Good Reasons for Bad Feelings

by Randolph M. Nesse

  Genes that come from mothers get an advantage by keeping a fetus a bit smaller to conserve the mother’s resources for a future pregnancy with her same genes and to ensure safe childbirth. Genes from the paternal line get an advantage if they make a baby a bit bigger and use more of the mother’s stored calories, since her subsequent offspring may have a different father.37 The details get complex fast, but Crespi has assembled evidence that excess dominance of paternal alleles may increase the risk of autism, while excess unopposed activity of maternal alleles may increase the risk of schizophrenia.38,39,40,41

  This predicts that babies born a bit larger than average will be more likely to get autism because of expression of genes from the father, while those born smaller will be more vulnerable to schizophrenia. Remarkably, the prediction is supported by a study of the medical records of 5 million Danes.42 I am unsure if this hypothesis will turn out to be correct, but it is a fine example of creative thinking and research inspired by an evolutionary perspective.

  Boys are many times more vulnerable to autism than girls are.43 Even for rats, females are better at social tasks and males are better at systematizing. This suggested to Simon Baron-Cohen and colleagues that autism is a product of an extremely male brain.44 Does the sex difference in rates of autism result from testosterone, genomic imprinting, the effects of genes on the X and Y chromosomes, or what? The answer to this question may provide the key to understanding autism.

  The huge fitness costs of these diseases have inspired suggestions that their symptoms, or the alleles that cause them, must offer selective advantages.45 Such ideas have spurred creative flights. One idea is that schizophrenics become shamans or charismatic leaders, and the resulting status gets them extra matings.46,47 This is inconsistent with data showing that people with these diseases have reduced numbers of children, although a recent report suggests that people with creative traits related to schizophrenia may have increased mating opportunities.48

  A more plausible possibility is that the same genetic tendencies that make a disorder likely also give other advantages. The association of creativity and intelligence with bipolar disorder has spurred enormous interest and many studies.49,50 My exceptionally creative academic friends seem especially likely to have children with major mental disorders, and the relatives of my patients with severe disorders also seem likely to be exceptionally creative. But this could be an illusion. Practice in a university setting gives increased exposure to creative people, and extraordinarily successful patients and their relatives are easier to remember because they fit the pattern. Also, people with dire disorders may choose creative occupations because they have a hard time getting and keeping other kinds of jobs. Or perhaps people with special abilities are especially likely to get social validation that encourages grand pursuits that escalate to mania. Some traits associated with bipolar disorder may give advantages, but I am unconvinced that creativity is a major influence; it may instead be a sometimes fortunate side effect of mood dysregulation and its complications.

  Several new studies support the idea that associated benefits preserve alleles that increase vulnerability. A study of twins found that the likelihood of developing bipolar disorder is associated with higher-than-average levels of sociality and verbal skills.51 A just-published article by Yale geneticists Renato Polimanti and Joel Gelernter found that alleles that increase the risk of autism spectrum disorder have been subject to positive selection, presumably because they give cognitive benefits.52 Another study found that the amount of a protein made by genes related to schizophrenia was related to verbal learning ability.53 As for why so many alleles with tiny effects can’t be added together to create big effects, it appears that brain development is influenced by their complex interactions.54 However, scores of suggestions about possible benefits from traits or genes associated with these diseases have been proposed without confirmation, so skepticism is warranted.

  New methods make it possible to estimate when genetic variations that influence vulnerability to schizophrenia first appeared. Most seem to have emerged sometime after our last common ancestor with chimpanzees, about 5 million years ago.55 A study of bits of DNA that enhance the expression of certain genes has found that the ones that influence brain development are evolving five times faster than others and that these variations also increase the risk of late-life diseases such as Alzheimer’s disease.56 This is a fine example of how alleles that cause disease late in life can be selected for because they give advantages earlier. Stephen Corbett, Stephen Stearns, and their colleagues argue that this phenomenon, called antagonistic pleiotropy, imposes much higher costs for organisms, such as humans, living in environments vastly different from those they evolved in.57

  If alleles that increase vulnerability to bipolar disorder really have offered a selective advantage across the course of human evolution, those alleles should have spread and become universal. Maybe they have. The psychiatrist Hagop Akiskal and colleagues have conducted a series of wonderful studies showing that full-fledged bipolar illness is just the tip of a spectrum of mood instability disorders.58,59 Mild versions of mood instability may be common because they increase reproductive success on average over the long run, despite their malign effects on health for some people. This could be because of productivity during bursts of manic energy or because such individuals get more sex partners.60 Mood disorders may offer another tragic example of how we were shaped for reproductive success at the cost of health.

  Another possibility is that the responsible genetic variations are not defects but genetic quirks, normal variations like those that cause eating disorders and substance abuse only in modern environments. The possibility that schizophrenia is more common in modern environments has been advanced several times in recent decades, but evidence to support it is limited.61,62 The prevailing view that schizophrenia rates are the same everywhere has been challenged by new studies showing slightly higher rates for immigrants and city dwellers.63,64,65 However, the evolutionary psychiatrist Jay Feierman tells me that he has seen many cases of clear-cut psychoses in his travels to cultures based on subsistence hunting and agriculture. More cross-cultural data would be useful, but these disorders are not diseases like eating disorders and substance abuse that are mainly products of modern environments.

  Infection offers another possible evolutionary explanation. Could dire disorders result from infections that influence brain development? The risk of schizophrenia is increased by infection during pregnancy with Toxoplasma gondii, a parasite associated with cats.66 Rates of schizophrenia are also increased in children whose mothers had influenza during the second trimester.67,68,69 The influence of such infections on brain development may help to explain some variations in rates across time and location, but such infections during pregnancy are rare, so their contribution to the total causation is small. They do, however, offer crucial evidence that the disruption of neural development due to diverse causes can cause similar syndromes.

  Many of the suggestions for why schizophrenia alleles persist consider the general idea that they were selected for in the process that shaped human cognition and language.70,71,72 This has long seemed plausible but untestable, but it is finding support in new genetic evidence for the effects of schizophrenia-associated alleles on cognition.73,74,75,76,77,78,79,80,81,82,83

  Minds Unbalanced on Fitness Cliffs

  The discussed ideas all help to explain the persistence of alleles that cause dreaded diseases, but I found myself still wondering why natural selection did not greatly reduce the risk of such devastating disorders. Each occurs with a frequency of about 1 percent. If they were 0.001 percent, that would be different, but 1 percent is relatively common. Linkage with helpful alleles before our progenitors left Africa seemed to offer a potential explanation,84 but the process of genetic recombination would have split such pairings long ago.85 I also found it mysterious that tiny effects from so many different genes could cause relatively consistent syn
dromes.

  After racking my brain on the problem for weeks, I finally found inspiration by rereading early studies by the British ornithologist David Lack.86,87 He wondered why birds didn’t lay more eggs in order to have more offspring and suspected that additional eggs would sometimes pay off but would sometimes result in fewer total surviving offspring. To test the idea, he moved eggs from some nests to others. As he suspected, adding one egg to a nest increased the number of fledglings somewhat on average, but beyond a certain number, added eggs decreased the total number of chicks fledged. His insight inspired me to wonder if a similar “cliff-edged fitness landscape” might explain vulnerability to schizophrenia.88

  Biologists use the metaphor of a “fitness landscape” to think about how variations in a trait influence Darwinian fitness. For instance, birds that have wings that are longer or shorter than average are less likely to survive a storm,89 so the fitness landscape for wing length is shaped like a hill, with a fitness peak in the middle at average lengths and smooth downward slopes on either side as fitness decreases for birds with shorter or longer wings. Long wings have both advantages and disadvantages; short wings have opposite advantages and disadvantages. The result is inevitable trade-offs, many of which are relevant to disease. Bernard Crespi has written profoundly about pairs of “diametric disorders” resulting from deviations to one side or the other.90,91

  The figure on the next page illustrates the standard model for genetic vulnerability to disease. Trade-offs are central. For example, risk-taking rabbits are at high risk of predation, but they have plenty of time to eat. Cautious rabbits are protected from predation, but they have too little time to eat. Rabbits with medium levels of cautiousness have the highest fitness, so selection shapes the population mean to the peak, where fitness for genes and individuals coincides with maximum health. Mutations spread out the distribution, resulting in some individuals having trait values far from the mean and lower fitness. Stabilizing selection eliminates such mutations, narrowing the distribution.

  The Standard Model

  The solid line is the fitness at each level of cautiousness. The points that maximize fitness for an individual (I) and a gene (G) and health (H) all coincide at the peak of the fitness landscape. If the distribution of degrees of cautiousness is narrow (the tall dotted curve), most individuals will have high fitness and good health. If the distribution is wide (the dashed curve), some will be at high risk of predation and others will be at a high risk of starvation.

  However, fitness landscapes can be asymmetrical. Sometimes fitness increases as a trait is pushed in one direction, but going one step too far results in going off the cliff, just like the bird nest that had one egg too many. Racehorses are prone to breaking the cannon bone in their legs. Why didn’t natural selection make it thicker? It did; wild horses are unlikely to break their legs. However, breeding only the fastest horses made their leg bones longer and longer, thinner and thinner, and lighter and lighter. Successive generations of racehorses have become faster and faster but also more and more vulnerable to breaking a leg, something that now happens about once every thousand times a racehorse starts a race.92

  Because all racehorses have been selected for speed, horses that break their legs and their relatives will not be much faster than other horses. The same logic may explain why it has been hard to find advantages experienced by the relatives of people who have serious mental disorders. Strong selection for extreme mental capacities may have given us all minds like the legs of racehorses, fast but vulnerable to catastrophic failures. This model fits well with the idea that schizophrenia is intimately related to language and cognitive ability.93 It also fits well with the observation that schizophrenia may be intimately related to the human capacity for “theory of mind,” our ability to intuit other people’s motives and cognitive abilities in general.94,95

  How Cliff-Edged Fitness Functions Make Disease Inevitable

  Traits with asymmetrical fitness functions are stabilized not at the level that maximizes individual fitness (I) or at the level that maximizes health (H), but at the level that maximizes gene transmission (G), despite dire outcomes for a few individuals.

  An individual at point I will have the maximum number of offspring, but inevitable variations among those offspring (the dotted curve above point I) will leave many off the fitness cliff with a high vulnerability to disease. An individual at point G will have almost as many offspring but only a few of them will have values off the cliff. Natural selection will stabilize the trait at this point. An individual at point H will have healthy offspring but fewer, so overall fitness will be lower.

  A mathematical model I created shows that whenever fitness for a trait peaks at a cliff edge, selection will shape the average trait mean to be a bit below what would maximize an individual’s reproductive success but above the level that would maximize health. A few percent of a population with that average trait value will have values off the cliff that put them at high risk of disease.96

  Diseases resulting from cliff-edged fitness functions should be highly heritable, observed in a few percent of the population, and the risk will be influenced by the complex interactions of many normal alleles that all have about the same small influence on disease risk. This matches the data for many diseases.

  Many traits are subject to catastrophic failure. Babies with larger brains and heads have advantages, but in environments without obstetric surgery, just one centimeter too large is fatal for both mother and baby.97 High uric acid levels protect against aging, but just a bit too high causes gout when uric acid crystals precipitate in joints.98 Having more stem cells slows aging, but increased numbers make cancer more likely.99 Some aspect of neuronal transmission may have been pushed to a cliff edge,100 making brains vulnerable to epilepsy from many causes, including mutations, infection, tumors, injury, and drugs.

  Competition between hosts and pathogens are especially likely to create steep cliffs.101 The price of not being able to defend adequately against an infection is death. In order to ensure the ability to counter such threats, the immune system is shaped to a level of aggressiveness that sometimes attacks normal tissues, causing rheumatic fever, OCD, rheumatoid arthritis, multiple sclerosis, and other autoimmune disorders.102 This makes the finding that many alleles with effects on schizophrenia are involved in immune responses especially salient.103

  A trade-off with immune benefits may also be involved in Alzheimer’s disease. Dying and dead neurons are usually surrounded by a protein called amyloid beta. Scientists have often assumed that this protein is a toxic by-product of metabolism. However, in a deep disappointment, drugs that prevent the synthesis of amyloid beta don’t slow progression of the disease.104 Furthermore, amyloid beta turns out to be a potent antimicrobial agent,105 and the system that prunes connections between neurons relies on a part of the immune system.106 Herpes virus remnants have just been discovered to be more common in the brains of people with Alzheimer’s disease.107 Our vulnerability to Alzheimer’s disease may be related to a trade-off involving immune system costs and several kinds of benefits.108

  These diseases all may result from natural selection stabilizing traits at the point close to a cliff edge that maximizes genetic fitness despite the dire outcomes for a few individuals. This idea is by no means widely accepted or even recognized, but I suggest it anyhow because it offers a potential explanation for why we have been unable to find specific genetic causes for specific mental disorders. In cliff-edge models, the problem results not from defective genes but from steep slopes on fitness landscapes that result from intrinsic trade-offs, like those that cause vapor lock. A two-dimensional landscape offers only a crude model; actual fitness landscapes are likely to be rugged in multiple dimensions. For some diseases vulnerability may turn out to result from the equivalent of sinkholes in a fitness landscape. If it accomplishes nothing else, however, considering cliff-edged fitness landscapes encourages looki
ng for traits and trade-offs that may be crucial for explaining dire diseases.

  Information Devices Fail in Special Ways

  Mental disorders are often thought to be fundamentally different from other medical disorders. Their vulnerability results from the same six evolutionary reasons that explain other medical disorders, but the brain is different from other organs in one important respect: it is a very general information-processing device. It receives information from many internal and external sources, uses chemical and electrical mechanisms to process the information, and produces output that adjusts physiology and guides behavior. Such systems fail in special ways.

  The analogy of brain as computer is easy to take too far. Engineers design computers with discrete components that serve specific functions. One part translates keystrokes into digital signals; another creates images on a screen; another allocates memory; another ensures that long strings of zeros and ones calculate what should be calculated. Airplanes and space shuttles have backup computers in case the primary one fails. We don’t have backup minds, but because our brains are organically complex integrated systems they carry on relatively well despite mutations and minor bits of damage.

  Software failures are somewhat different, but they provide a useful analogy for the different ways organic information systems can fail. Failure to receive adequate signals from the environment is serious, as you know if you have ever tried to log on to your computer when your keyboard is not working. Decreased sensory input in medical patients can cause delirium and hallucinations in similar ways. Software programs can reach a dead end. This is very much like the “thought blocking” experienced by some schizophrenics, who report that their train of thought sometimes just stops.