The Science of Evil: On Empathy and the Origins of Cruelty - Simon Baron-Cohen (2011)
Chapter 5. The Empathy Gene
Why should one person be Zero-Negative and another Zero-Positive? At the psychological level Zero-Negative occurs when an individual ends up all the way down at the far left end of the empathy curve of individual differences that we encountered in Chapter 2. That tells us nothing about how the person got there, a question related to the deeper level of causes. What we know is that the state of Zero-Negative can be caused by environmental neglect, such that the “internal pot of gold” is depleted. But the existence of people who are Zero-Negative and who have not suffered neglect, together with the existence of people who have suffered neglect but who have excellent empathy, shows that such environmental factors may be neither necessary nor sufficient to cause zero degrees of empathy.
Take Zero-Negative Type P, for example. We saw in Chapter 3 that, even though parental behavior can be to blame, it cannot entirely explain the making of a psychopath because among psychopaths parental style does not completely predict outcome.254 That is, there are parents who used the empathic, nonauthoritarian style of parenting, discussing things reasonably with their child, yet their child still turned out to be a psychopath. Equally, we all know individuals who have thrived despite growing up in difficult environments.
Dante Cicchetti grew up in the poorest, most dangerous neighborhood of Pittsburgh, yet he has ended up a professor of developmental psychopathology at the University of Minnesota. When I visited him at his research center in the 1980s, he told me he was lucky to be alive. Most of his peer group from childhood were either in prison or dead, a result of drugs, crime, or gang warfare. He is proof that what James Blair calls a “dangerous and criminogenic” environment does not totally determine outcome. In his studies he and his colleagues found as many as 80 percent of children who suffered abuse or neglect went on to develop “disorganized attachment.”255 But clearly it takes more than a harsh environment to make a psychopath. There must be a genetic element.
In this chapter we therefore explore the new evidence that environmental factors interact with “genes for empathy.” The word “interact” is, of course, key. I hope this book will not be misunderstood as arguing that empathy is wholly genetic because genes always exist in an environment, and we have seen buckets of evidence for the importance of early experience. Equally, I have put quotation marks around genes for empathy since genes cannot code for a high-level construct such as empathy. Genes simply blindly code for the production of proteins, blissfully unaware of their ultimate long-range effects.
But in this chapter we examine evidence that some genes are associated with your score on various measures of empathy. Even with these caveats in place, some people will be alarmed at the very idea of genes for empathy because they fear the deterministic implications of such a view. I would remind those readers that genes are not the only deterministic factor because the early environment also is. And I would ask them: Should we simply sweep such genetic evidence under the carpet just because it makes us feel uncomfortable? In the pursuit of trying to understand how human beings can end up doing awful things to each other, we have to look at all the evidence, not just the bits that suit our worldview.
The cause of Zero-Positive is rather different. As we saw in Chapter 4, for such individuals surfing down the empathy curve means also surfing up the systemizing curve. That is, they do not simply show zero degrees of empathy since at the same time they show high levels of systemizing. In their case, the genes that can leave them with zero degrees of empathy can also predispose them to extreme systemizing. We are therefore forced to conclude that different genes must be at work to produce Zero-Positive and Zero-Negative. Before we look at specific genes that can deplete empathy to make someone Zero-Negative and the different genes that can deplete empathy to make someone Zero-Positive, we should first look at the biggest clue that such outcomes are genetic at all. And that clue comes from twins.
If a trait or behavior is even partly genetic, we should see its signature showing up in twins. The key comparison is between twins who are identical (monozygotic, or MZ) and twins who are nonidentical (dizygotic, or DZ). If the trait or behavior in question does not differ much between MZ and DZ twins, then we are forced to conclude that genes play little, if any, role in the behavior because MZ and DZ twins are genetically quite different from each other: MZ twins are like genetic clones (they are genetically identical, so share 100 percent of their genes), whereas DZ twins are genetically no different from any other pair of siblings (they share on average 50 percent of their genes). In contrast, MZ and DZ twins are environmentally quite similar to each other: They are the same age and typically growing up in the same family. Expressed differently, if we discover that the trait or behavior in question shows a greater correlation among MZ twins than it does among DZ twins, then we can see that genes are at work.
Nearly all the studies of empathy in twins have found a greater correlation on empathy measures in MZ twins compared to DZ twins.256-258 As an example, the heritability of affective empathy (i.e., how much of the variation in affective empathy is genetic) has been estimated from a twin study to be 68 percent. That’s a lot. In contrast, one study looking at the heritability of “theory of mind” (or cognitive empathy) found that MZ and DZ twins were quite similar,259 suggesting environmental, rather than genetic, factors predominated. However, this conclusion was challenged by a later study.260
Estimating how big the environmental and the genetic contributions to empathy are varies depending on how empathy is measured. For example, some twin studies use questionnaire measures, whereas others use observational measures. In studies of very young twins observational measures include asking a mother to fake getting her finger caught while closing a suitcase as researchers filmed her child’s reactions. Studies using this observational method among toddlers have shown a strong genetic component to empathy.261,262 Observational methods are better measures of “affective empathy” (the responsive component) and may be suggesting that of the two main components of empathy (cognitive versus affective empathy), there may be a larger genetic contribution in the affective component. Alexithymia (difficulty reflecting and reporting on one’s own feelings) also shows heritability from twin studies.263
Regarding Type P, a twin study that used the Psychopathic Personality Inventory (questionnaire) found that two specific scales (“Machiavellian egocentricity” and “cold-heartedness”) showed moderate heritability. In a UK study of twins the callous and unemotional component of psychopathic tendencies at age seven showed even stronger heritability.264,265 Twins are not the only “natural experiment” to glimpse the importance of genes since the same clues are also seen in studies of children who have been adopted, again pointing to antisocial behavior being partly heritable.266
Adoption represents another opportunity for scientists to separate the effects of genes and environment because if a child—despite being raised in a different, genetically unrelated environment)—ends up being more similar to their “birth parents” than to their adoptive parents, then clearly genes are asserting their influence. Regarding twin studies of Type P, none of these show 100 percent heritability, but the genetic component is nevertheless substantial (the largest estimate being about 70 percent). This means there is still an environmental contribution to becoming a psychopath or developing some of the traits. In the “right” environment someone with the genetic predisposition to psychopathy could show this behavior.
Can we glimpse a genetic signature in Type N or Type B also? To date, there has not been a twin study of Type N, a gap in the literature that needs to be filled. Regarding Type B, family studies show that brothers, sisters, and parents of borderlines are ten times more likely to be Type B themselves.107,267-273 Family studies (unlike twin and adoption studies) don’t afford us the opportunity to separate environmental from genetic factors, so all we can conclude from family studies is that this form of Zero-Negative is familial (it runs in families). However, a twin study of Type B indeed showed higher “concordance” rates (35 percent) among MZ than among DZ pairs (7 percent). “Concordance” is a fancy word for correlation and is used when you count how often it is the case that when one of the twins in a pair has a condition (e.g., borderline personality disorder), the co-twin also has it. Although 35 versus 7 percent may look like a small difference, this tells us that becoming borderline is actually strongly heritable—about 70 percent of the risk for becoming borderline can be explained by genetic factors.274,275 So despite the clear environmental influence (principally abuse and neglect), to become borderline requires the individual to have some genetic susceptibility in the first place.
What about the genetic signature of Zero-Positive? Family studies show that brothers, sisters, and parents of people with autism or Asperger Syndrome (Zero-Positive) also show more than average levels of autistic traits.204,252,253,276-278 So we can glimpse that being Zero-Positive is familial. This is true when you use questionnaires such as the Empathy Quotient (EQ), and it is even true when you use psychological tests that measure emotion recognition from photographs or when you measure brain activity during such tasks.204,253 Equally, twin studies of Zero-Positive reveal that MZ twins show a higher correlation than DZ twins do on measures of autistic traits.279-281So given all this evidence for genes for empathy, which genes determine whether a person becomes Zero-Negative or Zero-Positive?
Genes for Aggression
Some scientists have focused their search for empathy genes on those that affect the neurotransmitter serotonin. Too much serotonin in the synapse has been linked to aggression. When you increase serotonin receptor activity, which clears serotonin from the synapse, this decreases aggression.282 An example of a gene involved in clearing serotonin (and other neurotransmitters such as dopamine, noradrenaline, and adrenaline) is the MAOA (monoamine oxidase-A) gene.
Now here’s the interesting bit. There are several forms of this gene. One is called MAOA-L because carriers of this gene produce low (L) levels of a key enzyme. The other form of the gene is called MAOA-H because carriers produce high (H) levels of the same enzyme. Low levels of MAOA often mean high levels of neurotransmitters in the synapses. Not surprisingly, people with the MAOA-H form are less aggressive. People with the MAOA-L form are overrepresented in warrior cultures (such as the Maoris in New Zealand). For this reason, controversially, it has been called the “warrior gene.” It won’t surprise you that this gene interacts with environmental factors. For example, Avshalom Caspi and his colleagues found that abused children with the MAOA-L form of the gene were more likely to develop antisocial problems compared to abused children with the MAOA-H form of the gene.283,284
Animal research backs this up: Male mice with a deletion of the MAOA gene show more aggressive behavior. In humans, male members of a Dutch family with a mutation of the MAOA gene showed high levels of aggression. And neuroimaging reveals the amygdala and the anterior cingulate (two key regions in the empathy circuit) are smaller in MAOA-L carriers. Carriers of this version of the gene also show increased amygdala and reduced anterior cingulate activity when matching facial expressions.285,286
Genes for Emotion Recognition
We know that at least three genes can affect how the brain responds to emotional expressions, and as we have seen, emotion recognition is a key part of empathy. Which version of the serotonin transporter gene (SLC6A4) you have affects how much your amygdala responds to fearful facial expressions (though not all studies confirm this).287,288 Genes that modulate the availability of other similar neurotransmitters (e.g., dopamine) also affect the amygdala’s response to fear faces.289-291 Recall that the amygdala is a key brain region in the empathy circuit. In addition, variations in the arginine vasopressin receptor 1A gene (AVPR1A), which has been linked to autism, also influence how much the amygdala responds to faces showing fear or anger.292
The third gene was one we discovered in our lab, so let me tell you the story. Plenty of research shows that happy faces are rewarding to look at. Just as we like to look at food or beautiful landscapes, we find happy faces rewarding. This is true from infancy onward, as demonstrated by the fact that the typical infant, from about eight weeks old, will smile at a happy face. This is called the “social smile.” We also know there are two key brain regions that are active when we are experiencing something rewarding: the striatum and the substantia nigra.293,294 So it comes as no surprise that these same brain regions are active when we look at happy faces.295
Bhismadev Chakrabarti wanted to know if there were genes that might influence how much your striatum responds to these happy faces. After all, we all know there are individual differences in how much we like to engage in people-watching. So he chose a gene that was already known to be involved in how we respond to reward: the cannabinoid receptor gene 1 (CNRI). This gene is strongly expressed in the striatum, a reward system of the brain.a This gene gets its name from the drug cannabis because its protein product is the main target for cannabis in the brain. Individual differences in this receptor are linked to how rewarding an individual will find cannabis. (Some people can smoke a joint and have no effects, others will find it pleasurable, and yet others will find it unpleasant.) We took a cheek swab from each of the people lying in the scanner, from which we could extract their DNA, so as to test Bhisma’s neat idea that the variation of this gene will affect how active your striatum is when you are looking at happy faces. Sure enough, this prediction was confirmed.
These three genes give us very clear examples of how genetic makeup can change how your brain responds to other people’s emotions.295,296 It is highly unlikely that these are the only genes that affect emotion recognition, but they are sufficient to show that there are genes involved in at least this aspect of empathy.
Genes Associated with the EQ
In 2009 Bhisma and I completed our second genetic experiment. We were interested to find out which genes were associated with individual differences on the Empathy Quotient. We had hundreds of people from the general population take the EQ, and, as we saw in Chapter 2, we observed the bell-curve (or normal) distribution, some scoring low, some scoring medium, and some scoring high on empathy. This is termed “normal” because these individual differences are just what we expect to see in any population, just as we see individual differences in height. The big question was: If we picked some plausible “candidate” genes, would variations in any of these genes be associated with variations in EQ scores?
The way we went about testing for empathy genes is worth a short digression because gene-hunting is a risky business. Given that there are an estimated 30,000 genes in the human genome, gene-hunters have two strategies available to them: a “whole genome scan” (i.e., to test all 30,000 genes), which is a costly business; or a test of plausible candidate genes, which is slightly more affordable (since you pay per gene per person). Essentially, this choice of strategy boils down to the difference between a fishing expedition where you have no particular hypothesis about where the fish will be (so you drop your line in “blind” at regular points all along the river) and a highly directed “hypothesis-driven” approach (where you know that the fish are likely to gather at a very specific point in the river). We chose this candidate gene, hypothesis-driven approach.
The next question was which candidates to choose. It is a high-risk strategy because, if you choose correctly, you can really strike lucky. But if you happen to encounter the bad luck of having chosen incorrectly, all you have to show for your troubles is a bunch of nonsignificant (and therefore unexciting) results. Bhisma and I sat down to plan our strategy. I was very keen on a group of genes involved in the sex hormones (testosterone and estrogen) and persuaded Bhisma this was worth a try because empathy shows clear sex differences at the level of psychology. On average girls and women score higher than boys and men on the EQ, for example, a result that is found across different cultures and that I discussed at length in my last book, The Essential Difference.12,15,17,297
Empathy also shows clear sex differences at the level of the brain. Women on average show more activity in many areas within the empathy circuit while reading emotional expressions in faces,253 and a recent study of structural differences between male and female brains indicated that many of the brain regions that differed between females and males included the amygdala and the “mirror neuron system,” which overlaps with parts of the empathy circuit.
I had one more reason for wanting to test the genes that regulate the sex hormones. For the last ten years we have been following a group of about five hundred children in Cambridgeshire in the UK whose mothers had had amniocentesis. This is where a long needle is introduced into the mother’s womb during pregnancy to draw off some of the amniotic fluid, a procedure carried out for clinical reasons. We had asked these mothers for permission to measure the testosterone, the so-called male hormone, in the amniotic fluid because males produce much more of it than females do. We found that the less testosterone produced by the fetus (fetal testosterone) before birth, the higher the score on the child version of the EQ taken later. So it seemed that this sex hormone might be involved in shaping the empathy circuit of the developing human brain.b
For all these reasons we selected genes known to be involved in sex steroid hormones.c We expanded our team to bring in two world experts in medical genetics research, Lindsey Kent and Frank Dudbridge. Bhisma wanted us to include a second group of genes that we loosely called “social-emotional behavior” genes. This was to follow up on the CNR1 gene but also to test the idea that there may be other genes that influence how drawn you are to other people. One such gene related to another hormone, oxytocin. Oxytocin has got a lot of press since it was discovered that males in one species of vole (a furry rodent) are less sociable (and likely to be polygamous) than males in another species of vole, who are more sociable (and likely to be monogamous). These species are largely identical apart from a dramatic difference in the expression of oxytocin and vasopressin in the brain.300-302 Oxytocin also hits the news a lot because if you inhale it through your nose so that it goes straight into your brain or if you inject it into your blood, it improves your score on tests of emotion recognition and empathy.303,304
In the popular press oxytocin has a variety of names. It is sometimes called the “love hormone” because we release it during intimate physical contact, including orgasm. It is also sometimes referred to as the “trust hormone” because if we boost our oxytocin levels, we tend to be more generous toward others as measured by how much money we would be willing to lend a stranger.305,306 And, finally, oxytocin is also sometimes referred to as the “attachment hormone” because it is released by new mothers during breastfeeding, promoting that well-being feeling driving mothers to fall in love with their infants, and vice versa.307 So we tested genes such as those involved in synthesis and the receptors of oxytocin, as well as the closely related peptide hormone arginine vasopressin.
Finally, we selected a set of candidate genes involved in what we loosely called “neural growth” simply because studies of the brain in Zero-Positive individuals (autism and Asperger Syndrome) had revealed atypical patterns in how the nerve cells (neurons) are wired up and how fast the autistic brain is growing in early postnatal development.
We waited with baited breath while the genotyping took place and wondered whether the substantial time and money we had invested would all be to no avail. Imagine our excitement when the results came through: Of the sixty-eight candidate genes we tested, four showed a strongly significant association with the EQ! One of these genes, CYPB11B1, was in the sex steroid group. A second gene, WFSI, was in the group related to social-emotional behavior.d The third and fourth genes associated with the EQ, NTRK1 and GARBR3, were in the neural growth group.e,f
So, all in all, we had found four genes associated with empathy. 308 This was only a first step because it is a big jump from finding genes to understanding how their functions have an impact on empathy. But a start is a start.
Genes Associated with Autistic Traits
Bhisma and I had taken the precaution of asking our volunteers from the general population to fill in the EQ and the Autism Spectrum Quotient (AQ), a measure of how many autistic traits you have. We included this second questionnaire because our earlier studies had shown this, too, to be replete with individual differences: Some people score low, having few autistic traits; others score in the average range for the population; and yet others score high, even though they don’t have a diagnosis of any kind. As we saw in Chapter 4, autistic traits are not all negative.
While reduced empathy can cause social difficulties, a remarkable attention to detail and an ability to concentrate on a small topic for hours, to understand that topic in a highly systematic way, can be positive and can lead the individual to blossom in certain fields, despite their relative difficulties with empathy and socializing. And if you are blossoming in a nonsocial academic field (such as mathematics or computer science, engineering or physics) or in a nonsocial practical line of work (such as car maintenance, map-reading, or railway time-tabling) or in an art or a craft (such as drawing, model-building, or design), you may find a niche and get by without a diagnosis. For these reasons, we might find genes related to autistic traits in the general population.
So would any of our candidate genes also show a significant association with the AQ? There were four more genes that were only associated with the AQ. These included one of the neurologin genes (NLGN4X), one of the homeobox genes (HOXA1) that regulate brain patterning, and ARNT2 (a gene involved in neurogenesis). It also included a monoamine oxidase gene (MAOB), similar to the gene we discussed earlier.
Finally, we looked at whether any of our candidate genes were strongly associated with being Zero-Positive (having a diagnosis of Asperger Syndrome), and we were very excited to find six genes that were. These included three sex steroid genes: one of the receptors for estrogen (ESR2), CYP17A1 (abnormalities in which cause irregularities of the menstrual cycle in women), and CYP11B1 (one of the genes that catalyze the production of testosterone from cholesterol and the gene that has also been linked to empathy). It also included the oxytocin gene (OXT), as well as ARNT1 and HOXA1 (also linked to empathy).
This was telling us that whereas some genes were just influencing EQ, others were influencing just AQ, Zero-Positive alone, or several of these. Such genetic work opens up more questions and avenues of research.g
The genes discussed in this chapter are by no means the complete list of genes that will turn out to play a part in empathy. There are more to be discovered, and some of the genes specified here may not stand the test of time in terms of independent replication. But the list is sufficient to show that genes for empathy and for some forms of both Zero-Negative and Zero-Positive are being discovered.
However, it is important to reiterate the caveats at the start of this chapter: While it is tempting to blame zero degrees of empathy on either genes or environment, a mix of factors is clearly causal. For example, babies who had difficult births in which they suffered anoxia (lack of oxygen) have a higher risk of developing conduct disorder, delinquency, or violence in adulthood. Boys with minor physical anomalies (e.g., having low hung ears) have a higher risk of later becoming violent offenders, especially if they live in an unstable home.310 The low hung ears can occur for genetic reasons or as a result of the mother having bleeding and infection during pregnancy, indicating that the baby was not in the best of health during early pregnancy. These increase a child’s risk of being violent (and therefore low in empathy) if they coincide with instability in the family environment. For example, in a Danish study 4 percent of boys had both a difficult birth and maternal rejection, yet these boys committed 18 percent of the violent crimes in adulthood.311 Once again, we see the complex interplay of biological factors (in this case birth trauma) and psychological factors (in this case a depleted “internal pot of gold”).
Other Animals and Empathy
And we can step back from humans, too, to ask if other animals have at least in simpler form some of the precursors of empathy. They should have if empathy is in part genetic because typically intermediate forms of evolved traits can be seen across the animal kingdom. Emory University primatologist Frans de Waal has argued that humans are not the only species to be capable of empathy, though he acknowledges that in humans it may have evolved to a higher level than that seen in other species.312 But in his view the precursors of empathy are evident in a number of behaviors. First, some monkeys and other animals will share food with other members of their own group. Why should they do this if they were entirely selfish?
It could be argued that this form of apparent altruism is in fact driven by genetic relatedness—that in helping members of your own group who may be cousins and who therefore share some of your genes, you are ultimately helping to protect copies of your genes (in others) that may enable their host to survive and reproduce, thereby perpetuating your (shared) genes. In a recent experiment at Emory University, when capuchin monkeys were given the choice to have food alone or food when another monkey was also given food, they chose the social option if they knew the other monkey. This suggests that food sharing is not just confined to genetically related members of a monkey’s group—it seems to extend to acquaintances too (see Figure 10).313
Second, there are other examples of animals of the same species helping each other, not just around sharing of food. For example, some chimpanzees have been observed to help each other to climb over a high wall. These are compelling examples of the ability to read each other’s needs and goals. Third, de Waal observed that after monkeys or apes fight, the loser is often shown consolation by other members of the group. As he licks his wounds, another animal will come over and touch him gently or even put an arm around the defeated animal as if to offer comfort. This use of touch strongly resembles what we would do to a person in discomfort, and at risk of anthropomorphism, we might interpret it as an emotion appropriate to someone else’s emotional state—in short, empathy.
Figure 10: Capuchin Monkeys Sharing Food
Finally, monkeys and apes show plenty of evidence that they can read emotional expressions from the face or vocalizations or body posture of another member of their species. For example, Northwestern University clinical psychologist Susan Mineka and her colleagues famously showed that young monkeys could learn to fear a snake if their mothers showed fear in their face and voice,314 and University of Wisconsin psychologist Harry Harlow also found that monkeys reared in isolation and then reintroduced into a social group tended to react as if a friendly approach by a fellow monkey were an aggressive approach, whereas those reared by their mothers and with their siblings could clearly distinguish another animal’s “intentions” (to be friendly or aggressive).315
In two remarkable early studies, if a rat learns that pushing a bar will lower onto the ground another rat who has been suspended, the rat will press the lever.316 That’s in rats, who are meant to lack empathy! And Northwestern University primatologist Jules Masserman and colleagues showed in 1964 that in our closer relatives, rhesus monkeys, those who were trained to pull a chain to get food refused to do so if pulling the chain also meant another monkey was given an electric shock. It was as if they refused to profit at the expense of another monkey’s pain.317 We can see why some people believe other animals, including monkeys and apes, have some level of empathy too.
However, there are limits to empathy in other species. Chimpanzees can, for example, fight “deadly turf wars” to expand their territory; in these wars large groups send out “patrols to strategically kill rivals” before moving into the new territory. 318,319 Such “cruelty” is common in other species as well as our own. As another example, whereas even human toddlers will use their index fingers to point to things, to share attention with another person, pointing is not seen in other species. Nor do other animals convincingly engage in deception, suggesting they do not think about another animal’s thoughts, even if they can respond to that animal’s emotions.320
And in a striking example from vervet monkeys, mother monkeys who are swimming across a flooded rice field to get to dry land might have their infant monkeys clinging to their furry underbelly. Even though each mother’s head is above water, many are blissfully unaware that their infants head is underwater, so that when the mother arrives safely at the other side of the field, tragically their baby has drowned. This vividly underlines how monkeys may not take another animal’s different perspective into account, with hard-hitting consequences, including the impact this behavior will have on the survival of their own genes. Clearly, whatever glimmerings of empathy we can discern (or imagine we discern) in other species, the level of empathy that humans show is qualitatively different from that seen in any other species.