The Marshmallow Effect | 2

The first part of this post provided some background to Mischel et al’s (1990) Marshmallow Test and started to question the idea it could be used to predict things like future academic performance.

This concluding post looks in more depth at exactly why this should be the case…

From A to B…

There are two significant problems with the claim that delayed gratification is a cause of both improved academic achievement (what we can term the “soft hypothesis” because it only makes a limited, but still significant, claim about the Test) and a wide range of adult life benefits such as better health, higher incomes and the like (the “hard hypothesis”).

The first is that delayed gratification itself can’t be a cause of something like academic improvement. There has to be a mechanism, such as “willpower” or self-control, through which delayed gratification is expressed. In other words, the claim here is something along the lines that children who have high levels of self-control are able to delay gratification in the Marshmallow Test. This, in turn, explains why these children become academically successful – presumably through things like showing greater determination to overcome problems or having the self-discipline to study rather than use their time in some other way.

As Michio Kaku, a strong advocate of the predictive powers of the Test, puts it:

“The ability to delay gratification is a key predictor of success because it reflects self-control, discipline, and the ability to plan for the future. These traits are crucial for academic achievement, as they help students stay focused, manage their time effectively, and persist through challenges.”

Assuming these traits really are “crucial for academic achievement” – something that ideally needs to be tested rather than (conveniently) assumed – a second problem involves showing how “self-control” directly connects to and determines improved academic ability. In other words, we don’t actually know if those students who delayed gratification and performed well academically did so because they developed greater self-discipline, determination, grit and so forth. Proponents of the Marshmallow test simply assume that they do.

Now you see it…

Leaving aside the argument that advocates of the Marshmallow Test are simply assuming relationships they should be proving, the general takeaway here is that will-power is a hugely-important trait that not only determines our chances of educational success, it also determines things like our future body-mass index (the extent to which we are under or over weight), general health, happiness and income. More importantly, will-power is not only something that can be set at a very young age, it is seemingly resistant to change, in the sense that once you have it, it remains with you throughout your life – which, although it seems improbable, is something we’ll let pass. There are more-significant problems with the Test – the first of which is does it actually measure what it claims to measure?

This is a slightly-odd question because when we normally consider the validity of something like an experiment we’re generally looking at the relationship between certain pre-defined variables.

For example, if we wanted to know how a plant (the dependent variable) is affected by changes in light (the independent variable) we need to define exactly what we mean by “a plant”. If we didn’t accurately define it we might accidentally measure the effect of light on something we mistake for “a plant”, such as television (not a great example, but you probably get the drift).

In other words we assume the variables themselves aren’t up for debate.

In the case of the Marshmallow Test, however, recent neuroscientific developments (that obviously weren’t available to the original researchers) have suggested that what the original researchers thought they were measuring may not have been what they were actually measuring.

And that presents a Very Big Problem.

We’ve known for quite some time that different parts of the brain serve different functions but it wasn’t until it became possible to scan the brain to see images of what was actually happening inside our heads in real time that we could see exactly which parts were active under what conditions.

Now You Don’t…

To this end, a study by Casey et al (2011) attempted to extend the longitudinal Marshmallow Test findings by scanning and imaging the brains of “a sub set of (26) individuals from the original study group” as they took-part in a test that replicated the conditions of the original Marshmallow Test using their responses to different faces because, as Casey et al. noted, “Confectionary items such as cookies would no longer be highly desirable with the participants grown up as adults by now”.

Long story short, what they discovered from the imaging was that the study participants were not using the part of the brain associated with “will-power” or self-control (the anterior midcingulate cortex if you’re really keen to know). Rather, they were using the nucleus accumbens, a part of the brain associated with impulse control.

Apart from the fact the original Marshmallow Test doesn’t seem to measure what its subsequent cheer-leaders assumed it measured – making the conclusions they drew about immediate and delayed gratification largely invalid – those participants who were able to delay gratification probably did so on the basis of a stronger impulse control.

To put this in terms of the initial Marshmallow test, while Mischel believed, quite reasonably given our knowledge and understanding of the brain at this time, they were measuring self-control, what they were actually measuring was each individual child’s reward response to something they saw as desirable: in this instance, food. Those who were able to delay their response to the food were not exercising greater self-control than those who immediately gratified their desires. They simply didn’t have the same level of desire (or craving if you prefer) for the food.

If you want to think of it another way, imagine a chocoholic (not a real thing, but you know what I mean) and someone who doesn’t like chocolate being presented with a piece of chocolate cake. The chocoholic is least likely to resist eating the cake not because they lack will-power but because their reward response to something they value highly will lead them to eat it before someone who doesn’t see the cake as desirable…

Casey et al summarise this general criticism of the test in terms of the idea that

“These results suggest that the Marshmallow Test may capture something rather distinct from self-control…Our results further suggest that simply viewing gratification delay as a component of self-control may oversimplify how gratification delay operates in young children.”

And if you fancy wasting a few precious minutes watching something that sensationally misunderstands this idea.  

Here’s Michio…

The Situation…

So far we’ve looked at the Marshmallow Test from a predominantly psychological viewpoint and to complete this toasting we can change the focus to look at it through a more-sociological, lens.

The first way we can do this is by looking at the concepts of delayed and immediate gratification  and the historical sociological context in which they’ve been used. This should illustrate a fundamental difference between individualistic psychological and sociological perspectives.

Barry Sugarman (1970) was one of the first UK sociologists to use the concepts when he argued that one of the defining features of working class subcultures was immediate gratification – an idea he contrasted with the delayed gratification characteristic of the middle-classes. The key difference between this usage and the way they’re used in something like the Marshmallow Test is their situational context. Immediate gratification in the latter is considered in terms of individual responses to specific situations (the Test) while in the former it’s considered in terms of general social responses to particular social situations.

To put it simply, in terms of the Marshmallow Test the rational response is to delay gratification in the certainty that if you can do this for a few minutes you are guaranteed a reward. Sociologically, however, if you’re working class the rational response is the reverse: you should choose immediate gratification because that way you are at least guaranteed some benefit in an uncertain world. To put it bluntly, you take what’s on offer now because there’s no guarantee anything better will be available at some indeterminate future point (and if you don’t take what’s on offer you risk losing it).

Thus, although the same concepts are present, they’re being used in very different ways and this produces very different meanings and interpretations. For sociologists (and, to be fair, situational psychologists), the key criterion here is the cultural situation in which people find themselves.

A second way to change the focus is to test whether the crucial variable in the original Test was the situational context in which it took place. While Mischel et al. were very careful to ensure their lab setting was made as welcoming and undistracting as possible, one crucial variable was not controlled: the fact the situation itself was reliable. When the experimenters said to the young children that if they could wait 15 or 20 minutes without eating the marshmallow they would receive a second marshmallow they were true to their word.

However, as we’ve suggested, one of the key sociological observations about gratification in relation to the working and middle classes is that the former tend to experience the world as unreliable: something, such as food, that is available now may not be available at some point in the future. It makes, therefore, perfect sense in unreliable situations to take whatever’s on offer now (such as a tasty treat) because there’s no guarantee it will be available later – a form of food insecurity frequently experienced by the poor. Middle class children, on the other hand, tend to live more-reliable lives; they rarely, if ever, experience something like food insecurity for example.

May Not Be Reliable…

In their replication of the Marshmallow Test, Kidd et al (2012) used an Art Project task to test the idea that the nature of the situation – whether the children perceived it as being reliable or unreliable – would significantly impact on the test results. And what they found probably won’t surprise you (although it’s worth watching nonetheless):

In the previous studies of Mischel and Ebbesen and Shoda et al., the children assumed (correctly) that the experimenters were reliable – if they promised an extra Marshmallow for waiting they delivered on this promise – and they waited, on average, around 6 minutes. Kidd et al, however, found that in their experiment “children waited twice that long in the reliable condition (12 min), and half as long in the unreliable condition (3 min)”.

As they conclude, “The effect we observed is consistent with converging evidence that young children are sensitive to uncertainty about future rewards” and that “The findings suggest that children’s ability to delay gratification isn’t solely the result of self-control. It’s also a rational response to what they know about the stability of their environment.”

Diversity…

Mischel and Ebbeson’s (1990) study has, probably fairly, been described as “A group of upper middle class researchers researching the behaviour of upper middle class preschoolers in an upper middle class setting” (Stanford University – one of the most prestigious Universities in America, if not the world) and so the final question we need to ask is what would happen if we introduced a bit of diversity into the proceedings?

Could the results of the test be replicated among, for example, lower middle class or working class preschool children?

This was a question – among many others – addressed by Watts et al. (2018) in their replication of the 1990 Marshmallow Test and the study threw-up three interesting conclusions:

First, they found the preschool children of mothers who had not completed college (i.e. they left education between the ages of 16 – 18, depending on the State in which they lived) did show a lower level of post-15 achievement than researchers such as Mischel and Ebbeson had found. However, they also found this relationship was much weaker than previous research had claimed.

Second, although Watts et al. did find a relationship between delaying gratification and subsequent academic achievements – albeit, as we’ve noted, at a very low level of significance – a puzzling aspect of the replication was that “for the children of non-degreed mothers, most of the achievement boost for early delay ability was gained by waiting a mere 20 seconds”.

As you may recall, one explanation for the claimed relationship between the ability to delay  gratification for 15 – 20 minutes at preschool level and subsequent higher levels of academic achievement was “driven by the ability to generate useful metacognitive strategies that will influence self-regulation throughout one’s life”.

In other words, what we are allegedly seeing in some preschool children was the early presence of study strategies (such as the ability to stick at a difficult task and see it through successfully) that would prove highly beneficial in later life. However, as they note, “Such strategies are unlikely to have played much of a role in a child’s ability to wait for only 20 seconds” – something that leads neatly to our third consideration.

When we’re dealing with a relatively homogenous group for children there may well be differences in intelligence, application and the like that result from a possible range of social-psychological factors.

However, once Watts et al. expanded the selection of participants to include children from different social backgrounds they found that “when factors like family background, early cognitive ability, and home environment were controlled for” the associationbetween the ability to delay gratification and academic success in later life “virtually disappeared”.

Once the researchers controlled for factors historically-known to hugely influence things like academic achievement, better health in later life and so forth the “Marshmallow Effect” ceased to be…

References

Mischel, W (1958) “Preference for delayed reinforcement: An experimental study of a cultural observation”: Journal of Abnormal and Social Psychology, 56.

Mischel, W and Ebbesen, E. B. (1970) “Attention in delay of gratification”: Journal of Personality and Social Psychology.

Shoda, Yuichi, Mischel, Walter and Peake, Philip (1990) “Predicting Adolescent Cognitive and Self-Regulatory Competencies From Preschool Delay of Gratification: Identifying Diagnostic Conditions”: Developmental Psychology Vol 26.

Watts, Tyler; Duncan, Greg;  Quan, Haonan (2018) “Revisiting the Marshmallow Test: A Conceptual Replication Investigating Links Between Early Gratification Delay and Later Outcomes”: Psychological Science

Casey, B. J., Somerville, L. H., Gotlib, I. H., Ayduk, O., Franklin, N. T., Askren, M. K., Glover, G. (2011) “Behavioral and neural correlates of delay of gratification 40 years later”: Proceedings of the National Academy of Sciences, 108(36).

Sugarman, Barry (19770) “Social Class, Values, and Behaviour in Schools”.

Kidd, Celeste; Palmeri, Holly; Aslin, Richard (2012) “Rational snacking: Young children’s decision-making on the marshmallow task is moderated by beliefs about environmental reliability”.