June 2013: Young Children’s Decision-Making on the Marshmallow Task

Kidd, C., Palmeri, H., & Aslin, R. N. (2013). Rational snacking: Young children’s decision-making on the marshmallow task is moderated by beliefs about environmental reliability. Cognition, 126(1), 109–114.

Summary by Dr. Jessica Church-Lang


The ability to resist an immediate reward (like playing video games) in favor of a long-term goal (like doing well on a test) or a later, bigger reward (like admittance to college) is considered a key aspect of self-control. Self-control is thought to be highly relevant to individual success, and high self-control has been shown to predict lower incidence of substance abuse (Ayduk et al., 2000) and a higher number of positive life outcomes, including higher SAT scores (Shoda, Mischel, & Peake, 1990). In the lab, one famous test of self-control is a delay-of-gratification task called “the marshmallow task” (Mischel, Shoda, & Rodriguez, 1989).

During the marshmallow task, a 4 year-old child is offered one treat (e.g., a marshmallow) to enjoy immediately or two treats if the child can wait until the researcher leaves and returns to the room. A marshmallow is then set in front of the child. The researcher leaves the room for 15–20 minutes, and the child is watched via video or one-way mirror. How long a child waits before tasting or consuming the marshmallow at age 4 has been shown to be highly predictive of future outcomes in school and beyond (Mischel et al., 1989).

One explanation for why some children wait a long time and others fail to wait for a reward is that the marshmallow task measures differences in children’s intrinsic self-control abilities. This is the favored, historical explanation, and the focus in research and education has been on fostering these internal motivations and abilities (e.g., grit, perseverance, self-control). However, another explanation, explored by the present study, is that children have different expectations and beliefs about the reliability of their environment (Kidd, Palmeri, & Aslin, 2013). Kidd and colleagues found that by first presenting children with either a reliable or unreliable set of experiences with the researcher, they were able to strongly influence children’s ability to wait during the marshmallow task. This finding suggests that a delay-of-gratification task not only assesses a personal, internal quality of self-control, but also is a reflection of the child’s beliefs about the reliability of the world. This finding has strong implications for child rearing and for fostering academic success.

Study Design

Hypothesis: The researchers thought that the marshmallow task reflected not just a child’s self-control abilities, but also a child’s rational behavior, based on experience with his or her environment. If this hypothesis were true, a child presented with an unreliable environment would wait less (perform worse) on the marshmallow task than a child presented with a reliable environment.

Reliability Training: A total of 28 children (age 4) were randomly split into two groups of 14. One group was exposed to the “reliable” condition, and the other group was exposed to the “unreliable” condition. In both conditions, children worked individually in the lab on a two-part art project (a create-your-own-cup kit) that provided evidence about the reliability of the researcher. The art project provided two important choices. In choice 1, children could either use worn crayons immediately or they could wait 2.5 minutes for a new set of art supplies. Then, in choice 2, children could either use one small sticker now or wait 2.5 minutes for a new set of better stickers.

All of the children in both groups waited for the better options. However, for the children assigned to the unreliable condition, the researcher returned after choice 1 and choice 2 without the promised better supplies, and the children had to use the worn crayons and the small sticker, despite waiting. In the reliable condition, the researcher returned with exciting art supplies after choice 1 and with stickers picked by the caregiver to be especially appealing to the child after choice 2. Each child then worked on the art project for 2 more minutes.

Test: The marshmallow task was performed immediately after the two-part art project, presented to the child as a snack break. A marshmallow was placed on the table in front of the child, and the researcher then left the room, promising two marshmallows when the researcher returned if the child could resist eating the single marshmallow. Two individuals who did not know to which condition (reliable or unreliable) each child had been assigned evaluated the video of the 15-minute marshmallow test for each child. They recorded when the child’s first taste of the marshmallow occurred.

Key Findings

Children in the unreliable condition waited an average of 3 minutes without tasting the marshmallow. Children in the reliable condition waited an average of 12 minutes. This large, significant difference indicated that children exposed to a reliable environment waited substantially longer than children exposed to an unreliable environment. In the unreliable condition, only 1 of the 14 children waited the full 15 minutes. In the reliable condition, 9 of the 14 children (64%) waited the full 15 minutes.

Importantly, children of the same age as those in the current study, but who were not exposed to the reliability manipulations, waited an average of 6 minutes (Shoda et al., 1990). This study used a randomized design; therefore, any differences in outcomes (the length of time children waited) can be assumed to be a result of the difference in conditions (reliability), rather than the result of other, unmeasured factors. Thus, in this study, researchers caused children to wait less (in the unreliable condition) and to wait more (in the reliable condition) than average.

It is important also to note that this study took place under lab conditions, rather than in a more complex environment, such as a classroom. Nevertheless, the authors of this study suggest that a child’s belief about the consistency of the world may be just as important to their future success as their capacity for self-control. A child raised in a resource-scarce environment, for example, may implicitly learn that the only guarantee in a situation is the immediate object at hand; thus, it may not be an advantage to this child to exercise self-control in a delay-of-gratification situation. This study suggests that success in life outcomes may rely heavily on a child’s worldview in addition to their self-control capacities, which is a useful message for educators.


Recommendation 1: Creating highly reliable situations in which students are asked to complete tasks may foster self-control.

An unspoken but essential part of this research is the concept that self-control is not a fixed, innate ability. Instead, qualities like perseverance, grit, and planning appear to be highly moldable, both in good and bad directions. For example, in another study, children who were given a cape and a description of how Superman is “very good at waiting” waited much longer during the marshmallow task than children without this mental strategy (Karniol et al., 2011). Thus, despite the years of experience in unreliable situations that a child may have at the start of schooling, Kidd and colleagues suggest that a reliable caregiver or teacher can shift a child’s rational decision-making process. Studies like these demonstrate how, even in a very short time frame, children’s self-control abilities can be improved or diminished.

Recommendation 2: Self-control abilities and academic performance may best be improved with a multiple-strategy approach.

The research discussed here shifts the allocation of self-control abilities from a purely internal view to a more nuanced interpretation. Teachers already realize that even young children have years of experience with reliable or unreliable people and with consistent or inconsistent environments. But we need to appreciate that the ability to wait and reach for long-term goals requires trust that the given situation will make good on its promise and that this trust comes from experience. Stable, reliable, and consistent school environments foster opportunities for students to feel safe in delaying their immediate goals in favor of planning for the future.

To foster this trust, educators can do the following:

  1. Create a predictable, safe school environment with consistent rules
  2. Create achievable opportunities to work for small, semi-long-term goals
  3. Demonstrate to children, both with words and actions, that their efforts yield positive results (follow through with promised rewards)


This research demonstrates children’s sensitivity to and deep awareness of the consistency of their environment and the reliability of the adults with whom they interact. Further, this study shows that impulse control, which is known to be associated with positive life and academic outcomes, is not intrinsic and unchangeable, as previously thought. By capitalizing on students’ sensitivity to and awareness of the consistency and reliability of their school environment, educators can foster self-control and put students one step closer to success.


Ayduk, O., Mendoza-Denton, R., Mischel, W., Downey, G., Peake, P. K., & Rodriguez, M. (2000). Regulating the interpersonal self: Strategic self-regulation for coping with rejection sensitivity. Journal of Personality and Social Psychology, 79(5), 776–792.

Karniol, R., Galili, L., Shtilerman, D., Naim, R., Stern, K., Manjoch, H., & Silverman, R. (2011). Why Superman can wait: Cognitive self-transformation in the delay of gratification paradigm. Journal of Clinical Child & Adolescent Psychology, 40(2), 307–317.

Mischel, W., Shoda, Y., & Rodriguez, M. I. (1989). Delay of gratification in children. Science, 244(4907), 933–938.

Shoda, Y., Mischel, W., & Peake, P. K. (1990). Predicting adolescent cognitive and self-regulatory competencies from preschool delay of gratification: Identifying diagnostic conditions. Developmental Psychology, 26(6), 978–986.