite Larrick and Soll. That book is not a route to them.
What it does not say
It is not a claim that people are bad at arithmetic. Given the two numbers and a minute, most people can do the calculation. It is a claim about which quantity a familiar unit makes salient, which is why the remedy is a change to the label rather than a change to the reader.
It does not rest on a fragile result. The effect is a straightforward consequence of the arithmetic combined with a well-documented tendency to read ratios as though they were linear, it was published in Science, and it was influential enough to change a national labelling standard. Nothing in the replication record bears against it.
It does not say the inverted unit fixes every decision. It makes fuel comparisons legible, which is a real gain, and it says nothing about the other things that determine whether a purchase is sensible.
Sources
- Larrick, R. P., & Soll, J. B. (2008). "The MPG illusion." Science, 320(5883), 20 June 2008, 1593-1594. External verification pass, 2026-09-15, which confirmed the paper's recommendation and its effect on the United States fuel economy label.
- Lee, H. The Art of Clear Thinking, pp. 48-49. The illusion is used accurately and presented as the author's own illustration; Larrick, Soll and Science are absent from the book's 43-item bibliography. Recorded in the vault as the book's clearest single attribution failure.
- Evidence status: robust. A consequence of the arithmetic combined with a documented tendency to read ratios as linear, published in Science and influential enough to change a national labelling standard.