Book: The Agile Gene

Previous: 3. A Convenient Jingle
Next: 5. Genes in the Fourth Dimension

The word “cause” is an altar to an unknown god.

William James

During most of the twentieth century “determinism” was a term of abuse, and genetic determinism was the worst kind of term. Genes were portrayed as implacable dragons of fate, whose plots against the damsel of free will were foiled only by the noble knight of nurture. This view reached its zenith in the 1950s, in the aftermath of the Nazi atrocities, but in some corners of philosophical inquiry it took hold much earlier. In psychiatry the fashion was turning against biological explanations around 1900 at exactly the time that Galton was winning the argu-ment for inheritance in human behavior more generally. In view of what happened later, it is ironic that this turn to nurture was happening first in the German-speaking world.

The central figure in the early history of psychiatry, before Sigmund Freud, was Emil Kraepelin. Kraepelin was born in 1856; he trained as a psychiatrist in Munich in the late 1870s, but he did not enjoy the experience. He had bad eyesight, and he disliked peering at slices of dead brain under a microscope. At the time psychiatry, a German speciality, was founded on the notion that the causes of mental illness would be discovered in the brain. If mind was the product of brain, then it followed that disorders of the mind could be traced to malfunctions of parts of the brain just as heart disease was caused by faulty parts of the heart. Psychiatrists were to become like heart surgeons, diagnosing and curing physical faults.

Kraepelin turned such reasoning on its head. After a period of academic migration, in 1890 he settled in Heidelberg and pioneered a new means of classifying mental patients not on the basis of their current symptoms, let alone the appearance of their brains, but on the basis of their personal histories. He collected records on separate cards for separate patients, so that he could see the individual’s history. Different mental illnesses, he argued, had characteristically different progressions. It was only by collecting information on each patient over a long period of time that you could begin to distinguish the separate features of each disease. Diagnosis was the child, not the father, of prognosis.

At the time, psychiatrists were seeing an increasing number of patients with a particular affliction. They were young, mostly in their twenties, and they suffered from delusions, hallucinations, emotional indifference, and social insensitivity. Kraepelin was the first to delineate this apparently new illness, calling it dementia praecox, or precocious madness. It is now known by an even less helpful name coined in 1908 by Kraepelin’s follower Eugen Bleuler—“schizophrenia.” There is much argument today about whether schizophrenia had indeed suddenly become more frequent or was just being noticed as mentally ill people for the first time emerged from the family and entered institutions. The balance of evidence suggests that despite such bias, there was a real increase in mental illness during the course of the nineteenth century and that schizophrenia in particular has been a rare disease before the middle of the century.

Schizophrenia takes many forms and varies in severity, but nonetheless the disease has remarkably consistent themes. Schizophrenics experience their thoughts as loud. In the old days, this was called hearing voices, but today it usually takes the form of believing, for instance, that the CIA has implanted a device inside one’s head. Schizophrenics also imagine that others can read their minds, and they are apt to personalize every event, so that they think a television news broadcaster is sending them secret messages. Paranoid schizophrenics develop baroque conspiracy theories and as a result are likely to refuse treatment. Given how many ways the brain can go wrong, such a consistent pattern suggests that schizophrenia is a single disease, not a collection of similar symptoms.

Kraepelin distinguished dementia praecox from a different syndrome, characterized by mood swings between mania and depression, which he called manic depression; nowadays it is called bipolar disorder. What was characteristic about each illness was its course and outcome, not its current manifestation. Still less could these illnesses be distinguished by visible differences in the brain. Kraepelin was saying that psychiatry should abandon anatomy and be agnostic about causes.

As long as we are unable clinically to group illnesses on the basis of cause, and to separate dissimilar causes, our views about etiology will necessarily remain unclear and contradictory.

But what is a cause? The causes of human experience include genes, accidents, infections, birth order, teachers, parents, circumstance, opportunity, and chance, to name just the most obvious. Sometimes one cause looms large, but not always. When you catch a cold the chief cause is a virus, but when you catch pneumonia the bacterium is only an opportunist—your immune system usually needs to have been run down first by starvation, hypothermia, or stress. Is that the “true” cause? Likewise, “genetic” diseases such as Huntington’s chorea are caused precisely and simply by a mutation in one gene; environmental factors have almost no influence on the outcome. But phenylketonuria (PKU), a form of mental retardation caused by an inability to digest phenylalanine, could be said to be caused by the mutation, or by phenylalanine in the diet—it can be seen as either nature or nurture, depending on your bias. How much more complex is the pattern when many different genes and many different environmental factors are almost certainly involved, as is probably the case with schizophrenia.

Therefore, in this chapter, by investigating the cause of schizophrenia, I hope to throw the whole notion of “cause” into confusion. This is partly because the cause of schizophrenia is still very much an open question, with many rival explanations covering all possibilities. You can still plausibly say that genes, viruses, diets, or accidents are the first cause of psychosis. But the confusion goes deeper than that, for the closer science gets to understanding schizophrenia—and it is very close—the more it is blurring the distinction between cause and symptom. Environmental and genetic influences seem to work together, to require each other, until it is impossible to say which is cause and which is effect. The dichotomy of nature and nurture must first confront the dichotomy of cause and effect.


The first witness I call to explain the cause of schizophrenia is the psychoanalyst. For much of the middle part of the twentieth century psychoanalysts dominated the subject. Kraepelin’s agnosticism about the causes of psychosis, which transfixed psychiatry at the turn of the twentieth century, left a vacuum which the Freudians were destined to fill. By apparently dismissing biological explanations of mental illness, and stressing life history, Kraepelin had opened the way for psychoanalysis, with its emphasis on childhood events as a cause of later neurosis and psychosis.

The extraordinary spread of psychoanalysis between 1920 and 1970 owed more to marketing than to therapeutic triumphs. By talking to patients about their childhood, analysts offered humanity and sympathy that had not been available before. This made them popular when the alternatives were a deep barbiturate sleep, insulin coma, lobotomy, and electroshock convulsions: all unpleasant, addictive, or dangerous. By emphasizing the unconscious and the repression of memories from childhood, psychoanalysts also gave psychiatry a ticket out of the asylum. Indeed, psychoanalysis could now offer its services to those who were not so much ill as unhappy, and who would pay well for the chance to recount their life story while lying on a couch. In the United States, thriving and lucrative private practice was the driving force by which psychoanalysts gradually took over the profession of psychiatry and made it their own. By the 1950s, even the training of psychiatrists was dominated by psychoanalysis. The key to each individual’s psychological problems lay in his own individual history, and specifically in a social or “psychogenic” cause.

The “talking treatment” was a great improvement on the contemporary alternatives. But, as is so often the case, psychoanalysis went too far and began to claim that other explanations were not only unnecessary but wrong—morally as well as factually. Biological explanations of mental illness became heresy. Like all effective religions, psychoanalysis ingeniously redefined skepticism as further evidence of the need for its services. If a doctor prescribed a sedative or cast doubt on a psychoanalytic story, he was merely expressing his own neurosis.

At first Freudians avoided severe psychosis, concentrating instead on neurosis. Sigmund Freud himself was wary of treating psychotic patients, believing them to be beyond his methods, though he did hazard a wild guess that paranoid schizophrenia was the result of suppressed homosexual impulses. But as the confidence and power of analysts grew, especially in the United States, the temptation to tackle psychosis was irresistible. In 1935, a refugee analyst from Germany, Frieda Fromm-Reichmann, arrived at Chestnut Lodge in Rockville, Maryland, an institution already devoted to Freudian treatment. She quickly developed a new theory of schizophrenia: that it was caused by the patient’s mother. In 1948 she wrote:

The schizophrenic is painfully distrustful and resentful of other people due to the severe early warp and rejection he encountered in important people of his infancy and childhood, as a rule, mainly in a schizophrenogenic mother.

Soon after this, a self-styled heir to Freud, Bruno Bettelheim, rose to fame with a similar diagnosis for autism: that it was caused by an indifferent “refrigerator mother,” whose coldness toward her son (boys are far more likely than girls to be autistic) destroyed his ability to acquire social skills. Bettelheim had been incarcerated by the Nazis in Dachau and Buchenwald, but he managed to bribe his way out of the worst parts of the camps and somehow arranged his own release in 1939, in circumstances that remain mysterious. He emigrated to Chicago, where he founded a home for emotionally disturbed children. His enormous reputation did not long survive his suicide in 1990. Twin studies have utterly demolished the “refrigerator mother” theory, which spread guilt and shame among a generation of parents: the heritability of autism is 90 percent. An identical twin with autism has an autistic co-twin in 65 percent of cases; the concordance for fraternal twins is 0 percent.

Then it was the turn of homosexuals. This time the blame fell on the emotional stiffness of the father or the dominating personality of the mother. Some Freudians still cling to such theories. A recent book asserted:

The father [of a gay man] is rejecting or withdrawn or weak or absent—emotionally, literally, or a combination of these—and the marital relationship is disharmonious. Gay men tend to have had negative relationships with their fathers, half of them (compared with a quarter of heterosexuals) feeling anger, resentment and fear towards fathers whom they deem cold, hostile, detached or submissive.

All of which is probably true. It would be a miracle if most straight fathers did not have a “negative relationship” with gay sons. But which came first? All but the most extreme Freudians have long since stopped assuming that the relationship causes the homosexuality, rather than vice versa. (The correlation tells you nothing about causality, let alone its direction.) The same is true of the parental theories of schizophrenia and autism. Mothers of autistic children, like fathers of homosexual boys, withdraw in frustration at the child’s behavior. Mothers of “schizotypal” children—that is, children with a mild version of the disorder—may indeed react badly to the child’s developing psychosis. Consequence had been confused with cause.

For the parents of schizophrenic young people—parents who were already under terrible stress—Freudian culpability was an additional blow. The pain it was to cause to a generation of parents would have been more bearable if there had been any evidence to support it. But it was soon obvious to any neutral observer that Freudian treatment was failing to cure schizophrenia. Indeed, by the 1970s some psychiatrists were brave enough to admit that psychoanalysis actually seemed to make the symptoms worse: “The outcome for patients who received only psychotherapy was significantly worse than the outcome in the no-treatment control group,” said one, bleakly. By then psychoanalysis had been used to treat tens of thousands of schizophrenics.

As often happened in the middle years of the century, the “evidence” was based on a broad assumption—that nurture, not nature, explained most of the resemblance between parent and child. With regard to schizophrenia, had the analysts not ignored the biologists, they would have known that such an assumption was unwarranted—because of studies of twins.

In the 1920s and 1930s a Jewish immigrant from Russia, Aaron Rosanoff, collected data on twins in California and used them to test the heritability of mental illness. Out of more than a thousand pairs of twins in which one twin had a mental illness, he identified 142 schizophrenics. In 68 percent of the identical twins, the other twin also developed schizophrenia, whereas this was true of only 15 percent of the fraternal twins. He found a similar difference in manic-depressive twins. Yet because genes were unfashionable in psychiatry, Rosanoff was ignored. According to the historian Edward Shorter:

Rosanoff’s twin studies arguably represent the major American contribution to international psychiatric literature in the years between the two world wars, yet the official histories of American psychiatry, dominated by psychoanalytically oriented writers, pass over his work in virtual silence.

Franz Kallmann, who had emigrated from Germany in 1935, did a similar study of 691 twin schizophrenics in New York and got an even stronger result (86 percent concordance for identicals, 15 percent for fraternals). He was howled down by the analysts at the World Congress of Psychiatry in 1950. Rosanoff and Kallmann, both Jewish, were even accused of Nazism for using twin studies at all. The maternal theory of schizophrenia was shielded from uncomfortable facts for two more decades.

The current consensus is that “psychosocial factors” have only a tiny effect if they have any effect at all. In one Finnish study of adoptees, it was evident that the offspring of schizophrenics were slightly more likely to show thought disorder if their adoptive mothers were also showing what was euphemistically called “communication deviance.” But there was no such effect for the offspring of unaffected biological parents. So if there is a “schizophrenogenic mother,” she can affect only those of her offspring with genetic susceptibility.


The second witness to be called believes that schizophrenia is caused by genes. This witness uses all the arguments of behavior genetics. Schizophrenia plainly runs in families. Having a first cousin with schizophrenia doubles your own risk from 1 percent to 2 percent. Having a half brother or an aunt with schizophrenia triples it again to 6 percent. Having a full sibling with the disorder puts you at 9 percent risk. Having a nonidentical twin with the disorder raises the risk to 16 percent. Having two parents with the disorder puts you at a 40 percent risk. And having a schizophrenic identical twin is the highest risk factor known for the disease: you then have roughly a 50 percent probability of also being schizophrenic yourself. (This number is considerably lower than that in Rosanoff’s and Kallmann’s studies, because of more cautious diagnosis.)

But twins share nurture as well as nature. Beginning in the 1960s, Seymour Kety gradually demolished this objection with a growing study of Danish adoptees. (Denmark has an unrivalled state database on children put up for adoption.) He found that schizophrenia was 10 times as common in the biological relatives of diagnosed schizophrenics who had been adopted as children as it was in their adopting families. The reverse experiment—children adopted by schizophrenics—is, of course, very rare.

All these figures reveal two important things. First, they show that the heritability of schizophrenia in western society is high: roughly 80 percent, or about the same heritability as body weight and considerably more than personality. But second, they reveal that many genes are involved. Otherwise the figure for fraternal twins would be much closer to the figure for identical twins.

The witness for genes is therefore remarkably convincing. Few diseases show such clear evidence of inheritance, except those that are caused by single genes. It ought to be a trivial matter, in this era of the genome, to identify the genes for schizophrenia. In the 1980s, full of confidence, geneticists set out to discover them. Schizophrenia genes were among the most popular quarry in the world of gene hunting. By comparing the chromosomes of people who have the disease with those of their relatives who do not, geneticists sought to pin down those bits of the chromosomes that were consistently different and so get a rough idea of where to look for the actual genes. By 1988, using the well-recorded pedigrees of Icelandic people, one team had a strong result. This team had found a piece of chromosome 5 that was apparently abnormal in schizophrenics but not in their close relatives. About the same time a rival team stumbled on a similar phenomenon: schizophrenia apparently associated with having an extra piece of chromosome 5.

Congratulations rained upon the winners. Headlines proclaimed that the “schizophrenia gene” had been found. It was one of many behavior genes announced at about this time—genes for depression, alcoholism, and other psychiatric problems. The scientists themselves were careful to acknowledge in the small print that the result was preliminary, and that this was only one gene for schizophrenia, not the gene.

All the same, few were prepared for the disappointment that followed. Others tried without success to replicate the result. By the late 1990s, it was acknowledged that the association with chromosome 5 was a “false positive”—a mirage. This has been the pattern with the genes affecting complex diseases of the mind: again and again over the past decade, they have proved illusory. Again and again, the initial excitement has faded. Scientists have learned to be much more cautious when they announce associations between a disorder and a chunk of one chromosome. Nobody now takes such an announcement seriously until it has been replicated.

Schizophrenia has now been linked to markers on most of the human chromosomes. Only six human chromosomes (3, 7, 12, 17, 19, and 21) do not have putative links to schizophrenia. But few of the links prove durable, and every study seems to find a different link. There could be good reasons for this. It could be that different populations have different mutations. The more genes are involved in predisposing people to schizophrenia, the more likely it is that there will be different mutations producing similar effects. Imagine, for example, that the light goes out in your bedroom. It could be a failure of the lightbulb, the fuse in the plug, or the trip switch in the circuit; it might even be a power cut. Last time it was the trip switch; this time it proves to be the bulb. Failing to replicate an association between the trip switch and the fault, you indignantly reject it as a “false positive.” Bulbs, not trip switches, are the cause of bedroom darkness.

Yet it could easily be both. In the brain, a system of far greater complexity, there are not three or four possible things that can go wrong, but thousands. Genes switch other genes on, which switch yet more genes on, and so on till there are scores of genes involved in even the simplest pathway. Knocking out any one could disrupt the whole pathway. But you would not expect the same gene to be knocked out in every schizophrenic. The more genes can cause the pathway to fail, the harder it will be to replicate associations between disease and gene. So false positives are not necessarily discouraging or even wrong (though some may be statistical flukes). Nor is the failure of linkage studies proof, as some have averred, that the whole concept behind “neurogenetic determinism” is wrong. The role of genes in schizophrenia is proved by the twin studies and adoption studies, not by finding or failing to find particular genes. But it is fair to say that linkage studies, which worked so well for the single-gene diseases like Huntington’s chorea, have largely failed for psychoses.


Call the third witness. Some scientists, instead of trying to find what was different about the genes of schizophrenics, set out to understand what was different about their brain biochemistry. From that they would then deduce which genes control this biochemistry and so investigate the “candidate genes.” The first port of call was the dopamine receptor, dopamine being a “neurotransmitter,” or chemical relay system between certain neurons in the brain. One neuron releases dopamine into the synapse between cells (a synapse is a special narrow gap), and this causes the neighboring neuron to begin transmitting electrical signals.

The focus on dopamine was inevitable after 1955, the year when the drug chlorpromazine was first widely used on schizophrenics. To psychiatrists forced to choose between the brutality of a lobotomy and the uselessness of psychoanalysis, the drug was a godsend. It genuinely restored sanity. For the first time schizophrenics could leave the asylum and return to normal life. Only later would the awful side-effects of the drug emerge, and with them the problem of patients’ refusing to take their medication. Chlorpromazine induced in some patients a progressive degeneration of the control of movement similar to Parkinson’s disease.

But if the drug was not a cure, it seemed to offer a vital clue to the cause. Chlorpromazine and its successors were chemicals that blocked dopamine receptors and prevented them from having access to dopamine. Moreover, drugs that increase dopamine levels in the brain, such as amphetamines, provoke or exacerbate psychotic breaks. Third, brain imaging shows that the dopamine-fueled parts of the brain are most atypical in schizophrenics. Schizophrenia must be a disorder of neurotransmitters, and in particular dopamine.

There are five different kinds of dopamine receptors on the receiving neurons. Two of these (D2 and D3) have proved to be faulty in some schizophrenics, but again the result is disappointingly weak and hard to replicate. Moreover, the best antipsychotic drug prefers to block D4 receptors. To make matters worse, the D3 gene is on chromosome 3, which is one of the six chromosomes that have never been associated with schizophrenia in linkage studies.

The dopamine theory of schizophrenia gradually fell from fashion, not least after the discovery of mice with faulty dopamine signaling, which do not behave at all like schizophrenic people. Attention has recently focused on a different signaling system in the brain, the glutamate system. Schizophrenics seem to have too little activity at one kind of glutamate receptor (called the NMDA receptor) in the brain, just as they have too much dopamine. A third possibility is the serotonin signaling system. Here there has been better success: one of the candidate genes, 5HT2A, does seem to be faulty quite often in schizophrenics, and it does sit on one of the chromosomes (13) most implicated by linkage studies. But the effect is still disappointingly weak.

As of the year 2000 neither linkage studies nor searches for candidate genes had cracked the problem of which genes account for the heritability of schizophrenia. By then the Human Genome Project was nearing completion, so all the genes were at least present, laid out in the innards of computers, but how to find the few that matter? Pat Levitt and his colleagues in Pittsburgh sampled the prefrontal cortex of dead schizophrenics to find out which genes had been acting oddly. They carefully matched their subjects for sex, time since death, age, and brain acidity. Then they used microarrays to sample nearly 8,000 genes and identify the ones that seemed to be expressed differently in schizophrenics. The first was a group of genes involved in “presynaptic secretory functions.” In plain English this means the genes involved in producing chemical signals from neurons—signals like dopamine and glutamate. Two of these genes in particular were less active in the schizophrenics. Astonishingly, these genes are on chromosomes 3 and 17—two of the six chromosomes where linkage studies had not found an association with schizophrenia.

But another gene also emerged from this study, which does map closely onto one of the right chromosomal spots (on chromosome 1). It is a gene called RGS4, and it is active on the downstream side of the synapse—that is, on the receiving end of the chemical signals. Its activity was dramatically reduced in the 10 schizophrenics Levitt’s group studied. In animals, the activity of RGS4 is reduced by acute stress. Perhaps this explains a universal feature of schizophrenics, that stress tends to bring on their psychotic episodes. In the case of the brilliant Princeton mathematician John Nash, an arrest and the consequent loss of his job plus despair at failing to crack a problem in quantum mechanics seem to have tipped him over the edge. In Hamlet’s case, seeing his mother marry his father’s murderer might be thought enough stress to drive anybody mad. If such stress depresses the activity of RGS4, and if RGS4 is already low in people who are vulnerable, then stress could trigger the psychosis itself. But this would mean not that RGS4 is a cause of schizophrenia but only that its failure is a cause of worse symptoms in schizophrenics following stress—it is more like a symptom.

But curb even this much speculation with caution. The microarray technique is picking up genes that have changed their expression in reaction to the disease, as well as genes that induce the disease. It could be confusing consequence with cause. Degrees of gene expression are not necessarily inherited. This is a vital issue that will recur throughout the book. Genes do not just write the script; they also play the parts.

However, the evidence from microarrays does at least support the hints from drug treatments that schizophrenia is a disease of the synapse, though this evidence does little to distinguish cause from effect. Something is going wrong at the junctions between neurons in parts of the brain, especially the prefrontal cortex.


Summon the fourth witness, who believes that schizophrenia is caused by a virus. The heritability of schizophrenia is high, this witness points out, but it is not total. Twin studies and adoption studies leave plenty of room for environmental factors to play a part. Indeed, such studies do more than that. They emphasize the role of nurture. No matter how many genes the geneticists eventually find, nothing will reduce the effect of the environment. Remember that nature is not at the expense of nurture; there is room for both, and they work together. Perhaps all that we inherit is a susceptibility, just as some people inherit a susceptibility to hay fever—but the cause of hay fever is surely pollen.

The twin studies reveal that an identical twin brother or sister of a schizophrenic has only a 50–50 chance of getting schizophrenia. Since the two have identical genes, there must be something nongenetic that halves the probability. Moreover, suppose the two identical twins have married different spouses and had children. As before, one twin then gets schizophrenia but the other does not. What will happen to the children? Clearly the children of the affected twin are at fairly high risk of schizophrenia, but what about the children of the twin who remains unaffected? You might expect that having escaped the disease, the unaffected twin is less likely to pass it on to his children. Yet this is not so. The children inherit the same risk from an unaffected parent, which proves that having the predisposing genes is necessary, but not sufficient, to develop the disorder.

The search for the nongenetic factors in schizophrenia goes even farther back than the search for genes. However, it took a dramatic turn in 1988, the same year that the first genetic link was apparently found in Icelanders. This story, too, is Nordic, for while Robin Sherrington was testing chromosomes in Reykjavik, Sarnoff Mednick was poring over medical records in the Helsinki Mental Hospital. Mednick was trying to explain a well-known fact about schizophrenia: more schizophrenics are born in winter than in summer. This is true in both hemispheres, despite the six-month difference in the timing of the seasons. It is not a large effect, but it is undoubtedly there, and it refuses to go away, however the statistics are massaged.

Mednick’s hunch was that influenza epidemics tend to occur in winter. Perhaps there is something about flu that predisposes mothers to give birth to potential schizophrenics. So he examined hospital records in Helsinki to discover the effect of an influenza epidemic that had occurred in 1957. He found that those who had been in the middle three months of their own gestation during the epidemic were more likely to have schizophrenia than those who had been in the first or last trimester of gestation.

Mednick then read the obstetric records of women pregnant during the outbreak of 1957 who gave birth to future schizophrenics. He found that they were more likely to have had the flu during the second trimester of pregnancy, the middle three months, than to have had it before or after. In Denmark, meanwhile, a historical approach produced a supportive result: in those years between 1911 and 1950 when influenza had been rife, more schizophrenics had been born. And the riskiest date for the mother to catch the flu was in the sixth month, and especially the twenty-third week, of her pregnancy.

So was born the viral hypothesis of schizophrenia: that influenza infection in pregnancy, especially during the second trimester, can cause some kind of damage to the immature brain which has the effect many years later of predisposing the affected person to psychosis. Of course, not all those whose mothers get influenza will become schizophrenics. The effect is bound to depend on genes: some people are genetically vulnerable to the impact of the virus, or infectiously vulnerable to the impact of their genes, whichever way you prefer to look at it.

An intriguing hint that may support the influenza theory comes from the study of “monochorionic” twins. About two-thirds of identical twins are even more intimately connected than the rest. They not only come from the same fertilized egg but develop inside a single outer membrane or chorion within the womb and share the same placenta. (A few even develop within a single inner membrane and are “monoamniotic.”) The later the twinning event occurs, the more likely the twins are to be monochorionic. Since monochorionic twins are bathed in the same fluid during pregnancy, perhaps they encounter the same nongenetic influences. They even share blood through the common placenta. Perhaps they encounter the same viruses. It would be especially interesting to know, therefore, if monochorionic twins are more concordant for schizophrenia than other identical twins. Such data, however, are hard to gather. You would have to find not just twins but schizophrenic twins whose birth records are available and sufficiently detailed to give an indication of whether they were in one bag or two. Not surprisingly, the data are just not available.

However, there are a few telltale signs. At least some of the monochorionic twins show mirror-imaging: their hair swirls and fingerprints are on opposite sides, and they write with different hands. Further, the details of fingerprints are more similar in monochorionic twins: fingerprints are created in about the fourth month of gestation. Using these features as admittedly crude signs of monochorionic twins, James Davis in Missouri discovered a much higher concordance for schizophrenia in monochorionic than in dichorionic twins. He speculates that this may be evidence for the role of viruses, because twins who share fluid are likely to share viruses as well. But the concordance of monochorionic twins might indicate a shared exposure to accidental events of all kinds, not just infections.

Other infectious agents, too, may be capable of triggering the chain of events leading to susceptibility to schizophrenia, among them the herpes virus and toxoplasmosis, a protozoan disease sometimes caught from cats. Toxoplasma can cross the placenta in a pregnant woman and blind or retard the fetus; this agent can also probably cause later schizophrenia. It has long been known that other insults to the developing fetus may be risk factors for schizophrenia, including especially birth complications. The facts are hard to interpret because schizophrenic mothers are susceptible to birth complications themselves. Nonetheless it seems that a fetus starved of oxygen in the womb by preeclampsia is at nine times the normal risk of schizophrenia. What the medical fraternity delicately calls hypoxic insults—near-suffocation—during birth is a definite risk factor. Again, it seems to interact with genes. You can endure a hypoxic episode better with the right genes, or you can outwit your genetic fate better with an easy birth.

Hypoxia may be a reason for the fact that twins do not have identical risks, even though they share the predisposing genes. During birth, or before it, one twin may be more likely to experience hypoxia than the other. That may be why they do not both show the disease in later life.

However, there is another, more intriguing possibility. The virus that causes AIDS is a retrovirus, which means that when you catch AIDS, the genes of the virus are literally incorporated into the DNA in the chromosomes of some of your cells. Because this happens in blood cells and not in sperm or egg cells, such genes cannot be passed on to your offspring. But sometime in the distant past—and more than once—a similar retrovirus has managed to infect germ cells. We know this because the human genome contains many different copies of complete retroviral genomes, recipes for making infectious viral particles. They are called hervs (for human endogenous retroviruses), and they sit among our own genes as parasitic intruders. We pass them on to our offspring. Indeed, simplified and abridged versions of these viral genomes are among the commonest motifs in our genome—they are the so-called jumping genes that make up nearly a quarter of our DNA. We human beings are, at the DNA level, substantially descended from viruses.

Luckily, the viral DNA is kept under a sort of house arrest, shut down by a mechanism called methylation. But there is always the risk that a herv will escape, making a virus and infecting our cells from within. If that were to happen, the medical effect would be bad enough, but consider what philosophical damage it would also do to the nature–nurture debate. This would be an infectious disease, just like any other virus, but it would start within our very own genes and be passed on from parent to child as a set of genes. It would look like an inherited disease but behave like an infection.

A few years ago, evidence began to emerge that precisely such an event might explain multiple sclerosis (MS). MS is quite unlike schizophrenia in symptoms, but the two share a few features. Both occur in early adulthood; both are more frequent in people who were born in winter. So Paromita Deb-Rinker, a Canadian scientist, analyzed the DNA from three pairs of identical twins in which one member of the pair had schizophrenia and the other did not. By comparing the DNA from the affected twins with that from the unaffected twins, she found evidence of a herv that might be more active, or present in more copies, in the affected twin. Robert Yolken and his colleagues at Johns Hopkins University also looked for evidence of herv activity in schizophrenics. They tested the cerebrospinal fluid from 35 people newly diagnosed with schizophrenia in Heidelberg in Germany, 20 people who had suffered from the disorder for many years in Ireland, and 30 healthy controls from the same two places. Ten of the German schizophrenics, one of the Irish schizophrenics, and none of the controls had evidence of active herv genes. Moreover, the retrovirus that was active was from the same family of herv as the one associated with multiple sclerosis.

None of this yet proves that hervs are relevant to the disease, let alone the cause, but the findings do suggest a connection. If hervs were indeed causing schizophrenia, perhaps themselves triggered by influenza infection in the womb, and perhaps by interfering with other genes during the development of the frontal cortex of the brain, that would explain why the disorder is both highly heritable and apparently associated with different genes in different people.


The fifth witness brings a mouse. This is no ordinary mouse but one that behaved rather oddly in its cage sometime in 1951. It moved with a strange “reeling” motion, as if dancing (but not in the same way as the Japanese waltzing mice I mentioned in ). A scientist duly noticed the phenomenon, and by backcrossing quickly proved that the cause was a single gene inherited from both parents. The brain of the reeling mouse is something of a mess, principally because certain layers of cells that should be on the inside are on the outside instead. The “reelin” gene was located in 1995 on the mouse’s fifth chromosome, and the human equivalent soon followed in 1997: a gene on chromosome 7 that produced a protein 94 percent homologous with the mouse protein. It is a very big gene, with more than 12,000 letters, divided into no fewer than 65 separate “paragraphs” called exons. Subsequent experiments have shown that reelin protein is vital to the organization of the brain in the fetus of both a mouse and a human being. It directs the organized formation of layers in the brain, apparently by telling neurons where to grow to and when to stop.

What has all this to do with schizophrenia? In 1998 a team at the University of Illinois measured the quantity of reelin in the brains of recently dead schizophrenics and found that it was half that of the brains of normal dead people. A new potential suspect entered the picture. Disordered neuronal migration is a characteristic of schizophrenia, and reelin is one of the organizers of neuronal migration. Reelin also helps to maintain the “dendritic spines” at which synapses form, so a shortage could lead to faulty synapses. For devotees of the influenza theory, it quickly became apparent that one way to cause a transient 50 percent reduction in reelin expression in the brain of a mouse was to give it a prenatal infection with human influenza. In other words, reelin seemed to tie together the other theories of schizophrenia.

The poor reeler mouse immediately became the focus of much attention: perhaps it would prove to be an animal model of schizophrenia. The reeling behavior is apparent only if the mouse has inherited the faulty gene from both parents. If it has only one faulty gene, a mouse seems superficially normal. But it is not. It learns its way through a maze much more slowly and never gets as good at the task as a normal mouse. It is less sociable than normal mice.

This is hardly rodent schizophrenia, though perhaps it has a few parallels. Hopes that reelin would prove to be the chief cause of schizophrenia began to fade, however, in the 1990s when human reelers were discovered in two separate families in Saudi Arabia and England. In both these families cousins had married each other and the marriage had brought together faulty versions of the reelin gene, causing a disorder called lissencephaly with cerebellar hypoplasia (LCH), which is usually fatal within four years of birth. If inherited reelin deficiency is the cause of schizophrenia, then you would expect that some of the apparently unaffected relatives of these unfortunate children would be schizophrenic, because they are carrying the mutation in one of their genes. But so far there is no history of schizophrenia in either family, though the Arab family has not been studied in detail. Once again, as so often with schizophrenia, a promising start leads to a dead end. Reelin reduction is part of schizophrenia, perhaps a crucial part, but probably not one of the primary causes.

Bizarrely, reduced reelin is not confined to schizophrenia but is common in patients with severe bipolar depression and autism as well. It is almost as if a reduction in reelin can cause different brain problems depending on where in the brain, or when during development, it occurs. Reelin and influenza both point toward events in the womb; at first sight, this is puzzling because the most characteristic feature of schizophrenia is that it is a disease of adults. Although children who will later become schizophrenics can be identified retrospectively as anxious, slow to walk, and poor at verbal comprehension, most are by no means ill until after puberty. How can a disease be caused in the womb and expressed in adulthood?

The neurodevelopmental model of schizophrenia attempts to explain this conundrum. In 1987 Daniel Weinberger argued that schizophrenia was unlike other brain disorders in that the cause was no longer there when the symptoms appeared. The damage had been done much earlier but became apparent only because of some later, normal brain maturation process: the early effects are “unmasked” by later development as adulthood approaches. Unlike, say, Alzheimer’s or Huntington’s disease, schizophrenia is not a disease of brain degeneration but a disease of brain development. For example, during late adolescence and early adulthood the brain is extensively altered. Many of its wires are insulated for the first time, and many of its connections are “pruned”: synapses between neurons are cut back, leaving only the strongest ones. Perhaps in schizophrenics either there is too much pruning in the prefrontal cortex in reaction to a failure of the synapses to develop properly many years before, or perhaps too few neurons have migrated or extended to their targets. There will be many genes that mitigate or exacerbate these effects, or possibly respond to them, and they might therefore be called “schizophrenia genes,” but they are more like symptoms than causes. It is among the genes affecting the original early development that one must seek for true “causes” of schizophrenia. (It is perhaps no coincidence that schizophrenia appears at the age when young men and women are competing most fiercely to gain a foothold in an unfamiliar adult world and win a mate.)

Most scientists are agreed that in this sense schizophrenia is an organic disease, a disease of development—a disease of the fourth dimension, the dimension of time. It is caused by something going awry in the normal growth and differentiation of the brain. It is another forceful reminder that bodies—and brains—are not made, like model airplanes. They are grown, and that growth is directed by genes. But the genes react to each other, to environmental factors, and to chance events. To say that genes are nature and the rest is nurture is almost certainly wrong. Genes are the means by which nurture expresses itself, just as surely as they are the means by which nature expresses itself.


But no lover of science should ever be happy with a consensus, and getting the sixth witness is determined to upset the consensual mood. This witness believes that genes, development, viruses, and neurotransmitters all play a part, but none is the really fundamental explanation of the cause. All are really symptoms. The key to understanding schizophrenia, he asserts, lies in what we eat. In particular, the developing human brain has a need for certain fats, known as essential fatty acids, and the brains of “schizotypal” people need more of these than usual. If they do not get these fatty acids in their diet, the result can be schizophrenia.

In February 1977, on a bright but bitterly cold day, David Horrobin, a British medical researcher was walking through Montreal when he had his own “eureka” moment. Horrobin had been trying to fit together pieces of a mental jigsaw of odd facts about schizophrenia. They all related to the often-forgotten nonmental aspects of the disease, and they were as follows. First, schizophrenics rarely suffer from arthritis; second, they are surprisingly insensitive to pain; third, their psychosis often gets much better, temporarily, when they have a fever (astonishingly, malaria was once tried as a cure for schizophrenia—it worked, but only temporarily). The fourth piece of Horrobin’s mental jigsaw puzzle was new. He had just noticed that a chemical called niacin, then used to treat high cholesterol, did not cause a flushing of the skin in schizophrenics as it did in other people.

Suddenly all the pieces fit together. The skin flushing, the inflammation in arthritis, and the pain response all depend on the release of fat molecules called arachidonic acid (AA) from the membranes of cells. These are converted into prostaglandins, which cause some of the signs of inflammation, redness, and pain. Likewise, a fever also releases AA. So perhaps schizophrenics were unable to release normal quantities of AA from their cells and this caused their mental problems as well as their resistance to pain, arthritis, and flushing. Only a dose of fever raised their AA levels to those seen in normal people and restored their normal brain function. Horrobin duly published his hypothesis in the Lancet and sat back to wait for applause. There was a deafening silence. The schizophrenia experts were too immersed in the dopamine hypothesis at that time even to notice a different theory, let alone consider it. Schizophrenia was brain disease, so what was the relevance of fats?

Horrobin likes to defy conventional wisdom, and he was undaunted. But it was not until the 1990s that evidence started to come in supporting his hunch. Deficits in AA in schizophrenics were soon reported, as was an increased rate of oxidation of AA. Details gradually emerged from the fog of ignorance suggesting that either AA leaks too easily from the cell membranes of schizophrenics, or AA once released cannot be incorporated back into membranes easily—or perhaps both. Both processes are the result of faulty enzymes, and enzymes are made by genes, so Horrobin is happy to allow a role for genes in predisposing people to schizophrenia. But in expressing the disease, or better still, curing it, he believes that diet may play a role.

A learned and lengthy disquisition on the nature and function of fats and fatty acids is probably necessary at this point. But I fear the readers did not buy this book because they are in love with biochemistry, so I am going to try to boil down the essential facts about fats into a few terse sentences. Each cell in your body is held together by an outer membrane, which is made largely of fat-rich molecules called phospholipids; a phospholipid is like a three-pronged fork, each prong being a long fatty acid. There are hundreds of different fatty acids to choose from, ranging from saturated to polyunsaturated. The key feature of polyunsaturated fatty acids is that they make a more flexible prong. This matters especially in the brain, because the membrane of a brain cell must not only adopt an intricate shape but also change rapidly as connections between cells are added or lost. So the brain needs more polyunsaturated fatty acids than other tissues need: about one-quarter of its dry weight consists of just four kinds of polyunsaturates. They are known as the essential fatty acids (EFAs) because our neglectful ancestors never invented the ability to make them from scratch; their precursors come from food, having worked their way up the food chain from the simple algae and bacteria that do know how to make them. People who eat a diet rich in saturated fats and poor in EFAs may end up with brain cell membranes that are less flexible than those of somebody who eats a lot of fatty fish. (This does not easily explain why schizophrenia is just as common in countries like Norway and Japan, where fish forms a large part of the traditional diet, as it is elsewhere.)

The obvious test of Horrobin’s ideas is to treat schizophrenics with EFAs. His colleague Malcolm Peet and others have begun to do so. The results are not spectacular, but they are encouraging. A large daily dose of fish oil—rich in EFAs—does produce a modest improvement in the symptoms of schizophrenics. In 31 newly diagnosed Indian schizophrenics, a dose of one of the four main EFAs, called EPA, had such an effect in a double-blind trial (where neither the doctor nor the patients knew which patients were getting the drug until afterward) that 10 subjects no longer needed to take antipsychotic drugs to control their illness; none of the 29 control subjects given the placebo saw any improvement. EPA inhibits the enzyme that removes arachidonic acid from neuronal membranes; it therefore preserves the AA in the membrane. Since most antipsychotic drugs have serious side effects, from listlessness and weight gain to the symptoms of Parkinson’s disease, this is exciting news.

The hypothesis about fatty acids is not a rival to the various genetic hypotheses. Many of the neural symptoms of schizophrenia could be connected to fatty acids. EFAs are known to regulate the pruning of neuronal connections at puberty. Women are better at making EFAs from their dietary precursors, and women are less likely to get schizophrenia. Starvation during pregnancy, hypoxia during birth, stress, and even influenza infection have all been shown to reduce the availability of EFAs to the developing brain. The flu virus actually inhibits the formation of AA, possibly because AA is needed as part of the body’s defense.

More direct evidence for the fatty acid theory comes from some of the actual genes implicated in schizophrenia. They include the gene for phospholipase-2, a protein whose job is to remove the middle prong of the phospholipid fork, the one that is usually an EFA. The gene for apoD, a sort of delivery truck that brings fatty acids to the brain, is three times as active in schizophrenics in the very part of the brain most implicated in the symptoms of the disease—the prefrontal cortex—but not in the rest of the brain or body. It is almost as if the prefrontal cortex, finding itself short of these fatty acids, cranks up the expression of the apoD gene in an attempt to compensate. (The apoD gene, by the way, is on chromosome 3, where no “schizophrenia gene” was detected by the linkage studies.) One of the reasons that clozapine is an effective drug against schizophrenia might be its ability to encourage the expression of apoD. Horrobin’s hypothesis is that for full schizophrenia you require two genetic faults: one that reduces your ability to incorporate EFAs into cell membranes, and another that takes them out too easily (each fault could be affected by several genes). Even with both these genetic faults, an outside event is also required to trigger the psychosis, and other genes can modify or even forbid the effect.


Schizophrenia is about equally common all over the world and in all ethnic groups, occurring at the rate of about one case per hundred people. It takes much the same form in Australian Aborigines and the Inuit. This is unusual; many genetically influenced diseases are either peculiar to certain ethnic groups or much commoner in one group than another. It implies perhaps that the mutations that predispose some human beings to schizophrenia are ancient, having occurred before the ancestors of all non-Africans left Africa and fanned out across the world. Since being schizophrenic is hardly conducive to survival, let alone to successful parenthood, in a Stone Age world, this universality is puzzling: why have the genetic mutations not died out?

Many people have noticed that schizophrenics seem to appear in successful and intelligent families. (Such an argument led Henry Maudsley, a British contemporary of Kraepelin, to reject eugenics, because he realized that sterilizing those with a taint of mental illness would wipe out a lot of geniuses, too.) People with a mild version of the disorder—as noted earlier, these are sometimes called “schizotypal” people—are often unusually brilliant, self-assured, and focused. As Galton put it, “I have been surprised at finding how often insanity has appeared among the near relatives of exceptionally able men.”

This eccentricity may even help them achieve success. It is perhaps no accident that many great scientists, leaders, and religious prophets seem to walk the crater rim of the volcano of psychosis, and to have relatives with schizophrenia. James Joyce, Albert Einstein, Carl Gustav Jung, and Bertrand Russell all had close relatives with schizophrenia. Isaac Newton and Immanuel Kant might both be described as “schizotypal.” One absurdly precise study estimates that 28 percent of prominent scientists, 60 percent of composers, 73 percent of painters, 77 percent of novelists, and an astonishing 87 percent of poets have shown some degree of mental disturbance. As John Nash, the Princeton mathematician, said after recovering from 30 years of schizophrenia and accepting a Nobel Prize for his work on game theory, the interludes of rationality between his psychotic episodes were not welcome at all. “Rational thought imposes a limit on a person’s concept of his relation to the cosmos.”

The psychiatrist Randolph Nesse of Michigan speculates that schizophrenia may be an example of an evolutionary “cliff effect,” in which the mutations in different genes are all beneficial, except when they all come together in one person, or evolve just too far, at which point they suddenly combine to produce a disaster. Gout is a “cliff disease” of this kind. High levels of uric acid in the joints protect human beings from premature aging, but a few people get too much of it and painful crystals of the stuff form in their joints. Perhaps schizophrenia is the result of too much of a good thing: too many genetic and environmental factors that are usually good for brain function all coming together in one individual. This would explain why the genes predisposing people to schizophrenia do not die out; so long as they do not combine, they each benefit the survival of the carrier.


During the twentieth century the ideological forces of nature and nurture often behaved like medieval armies laying siege to diseases as if to castles. Scurvy and pellagra, explained as vitamin deficiencies, fell to the forces of nurture, while hemophilia and Huntington’s chorea, explained as genetic mutations, fell to the army of nature. Schizophrenia was a vital border stronghold, held by nurture for much of the century as a fortress of Freudian theory. But although the Freudians—those Knights Templars of the nature–nurture war—were driven from the battlements decades ago, the geneticists have never managed to occupy the fortress convincingly, and they may be forced to call a truce and welcome nurturist forces back over the moat.

A century after the syndrome was first identified, the only two things that can be said for certain about schizophrenia are that blaming unemotional mothers was wrong, and that there is something highly heritable about the syndrome. Beyond that, almost any combination of explanations is possible. Many genes clearly influence susceptibility to schizophrenia, many may respond to it in compensation, but few seem to cause it. Prenatal infection seems to be vital in many cases, but it may be neither necessary nor sufficient. Diet can exacerbate symptoms and perhaps even trigger the onset of symptoms, but probably only in those who are genetically susceptible.

In tackling psychosis, neither nature theories nor nurture theories are much good at distinguishing cause from effect. The human brain is wired to seek simple causes. It eschews uncaused events, preferring instead to deduce that when A and B are seen together, either A causes B or B causes A. This tendency is strongest in schizophrenics, who see causal connections between the most patent coincidences. But often A and B are simply parallel symptoms of something else. Or, even worse, A can be both the cause and the effect of B.

Here then is a perfect illustration that nature and nurture both matter. I promised that schizophrenia would confuse the issue, and it does. Kraepelin was wise to be agnostic about the cause: even with all the weight of modern science behind them his successors have failed to find it. They have even failed to distinguish cause from effect. Instead, it looks highly possible that the ultimate explanation of schizophrenia will include both nature and nurture, neither of which will be able to claim primacy.

Previous: 3. A Convenient Jingle
Next: 5. Genes in the Fourth Dimension