Lead
You may have heard the marshmallow story. A four-year-old sits alone in front of a single marshmallow. The researcher says: wait 15 minutes and you'll get two. Children who waited, when followed up a decade later, showed higher academic achievement and stronger social skills than those who ate right away. That, at least, is the story that circulated.
For decades this finding anchored the idea that self-control in early childhood determines later success. In 2018, a study using a sample of more than 900 children re-examined the same question — and forced a substantial revision of that interpretation [2]. What changed, and what still holds?
What the original marshmallow study actually was
Mischel and colleagues (1989) followed children attending the Stanford University lab school in a longitudinal study: research design that follows the same participants over an extended period to track changes over time of the relationship between delay of gratification: the capacity to resist an immediate reward in order to receive a larger or better reward later — the ability to forgo an immediate reward in exchange for a larger later one — and subsequent intellectual and social development [1]. Among the same cohort tracked over time, children who waited longer in preschool showed higher SAT scores and stronger social and cognitive functioning in adolescence [1].
The finding traveled widely. In the process it was simplified into a message: self-control in early childhood is the key to success.
What the 2018 replication found
Watts, Duncan, and Quan (2018) revisited the question using a nationally representative sample of more than 900 children from the NICHD Study of Early Child Care and Youth Development [2]. The results were complicated.
A positive correlation between delay time and academic achievement at age 15 did exist. But its effect size was roughly half that reported in Mischel's original work. More consequentially, when the researchers statistically adjusted for family socioeconomic background, early cognitive ability, and home environment, the correlation shrank to roughly one-third its original size [2].
Among children from families without a college-educated mother — the economically disadvantaged subsample — the correlation between delay time and academic outcomes nearly disappeared entirely [2].
What this pattern suggests is not that "children who wait do better." The more accurate reading is: children who already live in stable home environments tend both to wait longer and to do better later — and a common third factor drives both outcomes. Remove the stability variable, and the predictive power of waiting mostly evaporates.
Put differently, the marshmallow test was not measuring pure self-control. It was also picking up signals about home stability, trust in adults who make promises, cognitive stimulation, and the accumulated experience of whether it is rational to believe that a delayed reward will actually arrive.
Executive function: the broader concept
The marshmallow test captures primarily one facet of self-control: delayed gratification. But what developmental researchers call executive function is a wider structure.
Diamond (2013) organized executive function into three core components [3]:
- Inhibition: suppressing impulses; sustaining attention on a specific target
- Working memory: the short-term cognitive system that holds and manipulates information needed for ongoing tasks: holding information in mind while using it
- Cognitive flexibility: shifting perspective; adapting when rules change
These components are not independent — they develop in interaction from toddlerhood through middle childhood. The marshmallow test taps mainly one form of inhibition, but adaptive behavior in real life draws on the balance of all three.
Diamond and Lee (2011) reviewed intervention studies designed to support executive function development in children ages 4–12 [4]. Multiple approaches showed evidence of effectiveness: computer-based training, martial arts and gymnastics, mindfulness practices, and specific classroom curricula. The consistent pattern across approaches was that activities involving repeated practice with progressively increasing difficulty were what produced gains in executive function [4]. The implication is direct: executive function is not fixed — it is developed through experience.
Moffitt's long-term evidence
Independent of the marshmallow research, Moffitt and colleagues (2011) conducted a longitudinal study that followed a birth cohort of 1,000 children in New Zealand from birth to age 32 — the Dunedin Study [5]. Self-control was assessed at ages 3–11 using multiple informants: observers, parents, teachers, and the children themselves.
This multi-decade assessment predicted health, wealth, and criminal conviction rates in the participants' thirties [5]. The relationship appeared as a continuous gradient: higher self-control associated with better outcomes, lower self-control with worse — linearly across the full range. The association remained after adjusting for IQ and socioeconomic background, suggesting that childhood self-control has some independent contribution [5].
Moffitt and colleagues themselves are careful about interpretation. This is observational research. It does not demonstrate that improving self-control will necessarily improve outcomes — the causal direction is not established by correlation, however large.
What this means in practice
A few things follow from the combined picture.
Before trying to train a child to wait, it may be worth asking first: is there an environment in which this child has reason to trust that waiting pays off? The replication evidence suggests that self-control is heavily contingent on environmental stability. The accumulated experience that promises are kept, that requests are answered — this is the foundation [2].
The intervention research from Diamond and Lee also points toward something more immediate than training willpower directly [4]. Activities that require holding rules in mind while adjusting behavior — games with complex structures, cooperative turn-taking, learning an instrument, gymnastics, martial arts — function as practice for the cognitive machinery of executive function. These are not optimizations; they are ordinary childhood activities that happen to support something real.
Paying attention in daily life to when a particular child can wait and when they cannot is also informative. Sleep deprivation, hunger, unfamiliar environments — executive function is state-dependent [3]. A child who struggles to wait at 5 p.m. after a short nap is not showing you their trait. They are showing you their current condition.
Summary
The marshmallow test was a striking experiment, but the simple reading — self-control in early childhood determines later success — has been substantially revised by the larger replication work [1,2]. Once family stability is accounted for, the independent predictive power of delay time shrinks considerably. The test was measuring something composite, not something pure.
Self-control is one component of a broader structure called executive function — and executive function is not a fixed trait but a capacity that develops through experience and environment [3,4]. Moffitt and colleagues' longitudinal research supports a real association between childhood self-control and long-term outcomes [5], but neither the optimist's reading ("any child's self-control can be improved and all outcomes will follow") nor the fatalist's reading ("it's set early") is what the current evidence actually says.
References
- Mischel W, Shoda Y, Rodriguez ML. Delay of gratification in children. Science. 1989;244(4907):933–938. doi:10.1126/science.2658056. PMID: 2658056.
- Watts TW, Duncan GJ, Quan H. Revisiting the marshmallow test: A conceptual replication investigating links between early delay of gratification and later outcomes. Psychological Science. 2018;29(7):1159–1177. doi:10.1177/0956797618761661. PMID: 29799765.
- Diamond A. Executive functions. Annu Rev Psychol. 2013;64:135–168. doi:10.1146/annurev-psych-113011-143750. PMID: 23020641.
- Diamond A, Lee K. Interventions shown to aid executive function development in children 4 to 12 years old. Science. 2011;333(6045):959–964. doi:10.1126/science.1204529. PMID: 21852486.
- Moffitt TE, Arseneault L, Belsky D, et al. A gradient of childhood self-control predicts health, wealth, and public safety. Proc Natl Acad Sci USA. 2011;108(7):2693–2698. doi:10.1073/pnas.1010076108. PMID: 21262822.