Patch 7.822: An Experimental Design Puzzle

The always-excellent @ctrlcreep eloquently states a puzzle in 140 characters that it took me lots of words to express previously. This puzzle is expressed at the outset of C. A. Soper’s article “Could some common mental diseases be evolved defences against suicide? – a theoretical enquiry.

Why do so few humans kill themselves?

This ultimate explanation for suicide, an unfortunate by-product of pain combined with human cognition, seems plausible as far as it goes, but it is evidently incomplete. So serious a design flaw would not have been left unchecked. In general the more severe the adaptive problem an organism faces, the greater the pressure that natural selection will exert in favour of adaptive solutions (Tooby & Cosmides, 1990a). It is hard to imagine a more severe adaptive problem for an animal than an unfettered capacity to extinguish itself. It is fair to assume that some restraints have indeed evolved because the posited co-authors of suicide – ‘pain and brain’ – can be presumed to constitute human universals (Brown, 1991): virtually all humans experience pain, and virtually all adults have the intellectual wherewithal to take their own lives. Close to 100% of humans could suicide, and yet ‘only’ about 1.4% do (WHO, 2014).

The proportion of mortality attributable to suicide has increased as mortality from disease (especially infectious disease) has subsided over this century. One of the most amazing things about suicide is that over the past 80 years or so in the United States, suicide rates have been extremely flat. The fact that suicide rates have not changed in response to changes in medical technology and other ways of life is astounding. Suicide rates among our ancestors are difficult to know, but suicide almost certainly accounted for a lower proportion of mortality in our various environments of evolutionary adaptedness. However, suicide mortality was probably still significant. And its rarity is fascinating, given the prevalence of human suffering.

Suicide can be maintained at such a small proportion of mortality, Soper says, because of various biological and cultural adaptations that keep suicide rates low. On the biological side, these adaptations include empathy and attachment to family, horror at bodily envelope violation, and pain from self-injury. It is difficult to kill a large animal, especially oneself. On the cultural side, adaptations include suicide prohibitions (in various forms), stories of eternal punishment, and insistence on the meaningfulness (and even sacredness) of life. Think about seppuku: Japanese culture allowed suicide in certain circumstances, but prescribed the most painful, horrifying method possible for its enactment.

Soper’s hypothesis – which I’ve been interested in since he first wrote about it in 2014 – is that many mental illnesses (such as depression and anxiety) are evolved defenses against the special risk that human cognition presents: the ability to conceptualize ending all of one’s problems by ending one’s brain.

Soper includes delusional disorders in his hypothesis; indeed, self-delusion appears to be a way of avoiding the horrors of reality and the mortality risk that they may present. But since the etiology of e.g. schizophrenia seems to be novel mutations, these disorders are probably terrible accidents, rather than protective adaptations. I will focus on depression here.

Depression is a kind of paralysis. Excessive sleeping is common (though inadequate sleeping may also be present). Whether sleeping is affected or not, many depressed people are unable to leave their beds or homes very much. They cannot form coherent plans and act on them. They ruminate instead of socializing or going out. A simple process with only a few steps is frequently beyond them (us). In considering “specifications” for a mechanism to protect against suicide, this kind of paralysis might work just well enough.

Depression is associated with a slightly increased risk of suicide (not nearly as high as the double-digit suicide rates seen in some other mental disorders, such as schizophrenia and certain personality disorders). Depression affects a huge proportion of humanity, and it is primarily a disease of the young. People of childbearing age are most at risk. Those 65 and older are at least risk for depression. However, suicide is the opposite: suicide risk increases linearly with age (especially for white men). Those in the oldest age groups are at the highest risk of suicide. Here we have an evolutionary story: depression protects the young from suicide, but the protections fade with senescence, as the fitness costs of suicide plummet to zero or are even negative (i.e., ceasing to sap resources from relatives).

I explained Soper’s hypothesis to a famous psychology professor recently, and his response was snickering. I understand that response, but I would rather understand why it is so clearly wrong, if it is.

But here is the puzzle: how can you design experiments or data analyses that might distinguish whether this hypothesis is true?

It seems fairly obvious to me that drug “abuse” (including alcohol) is frequently an attempted response to suffering of the kind that might cause suicide. On the musician Elliot Smith’s death:

Did Elliott Smith commit suicide? And if so, why? Many of Smith’s closest friends at the time of his death say yes, and suggest that his depression, alienation, self-loathing, and drug use were merely symptoms of an underlying trauma. To this inner circle, the fact that Smith died sober was no surprise, because as their testimonials suggest, Smith was not suffering from a drug problem — he was searching for a drug solution.

The many sources that claim that suicide is caused by mental illness or drug dependence rely on correlation that is just as suggestive of reverse causation. In Soper’s model, misery and the cognitive capacity to conceive of ending one’s life (“pain and brain”) cause both depression and suicide, and cause suicide less if they trigger depression. I think it’s obvious that misery also causes heavy drug use.

When phenomena travel together, how do we tease out causation? One experiment is treating depression without treating underlying suffering, as with modern antidepressant drugs such as SSRIs. Indeed, there appears to be a slight elevated risk of suicide in people taking some antidepressants, especially young people. But the effect is tiny. And some people do think antidepressants relieve suffering. If depression caused suicide, treating depression should reduce suicide rates, but of course it hasn’t. This could be due to the ineffectiveness of treatment, however.

Can we learn anything from group differences? Men commit suicide more than women (except in places like China), and women experience more depression and non-lethal attempts. Races differ in suicide rates; whites generally have the highest suicide rates, followed by Native Americans, and blacks and Hispanics have low rates (the black-white comparison appears to hold both the United States and South Africa). Evidence for prevalence of depression by race is mixed: one study found that acute, short-term depression (Major Depressive Disorder) was more prevalent in whites, but dysthymic disorder (long-term depression, lasting two years or more) was more prevalent among blacks and Hispanics.

But should we expect more depression in groups more affected by suicide mortality, or less? A highly suicidal, non-depressed group might suggest a lack of the depression “adaptation” in this group, compared to more depressed, less suicidal groups. Or a highly suicidal, depressed group might suggest the adaptation running at full speed against pain and brain. Groups may have had different needs for defenses against suicide in their environments of evolutionary adaptedness. I am not able to articulate a clear prediction about relative group prevalence of depression and suicide that would support or falsify this hypothesis.

And so, a puzzle: can we articulate any experiments or data analyses that would support or falsify this hypothesis? And what do the epistemic difficulties here say about the common belief that suicide is caused by mental illness?