The marshmallow test partly measures trust, not willpower

The delay-of-gratification studies, popularly “the marshmallow test”. · published 1 September 2026

Children do differ in how long they wait, and the waiting itself is real. What the wait reflects is circumstance and learned expectation as much as self-control, and the popular claim that the test predicts a child’s future is far stronger than the evidence supports.

The angle

The delay-of-gratification studies are usually told as a story about willpower, but Kidd, Palmeri & Aslin (2013) showed that children who had just met an experimenter who broke a promise waited far less — so the wait partly measures whether a child expects the second treat to actually arrive, and the 2018 large replication by Watts, Duncan & Quan found the link to later outcomes much smaller and largely explained by family background.

The notes this episode was written from

These are the working notes, unedited. Everything the video states as fact comes from here, and nothing else may be stated as fact. The sections on limits and on what the finding does not show are part of the brief, not an afterthought — see the method.

Walter Mischel and colleagues ran them at Stanford University’s Bing Nursery School, beginning in the late 1960s. The best known papers are Mischel & Ebbesen (1970) and Mischel, Ebbesen & Zeiss (1972).

The children were roughly three to five years old. A child was left alone with one treat and told they could eat it immediately, or wait until the experimenter came back and have two. The wait was on the order of fifteen minutes.

The treat was not always a marshmallow. Pretzels, cookies and mints were also used. “The marshmallow test” is a later popular name, not the original one.

Shoda, Mischel & Peake (1990) followed up on a subset of the children and reported that longer waiting in preschool correlated with parent-rated competence and higher SAT scores in adolescence.

Kidd, Palmeri & Aslin (2013) first gave children an experience of an experimenter who either kept or broke a promise. Children who had just met an unreliable experimenter waited far less time for the second treat. The wait therefore measures, in part, whether a child expects the promise to be kept.

Watts, Duncan & Quan (2018), in Psychological Science, ran a much larger and more demographically diverse replication. The association with later outcomes was substantially smaller than in the original follow-ups, and much of what remained was accounted for by family background and socioeconomic status.