Stanford Marshmallow Experiment

The Stanford Marshmallow Test is a series of studies conducted by psychologist Walter Mischel at the Bing Nursery School of Stanford University. The experiment aims to measure preschool children’s capacity for self-control and delayed gratification. Children are given a choice: they can immediately receive one reward, or wait for a specified period (typically 15 minutes) to receive two rewards. When the researcher leaves the room, the child may ring the bell to claim the single reward immediately, or wait to receive the double reward.

Marshmallow (Generated by Artificial Intelligence)

History and Purpose

The origins of the experiment stem from a 1970 study conducted by Mischel with Ebbe B. Ebbesen. The goal was to determine whether children’s ability to delay gratification constitutes a cognitive skill and whether it correlates with future success. In 1972, Mischel, Ebbesen, and Antonette R. Zeiss designed three experiments to investigate the influence of attention mechanisms on delay.

Experimental Design and Methodology

The design and methodology of the Stanford Marshmallow Test are based on Mischel’s studies from 1970 and 1972. In the 1970 study, it was determined that the visibility of the reward shortened the waiting time; when rewards were visible, children waited an average of 3.09 ± 5.59 minutes, whereas when they were hidden, they waited 8.90 ± 5.26 minutes. During the waiting period, children employed strategies such as looking at a mirror, covering their eyes, or talking to themselves to suppress their attention toward the rewards. This finding demonstrated that delayed gratification is linked to the ability to suppress attention toward the reward.

In 1972, Mischel, Ebbe B. Ebbesen, and Antonette R. Zeiss designed three experiments to examine the effects of active behaviors, cognitive activities, and their absence on waiting times.

In the first experiment, 56 children from the Stanford Bing Nursery School participated; researchers spent several days playing with the children to establish rapport. The children were randomly assigned to five groups (A–E), each taken to a separate room and taught to ring a bell to summon the researcher. Children in groups A, B, and C were shown a marshmallow and a pretzel, asked to indicate their preference, and informed that if they waited, they would receive their preferred item, but if they rang the bell, they would receive the other. The rewards were left in view, but children were allowed to ring the bell at any time.

Stanford Marshmallow Test (Generated by Artificial Intelligence)

Groups D and E received no reward and no instructions. Groups A and D were given a slinky toy and allowed to play with it; groups B and E were encouraged to think of fun things (for example, singing songs or playing with toys); group C served as the control group and received no distraction whatsoever (in this context, any activity or thought diverting the child’s attention away from the reward).

Six children who failed to understand the instructions were excluded, leaving 50 children for analysis. All children were told that the researcher would return regardless of whether they rang the bell and that they would be allowed to play with the toys. Waiting time was measured from the moment the door closed until the bell was rung or until 15 minutes had elapsed.

The second experiment differed from the first only in the following respects: 38 children participated, six who did not understand the instructions were excluded, and the remaining 32 were divided into three groups (A, B, C). All groups were shown a reward, asked to indicate their preference, and told that if they waited, they would receive their preferred item, but if they rang the bell, they would receive the other. The rewards were left visible. Group A was instructed to think of fun things, group B was instructed to think of sad things (with examples), and group C was instructed to focus on the rewards.

In the third experiment, 16 children participated; the rewards were hidden and no distraction was provided. The waiting time under these conditions was similar to that observed when children were given a fun-thinking task. This series systematically compared the effects of the presence and visibility of rewards, types of distraction, and cognitive instructions on delay.

Findings

The first experiment of the 1972 series compared the effects of reward presence and distraction on waiting time. Children who were shown a reward (groups A, B, and C) waited significantly longer than those who were not shown a reward (groups D and E); the promise of a reward motivated waiting. At the same time, children given distraction (group A with the slinky toy, group B thinking of fun things) waited significantly longer than the control group (group C) with no distraction; this demonstrated that diverting attention away from the reward facilitated self-control.

In the second experiment, all children were shown and left a visible reward. Children instructed to think of fun things (group A) waited the longest, while those instructed to think of sad things (group B) and those instructed to focus directly on the rewards (group C) waited for shorter durations. This revealed that the content of thought strongly influences waiting: positive and enjoyable thoughts increased waiting, while negative thoughts or constant focus on the reward made waiting more difficult.

Overall, across all experiments, successful delayed gratification depended on two key factors: suppressing attention toward the reward (not seeing it, not thinking about it, or focusing on something else) and trusting that the reward would actually be delivered. Children waited much longer when they redirected their attention away from the reward and trusted the researcher’s promise.

Stanford Marshmallow Test (Generated by Artificial Intelligence)

Limitations

Although the Stanford Marshmallow Test marked a turning point in research on self-control, the idea that the ability to wait at an early age alone predicts life success has largely lost its validity today. The original studies were limited to small and socioeconomically homogeneous samples; long-term correlations were reported without controlling for factors such as family background, cognitive ability, and home environment. The experiment underscores that, alongside willpower, critical factors in delayed gratification include trust, attention regulation, and environmental conditions.