Don't Bite: In Sum, Dear Reader

Writing this day by day, rather than all at once, this series has taken a number of twists, turns, and quite frankly, unexpected detours, some of which may have made the underlying logic hard to follow.  (Indeed, to the extent that these posts shed light on the inner workings of my mind, you can tell how frightfully coherent my day-to-day has become.  Frightfully, being the operative word…)

In any case, in this, my last ‘science’ post on the subject, I want to revisit some of the topics we’ve discussed, and clarify and expand on others, so that you may leave this series having learnt something – either about me, or self-gratification, or cookies, at the very least.

What was it we learned, after all?

#1  Performance in the cookie task is not well explained by cognitive control.

The way that we classically conceive of self control – as an ability to control our impulses and urges – is probably not the primary thing that's being measured in the cookie task, since delay-times only correlate weakly with measures of cognitive control.

#2  On the other hand, performance in the task may be explained by verbal ability.

Verbal ability is a strong predictor of individual performance in the task.  This raises the possibility that linguistic competence guides how well children are able to conceptualize and formulate distraction strategies, and thus determines how long they are able to delay.  If this interpretation is right, it would indicate that delay-time in the task may be an indirect measure of verbal ability.  This would have critical implications for how we interpret the results of Mischel’s longitudinal studies.

#3  However, it’s possible that while performance in the task is highly correlated with verbal ability, it is not caused by it.

The link between verbal ability and ability to delay isn’t necessarily causal.  It could also be that the two abilities arise from a similar source – such as the home environment, or an innate set of traits that tend to occur in tandem.  For example, having blond hair does not cause me to have blue eyes, but the two traits are linked genetically, and frequently appear in the same individual, as they do in me.

#4  That said, there are good reasons to believe that verbal ability can account for performance differences in the task.

For one, there are a host of experiments in the “delay of gratification” paradigm that have found that teaching children conceptual strategies for delaying makes them significantly better at delaying.  This suggests that learning and strategy do play a vital role in children’s “ability to delay.”

#5  Why give less credence to the other possibilities?

With regards to genetics:  Except in the case of severe learning disorders,  early environment has been shown to be a far more powerful determinant of language development than an individual’s genetics.  Given that, it seems highly unlikely that verbal ability and ability to delay would be tightly coupled simply on the basis of genetics.  While genetics undoubtedly has some part to play, it seems to add little explanatory value here.

With regards to environment:  We know that vocabulary acquisition is a function of environment, and that children from poorer families tend to have worse vocabulary scores and shorter delay times.  On this basis, it is tempting to conclude that lower-income parents are simply poor models of language and “delay of gratification” alike.  This prospect would be impossible to rule out without further study.  However, I think that there is reason to believe that class stereotypes about self control are both misconstrued, and also very likely irrelevant to behavior in the cookie task.

Further, and more critically, I think this is fundamentally the wrong way to look at the issue.  How people behave, in general, is intimately tied up with the culture they are a part of and the values they adhere to, which is in turn, tied to the language they speak.  In other words, even when we want to separate things out, and look to “environment” to explain something, we never get very far afield from language.

#6  What to make of all this?

When we speak of “self control” colloquially, it is important to understand that we are using an umbrella term that points to a range of behaviors, with various and sundry determinants.  Depending on the situation, those determinants may be cognitive, and may relate to the function of our prefrontal cortex and the wider workings of our mid-brain dopamine system; they may be linguistic, and may depend on the verbal and conceptual strategies we employ to stay our impulses; they may be cultural, and may be explained by our values and the situational and societal constraints we find ourselves working with; and that is just to name a few.

In all this, I think it is essential that we pinpoint what the causal mechanism is that best explains the behavior.  For example, if we think of someone coming to the doctor complaining of an ache in our side, it may be that they’ve partaken unwisely, and should be sent home with some anti-acids and a food advisory.  Or it may be that they have appendicitis, and need to be operated on right away.  Though the symptoms might present similarly at first, it would be critical as a doctor to look for the underlying cause, and not simply diagnose by the first label given.  In the same way, it is not entirely clear that the behavior that produces successful cookie-delayers is the same behavior that helps a kid make the grades to get into college, or resist drug use, or wait to have sex – even though we may call all these behaviors “self control.”

In the literature on self control, there is still much to be done to illuminate the underlying mechanisms that vie to produce “self control.”  In the meantime, I think it is helpful to look for useful correlates – like ‘cognitive control’ or vocabulary – that are empirically measurable, and point to mechanisms that are already fairly well understood.  In a similar vein, we might begin to assess how ‘self control’ behaviors relate to culture by plumbing people’s backgrounds, the idiosyncracies of their language use, and the kinds of attitudes they express; or similarly, look to how they relate to personality, by administering well-established personality measures.  It also seems vital that we assess behavior in tandem with these correlates, and not many years later, when the connection is invariably less clear.

These are what I would take to be important future avenues of research.  As for what we understand now : it seems very likely that the cookie task is, in its way, an indirect measure of language ability.  This would suggest that many of the ‘life outcomes’ that are predicted by the cookie task are related to verbal ability.  Given the relation between socioeconomic status and verbal attainment in this country, this link is not altogether surprising.  However, it points to just how important early-intervention and pre-K programs are, which can lay the foundation when it matters.  While the efficacy of such programs varies widely, and more research needs to be done on how to achieve the most effective interventions, it is clear that they can make a substantial difference (see e.g., Give Well’s report on the nation’s preschool programs for at-risk children).

#7  I’m not in research, or in a position to affect policy.  Is there anything I can do?

If you want to make a difference now, Give Well’s top charity recommendation in the states is the Nurse-Family partnership, which provides prenatal (and postnatal) support and training to young, low-income mothers, and which makes a significant difference in cognitive and social development.

# 8 ...

Which means we have learned, dear readers, that even the snarkiest of writers can polish off her posts with a tug at the heart strings, some money-grubbing organizations, and a sense of having, once again, taken the moral high road.  If what you expected was panache, best wait for tomorrow.

9 responses so far

  • Jason Kerwin says:

    Melody -

    In terms of the value of interventions you should look at Roland Fryer's work on the various programs in the Harlem Children's Zone (HCZ). givewell asserts that

    "any program addressing K-12 education as the key to inequality should be treated with skepticism, until and unless it can produce measurable results. "

    Fryer has demonstrated results from education programs as high as the seventh grade. In sharp contrast, the HCZ programs that attempt to do early-childhood interventions such as "baby college" (which aims to teach parents the correct way to raise their young kids) have no measureable effect.

    I don't doubt that effective early-childhood interventions are possible but it is hard to show that they work. Later when I have time I will look through the literature supporting NFP and see whether they're doing the evaluation right. The fundamental problem with most evaluations is that selection into the program is not random, so better parents end up getting the treatment and then you can't tell if it works. Fryer solves this problem for the HCZ using a random component to some of its student selection.

  • Dudley Filkins says:

    Hi Melody,

    I do wonder if the ability of a child to resist the temptation to take the cookie on the table rather than wait to get the two cookies later, might be related to previous experience the child may have had with the acquisition of delayed rewards: if the child has found that most likely the second cookie is a ruse, and delay may result in the loss of the first cookie, then the correct strategy for that child would be to take "the bird in the hand"; a child who has lived in an environment where self control actually does result in a reward would be more likely to exercise "self control" as that would have proven itself a more successful strategy.

    It's tempting, too, to think that children who have learned the strategy of waiting for the greater reward (which promise they have every reason to expect will be honoured) have been exposed to verbal negotiating skills, so the very act of hearing the explanation of what will happen, and putting the new information into the context of a strategy would be more familiar to them, and thus more likely to be implemented than would the same instructions given to a child whose success strategy is based on physical evidence only.

    (It seems this may even have correlates in later life, with lower socio-economic individuals tending to greater fecundity with more and earlier partners than their higher S.E. counterparts. In an environment where one's own, or one's partner's early death through violence or self harm is a significant consideration, the strategy of early reproduction makes a lot of sense, and may not be a consciously considered decision).

    • Jason G. Goldman says:

      Dudley: this is an interesting speculation. There is some evidence - from an evolutionary perspective - that the availability of resources in the environment may impact on risk-taking and decision-making. See this post for example.

    • melodye says:

      @Dudley This is an excellent point, which is very much in keeping with Karelis' work on how what constitutes rational choice behavior hinges on economic circumstance (see my "week in review" post, and his reply). If I understand correctly, you're saying that what constitutes a logical strategy or approach to the task, may depend on what has proven effective for the child in the past. --Such that, low-SES children may have learned a wholly different set of effective behaviors (given their and their parents circumstances) than high-SES children. And this then may explain why vocabulary and delay behaviors are so highly correlated -- since both the behavior and the vocabulary are tied to being low-SES.

      (Please correct if I got that bit wrong).

      I absolutely agree that this is a real possibility. The problem is that it gets bloody near impossible to disentangle behavior in the task that arises from the culture of circumstance from that which arises from linguistic exposure. What I was trying to get at above (perhaps badly) is that even if it is the case that distinct behaviors are being modeled and so on between classes, many of these behaviors will be modeled *in language.* I'm thinking you may have gotten that much out it from your point about verbal strategizing, but I'll try to clarify a bit further --

      From my perspective -- working in experimental psychology and language -- the particulars of our culture are everywhere reflected in the language we use. And when I say language I don't mean English versus Swahili; I mean English as it gets spoken in a professional household in New England versus English as spoken in a welfare household in the rural south. It is easy to imagine that which words (and expressions and structures) are used is massively different between the two -- not only because of the vocabulary-gap, but also because of the day-to-day topics of conversation, the values that get expressed, and so on. (Ewa Dabrowska, for one, has some crazy-cool work on SES-differences in English comprehension).

      It is (almost trivially) true that the language we speak reflects our cultural values. So, for example, if you look at how words distribute differently between American and British English, you can see that when Brits talk about people, they use class designators ("rich," "poor," "posh" etc) with much greater frequency than Yanks do, whereas Yanks talk way more about race ("black" "white") than Brits. This is what I mean when I say that things start getting muddied when we try to talk about "language" versus "culture," as if they were somehow separable entities. The words we use embody the culture we are a part of, and similarly, propagate that culture, as a bound on our ways of thinking and reasoning.

      When it comes down to it, then, I think that the relationship we're seeing between SES and delay performance, is telling us something important about how the language we speak (with all that entails, cultural baggage attached) predicts the lives we lead.

      • Jason G. Goldman says:

        Heh, but what about your participation in the culture of people who study language? What i mean is, people who tend to operate within certain perspectives tend to see the effects of their "pet perspective" everywhere. Linguists will see language effects. Ethologists will see continuities in behavior across species. Behavior geneticists see heredity. Social learning theorists see associative learning. And so forth.

      • anaxagoras says:

        I think there is a dichotomy that has been used several times in this and related threads, between children raised by parents on welfare and children raised by privileged parents. Wouldn't it be better to compare children raised by parents who are the working poor and children raised by privileged parents? The working poor are often not much better off financially than families on welfare, they do not have access to welfare-associated benefits programs, they have different problems and stresses, and they probably constitute a much larger portion of the bottom of the economic scale than do welfare recipients. By definition a parent on welfare has ample time available to interact with a child, whereas the working poor often have to hand their children over to substandard child care, and the long hours typically required at that end of the wage scale would seriously limit contact between child and parent. These factors might lead the children of the working poor to have a worse outcome than children of welfare recipients.
        Actually, that might be interesting, to compare all three groups: welfare, working poor, and privileged.

  • ed. policy says:

    One more thought on all of this....

    Economists would call this behavior "future discount" .

    I've read very little psych but a middling amount of behavioral economics which takes psych experiments and rewords them as economics.

    Future discount refers to the fact that we prefer 1 unit now over 2 units tomorrow as by tomorrow we could be dead, tomorrow might not come, promises of two might not be kept, a bird in the hand is worth more than 2 in the bush, and so on.

    I don't know if the work on future discount has been done with class issues or verbal issues in mind, but it is certainly at the forefront of economics, and so is clearly found in adults with enough frequency to affect the structures and functions of the economy.

    Again, thanks for the blog!

    • Jason G. Goldman says:

      Indeed, delay discounting is really interesting to me, but most of what I've read on it is related to addiction (one cigarette puff now, or 10 in an hour). I can't remember if I linked to this before, but there is at least some evidence that environmental resources do contribute to general decision-making strategies and risk-taking behavior, on an evolutionary scale. One can speculate about how SES might show similar patterns with respect to decision-making.