Can a cuttlefish pass the marshmallow test?

Yes, a cuttlefish can pass a version of the marshmallow test. In 2021, researchers found that nine-month-old cuttlefish were able to pass an adapted version of the test, tolerating delays of 50 to 130 seconds before eating, performances comparable to chimpanzees and crows; cuttlefish with better self-control even did better on separate learning tests. The original test came from a very different setting: the Stanford marshmallow experiment, a 1970 study on delayed gratification by psychologist Walter Mischel. A child was offered one small immediate reward or two small rewards if they waited, and the researcher would leave the child alone in a room with a single marshmallow for about 15 minutes.

Mischel and his colleagues later described the children's strategies in vivid detail: some covered their eyes with their hands, rested their heads on their arms, talked to themselves, sang, invented games with their hands and feet, and one even successfully fell asleep while waiting. The researchers concluded that effective delay of gratification depends heavily on cognitive avoidance or suppression of the reward, not on thinking hard about how great it will be. In follow-up studies, children who waited longer for the preferred reward tended to have better life outcomes later on, as measured by SAT scores, educational attainment, BMI, and other measures, a finding that made the marshmallow test famous.

That fame has run into real limits, though. A later replication attempt, using a sample more than ten times larger and drawn from a more diverse population than the original, found only half the original effect, and suggested that economic background, rather than willpower, explained the other half. Separate work done in 2018 and 2024 found that the marshmallow test "does not reliably predict adult functioning." Other studies found the result depends heavily on context rather than pure self-control: children whose experimenter had broken an earlier promise waited much less than children whose experimenter had kept one, up to four times less, suggesting trust, not just willpower, decides how long a child is willing to wait.