A possible link between pollution and crime—and marshmallows

At Wired, Jonah Lehrer delves into an interesting theory about why American crime rates have fallen so drastically over the last 30 years. Apparently, there is both a correlation and a mechanism that would seem to connect falling rates of a certain kind of environmental pollutant to the downward trend in crime statistics. It all comes back to one of my favorite experiments in the annals of behavioral psychology. I'm speaking, of course, of the marshmallow test.

In the late 1960s, psychologist Walter Mischel left pre-schoolers alone in a room with a marshmallow. He gave the kids a choice: Eat your marshmallow now, and it's the only one you get. Resist temptation, and you'll be given two marshmallows to eat later. It's a classic test of delayed gratification and self-control. And only 20% of Mischel's test subjects managed to get the second 'mallow. Their secret: Distracting themselves with other activities, like singing or playing a pretend game.

But here's the interesting thing I didn't know—Mischel has followed those marshmallow kids over the course of their lives. Today, we know that the 20% who could hold out for a second marshmallow also had higher SAT scores, more friends, and fewer anger management issues as teenagers. And, thanks to brainscans, we can actually see differences between the adult brains of the 20% and their less self-controlled counterparts. In particular, Lehrer writes, the 20% demonstrate more activity in "the frontal cortex, [including] areas such as the dorsolateral prefrontal cortex, the anterior prefrontal cortex, the anterior cingulate, and the right and left inferior frontal gyri."

How does this tie back to crime and pollution. Turns out, those are also regions of the brain known to be particularly (and detrimentally) affected by early childhood exposure to lead. Medical researchers have long known that decreased impulse control is a common side-effect of lead-related brain damage. And, during the time that America's crime rates have fallen, so too have the levels of lead in our bloodstreams. According to this theory, environmental regulations that banned things like leaded gasoline and paint might be partly responsible for the fact that the murder rate in many American cities has fallen by 50%.

What Mischel's data demonstrates is that attention isn't just about information. Instead, it's also what allows us to blunt the urges of our errant emotions, allowing us to look past the desire to stuff that yummy marshmallow into our mouth. While we can't always control what we feel - many of our urges are ancient drives, embedded deep in the brain - we can control the amount of attention we pay to our feelings. When faced with a tempting treat, we can look away.

This returns us to aggression. Let's say you're being teased by a bully at school. You can feel your anger rising; the hot emotion is vibrating in your veins. It would feel so good to punch that bully in the face, to vent your frustration with a fight. However, you also know that such violence will get you suspended from school, which is why throwing a punch is not a good idea. If you can't strategically allocate your attention - and this is a skill that requires a solid prefrontal cortex - then you're not going to be able to resist your anger. You're going to get in a fight and get suspended. However, if you can properly look past this negative emotion - perhaps by counting to ten, or just walking away, or finding something else to think about - then your anger will subside. The hot feeling has been cooled off.

Video Link


    1. Which is why I specified that this was a theory supported by correlation and mechanism, but not causation. It’s not proven. But it’s still interesting. Especially given the fact that, as you’ll see in Lehrer’s article, nobody really knows why crime has fallen so much over the last 30 years.

      1. If this is true, then the % of preschoolers who can hold out for more in the marshmallow test should be increasing.

        I can see it now: every kindergartner in the country is given the marshmallow test as a predictor of their ability to achieve as an adult. Oog.

      2. Crime is down in the last thirty years because abortion became legal and there are less unwanted children in the world. Better access to birth control in general has probably helped as well. Still technically a correlation, but a very compelling one.

    2. Right, that’s why the bulk of the post details a plausible causal mechanism that explains the correlation.

    3. But without recognizing the correlation between two facts causation cannot be proven.

      Of course, proving a definite causal link between human murder and lead exposure sounds like it won’t pass through any IRB I know.

    4. More specifically, correlation doesn’t *have* to be causation — a slightly different formulation. While I’m as annoyed as anyone by the mass media picking up the latest barely statistically significant correlation and spinning it as proved causation, the fact is you can’t find causation without finding correlation first. Most correlations will later be shown to be artifacts. But not all.

    5. No, but correlation can make frantic gestures while nodding its head in the direction of causation.

      The whole Correlation =/= Causation thing gets trotted out way too often. Yes, random facts that happen to share some vague mathematical relationship are probably meaningless. But this is an example where they have a proposed connection between the two variables and the whole thing is built on relatively settled science.

      Interesting follow-ups would be doing marshmallow tests in different areas of lead exposure and doing follow ups to track changes over time.

      1. The whole Correlation =/= Causation thing gets trotted out way too often.

        Yeah, but that’s just an anecdote, and therefore you are wrong.

    6. Yes, but as Randall Munroe said, “Correlation doesn’t imply causation, but it does waggle its eyebrows suggestively and gesture furtively while mouthing ‘look over there’.”

  1. Will the correlation is not causation people please go back to sulking behind the lonely green glow of their Ubuntu terminals and leave us happy mutant people alone to squee in glee over the wild and wonderful things that are (possibly) abound in this world?

    Seriously, we get it. Move on to a new trick now. Kthxbye!

    1. That and the continuing decline of surviving Nazis from WWII. But there’s stronger evidence that lead could be an actual factor in the decline. A large number of people just happen to not like Nazis.

  2. Aren’t crime stats heavily manipulated? I want to read a study about the accuracy of crime stats… the one thing all police chiefs can hold up to prove their effectiveness. Plus, from what I’ve read, crime is going up in smaller cities.

  3. @Maggie Koerth-Baker the freakinomics duo says that it lowering of crime correlates with roe vs. wade.

  4. Where are the results of testing lead-free kids today? Seems like they are implying that there would be >20% compliance.

  5. The marshmallow test is bogus. So, whatever.

    Declining lead exposure due to non-lead based paint and unleaded gas regulations are brought up in this WSJ piece. Declining lead exposure is real.

    But as a whole, we are exposed to a wider variety of contaminants (some at ppm, ppb, and ppt) than people in the past. And it’s not just one contaminant, but a whole cocktail of a mess. All that peed out prozac, hormones, and caffeine that our waste water treatment plants don’t filter out ends up someplace. Who knows what those affect will be.

    1. It sure doesn’t sound like the marshmallow test is good science (replicating results is essential, and it sounds like that hasn’t happened) — but this paragraph, right at the beginning of that article: “First, it’s the easiest test in the world to fool. Parents can just promise their kid a pony if they don’t eat any marshmallows or cookies during the evaluation session” — misses the point of the test entirely. If a kid doesn’t have self-control, why is promising them a larger reward going to make a difference? They don’t have self-control!

      1. From what I saw in that Daily Beast article, it’s not necessarily that the test isn’t good science, just that its results weren’t confirmed by another, similarly small study. Not promising, but not the same thing as “bogus.”

    2. That’s interesting. I didn’t realize that the evidence for the marshmallow test being a good predictor was so shaky.

      In this day and age, with so many reports on previous experiments being unreproducible, I had hoped we’d stop relying on single tests with a small handful of subjects.

      From your link it appears that the theory that the marshmallow test is a good predictor of success is based on a shockingly-tiny sample of 35 kids. That’s not good.

      It doesn’t mean that the data aren’t real, but this test really needs to be repeated for us to put any faith in it.

  6. I want them to run the same tests on people before and after they have habitually taken cocaine or ecstasy (or similar dopamine/serotonin dumping drugs). You know there’s an affect.

  7. Politicians and pundits often make claims that they or their policies are responsible for certain changes in society, like crime rates or literacy, and in some cases they may be correct, but I think most of the time they are taking credit for something they had no control over.

    I suspect there are many other aspects of modern life that are controlled by chemicals we have introduced into the environment, but that we have no clue about.

  8. So only 20% made it to a second mallow in the original test performed in the 60s. What about test subject now, 50 years later? Do a higher percentage pass now?

    1. Yes, what’s with those marshmallows? If the article is right, then marshmallow abstinence rates should have increased. Have they? Seems a critical piece of data without which the link is tenuous at best…

  9. “But without recognizing the correlation between two facts causation cannot be proven.”

    “the fact is you can’t find causation without finding correlation first.”

    Now to get super nerdy on ya’ll, while everyone knows that correlation does not imply causation, did ya’ll know that causation does not imply correlation?

    That is, you can still have causation even with no correlation. And, as such, even if you find a lack of correlation you can’t necessarily assume a lack of causation.


    1. Nice. Looks like I have some reading to do. If I don’t nothing will be certain. Or is that nothing is certain even with that knowledge? Or is that depending on what I observe the certainty of my idea may or not be correlated to what I may not have observed? Dogs and cats, living together, mass hysteria!

  10. Exposure to lead is probably a factor, but not the main cause of lowered crime rate. That’s another thing; the carefully worded qualification of “lower crime rates in many cities.” What about other cities? Have we looked at who’s moved from city to city?

    Maybe scumbags flock to a place like Atlanta, prompting the honest people to move away to, say, Asheville, NC (or to another town within Georgia). Over time, this kind of large-scale self-organization of social groups would probably produce the same results being attributed to lead exposure.

    For example, over the course of just two generations, Detroit went from being a powerful and desirable city – “Motor City” – to a crime-infested, half-deserted ruin. What happened to the honest people? Well, after years of prosperity attracting the criminal element, it seems most of the decent people left to find security elsewhere. Without those honest ones to prey upon, the parasitic crooks eventually went broke and now can’t afford to leave the area.

    I don’t expect anyone to take my idea seriously, but hey, it’s just an idea on the interbutt. Poor impulse control is absolutely a factor in criminal behavior, so I have no doubt lead is an issue. But I don’t think lead is a very significant factor in country-wide crime statistics.

  11. It’s things like this that make the public so suspicious of cell phone radiation (or immunization or whatever). Lead in paint and gas for years was great science making our world better, until it turned out that it caused all kinds of problems. Shoe stores used to have X-Ray machines to size your feet, until we realized that radiation causes all kinds of problems.

    These historical incidents have undermined public faith in technology and in industry’s ability to regulate itself and tell the public the truth.

  12. All this would take is the marshmallow test and blood lead tests. I mean, the children theses days most often exposed to lead either live in substandard housing pre-1970s that has been unabated of course, or are children of relatively well to do parents who are remodeling and rehabbing historic structures. I think it would be interesting to look at the two populations given the stratified economic strength and impulse control as correlated to lead exposure.

    1. Current lead problems are not just an issue with houses and paints. Inner cities, especially those in the rust belt with large manufacturing bases, still have issues with lead content in soils. All those leaded gas vapors plus old paint breaking down into dust have left unusually high lead concentrations in soil.

      There are people who can’t (or shouldn’t) grow gardens in Detroit because lead concentrations in soils are so high. New industries can’t move into old industrial centers because banks won’t give loans if environmental risk is too high. Why move your facility into an old industrial armpit when you can go out into the country and plop your facility down on old agricultural land? It’s another reason for sprawl.

      1. “There are people who can’t (or shouldn’t) grow gardens in Detroit because lead concentrations in soils are so high.”

        One word : phytoremediation.

        “Lead: using Indian Mustard (Brassica juncea), Ragweed (Ambrosia artemisiifolia), Hemp Dogbane (Apocynum cannabinum), or Poplar trees, which sequester lead in their biomass.”



        Just give us some time, it can get better!

      2. Heck, we can’t grow gardens in Cambridge Massachusetts, so forget about Detroit. That is to say, of course, we can grow gardens, but only in raised beds.

        @Ugly Canuck: So what do you do with the Indian Mustard or whatever once you’ve grown it? Do you just grow it and throw it away? I assume you can’t eat or compost it. Hmmm….. do I spend ten years growing inedible mustard greens and throwing them away, in the hopes that it mitigates the lead in my soil, or do I simply replace large swarths of my soil with fresh organic matter as need be?

        1. Hi SamSam. I work for an environmental engineer, so the short answer is: you grow a season’s worth of plants (in post-Katrina New Orleans, the boss encouraged people to grow sunflowers to suck up the heavy metals out of the soil), and at the end of the season, with the plants at their largest, you pull them, bag them, and send them to the landfill. The plants themselves then hold the toxins, and would release them were they to rot or turn to dust.

          Test–or re-test–the soil (a county or university extension service does this for nominal fee) at the end of each growing season until the contamination levels are within “acceptable” ranges, then plant food crops.

          I realize the trash-bag approach is not an ideal procedure, because among other things, it relies on landfills for “final” sequestration of heavy metals (bound up in plant material). But growing plants is cheap, and it’s more workable and cheaper than removing all the contaminated soil and replacing it with safe soil. Growing mustard, sunflowers or other plant “hyperaccumulators” can be a very effective low-tech solution in a place that may not have other options at hand. More, briefly, here: http://www.starhawk.org/permaculture/NOLA_bio_basics.html

          Wrt getting the lead out in particular: http://isebindia.com/01_04/04-04-4.html

          There are plenty of lists if you’re serious about using bioremediation, which includes both plants and fungi, to clean the soil. Thank goodness the science on this has been cooking for many years now, and the solid data is in. It works.

          1. I used to think phytoremediate and treatment wetlands were a create answer. Now I’m not so sure.

            You have to monitor phyotremediation quite closely. Not all plants remediate the same contaminants. Certain contaminants simply won’t be taken up by plants. Never mind the fact that I could clean up my acre of land, but surface water and groundwater coming from my neighbor’s property could be hot. If you’re looking at phytoremediation as a way to prep soil for food bearing plants, good luck with that. You’d have to do a heck of a lot of testing to assure me that all hotpots are cleared.

            Phytoremediation sounds good, and can work in small, isolated, well defined areas. But when you’re dealing with sprawling metropolises that have decades of industrial grime sprinkled over them link fairy dust, it’s just not a good approach unless you have decades alloted for cleanup. Most people don’t.

  13. One important step in science is hypothesis formation. Interesting correlations inspire curiosity and hypothesis formation. If we are clear that we are asking a new question that merits investigation, rather than claiming a causal link, then we won’t be in error on the admittedly-important “correlation=/=causation” issue.

    One of the most impressive correlations in environmental health is the chart that shows blood lead levels next to total lead used in gasoline. Sometimes regulations really work. See the pretty lines: http://www.epa.gov/glnpo/bnsdocs/98summ/alead/Figure1.gif (USA 1974-1992)

    It would be interesting to compare blood lead levels and crime rates in various countries.

    Unfortunately we do have opportunities for ongoing field testing of lead-exposed populations, as many places in the world still use leaded gasoline.

    Even better than more research would be an immediate global ban on leaded gasoline.

  14. I work in medicine. It’s always interested me that the typical “low functioning” patient population, ie, the repeat customers, tend to have a lot more individuals who display problems with delaying gratification. Unfortunately, their doctors tend to have a lot of individuals who are above the median in delaying gratification (a requirement to proceed through all 12-16 years of undergrad, then med school, then residency selects for or entrains this) and therefore have trouble understanding why their patients simply can’t check their sugars, stop eating the twinkies so much, remember to take their HCTZ, or lay off the meth for a while. It’s mutual incomprehension.

  15. Rick Nevin, an economist, published a paper using data from several countries that demonstrates a strong link between lead exposure and crime: http://www.sciencedirect.com/science/article/pii/S0013935107000503

    “Nevin says his data not only explain the decline in crime in the 1990s, but the rise in crime in the 1980s and other fluctuations going back a century. His data from multiple countries, which have different abortion rates, police strategies, demographics and economic conditions, indicate that lead is the only explanation that can account for international trends. ” — From http://www.washingtonpost.com/wp-dyn/content/article/2007/07/07/AR2007070701073_pf.html

  16. Huh. I’m just finishing my second quarter of introductory psychology, and they’re still using that marshmallow study in textbooks. As my instructor is one of the authors of this quarter’s book, maybe I should give her a heads-up for future editions!

    (And this is one of the reasons I love to read the comments here. You guys have a lot of useful info!)

  17. I’m doing some research to post a bit on that report in the next couple weeks.

    Right now, there is very, very, very little known definitively about what is the relationship between pollution and criminality. And what we do know for sure cannot (at this point) be tied to lead exposure.

    I’m all for calling out correlations when they happen. But I also happen to think it’s pretty unnecessary (and, in fact, somewhat discrediting of situations where the evidence is strong) to take largely undocumented phenomenon of unknown origin and call them correlation.

  18. I’d be curious to see what percentage of the population was in prime age for crime (teens through maybe 35) during that 30-year span. I suspect it’s been falling as the country’s demographics concurrently shift up the population pyramid.

  19. The video link is actually cut off somewhat.
    The original is here:

    My nephew ended up with two marshmallows.

  20. Increased self control and better delayed gratification has been used by economists to explain the differences in wealth and success between individuals and countries. If both of these traits were on the upswing and causing the drop in crime rates, I would also expect to see an improving economy and more equal distribution of wealth (since more people will take the delayed gratification actions that will make them successful). Instead we see precipitous economic decline and increasing wealth concentration. Of course, a myriad of other factors are at play here. But if those traits are, in fact, increasing and causing our drop in crime rates, and we’re declining economically anyway, then we’re really in a terrible spot.

    I think the much more likely pollutant explanation is plastics. The xenoestrogens in plastics, and the phytoestogens in our plant sources (like soy) have been affecting male reproductive health, causing less males to be born, and also causing testosterone levels to drop. It is well known that testosterone is linked to violence and aggression in humans. So a combination of less men, and men and women with lower levels of testosterone seems to me the more likely cause of the lower level of violence (and metrosexualism to boot).

    The lead theory would also provide and interesting explantion for the decline of the Roman Empire, who used lead extensively in pipes. I’ve heard that theory put forward in the case of lead poisoning, but not in relation to changes in the frontal cortex. Perhaps the late Empire fell apart because its wealthier subjects could no longer delay gratification due to their plumbing.

  21. Wait, what? The recommended response to being bullied is just to suck it up? And learn how to ‘resist your anger’?

  22. I was one of the marshmallow kids. Here’s the thing: I don’t like marshmallows and never have.

  23. Of course it has nothing to do with a population bubble. This is exactly the same time a huge portion of our population was moving into an age when crime is a more likely option. These poeple aged past their crime tendancy years and crime has drastically fallen since. We call these people the baby boomers.

  24. If I recall correctly, this video was originally produced with an agenda, endorsing the conclusion. It’s not a scientific study.

    I believe it’s actually a production of the Watermark Community Church (http://www.Watermark.org) and the theme is the deferred reward of heaven.

    I think it’s interesting the video linked here is entirely sourceless – no one investigated where it’s from, why it was made. This link is to a copy. Journalism is hard!



Comments are closed.