Friday, February 13, 2015

Waiting for Marshmallows

The Marshmallow Experiment was a series of studies performed at Stanford starting the late sixties (and also not a bad band name) that purported to study self-control and the ability to defer gratification. It comes up these days in many contexts, including discussions of grit and prudence and character education.

It also shows the flaws in some models of human behavior.

In case you slept in that day in Psych 101, here's the basic layout. Put a child and some marshmallows in a room together. Promise the child even more marshmallows if she'll refrain from eating the ones in front of her. Then leave the room. The child's subsequent behavior provides a measure of how much ability the child has to delay gratification.

In 2012, a new variation on the study was conducted at the Rochester Baby Lab, and it revealed a whole aspect of the problem that was not covered in the original experiments.

What if the child's ability isn't the only important variable? What if environment also matters?

The new experiment varied the environment. Some children dealt with a reliable environment, and some dealt with an unreliable one in which the adults they dealt with were not as good as their word.

The format was a one-two punch. The children were given a box of lame crayons and told, "If you can just wait a minute, I'll bring back some better art supplies." Then they were left alone with a sticker and told, "Don't touch this sticker and I'll bring you a bunch of better ones."

In the reliable environment, the adult followed through as promised. In the other environment the adult returned empty-handed with excuses. And then it was time for the marshmallow.

The effect was huge. The mean wait time for children in the unreliable environment was about three minutes. For those in the reliable environment, about twelve. Compared to previous research, that's half as much waiting for unreliables and twice as much waiting for reliables.

In other words, the quality of deferred gratification is not just an innate immutable quality that the child possesses in some sort of vacuum-- it's a rational reasoned response to what one knows about conditions in the environment. Put another way, this quality of "self-control" is really about the relationship between the person and the environment (particularly the parts of that environment shaped by other people).

The broadest conclusion I can draw from this is that what we often ascribe to deficiencies in a person's character are actually behaviors developed in response to that person's environment. We are focusing on the person when we should be focusing on the relationship between that person and the surroundings.

Say your engine is running hot. Should you be looking for a particular engine part that is running with too much friction, or should you check the oil? Say your child has developed hives all around his upper torso and arms where his shirt touches his skin? Should you worry about why his skin has such a hivey quality, or should you be checking to see if he's having an allergic reaction to something in the shirt?

Say your kid won't wait long when you set a marshmallow in front of her. Should you declare the child character-deficient, with  a sad lack of self-control? Or should you look at the environment that child lives in every day and ask how it has taught her that waiting is a fool's game.

Children are learning machines. They are learning all the time, and they are learning lessons like whether or not the world around them can be trusted or counted on. When they arrive at school, they have already earned a PhD in Human Behavior, and they operate with a set of assumptions based on what they've learned.

It is not helpful to say that children who have learned certain lessons from their environment, and who now make choices based on what they've learned-- it is not helpful to label those students as character-deficient because the lessons they've learned are different from the lessons we wish they had learned.

If it helps, think of the conclusions you reach about students as marshmallows. You can reach some conclusions quickly and easily right now. Or you can wait, and you'll get more to work with. Show some self-control.


  1. A wonderful rebuke to the know nothings that make simplified assumptions. Their analysis is missing a few things as this shows. Thanks again. :)

  2. aw- context. We know real world context is irrelevant in David Coleman's mind.