The Marshmallow Test Revisited(washingtonpost.com)
washingtonpost.com
The Marshmallow Test Revisited
http://www.washingtonpost.com/blogs/ezra-klein/wp/2012/10/13/the-marshmallow-test-revisited/
10 comments
Perhaps what the first study actually discovered, unbeknownst to the original researchers, is that children from homes in which the guardians are reliable turn out to be more prosperous in life. Therefore, what was actually being measured, wasn't the self control of the children, but a proxy for the reliability of the adults in their life.
While that's a tempting conclusion, it's a tough to justify. The original experiment didn't study or account for the parental variable, so it's impossible to tell what effect it may, or may not, have had "behind the scenes."
It's entirely possible that the reliability of adults is a factor in developing a child's impulse control. It's also possible that it's a significant factor. But we'd have to design experiments to study its significance. Furthermore, it seems unlikely that the variable being studied in the original experiment, impulse control, was actually just a red herring disguising parental reliability as the true factor.
It's entirely possible that the reliability of adults is a factor in developing a child's impulse control. It's also possible that it's a significant factor. But we'd have to design experiments to study its significance. Furthermore, it seems unlikely that the variable being studied in the original experiment, impulse control, was actually just a red herring disguising parental reliability as the true factor.
>Furthermore, it seems unlikely that the variable being studied in the original experiment, impulse control, was actually just a red herring disguising parental reliability as the true factor.
The variable being studied was weather or not the children would wait for the second marshmallow, and that the children who did wait did better in life. I suspect that everyone would agree that this is not an casual relationship, which would imply that the children did better because they waited for the marshmallow. Therefore, we know that we are looking for what factor(s) lead to the correlation between waiting and success in life. In the original experiment it was assumed, I believe without justification, that the common cause was impulse control. This experiment shows that the trustworthiness of the environment plays a significant role in determining how the children make their decision. This result support the hypothesis that the original marshmallow experiment was a proxy for the living environment of the children, not their impulse control. I'll leave it up to researchers to determine how to answer the question of what is going on behind the scenes, however I think the take home message for the rest of us is that experiments and statistics only prove exactly what was being looked at, and any conclusion we draw from that is interperatation that is subject to human error.
The variable being studied was weather or not the children would wait for the second marshmallow, and that the children who did wait did better in life. I suspect that everyone would agree that this is not an casual relationship, which would imply that the children did better because they waited for the marshmallow. Therefore, we know that we are looking for what factor(s) lead to the correlation between waiting and success in life. In the original experiment it was assumed, I believe without justification, that the common cause was impulse control. This experiment shows that the trustworthiness of the environment plays a significant role in determining how the children make their decision. This result support the hypothesis that the original marshmallow experiment was a proxy for the living environment of the children, not their impulse control. I'll leave it up to researchers to determine how to answer the question of what is going on behind the scenes, however I think the take home message for the rest of us is that experiments and statistics only prove exactly what was being looked at, and any conclusion we draw from that is interperatation that is subject to human error.
"This experiment shows that the trustworthiness of the environment plays a significant role in determining how the children make their decision."
No, the experiment shows that the trustworthiness of the person establishing the immediate challenge plays a significant role in the outcome. It's a huge leap to go from that relatively mundane finding, to the conclusion that the randomly selected participants in the original study were biased in a systematic way toward mistrust of adults.
"This result support the hypothesis that the original marshmallow experiment was a proxy for the living environment of the children, not their impulse control."
It provides only extremely weak support for that hypothesis. A reasonable prior expectation is that it is unlikely for a randomly selected group of young children to share a mistrust of adults that would carry on throughout their lives.
No, the experiment shows that the trustworthiness of the person establishing the immediate challenge plays a significant role in the outcome. It's a huge leap to go from that relatively mundane finding, to the conclusion that the randomly selected participants in the original study were biased in a systematic way toward mistrust of adults.
"This result support the hypothesis that the original marshmallow experiment was a proxy for the living environment of the children, not their impulse control."
It provides only extremely weak support for that hypothesis. A reasonable prior expectation is that it is unlikely for a randomly selected group of young children to share a mistrust of adults that would carry on throughout their lives.
"It provides only extremely weak support for that hypothesis. A reasonable prior expectation is that it is unlikely for a randomly selected group of young children to share a mistrust of adults that would carry on throughout their lives."
I think you misunderstand gizmo686's point. My reading (of gizmo686's post as well as the study) is that this is not about a question of merely the individual child's responses but also about their environment, which at the age of 3 is going to be primarily controlled by their parents. It's not just a matter of whether a child who grows up mistrusting the world is at a disadvantage. It's more that a child who grows up in a world that is untrustworthy is at a disadvantage.
Let me be blunt, we are talking about shitty parents. Or, to tone it down one notch, less capable parents. On the one hand you have parents who can provide for their children, who are attentive to their needs, etc. Those parents will tend to raise children better, and will tend to raise children who grow up to be more successful. On the other hand you have parents who let their children down often, are unable to provide for them as well, make promises that go unfulfilled, etc. Those parents will tend to raise children less well, and will tend to raise children who grow up and have more problems in life, problems finding jobs, problems with law enforcement, problems with relationships, etc.
In broad strokes this is essentially what the second experiment is trying to show that the first experiment is testing. By intentionally priming children to either be trusting or untrusting of their immediate environment and of adults they show that they can control the outcome of the marshmallow experiment. And that in turn leads to the argument that perhaps it is a similar degree of priming (through the ongoing experiences with their family) and the underlying level of trust for adults that is being tested in the first experiment.
It's not an ironclad experiment but it's an interesting hypothesis, certainly, and at the very least it unravels some of the overly simplistic interpretations of the first marshmallow experiment.
I think you misunderstand gizmo686's point. My reading (of gizmo686's post as well as the study) is that this is not about a question of merely the individual child's responses but also about their environment, which at the age of 3 is going to be primarily controlled by their parents. It's not just a matter of whether a child who grows up mistrusting the world is at a disadvantage. It's more that a child who grows up in a world that is untrustworthy is at a disadvantage.
Let me be blunt, we are talking about shitty parents. Or, to tone it down one notch, less capable parents. On the one hand you have parents who can provide for their children, who are attentive to their needs, etc. Those parents will tend to raise children better, and will tend to raise children who grow up to be more successful. On the other hand you have parents who let their children down often, are unable to provide for them as well, make promises that go unfulfilled, etc. Those parents will tend to raise children less well, and will tend to raise children who grow up and have more problems in life, problems finding jobs, problems with law enforcement, problems with relationships, etc.
In broad strokes this is essentially what the second experiment is trying to show that the first experiment is testing. By intentionally priming children to either be trusting or untrusting of their immediate environment and of adults they show that they can control the outcome of the marshmallow experiment. And that in turn leads to the argument that perhaps it is a similar degree of priming (through the ongoing experiences with their family) and the underlying level of trust for adults that is being tested in the first experiment.
It's not an ironclad experiment but it's an interesting hypothesis, certainly, and at the very least it unravels some of the overly simplistic interpretations of the first marshmallow experiment.
I believe timr's point is that this is a hypothesis based on the outcome of the study, but the study did not measure it. The study measured the response to the experimenter.
I would add that it's possible that the two marshmallow experiments have very little to do with each other besides using children and marshmallows to perform. Further study is needed to make further conclusions.
For example, we might be tempted to try to combine the results and say something like "kids who trust and respect authority do better in life as adults." But this is not what the study/studies measured, and while this study may inspire interesting hypotheses for future work, that work needs to be performed before we can make conclusions.
I would add that it's possible that the two marshmallow experiments have very little to do with each other besides using children and marshmallows to perform. Further study is needed to make further conclusions.
For example, we might be tempted to try to combine the results and say something like "kids who trust and respect authority do better in life as adults." But this is not what the study/studies measured, and while this study may inspire interesting hypotheses for future work, that work needs to be performed before we can make conclusions.
The article doesn't actually say the unreliable adult was the same one offering marshmallows. Maybe the subjects (children) extrapolated from one recent adult interaction to all adult interactions. You'd then have to hypothesize Every adult in contact with a child, even incidentally, had a profound effect on a child.
Fortunately its not clear the effect is long-lasting. Small children have the attention span of a fish.
Fortunately its not clear the effect is long-lasting. Small children have the attention span of a fish.
To be clear (and I don't think you are implying this, I just want to be explicit) -I'm not suggesting that the adults impacted the development of the child's impulse control - I'm extrapolating from the cited paper towards a quite different conclusion. If their conclusion is accurate, that children are rational actors in the face of uncertainty (that is, if the children asses that future rewards are more or less likely to happen based on their assessment of the other adults in the children's life) - then what I'm suggesting is that the original study had a group of rational players, and what the scientists were measuring, wasn't impulse control of the children (which I'm suggesting might be innately consistent across the group, or at least not revealed in this study) - but instead, was identifying how reliable the guardians and adults in these children's life were. It stands to reason, that children growing up in stable households, would have a better chance of being prosperous in life, than those who grew up in unstable, unreliable homes.
Not to mention that there is probably a relationship between reliability and impulse control. And parents with poor impulse control tend to have children with poor impulse control(whether the relationship is genetic or learned I'll leave to others to decide). So it stands to reason that children with poor impulse control will also have parents that score low on reliability, even if the reliability in itself is not the cause.
From the actual paper:
"To be clear, our data do not demonstrate that self-control is irrelevant in explaining the variance in children’s
wait-times on the original marshmallow task studies. <snip> Rather, an unreliable worldview, in addition to
self-control, may be causally related to later life outcomes..."
In watching my kids grow up I'm always struck by my inability to figure out why they did certain things. They obviously had reasons but when questioned those reasons were regularly things I would have never thought of. I always felt it was a bit like anthropomorphizing animals to make deductions about their behavior. You might think you have a good hypothesis but it's very difficult to know that you're working with the same variables they are.
The more I read about psychology/rationality, the more I realize that what you're saying is also true about ourselves. We regularly do things that don't make sense, and have only vague, and often wrong, ideas about why we did what we did. Except that our brain is amazingly good at self-delusion, making us believe that there were reasons behind it all.
There are almost always reason, but not always conscious rationality. Emotions in the limbic system are sort of like the analog machines of old, before digital logic computers, as a weak analogy.
What is often ignored is that rational decision making depends on priorities and estimations of prior probability, and these vary wildly based on circumstance.
What is often ignored is that rational decision making depends on priorities and estimations of prior probability, and these vary wildly based on circumstance.
I'd like to agree; but the conclusion "Irrationality is observed; sometimes I find out its based on emotion: All irrationality is based on emotion" is not warrented. Its possible people make bad decisions because they're bad-decision-makers. Or maybe something else, I don't know.
I hear you. There's always a method to the madness, but sometimes it's hard to figure out in advance.
The interesting bit in the original experiment seemed to be the kids who'd say they'll wait for 15mins, but then succumb to the temptation of the marshmallow in front of them and eat it while waiting. Any rational (inductive) bit ought may be expected to be completed before the wait started.
PS: Will be good if such article writers link to the paper, or at least the abstract.
PS: Will be good if such article writers link to the paper, or at least the abstract.
When I saw the title, I originally thought of:
http://marshmallowchallenge.com/Welcome.html
I think the above is a more interesting challenge. Particularly because there is a magical point where the kids or even adults realize that they just have to start playing with the spaghetti - can't just stare at it.
http://marshmallowchallenge.com/Welcome.html
I think the above is a more interesting challenge. Particularly because there is a magical point where the kids or even adults realize that they just have to start playing with the spaghetti - can't just stare at it.
The more interesting part of that is that, next to engineers, kindergardeners are the best performing group at the marshmallow challenge. [1]
[1] http://marshmallowchallenge.com/TED_Talk.html
[1] http://marshmallowchallenge.com/TED_Talk.html
I thought of the parody: http://www.youtube.com/watch?v=jQvBrEEYS20
Which was actually the first time I had heard of this test.
Which was actually the first time I had heard of this test.
It would be delightful if the secret undercurrent of that game is that *someone always eats the marshmallow", and the importance of the exercise is in who eats the marshmallow, when it is eaten, and how everyone reacts.
Did any of the kids eat one marshmallow and then trick the study administrators into giving them the other two marshmallows? Those are the kids that are going to be founders.
Maybe I'm reaching.
Maybe I'm reaching.
I think I remember reading about some of the kids (in the original study) who hollowed out the marshmallow and ate the insides, then left the shell on the plate to trick the adults.
Would be interesting if there were. I remember always having a knack for having the cake and eating it too as a child.
What if the child didn't much care for marshmallows, and reasoned that spending 15 minutes of their life waiting for an extra marshmallow was an incredible waste of time?
Funny; but seriously the kids enter the room knowing "Im taking part in a serious Adult study of my behavior." They likely try to behave as they're expected to behave. The real test is, how soon does the feeling of obligation decay to the point that the appeal of a marshmallow (admittedly insignificant) outweights the desire to please.
My Sister took her youngest to an experiment where, in the waiting room they put the kid in a room with toys on a table, and took her to 'do paperwork'. The boy was told to leave the toys alone.
Really she went to the observation room. They simply timed how long it took for the bored child to play with the toys.
See, by testing the child 'before the test started' they avoided pre-loading the kid. They changed-up the test by having other kids and adults enter the room and play with the toys, or repeat the admonition, or whatever.
The
My Sister took her youngest to an experiment where, in the waiting room they put the kid in a room with toys on a table, and took her to 'do paperwork'. The boy was told to leave the toys alone.
Really she went to the observation room. They simply timed how long it took for the bored child to play with the toys.
See, by testing the child 'before the test started' they avoided pre-loading the kid. They changed-up the test by having other kids and adults enter the room and play with the toys, or repeat the admonition, or whatever.
The
My issue is this: I do not know at what age the ability to prioritize is developed(this may be ignorant). I feel prioritizing and weighing pros vs cons are both much more valuable than delaying gratification.
I am not a young child, anymore, but I could imagine a child (me) taking the marshmallow that is sitting in front of me, because maybe I'd rather have a marshmallow now, and waiting 15 minutes is just too long for a second marshmallow. Maybe I'm wrong and I was just a kid who didn't think about candy all the time.
My point is even when I was really young, I still may value my time (not consciously), and would rather go play with legos and have a marshmallow, than spend my time concentrating on getting a second.
I am not a young child, anymore, but I could imagine a child (me) taking the marshmallow that is sitting in front of me, because maybe I'd rather have a marshmallow now, and waiting 15 minutes is just too long for a second marshmallow. Maybe I'm wrong and I was just a kid who didn't think about candy all the time.
My point is even when I was really young, I still may value my time (not consciously), and would rather go play with legos and have a marshmallow, than spend my time concentrating on getting a second.
Good point. I'd hope the study would control for this by locking you in a room for 15 minutes regardless of marshmallow logistics.
Whenever I meet a new kid, I mentally classify them as either a "one marshmallow child" or a "two marshmallow child" based on what I think the result woud be if they took the marshmallow test. One thing I have noticed is that there is a strong correlation between one marshmallow kids and one marshmallow parents.
The question then is then: is it a genetic predisposition or is it more behaviorally learnt?
This calls for a twin study!
For anyone interested, here is the original paper:
http://www.bcs.rochester.edu/people/ckidd/papers/KiddPalmeri...
It was hyperlinked in the 3rd paragraph.
It was hyperlinked in the 3rd paragraph.
Notice what they downplay in this article: That almost half of the kids who had just been in contact with a reliable adult still ate the marshmallow.
Also, were the kids allowed to keep playing with the fun art supplies while they waited?
A link to the actual study would help.
Also, were the kids allowed to keep playing with the fun art supplies while they waited?
A link to the actual study would help.
In the third paragraph, "revisited" is a hyperlink to the original paper.
For anyone interested, here is the URL:
http://www.bcs.rochester.edu/people/ckidd/papers/KiddPalmeri...