Unlike Coyne, however, I do not believe any of these criticisms come close to undermining the adaptive rumination hypothesis. At most they suggest avenues for the future development and testing of an hypothesis that I strongly suspect will turn out to be true.
(Here are links to part 1 and part 2 of Coyne's critique, and here are links to my first and second posts on the critique.)
Does depression help solve complex problems?Coyne dismisses the evidence that depressive rumination actually helps solve tough life problems, concluding
the evidence is very, very weak. The authors cite only one “pertinent mood induction experiment”, which may not be relevant at all.And:
as a strong piece of evidence for depression helping solve thorny social dilemmas, it’s weak. It’s certainly not enough evidence to tell people to go off their meds!Coyne's first error, as I noted in my previous post, is failing to mention the overwhelming evidence that depressed folks ruminate a lot, they ruminate about their problems, and they themselves say they are trying to understand and solve their problems (Nolen-Hoeksema et al. 2008 and references therein). The question is whether this rumination actually helps solve the problem that triggered the depression, or whether it impedes solving the problem.
The current view is that, when depressed, thinking intensely about one's problems impedes solving them. But A&T point out, as I also noted earlier, that all studies supporting the current view involved depressed people attempting to solve tasks unrelated to their life problems. It's not too surprising that performance on cognitive puzzles, such as random number generation, will be impaired if one is ruminating about the death of a close relative or abandonment by a spouse.
Because no one has yet explored whether depressive rumination helps solve the problem(s) that actually triggered the depression, A&T have been forced to build their case on common sense; the weak empirical case for the current view; evidence that depression is conducive to rumination and analysis; and what they themselves readily admit (e.g., p. 633) is the limited direct evidence supporting the ARH.
Coyne's second, and more egregious error, is to ignore the lengthy discussion in A&T (pp. 628-633) of the evidence that depression promotes an analytical style of reasoning. As A&T summarize:
Although A&T, Coyne, and I all agree that currently there is little direct evidence that depression actually helps solve the triggering problem (because the appropriate studies simply have not yet been done), A&T have assembled considerable empirical evidence that depression promotes intense, focused analysis. The most obvious explanation for these results is that depressed people have very important yet complex problems to solve.The analytical rumination hypothesis predicts that depression coordinates changes in body systems to promote an analytical problem-solving approach that is concordant with the analytical, difficult problems that trigger it. Consistent with this, studies of preexisting and experimentally induced mood have consistently shown that depressed affect promotes an analytical processing style in which information is processed more carefully, thoroughly, and methodically and is processed in smaller chunks (Ambady & Gray, 2002; Edwards & Weary, 1993; Gasper, 2004; Gasper & Clore, 2002; Schwarz, 1990; Schwarz & Bless, 1991; Yost & Weary, 1996).
Cost-benefit analysesIn his first post, Coyne implies that A&T do not consider the costs of depression:
Against [the proposed benefits of depression], however, must be set the maladaptive consequences of depression....[and] we do know that there is one hugely maladaptive consequence of depression: suicide...It’s telling that in the entire 34-page article by Andrews and Thomson, the word “suicide” is not mentioned once.In his second, Coyne makes this explicit:
The authors give no cost-benefit analysis for depression, despite the fact that the costs are certainly severe. At present they include an appreciable frequency of suicide. And, as one alert reader pointed out, a loss of appetite or desire for sex would have been seriously maladaptive in our savanna-dwelling ancestors. Imagine an ancestral H. erectus, curled up on the floor of his cave, ruminating obsessively because he suspects his mate of infidelity. He doesn’t sleep; he doesn’t eat; he doesn’t have sex or go hunting with his mates. Does this really give him an adaptive advantage? This scenario is a bit facetious, but the point is serious.In fact, A&T do mention "suicidal behaviors," on p. 644, and they do discuss the trade-off between costs and benefits multiple times (I count 14 uses of the term "trade-off"), e.g.,:
Stress response mechanisms often require trade-offs between fitness-related goals to produce an effective response to the triggering problem. Like fever, then, the impairments associated with depression are usually the outcome of adaptive trade-offs rather than disorder. (p. 623)Nevertheless, A&T leave the distinct impression that most depression symptoms, even quite costly ones, play a functional role in rumination. I agree that depression has substantial costs that are not well-explained by the ARH, of which suicidality is only one. Another is the loss of interest in virtually all activities that, along with depressed affect, is the core defining feature of major depression.
The evolutionary benefits of depressive rumination must have been great enough to compensate for the substantial costs, many of which we have discussed. (p. 643)
A&T persuade me that many depressive symptoms, including anhedonia, function, at least in part, to promote an analytical mode of reasoning and to reduce distraction from the analysis required to solve major life problems. Yet they haven't convinced me that reducing distraction is enough of a benefit to explain loss of interest in virtually all activities. Individuals with major depression often don't groom, bathe, and sometimes don't even use the toilet. Depressed parents reduce care of their own offspring, including highly vulnerable newborns. Over the weeks and months that major depression can last, a failure to groom or bathe would probably have exposed a depressed person to a significantly increased risk of infection, and reduced investment in childcare, as occurs in postpartum depression, would have detrimental effects on child survival, growth, and development. Each of these would plausibly have had a serious negative effect on fitness that would not easily be offset by more uninterrupted time to think.
Loss of appetite would have been another serious cost of depression, a fact mentioned by a reader of Coyne's blog. On loss of appetite A&T again speculate that eating distracts from thinking, and that's why depression is associated with a loss of appetite:
Depression may also decrease disruption of analysis through psychomotor changes, including a preference for solitude, fatigue, changes in appetite, and changes in sleep or activity patterns....[N]eurobiological evidence indicates that oral or buccal activity interferes with the processing of stimuli (Jacobs & Fornal, 1999). The reduced appetite often seen in depression may sustain processing by reducing oral and buccal activity. (p. 632)I find this pretty unconvincing, at least for depressions that last more than a day or two. Cognitive processing, the very heart of the ARH, takes a lot of energy. The brain is one of the biggest energy consumers in body, consuming about 20% of calorie intake though comprising only about 2% of body mass. Among hunter-gatherers, hunting success rates are often pretty low (see, e.g., Kelly 1995, p. 103), which means that human ancestors should have eaten whenever food was available, especially if they were in the midst of analyzing a complex life problem that required sustained, intensive cognitive functioning.
If depression is an adaptation then plausible explanations must be found for its costliest symptoms: loss of appetite, loss of interest in virtually all activities, including grooming, bathing and care of one's own children, and suicidality.
In a series of papers my colleagues Paul Watson, Peter Hammerstein, and I have proposed that these extremely costly features of depression function to deliberately hurt oneself as a means to (a) credibly signal need (a la costly signaling theory) and (b) threaten and hurt interdependent others to force concessions from them (see here, here, here, and here).
This work dovetails nicely with the ARH because it also sees depression as a functional response to a complex life problem. The difference is that we argue such problems can often be solved only if other people change their behavior. In many cases, other people would only change their own behavior if they were convinced that the complex problem was genuine, which would require a credible signal of need. Moreover, others would often only change their behavior if they were compelled to do so. We posit that the costly symptoms of depression serve both functions.
Andrews and Thomson cite this body of work, and, if Coyne had bothered to dig a bit deeper, he would have found that Andrews is a co-author on one of these papers (this one). So, elsewhere, Andrews (and Thomson) endorse other evolved functions for some of depression's most costly symptoms, and I think the current paper would have been strengthened if A&T had conceded that the ARH probably does not explain them.
Final thoughtsCritiques of the illness model of depression are especially urgent because antidepressants, some of the most widely used medications (and also some of the most profitable), have profound, and largely unknown, effects on the brain. These compounds have certainly not been shown to correct a brain "chemical imbalance."
We do know, however, that fluoxotine (Prozac) and imiprimine (a tricyclic antidepressant) interfere with adaptive stress responses in at least one species. Guinea pigs pups, who possess central 5-HT1D receptors similar to humans, cry when they are separated from their mother, except when, in studies screening novel compounds for antidepressant effects, they're given doses of fluoxetine, imipramine and other known antidepressants. These antidepressants suppress the guinea pig pup separation cries, interfering with a clearly adaptive stress response. As hard as it is to believe, compounds intended for human use as antidepressants are screened based on their ability to interfere with stress responses in other animals.* It was wise, not reckless, for A&T to raise the possibility that antidepressants might also interfere with a functional stress response in humans.
In summary, A&T use a standard method in evolutionary biology to argue that depressive rumination might be functional. They base this hypothesis on the substantial evidence (though not proof) that severe negative life events, in part, cause major depression; the overwhelming (and undisputed) evidence that depressed individuals ruminate a lot about their problems, trying to solve them; evidence that depressive symptoms, at least in part, are conducive to rumination and analysis; and the admittedly limited evidence that depressive rumination helps solve the problem that originally triggered the depression.
Combine this with the work of Watson, Hammerstein and myself, and the strategy for treating depression becomes straightforward, though rarely simple: if most depressions are caused by genuine life problems then "treating" depression will involve solving these problems -- sometimes this will require changes on the part of the depressed person as a result of analytical rumination, sometimes this will require changes on the part of others as a consequence of credible signals of need and arm-twisting, and often it will require both. There's nothing at all kooky about this hypothesis and nothing reckless or unwise about its clinical implications. As for its truth, only future empirical research can establish that.
*Unlike the guinea pig pup vocalization test paradigm, it is difficult to tell whether the stress responses exhibited in other animal models of depression, such as the forced swim test with rats or mice, are functional.