Skip to content

Is It Really Self-control: A Critical Analysis of the “Marshmallow Test”

November 10, 2013

By Angela Duckworth (University of Pennsylvania)

Marshmallows are everywhere. From the Vanguard newsletter to Sesame Street to the Colbert Report, direct references to the “marshmallow test” suggest this simple task, invented by Walter Mischel and his colleagues a half-century ago to assess delay of gratification in preschoolers, has captured popular interest (Mischel, Ebbesen, & Zeiss, 1972; Mischel & Metzner, 1962).

The challenge of managing impulses when what’s good for us in the long run conflicts with what feels good right now is hardly new. Virtually every philosophical and religious thinker has opined on self-control and its importance to both virtue and worldly success.  Of course, these opinions have varied. For instance, both Plato and Mill contended that battles of will were always adjudicated in favor of long-run interests when sufficient reflection and reason were brought to bear on the conflict. Aristotle and Frankfurt, on the other hand, wholly acknowledged the possibility of akrasia, weakness of will in the face of alluring but fleeting pleasures. Empirical research on self-control, which has expanded exponentially in recent years, supports the latter view.

It is, in fact, this burgeoning research literature on self-control that explains, at least in part, why non-psychologists are familiar with the marshmallow task. The Palo Alto preschoolers Walter studied in the 1960s have now passed through adolescence and young adulthood. It is now known that he number of seconds children were able to wait for a larger, preferred treat predicts an astounding array of life outcomes. Separate research using different measures of self-control by Moffitt and colleagues (2011) also supports the importance of self-control for adaptive maturation over the life course. In other words, what Freud conjectured in his Introductory Lectures on Psychoanalysis, has now been affirmed by rigorous empirical research. Specifically, his supposition that successful development hinges upon the capacity to “postpone the obtaining of pleasure, to put up with a little unpleasure and to abandon certain sources of pleasure altogether” (Freud, 1920/1966, p. 444).

While non-psychologists find the delay of gratification research intuitively sound, I have been to more than one meeting where the validity of this task was questioned. Does the task really measure self-control, or does it, for instance, simply index general intelligence? Does it measure approach or reward-oriented impulses more than the ability to manage such impulses? Or, is the task really an index of compliance with adult authority?

To answer these questions, my collaborators and I (Duckworth et al., 2013) adapted the task slightly and administered it to a sample of 56 fifth grade children. We found that the longer children were able to wait for a preferred treat correlated with parent and teacher ratings of self-control but not with other personality traits, intelligence, or reward-related impulses. In a second sample of 966 preschool children in a longitudinal study conducted by NICHD, we found that delaying gratification was related to concurrently measured intelligence. However, the predictive power of the delay task for academic achievement, physical health, and risky behavior in adolescence was more consistently explained by ratings of self-control by parents and caregivers (see Figure 1).

In sum, the evidence we gathered in these two studies suggests that the marshmallow task really does measure self-control.


Figure 1. Relationship indicators of self-control in Study Two.

All measures have their limitations, of course, but since psychological science benefits tremendously from behavioral assays, one might ask why there are so few of them? Why do we researchers rely so frequently on self-report questionnaires when, for many research questions, a behavioral measure is better suited? My guess is that the weeks, months, and years of iterative prototyping required to construct, refine, and validate such tasks is itself a delay of gratification test. I hope that as a field we follow through on our intentions to forgo more immediately rewarding temptations to instead do what is best for science in the long-run. That is, I hope we can exercise the collective self-control to invest in research measures that improve on those of our forebears.

Author Information

duckworthDr. Angela Lee Duckworth is an associate professor of psychology at the University of Pennsylvania. Angela studies non-IQ competencies, including self-control and grit, which predict success both academically and professionally. Her research populations have included West Point cadets, National Spelling Bee finalists, novice teachers, salespeople, and students. Angela received a BA in Neurobiology from Harvard in 1992 and, as a Marshall Scholar, a Masters in Neuroscience from Oxford. She completed her PhD in psychology at the University of Pennsylvania. Prior to her career in research, Angela founded a non-profit summer school for low-income children which won the Better Government Award for the state of Massachusetts and was profiled as a Harvard Kennedy School case study. Angela has also been a McKinsey management consultant and, for five years, a math teacher in the public schools of San Francisco, Philadelphia, and New York City.


Duckworth, A. L., Tsukayama, E., & Kirby, T. A. (2013). Is it really self-control? Examining the predictive power of the delay of gratifcation response. Personality and Social Psychology Bulletin, 39, 843-855.

Freud, S. (1920). Introductory lectures on psychoanalysis. New York, NY: Norton.

Mischel, W., Ebbesen, E. B., & Zeiss, A. R. (1972). Cognitive and attentional mechanisms in delay of gratification. Journal of Personality and Social Psychology, 21, 204-218.

Mischel, W., & Metzner, R. (1962). Preference for delayed reward as a function of age, intelligence, and length of delay interval. Journal of Abnormal and Social Psychology, 64, 425-431.

Moffitt, T. E., Arseneault, L., Belsky, D., Dickson, N., Hancox, R. J., Harrington, H. L., & Caspi, A. (2011). A gradient of childhood self-control predicts health, wealth, and public safety. Proceedings of the National Academy of Sciences of the United States of America, 108, 2693-2698.

Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: