Unit Bias: Why One Serving Feels Like the Right Amount

A unit answers a question that would otherwise need answering every time: how much is the right amount. One cookie, one bottle, one bag, one plate, one serving. Within a wide range of sizes the unit itself reads as the proper portion, so the decision becomes whether to take one rather than how much to take. That is why the size of a unit can change the amount a person eats while the person's appetite, mood and intentions stay exactly as they were.

Paul Rozin and his colleagues put the heuristic in cultural terms as well. The unit people treat as one serving is partly learned, which is why it travels with a food culture: the typical single-serving yogurt container in French supermarkets held 125 grams against 227 grams in American ones, and the French were not compensating by eating two. A norm of what one portion is, carried in packaging and plate size, does the work that deliberate restraint would otherwise have to do.

The effect is real and partial, and the original paper was careful to say so. If unit size were the only thing driving intake, quadrupling the unit would quadruple the amount taken. It did not. Across the three studies the amount taken rose by factors of 2.27, 1.69 and 1.67 where pure unit control predicted 4, 2 and 4. People do notice size and do adjust; they simply do not adjust all the way, so bigger units still mean more food.

Two limits are worth stating. The 2006 studies measured what people took from a tray or a bowl, not what they swallowed, and they measured it as a daily total in a shared building, so there is no record of individual servings and some of the same residents appeared across days. The broader portion-size literature supplies the controlled evidence: a meta-analytic review found that doubling a portion raised consumption by about 35 percent on average, with the effect flattening as portions grow very large, and a Cochrane review of 86 comparisons from 58 studies with 6,603 participants found a standardized mean difference of 0.38 for larger portion, package or tableware size, 0.46 in adults.

Andrew Geier, Paul Rozin and Gheorghe Doros published "Unit bias: A new heuristic that helps explain the effect of portion size on food intake" in Psychological Science in 2006. The three studies were run in public buildings in Philadelphia, on people who did not know they were in an experiment, with free food and no limit on how many units anyone could take.

In the first, a mixing bowl of Tootsie Rolls sat in a busy spot on the ground floor of an office building. Every morning it was filled with 240 grams of candy: on alternating days that was either 80 small pieces of 3 grams or 20 large pieces of 12 grams, so the weight on offer was constant and the unit differed fourfold. The pieces left were counted at the end of each day, for ten working days.

The second study is the one the demo reproduces. At an apartment building, management already left large Philadelphia-style soft pretzels in an alcove off the lobby on Mondays, Wednesdays and Fridays. For alternating weeks the researchers either left them whole, at 3 ounces each, or cut them carefully in half, which doubled the number of pieces to 120 and kept the total weight the same. This ran for 12 weeks, giving 18 days of data for each unit size. The third study replaced a tablespoon scoop with a quarter-cup scoop, four times larger, in a one-pound bowl of M&M's on the concierge desk of the same building, under a sign reading Eat Your Fill.

In all three, more food left the bowl when the unit was bigger. Pairing corresponding days, the weight taken was 2.27 times higher for the large Tootsie Rolls, 1.69 times higher for whole rather than halved pretzels, and 1.67 times higher with the large scoop. The pretzel result was the most consistent: all 18 pairs of days favored the whole pretzels. The authors drew the conclusion that gives the paper its title: small portions work not because people eat the same weight in more servings, but because something stops them from going back for a second unit.

The rows below are illustrations, not findings from the studies cited on this page.

Unit bias is hard to argue with and easy to work around, because the unit is a physical thing you can change.

• Set the unit before you start eating. Serve a portion onto a plate or into a bowl and put the container out of reach; the served amount becomes the unit.
• Buy the smaller package when the food is one you tend to finish. The unit price is worse and the amount eaten is usually lower.
• Divide on the way in. Repackaging a large bag into single portions at home does the work the manufacturer did not.
• Change the vessel rather than the resolve. Smaller plates, bowls, glasses and serving spoons set smaller units without requiring a decision at the moment of eating.
• Cut things in half. The pretzel study is the clearest demonstration that a unit cut in two is still one unit to the person taking it.
• Read servings per container, not calories per serving. A package that holds two and a half servings is still one unit on the sofa.

This is a page about judgment, not about diet or health. What the research supports is that unit size changes the amount selected and eaten, not that any particular eating pattern follows from that.

Unit bias is about the size of one serving of a thing. Its neighbors deal with other ways that quantity and form change a decision.

• It is not the denomination effect, in which the same sum of money is spent more readily as several small bills than as one large one. That one works in the opposite direction: breaking money into smaller units makes it easier to use, while breaking food into smaller units makes it easier to stop.
• It is not the restraint bias, the overestimation of your own ability to resist an impulse. Restraint bias is a belief about willpower; unit bias is about what the environment presents as one portion.
• It is not the less-is-better effect, where a smaller option is preferred when the two are judged separately rather than side by side. That is about how options are valued, not about how much of one is taken.
• It is not simple appetite. The 2006 studies varied nothing but the size of a unit, with the same food, the same place and the same price of zero.