David Dobbs’ Social Life of Genes was racing past 6000 words when it screeched to a halt with an exhortation from Steve Cole. The quote could have come right out of a positive psychology motivational talk:
“Your experiences today will influence the molecular composition of your body for the next two to three months,” he tells his audience, “or, perhaps, for the rest of your life. Plan your day accordingly.”
The message is that we had better be careful how we think, feel, and act because irreversible, life-altering change could result from the next move we make.
Of course, there is a potential self-contradiction or two here hiding in plain sight. If today is so important, so is tomorrow, which might be able to reverse anything that happens today. If genomic expression is so mutable, maybe it can be changed back. But that is not the point that Cole is going after. He wants to emphasize that “our social lives can change our gene expression with a rapidity, breadth, and depth previously overlooked.”
Readers who have been following Steve Cole’s work may recognize the connection to his recent article in Proceedings of the National Academy of Sciences (PNAS). His co-authors included Barbara Fredrickson, a psychologist professor, but also known as a rising positive psychology coach and author of the pop psychology advice book, Love 2.0.
The PNAS article had the immodest goal of providing “an objective approach to moral philosophy.” It tackled a question posed by ancient philosophers: whether people should just pursue happiness (hedonic well being) or go after meaning (eudaimonic well-being). The article claimed identification of “distinct gene regulatory programs” associated with pursuing these different pathways in the past week. Eudaimonic striving won, they concluded. So there, question answered.
Unfortunately, as I showed in a recent blog post, these claims are based on psychometric malpractice and voodoo statistics. These authors started with an off-the-shelf psychological measure. They ignored its scoring and factor analyses in picking items that they hoped would measure distinct strivings. They did not check how the items actually performed before mounting the study. They then popped their data into inappropriate multivariate analyses that were guaranteed to produce a particular pattern of results.
Measures of these supposedly distinct strivings were essentially tapping statistically indistinguishable characteristics. The authors’ not able to identify distinctive strivings, and so they were prevented from identifying distinctive profiles in genomic expression that go with them.
Dobbs presents dramatic, even if highly selective examples of changes in gene expression from studies of the birds and the bees and fish.
He then describes meeting with Steve Cole for sushi. I tried to imagine the frenetic waving of chopsticks and splattering wasabi and soy sauce in the rapid fire conversation. I conjured up scenes from the 70s cult classic My Dinner with Andre.
The three hour movie stars Andre Gregory and Wallace Shawn, and involves the two discussing aesthetics, the meaning of life, postmodernism, science versus the supernatural, and Halloween on Long Island. The film has been described as one which many wish to have seen, but few wish to see.
I actually sat through it twice, but fell asleep both times. If you want to pass for knowledgeable, but not have to see it, you can view the 9 minute clip here.
Steve Cole’s conversation with Dobbs is rambling, but much more interesting than My Dinner with Andre. Ultimately, though, the talk is oppressive in ways that movie is not.
The article falls into some familiar traps with repeated attempts to enlist what are portrayed as scientific findings to strengthen messages from positive psychology. Namely, all of us need to assume a heightened sense of personal responsibility for how our circumstances affect us and for our health. If we do, we can triumph over adversity.
If applicable anywhere, the message is much more appropriate for rich people. Their wealth affords them better circumstances and better chances for health and longevity. The message of positive psychology gives the socially advantaged the smug reassurance that it is not their circumstances, but what they make of their circumstances that accounts for their superiority.
David Dobbs and his sushi eating companion preach a reductionistic genomic expression ueber alles. They neglect the rich context of linked biopsychosocial signaling systems in which gene expression occurs. There are lots of complex, reciprocal feedback loops within which gene expression can be dampened or amplified, bypassed, or turned on or off by other processes.
Only by imposing artificial punctuation, can a sentence begin “Gene expression causes…” The endless, uninterrupted flow of the sequence is more accurately “…and then something prompts gene expression, the results of which can in turn be further amplified or dampened by…. and provisional outcomes can be temporarily or permanently undone by…”
Many readers come to the article with limited examples and a distorted understanding of genomic expression. They are inclined to focus on gene expression as producing blue eyes. Dobbs and Cole want to shock and confront them with evidence that maybe social experiences can change blue eyes into brown. Dobbs sets the reader with a description of an altered genomic expression in the African cichlid Astatotilapia burtoni fish:
The fish underwent massive surges in gene expression that immediately blinged up his pewter coloring with lurid red and blue streaks and, in a matter of hours, caused him to grow some 20 percent. It was as if Jason Schwartzman, coming to work one day to learn the big office stud had quit, morphed into Arnold Schwarzenegger by close of business.
I haven’t bothered to check Google Scholar about African cichlids
and their displays of color, but I am quite sure that Jason Schwartzman won’t be confused with Arnold Schwarzenegger anytime soon.
My expertise is more in social and behavioral sciences, epidemiology, and methodology. But I have enough tools to identify in the Dobbs-Cole conversation lots of hyperbole, good science being given distorted interpretation, and mediocre or small-scale, preliminary science being given exaggerated importance.
Let’s jump to what Coles says through Dobbs in a blazing blue-lettered sidebar:
“If you actually measure stress, using our best available instruments, they can’t hold a candle to social isolation. Social isolation is the best-established, most robust social or psychological risk factor for disease out there. Nothing can compete.”
This is a weird boast from a social epidemiological perspective. It borders on nonsense when it is put to use in the many correlates of social isolation to a matter of specific genomic or immunological mechanisms.
A nerdy epidemiologist might have yawned: “Sure, so your measurement of social isolation is more confounded than my measure of stress. Too bad. And we cannot really neatly separate stress from social isolation neat and cleanly, anyway. So what?”
Community surveys of correlates of social isolation are observational studies, not experimental manipulations, and so any causal inference always involves some speculation. Social isolation is correlated with a lot of background characteristics that have identifiable associations with health themselves. In turn, health problems result in social isolation.
Being socially isolated is associated with stress and deprivation, poverty, unemployment, recent losses such as divorce or bereavement, poorer adherence, restrictions in function due to health, delay in help seeking for acute and chronic health problems, etc. etc. Being sick is associated with socially isolation and can cause it.
Technically speaking, problems await anyone trying to make sweeping statements about the power of social isolation versus stress: reverse causality, confounding variables; unidentified, poorly measured, or incompletely controlled confounds; residual confounding, spurious associations, and a host of mechanisms relating social isolation, as well as competing explanatory variables like stress to health outcomes.
Bottom line: causal inference in epidemiological studies is complex and depends on imperfect efforts to rule out a variety of alternative interpretations.. You just cannot validly make sweeping statements like Cole does here and expect to remain credible.
A 1988 Science article by Jim House, Kevin Landis, and Deb Umberson has been cited over 4000 times as showing that the evidence linking low social support and social isolation was as strong as the evidence that had first linked smoking to lung cancer. Yet, just a few years earlier, Jim House had been unable to demonstrate the importance of social relationships for mortality among women in a large prospective community study. Epidemiological studies can be frustrating that way.
One of the more reasonable things that Cole admits to Dobbs: “Epidemiology won’t exactly lie to you. But it’s hard to get it to tell you the whole story.”
Cole led into his declaration by citing an earlier study in which he says showed that closeted HIV men succumbed to the virus much more readily. He claimed to have found
that HIV-positive men who were lonely also got sicker sooner, regardless of whether they were closeted. Then he showed that closeted men without HIV got cancer and various infectious diseases at higher rates than openly gay men did.
Something about feeling stressed or alone was gumming up the immune system—sometimes fatally.
Note the shift from social isolation to stress and loneliness. I imagine his voice suddenly booming his next statement, startling anyone else eating along the sushi bar into drop their chopsticks.
“You’re besieged by a virus that’s going to kill you,” says Cole, “but the fact that you’re socially stressed and isolated seems to shut down your viral defenses. What’s going on there?”
Robert Gross, an infectious disease expert at University of Pennsylvania enlisted me as co-investigator in a clinical trial in which we would utilize psychological interventions to reduce viral load in HIV/AIDS patients. I would have had to flee the meeting in embarrassment if I have proposed harnessing any of the mechanisms claimed by Cole. Whatever their theoretical interest, they do not have demonstrated clinical significance.
We instead developed a problem-solving intervention to improve adherence to HIV-medication. Comparing the intervention to routine care in a randomized trial, we found that we had succeeded in actually reducing viral loads.
The nice thing about adherence was that we could establish its importance in observational and quas- experimental studies, quantify the likely effect size, and then manipulate adherence in a clinical trial. Not as sexy as genomic expression or psychoneuroimmunological mechanisms, perhaps, but at least testable and in a way that our hypothesis could be disconfirmed.
At another point, Dobbs describes Cole’s work with John Cacioppo in which they picked from 153 healthy people
the eight most socially secure people and the six loneliest and drew blood samples from them. (The socially insecure half-dozen were lonely indeed; they reported having felt distant from others for the previous four years.) Then Cole extracted genetic material from the blood’s leukocytes (a key immune-system player) and looked at what their DNA was up to.
He found a broad, weird, strongly patterned gene-expression response that would become mighty familiar over the next few years. Of roughly 22,000 genes in the human genome, the lonely and not-lonely groups showed sharply different gene-expression responses in 209.
This passage conjures up Andrew Kaufman’s My Breakfast with Bassie, a spoof of My Dinner with Andre. The film is a lot more watchable, because it is not as pompous and pretentious. In the film, Kaufman who in real life mostly wrestled women, goes to breakfast with Freddie Blassie, who is a professional wrestler and now self-proclaimed King of Men.
Andy and Freddie say outrageous things, which eventually draw in some other people who challenge them. If I were at the sushi bar, I might have felt compelled to intervene here with Cole:
What?!! You did a Genome-Wide Association Study (GWAS) with 14 subjects and claim that you isolated a pattern involving 209/22,000 genes? Ever think about having capitalized on chance as an explanation? Don’t you know that a GWAS requires hundreds more participants in order to obtain a robust, replicable findings? And that most GWAS findings don’t replicate? Maybe you got away with the study back when you published it, but surely would not expect to publish it today! Why are you still bragging about it?
I’m glad I wasn’t there, because then he went on to discuss a study he did with my friends Greg Miller and Edith Chen, who are now at Northwestern University. I would have disputed his interpretation of their work. Cole reports selecting 16 poor and 15 well-off children who had asthma from a larger sample and running genomic expression analysis. Not surprisingly, the poorer children had poorer functioning immune systems.
Forget the problems of trying to compare such small samples and the need to ignore the many other ways in which these groups of children differ. Cole is on his way to making a stream of points:
The poorer kids perceived more threat; the well-off perceived less…..
To an extent that immunologists and psychologists rarely appreciate, we are architects of our own experience. Your subjective experience carries more power than your objective situation. If you feel like you’re alone even when you’re in a room filled with the people closest to you, you’re going to have problems. If you feel like you’re well supported even though there’s nobody else in sight; if you carry relationships in your head; if you come at the world with a sense that people care about you, that you’re valuable, that you’re okay; then your body is going to act as if you’re okay—even if you’re wrong about all that.”
Cole was channeling John Milton: “The mind is its own place, and in itself can make a heaven of hell, a hell of heaven.”
If I go for sushi, I had better make sure that David Dobbs and Steve Cole are not there. Otherwise, I would risk having to hear them talking this insensitive, classist positive psychology trash. I might feel compelled to ruin the dining experience for all three of us by calling them out about their trying to pass off their ideology as science.