June 05, 2003
Altruistic Punishment And Genetic Engineering Of The Mind

One of the recurring themes on FuturePundit is that the greatest danger from human genetic engineering will come from genetic engineering of the mind. We will develop the ability to create minds that will be dangerous or simply not compatible with the kind of societies that most of us prefer to live in. At one extreme, imagine genetically engineered minds devoid of conscience or empathy and at the same time highly calculating and ruthless in the pursuit of their own desires. Or, at a different extreme, imagine minds that so desired to fit in and to serve that they'd make ideal members of a communist collective ruled over by personalities genetically engineered to lead the masses.

The Debate About Genetic Engineering As A Threat To Human Nature

A number of commentators voice worries about human genetic engineering. Others consider those worries exaggerated in part because they think human genetic engineering is unlikely. Let me state up front that I think human genetic engineering is inevitable and that it will become a widespread practice. Furthermore, just because one believes that some specific objections voiced by particular worriers seem naive or unlikely does not discredit the very idea of being concerned about the consequences. It will eventually become possible to use genetic engineering to raise intelligence, to alter personality and to change the behavioral tendencies of our progeny. Whatever alterations parents or governments choose to make to future progeny will have profound effects upon human society. It would be irresponsible to simply dismiss the fears of those who are frightened by this prospect.

Some of those who are opposed to the practice of human genetic engineering are afraid that something vital about human nature will be lost by genetic engineering. Some are afraid that humans will be genetically engineered to be perpetually happy and that this happiness will somehow leave humans spiritually impoverished and devoid of the capacity to understand the deeper meaning of life. Curiously, such critics rarely seem to offer examples of how humans could be made less able to respect the rights of others. I suspect this particular danger from genetic engineering is not cited more often because the idea that humans can be made to have wildly different moral capacities and behavioral tendencies undermines the religious view of humans as moral actors possessed of consciences and capable of judging right from wrong according to some universal God-given standard. Well, the day is approaching within 10 or 20 years when it will become possible to do genetic engineering of offspring in such a way that they will have different behavioral tendencies and different innate conceptions of right and wrong. Therefore we can not afford to continue to avert our gaze from the biological basis of conscience, of the tendency to form moral judgements, and of the biological foundations of human values and normative beliefs. These basic attributes of human nature already vary considerably between humans. Genetic engineering will make these attributes more mutable in ways that constitute a substantial potential threat to the continuation of human civilization.

It seems likely that there are many genes in the human population which have variations that cause people to differ in their personality characteristics. Therefore the large number of different combinations of genetic variations found in human populations contribute to the large variety of personalities and behavioral tendencies also found among humans. Because of the genetic variations that are so largely responsible for the existing variety of personalities it should be possible to use only currently existing genetic variations to create a human population which is much different in average behavioral tendencies from existing human populations. A large change in the average of human behavior could be accomplished just by increasing the frequency of some genetic variations while decreasing the frequency of other variations which influence cognitive processes. Because there are already fairly extreme outliers in behavior and personality in the human population and since in at least some cases part of the reason for their extreme desires and behaviors is genetic it probably will not be necessary to create new genes or new variations of existing genes to do embryonic genetic engineering to create humans that differ considerably from the vast majority of existing humans. To get a sense of just how radically the human population could be altered without developing new genes or new genetic variations one has to look no further than the behavioral differences already existing in the human population.

Psychopaths Demonstrate The Danger Of Existing Extreme Human Outliers

Consider more extreme deviations from the human norm. One of the worst forms of deviations from human norms of behavior is found in psychopaths.

"The murdering psychopaths showed a much more positive association to violence. Psychopaths who were not murderers had a much more negative view of violence," Gray explained.

Unrestrained by the guilt that most humans would feel from harming others psychopaths do not even appear to have memory associations that categorize violence as unpleasant.

Normally, when shown a word on the screen, people take longer to figure out which button to press when non-related words -- such as "violent" and "pleasant" -- are on the same button, Snowden said.

However, psychopathic murderers responded differently, and completed the test "as if they do not associate violence and unpleasant," Snowden said.

Will it some day be possible to genetically engineer violent psychopaths? Why not? After all, a number of non-human predator species enjoy killing and in some species in some circumstances they even kill members of their own species. Surely these behavioral traits are somehow coded for by the genomes of these species. It may well be that there are genetic variations which influence personality that predispose the existing psychopaths to be psychopaths.

You might argue that very few people will want to choose genetic variations for their children that would increase the odds that the children will be psychopaths. True enough. But these outliers in human behavior and human cognition demonstrate just how far existing human nature extends without the use of genetic engineering. Genetic engineering will make it possible to create humans whose emotional make-up will differ substantially from what we see in most humans today.

Could Subtle Changes In Personalities Cause Huge Societal Changes?

Are there changes in human nature that at first glance might strike people as less extreme and less threatening than the creation of psychopaths and yet that could still cause huge problems for the healthy functioning of human societies? Are there changes in personality and in behavioral tendencies that people might want to give their offpsring that would have profound and negative consequences if a sufficiently large percentage of the populace opted to do genetic modification of embryos to cause personality changes in their offspring?

An accurate answer to those questions would give us a better idea of whether the ability to do genetic engineering in the embryonic stage of our future progeny could lead to disastrous consequences for the future of human civilization. One way to attempt to answer these questions is to look for evidence of characteristics of human nature that are beneficial for society, which may be genetically based, which are not equally shared by all humans, and for which we could imagine reasons why at least some prospective parents would want to modify those characteristics in their future offspring. This brings us to the topic of altruistic punishment.

A Study Of Altruistic Punishment

Ernst Fehr of the University of Zurich and Simon Gächter of the University of St. Gallen in Switzerland published an interesting study "Altruistic punishment in humans" in the January 2002 issue of Nature. This study has occasioned a great deal of discussion about the implications it holds for human nature. Fehr and Gachter showed that many people will pay to punish those who do not cooperate even though the punishers derive no other benefit from punishing aside from the satisfaction of carrying out the punishment.

In an investment game with shared profits, players punish those who do not contribute to the group's good, despite the personal cost. The emotional satisfaction of dispensing justice seems to spur them on: "People say, 'I like to punish'," says Ernst Fehr of the University of Zurich.

The punishment was doled out to people who the punishers knew they would not play again. The ability to dole out punishment caused people to cooperate to mutual benefit.

Investment climbed to four times the previous level as the threat of punishment encouraged cooperation.

Researchers said that anger was the reason the players handed out punishment, even though it cost them money to do so.

"At the end of the experiment, people told us they were very angry about the free-riders," said Fehr. "Our hypothesis is that negative emotions are the driving force behind the punishment."

These people doled out punishment at cost to themselves even though one rule of the game was that players never played with other players more than once. The punishment therefore did not benefit the punisher by causing the punished person to be more cooperative toward the punisher in future rounds of the game.

In a separate series of games that Fehr and Gachter conducted where it was not possible to inflict punishment the amount of cooperation quickly declined. However, in game series where it was possible to inflict punishment on non-cooperating free riders the amount of cooperation rose in successive rounds even though each person played a completely new set of people in each round.

"It's a very important force for establishing large-scale cooperation," Dr. Fehr said in a telephone interview. "Every citizen is a little policeman in a sense. There are so many social norms that we follow almost unconsciously, and they are enforced by the moral outrage we expect if we were to violate them."

People expected to be punished based on their previous experience and they adjusted their behavior accordingly. This expectation that others would punish them even though others had nothing to gain from doling out punishment was key to increasing cooperation in successive rounds of games.

You can read the full paper "Altruistic punishment in humans" in PDF format.

Altruistic punishment took place frequently. In the ten sessions, subjects punished other group members a total of 1,270 times; 84.3% of the subjects punished at least once, 34.3% punished more than five times during the six periods, and 9.3% punished even more than ten times. Punishment also followed a clear pattern. Most (74.2%) acts of punishment were imposed on defectors (that is, below-average contributors) and were executed by cooperators (that is, above-average contributors), and punishment of the defectors was harsh (Fig. 1). For example, if a subject invested 14±20 MUs less than the average investment of the other members during periods 5 and 6, the total group expenditures for punishing this subject were almost 10 MUs. Moreover, the more a subject's investment fell short of the average investment of the other three group members, the more the subject was punished. The pattern and strength of punishment was also stable across time (Fig. 1). A Wilcoxon signed rank test of punishment in periods 1±4 versus periods 5 and 6, with 10 matched observations, yields z = -1.07, P = 0.285 (two-tailed). The same test for periods 1±5 versus period 6 yields z = 0.178, P = 0.859 (two-tailed).

Note that the most enthusiastic cooperators were also the ones most likely to punish. Those people who most enjoyed working in a cooperating group also had the strongest drive to make others cooperate as well. It may be that the anger that came from observing free rider behavior came as a response of being denied the joy humans experience from working in a cooperating team. Are there genetic variations that make people feel greater or lesser amounts of pleasure from working in cooperating groups? If there are (and this seems likely to be the case) then imagine how much human societies would change if a substantial portion of the population chose to give their offspring genetic variations that increased or decreased their desire to work in cooperating groups or to punish those who didn't.

The chain of cause and effect that leads to the infliction of punishment probably has a few different parts that are each separately variable from person to person. The participants in this study were motivated by anger. But in order to feel anger they first had to perceive unfairness. In order to do that they had to believe that people in a group have an obligation to cooperate for joint benefit. This desire to work together is an important human desire. Is there a genetic basis for just this desire? Well, look at other species. Some like to work together in groups. Others prefer solitary existences. Surely there must be a genetic basis for this inter-species difference in behavior.

The participants also had to be willing to act on their anger, pay a price for that action, and to act even when they stood to gain nothing personally from acting. It is likely that different groups of genes and genetic variations separately influence different parts or stages of the response that leads to infliction of punishment. Though it is not clear just what those parts are.

Is Altrustic Punishment The Result Of A Lack Of Discernment?

Cooperation is encouraged by the ability of people to reward each other for cooperating. But what Fehr and Gachter found was that the ability to punish non-cooperators encouraged cooperation and, crucially, that most people are willing to incur costs in order to punish non-cooperators. Those who elected to pay to punish must have derived satisfaction from the ability to punish those who angered them by acting in what the punishers saw as an unfairly selfish manner.

Why would altruistic punishment be selected for by evolution? One possible explanation is that the behavior that was selected for caused benefit to those who had the trait in ancient environments but that in modern environments the trait frequently causes humans to engage in altruistic punishment. In this view we are seeing it because humans are living under conditions which are far from the conditions in which we evolved. It is quite possible that historically humans were far more likely to benefit from punishing those who did not cooperate with any group they were members of because people were members of fewer groups and for longer periods of time per group. Anyone who was punished was almost always someone with whom the punisher would have future dealings. Therefore humans may not have been under enough selective pressure to become more discerning about whom to punish. There wasn't as great of a need to be able to accurately judge when the costs of inflicting punishment would be a net benefit to the punisher because of the longer term nature of most relationships. Therefore the willingness to mete out punishment to noncooperators probably didn't need to be complex enough to make humans draw distinctions between people they would or would not have future dealings with.

If you think that humans do not have traits that are expressed in ways that show insufficient use of cognitive processes to discern the appropriateness of emotional responses then consider sexual jealousy in human males. It was probably selected for in men so that men would have a motive to prevent their women from mating with someone else. A man unknowingly who raised another man's child wasted his own precious resources and decreased his reproductive fitness. Emotional responses that decrease the likelihood of that happening were selected for. But in the modern era sexual jealousy happens in men who are in relationships with women who are incapable of having children or unwilling to do so. So why should jealousy happen under those circumstances? Because the emotional response of jealousy was never selected for to use a cognitive process that is sufficiently discerning to be able to take into account mating that did not have the possibility of causing reproduction. That kind of mating is far more common today than it was in our evolutionary past when our traits were selected for. Also, people who are not going to reproduce are not going to pass along a greater or lesser tendency toward sexual jealousy and therefore there is not much of a mechanism available to even select for a more complex sexual jealousy response in the modern world.

Not everyone in the Fehr and Gächter study meted out punishments. There are, broadly speaking, two possible major reasons why some did not pay to punish. Some people may simply be less easily roused to punish uncooperative people in general. Some step in the process leading to the act of punishment may be harder to stimulate in them. Another possibility is that some may be far more discerning (either for genetic or environmental/educational reasons) in evaluating when paying to punish is worth it to them. It is likely that both of these factors cause differences in how people respond to non-cooperators and that genetic variability has an effect on both factors.

Change The Desire To Punish Free Riders Or The Ability To Discern One's Interests?

How does all this matter to the genetic engineering of offspring? Suppose genetic variations will be discovered that affect how easily people become angered by uncooperative behavior. Imagine that some people choose to give their offspring genetic variations that decrease their tendency to be angered by noncooperation. It is possible to conceive of plausible reasons why some people will make these choices for their progeny. Parents might decide they want their children to go thru life feeling less anger about perceived injustices in their lives. If that happened then future generations would be less inclined than current generations to enforce cooperation. The consequences for how human societies functioned would be profound.

It is also possible that there are genetic variations that make a person more able to evaluate whether paying to punish someone is worth it. One can easily imagine why a parent would want to make their children more capable of subtle discernment of where their real interests lie. This ability would give their kids an edge in dealing with other people in business negotiations and in other settings. But that enhanced capacity to discern where one's own interests lie might come at the expense of making society function less well as a whole - at least in some respects. In a society where people get less riled up when they are able to more accurately calculate their own self-interest then there would be less altruistic punishment doled out. This would effectively lower the amount of informal policing of norms in a society. Therefore those who would go unpunished would, as a consequence, be more willing to be uncooperative and to free load off the efforts of others. Again, the consequences would be profound and problematic.

The More General Desire To Punish Perceived Unfairness

Avoidance of cooperation in working toward a group goal is just one way that individuals can cause problems for others in a group. People can also take the possessions of others, hurt others, and deceive others for a variety of reasons. These other types of perceived unfair behavior are all also capable of eliciting an anger response and a desire to punish.

The desire to punish perceived unfairness is important. That desire causes behavior that is altruistic and that is necessary to maintain cooperation between members of groups. The desire to punish the unfair among us probably motivates police officers, prosecutors, soldiers, government and corporate whistleblowers, and a great many others as well. Imagine a society where either a smaller percentage of the population would ever feel angry enough to perform altruistic punishment or where those who did feel the desire didn't feel it as strongly and didn't act on it as often. The resulting society might have more crime for a number of reasons. Law enforcement personnel might be less motivated. Fewer would be willing to work at the most challenging law enforcement jobs since job satisfaction from meting out punishment would be felt to a much lesser extent. Witnesses to crimes would be less motivated to come forward to testify or to intervene to stop a crime. An assortment of other behaviors would change in ways that reduced restraints on law-breakers.

But the effects would not be limited to law enforcement. Members of groups punish each other in an assortment of ways in a variety of environments including businesses, volunteer groups, militaries, and families. Imagine every kind of situation where you've wanted to punish someone for something they did. Genetic engineering that affected that desire would change human behavior in all of those situations.

A person making a purely selfish economic calculation would probably not choose to punish unfairness in cases where the bulk of the benefits of meting out the punishment would flow to other people. Witnesses to crimes, to unfair acts in the workplace, and to unfair behavior in general are frequently in the position where they have little at stake and yet often are willing to intervene or testify or otherwise pay a price to prevent or punish unfairness that is not directly aimed at them personally.

Another possible consequence of a reduction in the desire to perform altruistic punishment might be that governments would be more likely to abuse a small fraction of the populace. The rest of the populace would be less inclined to get angry about it and to make sacrifices to protest and oppose such government actions. Therefore governments would be less constrained. On the margin a large number of decisions would be made differently in ways that would make a society function less well and a society whose populace was less motivated to dole out altruistic punishment might well become less free as a consequence.

Fehr and Gachter have uncovered a human behavior that is most likely the product of natural selection. The fact that people desire to punish others even though they have to pay to mete out the punishment suggests that the punishment behavior is deeply built into human minds. This desire to punish those who are viewed as unfair is probably an essential element of human nature needed to maintain a civilized society.

Fairness In Real Life Situations Is Harder To Judge And Open To Dispute

The desire to mete out justice is problematic because determining what is fair is difficult and open to dispute. Fehr and Gachter defined the rules of simple games that their experimental subjects played. The actions of each of the players were easy for the other players to understand. There was no uncertainty as to the number of players, the actions taken, or their ramifications. There was no dispute as to the legitimacy or interpretation of the game rules. There was no need for reference to events of previous days, months, years, or centuries. By contrast, real human societies have all these complications and much more.

In real life situations disagreements over what is fair and over what are the relevant facts in a given situation make many acts of punishment itself seem unfair to those receiving it or to observers. One reason people differ on whether any particular act of punishment is justified is that people can be and frequently are misled by others or by their own flawed cognitive processes into reaching false conclusions about who did what and why. The desire to punish unfairness can occur in situations where the real facts of the matter do not justify the response. Also, some react with to perceived unfairness with what others see as excessive anger and their response can seem a disproportionate act of punishment compared to the original act that evoked the perception of unfairness. It is easy to see how that can get out of hand. For instance, if members of a nation, religion, or other grouping become convinced that they have been on the receiving end of a great injustice (e.g. the famous Nazi myth about being stabbed in the back by Jews in World War I which contributed to World War II) this sort of belief can be used to motivate them to commit all manner of violent acts individually and collectively. But incorrect beliefs about unfair treatment and excessive responses to perceived unfairness are common well below the level of grand historical events. Such beliefs can be found everywhere in school playgrounds, work places, and marriages. Surely, the impulse to punish unfairness is not an unmitigated benefit to the human race.

Still, in spite of all the problems that arise from the desire to punish a bigger problem would occur if people had a weaker desire to punish the unfairness of others. Societies absolutely need cooperation and the ability and desire to inflict punishment are essential to the maintenance of a sufficient degree of cooperation to make societies function well.

We Need To Understand The Genetic Basis Of Human Nature

The most important missing element in research on the intersection between economics and psychology is the genetic link. But at this point in time it is hard to make that connnection. The cost of DNA sequencing is still in the millions of dollars per person. It is too expensive to find connections between genetic variations and variations in behavior. Surely progress along that front is being made. But it would be far easier to do if every experiment on human behavior could include complete DNA sequence information on each study participant. Then genetic variations could be compared with behavioral differences. The inability to effectively control for genetic differences when doing experiments is one of the biggest factors holding back the advance of a more accurate social and psychological science of human nature.

Science is starting to supply us with information about how genetic variations affect human nature. The coming abililty to do make use of this information when doing germ line genetic engineering will cause a huge conflict between the desires of parents to give their offspring characteristics that the parents prefer versus the interests of the larger society on how members of future generations act toward the rest of us. The ability to affect how and when future generations will act in altruistic fashions will be politically far more contentious than current issues such as abortion or embryonic stem cell therapy.

The problem with allowing parents alone to decide on what future generations will be like is that we all have to live with the consequences of their decisions. Currently the effects of decisions that people make over who to mate with can not be easily measured or predicted. Also, currently there are limits to how much a difference each person can make in the genetic make-up of their progeny because they can only pass down what they have. What is going to change is that much of the uncertainty will be eliminated and the degree of control on the outcome will rise enormously. This will allow much larger changes in distribution of behavioral tendencies in populations. Averages and extremes will shift in ways that we can only begin to guess at today.

If one wants to have a relevant debate about the dangers of genetic engineering of humans then the central issue must be genetic engineering of the mind. The biggest benefits and greatest dangers come from the decisions people make when they start genetically engineering the minds of future generations.

Update: See Kenneth Silber's essay on Tech Central Station entitled Genetic Paradoxes.

And yet, it is precisely such automatons that would be most readily manipulated by genetic engineering. Genetic engineering's conservative critics seem to believe both that humans are beings of depth and complexity, and that humans can be transformed by science with relative ease. These positions are in considerable tension with each other. Could it be that conservatives who worry about genetic engineering actually regard dignity as an illusion that must be protected from scientific probing?

Silber sees contradictions with the conservative critique of genetic engineering by Leon Kass, Bill McKibben et.al. that are similar to the problems I see in their arguments. If there is something irreducibly magic about being humans then we can't possibly genetically engineer humans who are missing the magic elements. Are we just going to reduce human dignity and make life less meaningful if we genetically engineer people? Or are we going to gain the ability to change human nature? Also, when people talk about "human nature" it is only sensible to mean charactertistics of the mind. Are there characteristics of the mind that are inherent to human nature that will be changeable some day with genetic engineering? I think so. Others differ.

There are others who are attacking the conservative critique who also seem quite wrong to me. Charles Murtaugh recently made an argument in a recent article on Tech Central Station arguing it is all so complex we may never figure it out.

When the trait is complex, and the genetics are complex, their interaction may well never be unraveled.

But to illustrate his point Charles uses an example where the womb environment differences between lab mouse strains turned out to be important. But the scientists investigating did find the factor that accounted for a developmental difference even though the genetic difference between the strains operated by changing the womb environment in a way that changed behavior. The mechanism of action was seemingly obscure and yet it was identified. The cause did not remain unknowable. Well, I'm reminded of a quote spoken by the character Yama-Dharma the Death God in Roger Zelazny's Lord of Light:

"It is the difference between the unknown and the unknowable, between science and fantasy - it is a matter of essence. The four points of the compass be logic, knowledge, wisdom and the unknown. Some do bow in that final direction. Others advance upon it. To bow before the one is to lose sight of the three. I may submit to the unknown, but never to the unknowable. The man who bows in that final direction is either a saint or a fool. I have no use for either."

(You can see the chapter context here)

Yes, we will figure out what all the genes do and how they interact with each other and with the environment. Long before we understand what they all do we will understand what many of them do and how differences between them create differences in personality, intelligence, and behavior. We will be able to figure that out by sequencing millions of people and comparing their genetic sequences and their mental qualities (as well as medical histories and assorted other things about them).

Aside: For more on how it will become possible to do massive comparisons of genetic sequences see reports on efforts to drive down the costs of DNA sequencing by many orders of magnitude in my Biotech Advance Rates archive.

What annoys me about the arguments by Steven Pinker, Charles Murtaugh, Kenneth Silber, and Amy Greenwood (see my pleasant debate with Amy in the comments section of this post) is the way they try to down play the fears of the spiritual conservatives of the Kass school of bioethics (no, there is no such formal school but close enough) by arguing that it will not even be possible to abuse biotechnology to create intelligent creatures that are far from the human norm. They sound like Officer Barbrady of South Park: "Move along folks. Nothing to see here".

An argument that it will be hard to develop the knowledge needed to do germ line genetic engineering is not an argument that it will never happen. Even if it takes 20 years (as Amy estimates for single location genetic changes to fix genetic diseases - see the comments of this post) to develop the ability to do germline genetic engineering, well, 20 years is not all that long a time. Humans seem to have a tendency to think that 20 years into the past is more real than 20 years into the future because they can remember back 20 years. But while we can not go back in a time machine 20 years into the past most of us now alive will slowly but surely travel in the unidirectional time machine of this universe to 20 years in the future (at least barring global thermonuclear war or a massive asteroid strike). We will live in that day when we will have the ability to do a great many more things with biotechnology than we can now. Even if germline genetic engineering to change many genetic locations to affect a complex trait such as the mind is 30 year out it will happen. It is real kids. It is coming. We are moving along into a future where many more things will become possible.

Another argument these folks make is that so many genetic variations are involved in causing different types of minds that we will not be able to manage to make changes complex enough to change the desired traits while also avoiding undesireable side effects. But this is an argument that the task is difficult, not that it is impossible. We are gaining the ability to manage more complex processes all the time. Our tools for understanding and modelling complex processes continue to improve and in all likelihood will continue to do so. We will eventually gain the ability to genetically engineer offspring to intentionally cause them to have minds which will be substantially different than what their minds would have been like had they just gotten the unmodified genes of their parents. We ought to discuss what problems may arise as a consequence of this inevitable development and what we will need to do to try to avoid the dangers that unwise and malicious uses of these technologies will make possible.

Share |      Randall Parker, 2003 June 05 05:57 PM  Brain Altruism

amy greenwood said at June 6, 2003 4:31 PM:


This is ridiculous. We are NOT going to be able to genetically engineer morality, personality, or behavior into people, no more so than the effects we can acheive with drugs. It is extremely important to me that people think critically about science, biology, and genetic engineering and be able to debate the effects of technology on society. However, its irresponsible to promote irrational fears.

"Furthermore, just because one believes that some specific objections voiced by particular worriers seem naive or unlikely does not discredit the very idea of being concerned about the consequences"

Yes, in fact, it does. If someone believes that the world is going to melt next spring, I am not obliged to entertain such foolishness. The argument that human behavior, culture, and morality are in many ways independent of biology AND that the biology that does affect these processes is too complex for us to control on a genetic level is a GOOD ARGUMENT.

Dienekes Pontikos said at June 6, 2003 6:12 PM:

Very interesting summary of research. It's certainly interesting to think about the philosophical issues associated with the _potential_ arrival of genetic engineering of personality. One should perhaps one day write a rigorous study on the feasibility of this prospect. IMO there is just too little data in the six billion or so human beings that upper bound our sample space to allow us to figure out patterns of sufficient complexity, the kinds of patterns that are likely to have significant effects on the personality of human beings.

Randall Parker said at June 6, 2003 6:54 PM:


We sure will be able to do all of those things. We won't be able to totally control the outcome. But we will be able to produce people whose average chance of being murderers will be higher or lower by orders of magnitude than would otherwise be the case. We will also be able to make people who tend to be very shy or very out-going. We will be able to make people who are placid or easily aroused to aggression. We will be able to make people who are easily distracted or who can ignore a great deal of what is going on around them and concentrate.

The idea that human behavior is independent of biology is hard to credit. Genes that strongly influence behavior are being identified all the time. Look at Pet-1, the gastrin-releasing peptide (GRP) gene, the influence of maternal testosterone levels on female tomboy behavior, and the influence of MAO-A variants on adolescent violence. The MAO-A variants are surely just the tip of the iceberg in terms of genes that influence criminal tendencies and of how well people fit in with others. There are surely many genes that come in variations that influence impulsiveness, ease with which people can be angered, and a great many other mental qualities. Why wouldn't there be?

As for the effects we can achieve with drugs: tell that to a cop who has had to take down some madman on PCP. People on drugs become more or less inhibited, more or less polite, more or less shy, more or less morally constrained. What drugs can accomplish is far less than what genetic manipulation will be able to accomplish.

However, as more about how the mind works is discovered new targets for pharmacological intervention will be discovered as well. Take for instance L-type voltage-gated calcium channels (LVGCCs) which can be blocked in mice to prevent them from overcoming fear.

Can we intervene in the mind to change what drives and desires a person has? There are just too many examples of weird diseases that caused very specific behavioral problems that suggest this will be possible. How about the example of the guy who had a brain tumour that made him into a pedophile? Does that not suggest that human desires are very physically based and that our ability to control them has a very physical basis that will be able to be manipulated by sufficiently advanced technology?

amy greenwood said at June 6, 2003 8:46 PM:

No, Randall, I think you are misunderstanding me. Of course human behavior is influenced by genetics...the brain is the seat of behavior and genes instruct the development and maintenance of the brain. But much of behavior is influenced by experience...and experience ITSELF can change gene expression in the brain. All the examples you site are simply examples where a whole neural system is ramped up or down. This is a very crude way to influence behavior. We can do that with drugs, some of which are very specific. Even so, the overall influencing of proclivities is a very different thing than instructing someone WHAT to think, e.g., morality, personality, religiosity. You can't take experience out of the equation, and you can't take choice out of the equation.

What it is seems you are implying, the part I find objectionable, is that when we get better at understanding genetics we will be able to do MORE than just influence proclivities, we will have control over motivations and thoughts. Behavior and genetics don't parse the same way.

It is notoriously tricky in neuroscience to draw conclusions about what a gene, drug, training paradigm is actually doing. You could, as you have, draw the conclusion that the pedophile guy with the tumor is suffering a moral laspe. But maybe if he'd had a set of different experiences, he would have been just really aroused by his wife. Or had affairs with other women. Or sheep. Whatever. The point is that his arousal system was BROKEN. The tumor (gene, or drug) did not put specific thoughts in his head. And that same tumor in someone else's head may have caused them to commit suicide. Behaviors have many motivations.

I think it is instructive to remind oneself that genes are not traits: they are the tools and construction material and instructions that build traits. By analogy, you can influence the construction of a house by taking away the philips head screwdrivers (enzymes) or replacing the glass window panes with acrylic (structural proteins) or even by messing with directions of how to wire the lights (regulatory DNA). In general, you are just going to ruin things (usually more than you expect) or create subtle qualitative differences (acrylic vs. glass), but you're generally not going to have much control or get anything really new or useful. And you would have to understand the whole system of instructions and know all the tools in the tool box to do anything great, like build and addition. Most importantly, you can't control who moves in to the house (experiences), people who will surely remodel like fiends.

Randall Parker said at June 7, 2003 12:52 AM:


Why are most people heterosexual? Genes code for it. There might be people not strongly oriented one way or the other who respond to environment and experience to get pushed one way or another. But the vast bulk of humans are hetero because genes make them that way. Would you deny that it will become possible to change a person's sexual orientation by genetic manipulation?

Humans also find symmetrical body shapes more attractive. Males find younger appearances more attractive. There are genes that are causing all this.

Abstract ideas get taught. Some moral ideas are taught. But basic desires are mostly not taught. If two people have a desire that is forbidden by society and yet one has that desire to a greater extent then the one who has it to a greater extent is far more likely to do the forbidden thing to get it. A person's willingness to behave in a way that stays within the rules of a religious or cultural belief depends on the strength of the desire have some forbidden thing. So, yes, genetics can influence whether a person violates a moral code.

Empathy is also probably heavily influenced by genetic coding. Look at the average differences between men and women wrt empathy. Empathy is another factor that constrains behavior. Someone who empathises strongly is less likely to harm others. Turn empathy up and it is likely that people will be less violent.

Yes, yes, genotype is not phenotype. Genes interact with the environment. In different environments the same genes can produce different phenotypic expressions. But genes that code for the brain do not just code for logic gates. They code for behavioral tendencies, tastes, and desires.

You keep telling me we will have a hard time getting enough control to get anything usefully different out of genetic engineering of the mind. But I keep telling you that useful differences (or dangerous differences in some cases - some people really are genetically more prone to be criminals for instance) already exist in the human population. If we can identify the variations that produce differences then some day people will choose among those variations when having offspring.

In the book Behavioral Genetics In The Postgenomic Era edited by Robert Plomin, John C. Defries, Ian W. Craig, and Peter McGuffin in the chapter by K. Peter Lesch on Neuroticism And Serotonin Lesch states "...based on estimates from twin studies using these and related measures that have consistently demonstrated that genetic factors contribute 40%-60% of the variance in neuroticism and other related personality traits.". That's a high percentage. But that probably understates the real effect on personality for a simple reason: Some people may be genetically on the boundary between, for instance, being neurotic or not being neurotic. Wherease others may have genetic variations that make them very firmly neurotic. In other words, in the cases where two twins have significantly different levels of neuroticism they probably do not have as many genetic variances tilted toward neuroticism as in the case where both twins are neurotic.

Is neuroticism important in influencing morality? Well, it contributes to anxiety, hostility and depression. Surely at least the tendency toward hostility will make someone less willing to respect some moral code that says not to, say, slap or shoot another person. Other personality traits are also shown by twins studies to have major genetic compoents. Let me repeat again what I said about the contribution of genetic factors: the argument that genetic factors do not foreordain a particular personality outcome is flawed because what is really going on there is that some people are, metaphorically speaking, genetic fence-sitters. There are lots of loci contributing to any given personality trait. So twins who have a mix of loci not all pushing firmly in a single direction may be different from each other. But I'm willing to bet money that for some people the loci all line up so firmly that they are bound to be neurotic. Others are bound to be extraverts. And so on.

I really do not understand your meaning when you say that the examples I cite are where whole neural systems are ramped up or down. LVGCC channel blocking ramps a whole system up or down? Homey don't think so. Turning a female into a tomboy is a big change in behavior in any case. But you are so hung up on the idea that we have individual choice that you are not seeing the larger forest: if more people had the MAO-A variant that makes them more likely to be violent if their dads beat them then there will be more violence in the world. Change the genetic variations that offspring get and group average levels of behavior will change. Whether you think they will still have choice or have the ability to be morally judged by God or the cosmic muffin is really besides the point.

Are you religious? Do you object to the idea that some people may have genetic coding that is so strongly pushing them toward behaving in some objectionable way that they can't be said to be moral agents in conscious control and responsible for their actions? Is your problem that you don't like a view of the mind that undermines the Western conception of free will?

I have a problem with this religious view because some (not all) religious folks are ignoring the real dangers of genetic engineering. They just don't want to see how much minds will be able to be changed by genetic engineering. There is a religious blindness to the biological basis of human nature that is preventing at least some religious folks from accurately seeing the nature of the approaching danger. At the same time, the lefties who believe in the Tabula Rasa model of infinitely moldable human nature (got to make that New Soviet Man) are equally blind. They get the blindness by embracing a secular faith instead of a religious one. But they are blinded all the same. Then there are people who delude themselves by building their blindness from a mixture of both viewpoints.

If you think your viewpoint is not influenced by a secular or religious faith then I don't get where you are coming from. Are you familiar with the twis studies results on behavior? Or are you just thinking that it will be too hard to figure out the human genome? I think you are underestimating the steadily growing human ability to understand and manage complexity. Moore's Law and similar laws with hard disks and fiber optics are making computers steadily faster. We are going to have the ability to model systems orders of magnitude more complex than anything we can model today. Computer programs are getting bigger and more complex. They will become orders of magnitude more complex in the future.

Meanwhile science is going to advance and matter - including mind matter - is going to become ever more mutable to our will.

amy greenwood said at June 7, 2003 10:43 AM:

Hi Randall,

Its funny. I think we are seeing a lot of the same data, but giving it interpretations. Mine are strongly influenced by my background: practical experience with mammalian developmental biology and a background in neuroscience. To answer your other question, I'm a devout atheist. So, no. Etherial moral agents don't motivate my arguements.

Biologically, speaking. LVGCC channel blocking DOES change a whole neural stystem. In fact, L-gated voltage channels are expressed quite widely througt the brain, spinal cord, peripheral nervous system, and even in the heart. By injecting blockers that can cross the blood brain barrier the authors can block them everywhere. What if the primary effect is on heart rate? A neural system involved with motivation? Perception? Fear? All of the above? That was my point in the last post about being able to interpret the effects of a manipulation on behavior. (Ectopic expression of a hormone is even worse. Talk about somethign that is designed to affect just about every tissue in the body.)

My main thesis on this subject is this: The facts that 1. most traits are controlled by many genes, 2. individual genes have many functions, 3. compensation and redundancy occur during development, 4. behavior arises from the network (the network part is important) activity of neurons, which is affected BOTH by inherited genetic factors (usually many) and by EXPERIENCE, means that germ line engineering of traits, especially ones related to behavior, will be extremely difficult AND will often be dangerous.

That's it. I have no moral problem with the biological basis of the mind.

I am familiar with some of the twin studies. Its clear that behavior is affected by genetic background and that we will be able to identify some the loci that INFLUENCE some behavioral traits. I do not think that by manipulating these loci, individually or in small groups, will we get the desired effect (engineering) and I think such a think would often have nasty side effects.

I have a feeling that you are going to argue that the ability to manipulate genes will get better and better. While I agree with you on one level, I personally can't see the technology that will allow for the large scale manipulation of many loci in a conditional way. I am in the process of trying to engineer mice in which I can conditionally manipulate the expression of a single gene. I can tell you that its more difficult than adverstised, and there are many many many caveats of which I must be aware if I want to interpret the effect correctly. Thank goodness I'm only concerned about the development of a single tissue and I don't have to worry about the effects on the whole animal, or shudder to think, behavior.

Anyway, I can't prove a negative. All my training tells me you are greatly underestimating the complexity of this problem. But, I gues we'll just have to wait and see in 20 years.


michael vassar said at June 7, 2003 10:59 AM:

Obviously the interaction between technology and behavior will be important in a generation or two, but for a trait as esoteric as desire to punish to be substantially altered in the majority of the population, as it would have to be before it would cause social pathology, would almost certainly take many decades. You are interested in MNT (molecular nanotechnology). It's hard for me to understand why the consequences of genetic changes a century out would be important to you. There are many ways in which genetic engineering might effect society a few decades down the road. Genetic engineering might drive better neurology. It might save medical expenses. It might increase the fraction of the population that contributed to cultural and technological innovation. Most importantly, it might lead to life-extension before MNT does, giving the world a decade or two's warning that it was headed towards radical change, and possibly even increasing people's planning horizons. What it won't do is create the sort of qualitative change in learning, creative, and analytical abilities that leads to a whole new sort of existance. It could do that, if we had centuries, but MNT can almost certainly do it WITHIN a century. Ultimately, biotechnology as a whole is just a bridge to MNT and AI, so it is only the immediate effects of biotech on MNT and AI that matter.

Randall Parker said at June 7, 2003 1:17 PM:


You are a post-doc right? I'm guessing you are in your late 20s. I wish you could see what molecular biology used to be like back in the dark ages of 25-30 years ago. Back in the late 1970s I was talking to an organic chemist (actually, she taught the o chem course I took) and she commented that a discovery that a couple of people had won a Nobel Prize for a few years previously had taken them decades to accomplish (might have been Guillemin and Schally for LHRH and FSHRH) but could at that time (late 70s or maybe 80 or 81) then be accomplished by a grad student in a few months time. Technique advanced. Heck, I remember as an undergrad the amazement when the Maxam, Gilbert and Sanger papers were published on how to sequence DNA in the first place.

I also went to one of the first seminars where Leroy Hood introduced his then new radical DNA sequencing machine (80? 81? not sure). He'd taken a mass spectrometer that had been developed for the Mariner mission to Mars and adapted it to make a DNA sequencer. He had to have someone working full-time purifying reagents to feed the thing because it was so sensitive to purity of reagents. Everyone was amazed. Well, that instrument looks like an antique today and today's instruments are surely going to be obsolesced by nanopore sequencers and other approaches that lower costs and increase speed by orders of magnitude (one of the guys working on the nanopore approach is at Harvard btw). A similar tale could be told about microarrays for monitoring gene expression. Just measuring the expression of a single gene used to be so difficult and now people watch dozens or hundreds at a time. Truly amazing for anyone who saw the insides of a molecular bio lab in the 1970s when even a personal computer was a rarety. Heck, I took my first programming class in a math department because the C.S. department students all had to use cards but the smaller group in the math dept could afford to use IBM terminals for full screen editing. Such was life in the dark ages.

Step back from the techniques you are using today and look at the work that is being done to make new kinds of tools. Look at the work on microfluidics and other approaches to miniaturizing and automating lab work. These efforts are going to automate what you now do by hand. They will speed up and make the procedures have higher success rates. New types of procedures will be developed that are more reliable.

Large scale manipulation of loci in a conditional way: if I understand you correctly I think you are referring to the ability to turn genes on and off (or at least modulate to some extent) selectively in only some parts of the brain. Well, I'm talking about germ line genetic engineering. No need to put promoter sites in that are meant to be controlled by drugs when a person is an adult to control conditional expression. No need to send gene therapy into only some neurons.

Surely groups of genes affect personality. Say, for the sake of argument, 100 genes control whether a person is neurotic (Robert Plomin thinks locis which have variants that contribute to pesonality traits each contribute between 0.1% and 5% of the total control of them and so 100 genes is a ballpark number). Well, suppose a couple want to have a kid that is not neurotic (a reasonable desire IMO). But suppose they each have about half their 100 genes tilting in the direction of neurosis and that they both know they have problems (surely you've met such people). Do they have to change every gene in a fetus to get a non-neurotic offspring? No, and for a couple of reasons. First of all, they can do pre-implantation selection from multiple embryos. Secondly, they might have an embryo that has 40 genes tilting toward neurosis and feel they want better odds. So they do gene therapy (say 20 or 30 years from now) and change 20 of them. Now they have a tilt so strongly away from neurosis that the odds of getting a neurotic child are slim (assuming that the odds do not scale linearly for each gene that contributes to a trait - a fair assumption). Also, out of the 40 genes that contribute some probably contribute percentage-wise more than others. So they can just aim to modify those which contribute the most.

Nasty side effects: If we manipulate the loci to change them to variations that already occur naturally and if we can even say that we know of people who have the combinations we are aiming for then the risks will be pretty low.

But I think you are missing something really big here: Some people already have genetic variations that put their potential offspring at significant risk of neurosis, depression, and other mental disorders. There are a lot of mentally unhealthy people walking around. You can't compare a no-risk lack of manipulation to a risky manipulation because for many people the lack of manipulation comes with substantial risks. Therefore millions will elect to have gene therapy done to their fetuses to change the genes that influence the brain. They will not see their choices as increasing risks because once the techniques for doing the modification of genetic variants in embryos become refined enough the odds will be that they will be decreasing risks.

There are two risks from the genetic manipulations of embryos that need to be distinguished:

  • 1) Dangers from the manipulations that are independent of which genes are changed. The very fact that one is going into embryos and putting in genes and knocking out genes runs the risk of messing up other things besides the desired targets (eg epigenetic state).

  • 2) The risks of putting together variations that work together in unexpected and undesired ways.

I can't say when the first problem will be solved. Surely a lot of work is going into developing new gene therapy techniques. Possibly it will be solved by doing gene therapy before even meiosis takes place. But then we need to get control of meiosis. I think you are overstating the second risk for a few reasons:

  • A) We will know a great deal about all the variation combinations that already occur naturally once we can sequence large numbers of people cheaply.

  • B) people make new variation combinations every time they have kids anyway. Life is a risky business.

  • C) Some people are already at considerable risk of having messed up kids because they already have bad variants and considerable (even 100% in some cases) risk that their children will be as messed up mentally as they are (I could introduce you to families... :>).

One other point: do you accept that whether it takes us 20 or 40 or 50 years we will eventually achieve the abiity to safely (in the sense of causing no side effects from the manipulation itself) change many genes in an an embryo? Also, do you accept that we will eventually know all the genetic variations that contribute to personality and that we will eventually know all the major side-effects of putting together a large number of different combinations of variations? As I see it we can debate when something will happen and that is hard to predict. The biggest reason I see things happening sooner is that I see Moore's Law-like phenomena at work in biotech and the effects of these phenomena cause the rate of progress to accelerate.

But then there is the separate argument about whether something will happen at all. I've already made my point about why I think the twins studies understate the extent to which genetic variations can control personality: some people have variations that make them fence-straddlers more susceptible to environment. Others have variations that line up to push them strongly in one direction. Until the DNA of twins studies members can be sequenced and we can compare the cases where two twins have the same personality type with those who do not that point will probably not be provable or disprovable. So are you doubting that we will ever be able to reliably produce particular personality types? If so, we will just have to wait for DNA sequencing costs to fall to settle that one by doing twins studies with DNA sequencing.

Randall Parker said at June 7, 2003 1:33 PM:


Why am I worried about the world 100 years from now? Partly because we might still be alive then. I think engineered negligible senescence is achieveable within the next 50 years and maybe much sooner. Also, because I think that it is ethical to try to not make the world a worse place for future generations. However, I believe the kinds of dangers I'm discussing could become problems much sooner than 100 years from now.

As for the benefits of technological advance: I certainly want all the wonderful things that advances in science and technology promise to bring. I'm not a technological Luddite. But I also see big dangers as well and think we should try to identify those dangers more clearly. As for how rapidly genetic engineering on embryos could cause large changes in a society: a majority would not have to be changed in order for large problems to occur. A single serial killer can kill a lot of people and frighten a great many more. The ability of small numbers of people to kill larger numbers of people will only increase.

amy greenwood said at June 7, 2003 3:55 PM:

Hi Randall,

Post-doc, yes, though early 30s. What you say about the rapid advancement in biology is quite true and astonishing even from my perspective. Beautiful, really. But the future is notoriously hard to predict. Nixon declared a war on cancer in '71 and where are we now? Although there has been some very important progress, we realize now that cancer is not something that can be eradicated like polio. I think genetic engineering is going to be a bit like that. Harder than many peole initially think due to inherent complexities in the biology. Meanwhile, astonishment will come from many other places and mostly we'll be astonished that we didn't see it coming.

Details: "Large scale manipulation of loci in a conditional way: if I understand you correctly I think you are referring to the ability to turn genes on and off (or at least modulate to some extent) selectively in only some parts of the brain. Well, I'm talking about germ line genetic engineering. No need to put promoter sites in that are meant to be controlled by drugs when a person is an adult to control conditional expression. No need to send gene therapy into only some neurons."

Germ line engineering and conditional mutants are not mutally exclusive. In fact, there is a great need to make germ line mutations conditional (regulatable in space and time by tetraycycline, Cre-ER, ecdysone, other genes, etc.) for the very reasons I have been talking about. When you alter a expression of a gene to see what effect it has on the brain, it's rather a drag if you also cause a lethal heart defect. Or diabetes. Or impede development.

Your other point about making mutations that are already found in the human population is a good one (and, yes, I agree that each mutation likely carries its own vulnerabilities, e.g. sickle cell anemia/malaria restance.). But I think its important to remember that there are many other genes that "buffer" the effect of the one you are interested in. People who inherit that mutation normally also inherit the buffering genetic background. I think you run a higher risk of side effect by moving one (or 100) genes from one genome to another. With normal inheritance you normally get half.

How long do I think it will take to learn how to manipulate the expression of many genes (100+) via germline mutations... I think in animals we will learn how to do this within maybe 20 years. And I think these experiments (like with cloning) will prove to be cautionary tales about why we shouldn't proceed pell mell with genetic engineering of humans.

On a more postive note, I see a lot of promise for gene therapy and cell replacement therapy in adult tissues, though I still think we have a long way to go.

Randall Parker said at June 7, 2003 4:56 PM:

Nixon's War On Cancer: That predictably failed. In fact, there were scientists involved in the initial meetings who predicted it would fail and why it would fail. I heard this first hand from Ephraim Racker (was then an old biochemist at Cornell about to retire and NAS member but I don't know whether he's alive now) in a dinner conversation that probably took place in 1980 (give or take a year). He told me that at the time he and many other scientists viewed the War On Cancer as a big mistake. He was called in to the very first NIH conference to organize this war in the early 70s to do the initial planning of Nixon's big initiative. He said that cancer wasn't going to be solved until we figured out a lot more about cellular regulation and that the attempts to test lots of treatments (there were just massive screenings and testings of compounds) were huge wastes of money. The correctness of his argument seemed obvious to me at the time. Then other profs told me how people would count the number of times they used the word "cancer" in grant proposals where they were just trying to study how genetic regulation worked. Government biomedical research policy seemed dumb to me (still does).

But things are changing. The biggest trend changing molecular biology is the development of faster ways to take apart systems. DNA sequencing is one big visible example. You surely know many others from first-hand experience. I think that nanotech is going to revolutionize how molecular biology is done because tools that work at the scale of cells and molecules will be developed that are more automated, more sensitive, cheaper, faster, etc. This all could have happened sooner and faster. But funding continues to be tilted too much toward using the existing tools to figure out systems rather than toward the development of better tools. I'd have stayed in biology as a career if back then it was possible to get much funding to develop better tools. But everyone told me it wasn't and they were all amazed that Hood could even get the funding to do what he did.

Buffering genes: But look, suppose a person wants a certain characteristic produceable by a certain combination of variations. If millions of people have had their DNA sequenced then the thing to do would be to check on whether existing people already have genomes that mostly look like one's own genome but which have combinations that provide the feature you want. Say you are a Scottish Highlander and have particular HapMap groupings. Find other people who have those groupings who have the kind of personality or level of intelligence you want your child to have and see if they have health problems as a result. The key is to get DNA sequencing cheap enough and then build huge databases for millions of people of DNA sequences combined with medical records and assorted psychometric and personality and behavioral measures so that the correlations can be found. There are enough existing combinations yielding enough different phenotypic outcomes that if large enough numbers of people could be fully sequenced then lower risk alterations of one's own combinations of variations could be identified. This is one reason why I think the development of nanopore sequencers will lead more quickly to human genetic engineering (granted we need the techniques to manipulate germ line genes safely and easily too).

Some mutations that produce certain personality or intelligence outomces will be rare and it will be hard to say what other genetic combinations they can be used with safely. But for the ones that are more common we will be able to find many different combinations of background or buffering genes that a given variation will be compatible with. In the process we are also going to discover a large number of subtle incompatibiites between genetic variations that are already existing combined into the same humans. It is just too easy for two different populations to each get mutations that give them advantages but which when combined cause at least mild incompatibilities where the resulting offspring do not function optimally in some fashion.

Then there are also mutations that we have that have no benefits at all. I've read estimates upward of 500 for the number of genetic variations that the average person has that are harmful in some way with no benefits that were ever selected for in any past environment (as distinct from mutations like sickle cell or beta thalessemia which are harmful but confer protection against malaria and therefore were selected for). Most mutations are harmful or neutral. Very few are beneficial even with a mix of benefit and cost. But since many harmful mutations only very mildly reduce reproductive fitness it takes many generations to eliminate them. Therefore some genetic theorists run models with various assumptions of the rate of new mutations occurring in each generation, how harmful each mutation is on average, and they project from these models that we each on average have hundreds of mutations that are purely harmful.

Yes, I understand the value of making germ line engineering conditional. That even has neat therapeutic applications in humans. Aubrey de Grey proposes using gene therapy (once it becomes possible of course) to put xenohydrolases into human cells that only get turned on once every 5 or 10 or 20 years to clean out lipofuscin and other junk that accumulates. But I'm arguing that we will be able to make significant alterations in average personality outcomes without using conditional mutants.

Germline genetic engineering in animals in 20 years: You really think it will take that long? There are guys at ORNL who are working with nanofibers to deliver genes directly into the nucleus. Granted, that doesn't get the genes safely integrated into chromosomes. But suppose it takes 20 years in animals: will it take even 10 more years to replicate that in humans? I figure within that time frame personal DNA sequencing will be cheap. So we will have the data to use to identify the effects of many thousands of genetic variations.

godlesscapitalist said at June 7, 2003 7:03 PM:

Amy -

You say:

My main thesis on this subject is this: The facts that 1. most traits are controlled by many genes, 2. individual genes have many functions, 3. compensation and redundancy occur during development, 4. behavior arises from the network (the network part is important) activity of neurons, which is affected BOTH by inherited genetic factors (usually many) and by EXPERIENCE, means that germ line engineering of traits, especially ones related to behavior, will be extremely difficult AND will often be dangerous.

So since we all agree that the brain is a biochemical system, and that there is no ghost on the machine, what you are really arguing about is whether we will have fine-grained enough tools to manipulate behavior consistently - and not just to initialize genes. As Randall says (and I agree), I think that you underestimate the rapidity at which tech is going to develop. For example - things like artificial chromosomes will be really big in terms of introducing large numbers of traits in a packaged way (kind of like an integrated circuit).

More importantly, I think the barriers you bring up are ethical rather than technical. If you're the Chinese government, say, and willing to screw up on humans like you screw up on mice - you can make very rapid progress indeed. The first step would be to make a lab strain of humans with short generation times, probably inbred and mentally retarded so that they don't escape. Improvements in cloning technology would then allow truly controlled experiments on humans. Of course that's horrific by Western ethical standards, but as our species has sanctioned slavery, communism, nazism, and plain-ole-non-ideological mass murder - it is definitely not out of the question.

Second - as far as I understand it, your objection is that pure genetic engineering is kind of like open loop control - all we're doing is initializing the genetics and letting it fly, and doing this lets downstream stimuli (e.g. Experience) have a substantial impact. The problem with this notion is that you can program negative feedback control into genetic systems (as I'm sure you're aware). You probably also know about the interest in systems biology; that happens to be my field, and if you read papers by Leibler, McAdams, Arkin, Hasty, etc. on PubMed you'll see that we're improving our ability to simulate mesoscopic scale genetic control circuitry.

So the point being that you can make downstream reactions insensitive to input stimuli given good enough control engineering. Even if you don't believe that it is possible to program sophisticated feedback control into genetic circuits, I think you have not considered the possibility of implanted circuitry working in concert with biology to damp external inputs (or directly alter internal chemical levels). Crude example for the sake of argument - a device implanted on the back of your neck administers an imperceptible shock whenever your adrenaline levels get too low, keeping you in a constantly mildly agitated state.

amy greenwood said at June 8, 2003 9:12 AM:


Scheduled toxin cleanup...that's a cute idea. Maybe we could set it to happen while we're sleeping like hard disk maintenance.

But as far as even basic germline engineering to fix even single-hit diseases, yep I think 20 years. Technical considerations: first, to be safe you want to make the fix by homologous recombination rather than by random transgenesis to avoid inactivating any important gene. So, you introduce the change in ES cells, which is easy. (And let's say for convienience's sake that you don't need to fix both copies of the alelle.) But now you have to make these ES cells into a whole person, whereas normally you'd allow them to contribute to a chimera. That can be done by tetraploid complementation, but that means that any epigenetic problems or chromosomal instability that has occured in the ES cells can cause problems for the development of the child. In mice, we don't much care, so long as that first generation can mate. Its not a risk I'd care to take with any child of mine. Since I think we're not nearly as far along at understanding epigenetic control of gene expression, even at analyzying it on a genome wide scale, I think we have a ways to go.

Randall Parker said at June 8, 2003 1:16 PM:


I agree that homologous recombination is necessary for doing germline gene therapy. I've long thought that the researchers trying the random insertion approach are being excessively optimistic and wasting their time when they should be working on the homologous recombination problem (or at very least on artificial chromosomes). I wasn't surprised when those folks in France found leukemia cases in the small group they treated for some other disorder (some sort of immune deficiency if memory serves) when they used random insertion of a gene therapy.

However, artificial chromosomes will work for some applications. Recently some researchers showed in an animal model (mice? in Ohio somewhere? I should have posted on it) that they could do RNAi with just a single nucleotide difference in the target they wanted to shut down. So in theory artificial chromosomes could be used both to up-regulate (just add more) and even to down-regulate.

One problem with artificial chromosomes is whether they can be made to reliably replicate and separate into two cells in mitosis. I wonder if anyone has tried to do that yet.

Artificial chromosomes will probably first be used to deliver gene therapy to adult bodies. It ought to be useful in post-mitotic cells if only some way can be developed to get the artificial chromosomes into enough cells. Take the example of the Aubrey de Grey's proposal to putting xenohydrolase genes into cells to to clean up the accumulated waste. If one wanted to do that in brain cells one would have to have a way to reliably get the artificial chromosomes into a large fraction of the cells. But one might get many copies in some cells and none in others. It seems a hard problem.

As I see it, the ability to do gene therapy with homologous recombination and the ability to cheaply sequence individual DNA will be the two biggest enabling technologies for human germline genetic engineering. My guess is that the DNA sequencing cost problem will be solved first.

Granted, there is the epigenetic info problem. But that might be side-steppable if we can reliably convert adult cells into germ line cells. Basically, do the gene therapy to adult cells, get them to go thru meiosis, then do fertilization. We may be able to use the conversion into a meiotic replication cycle to set the epigenetic info to the right state (after all, that has to be what happens in the body). We might get lucky and find that there is a single gene (or even a small group of genes) that can be regulated to cause meiosis. Think of this theoretical gene as the logical equivalent of the recently discoverd nanog gene which appears to make cells become stem cells.

I see all the work going into lowering the cost of DNA sequencing and the predictions of Venter and others of below 1000 dollars per person DNA sequencing costs (Venter is saying in less than 10 years) and am thinking that problem is going to get solved. There are enough venture capital start-ups and university labs making encouraging noises. I have a bunch of posts about US Genomics, Solexa, and other efforts to do this in my Biotech Advance Rates archive. These folks see an engineering problem to solve and are going at it with a number of approaches. Also, I presume you are aware of the efforts of Daniel Branton and Harvard Nanopore Group as well. There are other labs working on the nanopore approach too.

So to do germline genetic engineering I come down thinking the rate-limiter is going to be the dfficulty of developing a technique to do controllable homologous recombination. Seems like as part of the solution we need enzymes that can be reprogrammed to match particular target DNA sequences for where one wants to replace the existing sequence. But there are many more parts of the puzzle as well. For instance, to even get to a site to do the recombination one may need to get a chromosome to change shape to expose a target area. I agree that homologous recombination control is going to be a hard one to solve.

Bob said at June 9, 2003 4:26 PM:

Unwise and malicious? Heck, we need to worry about the unintended and inevitable.

You know, I used to know this guy who had "Beware the law of unintended consequences!" in the sig line of all his emails. ;-> I guess it kinda rubbed off after a while.

amy greenwood said at June 9, 2003 8:45 PM:


You lost me. Why is homologous recombination hard? I was imagining the standard drug selection scenario in ES cells. (Even so, I still think that the risk of using such cells to make a whole person is huge.) Are you thinking about recombination within an intact fetus?

BTW, I estimated 20 years until even the most basic form of engineering for diseases caused by single mutations. And my gut feeling tells me that's too soon, but I acknowledge that unexpected jumps in technology can happen. I definitely don't think that genetic engineering as a technique will be worked out or common in 20 years. I think 20 years is a minimum...


While its true that inhumane state sponsored medical experimentation has existed, such programs are usually directed at a punishing someone or using people who are "dispensible". Since the people who will suffer from tinkering with genetic engineering are CHILDREN, and since the goal would be to enhance those children, I find it hard to imagine such a state sponsored program, even on a guinnea pig group of kids. Also, since I've already said that I don't think it would work for the kind of enhancements that you and Randall are talking about, I can't see that there will be much incentive. Same with cloning. Fringe technique.

I like the idea of artificial chromosomes, but even if such a technique existed, its not an easy thing to make a genetic circuit that makes a predicable and subtle change. I mean, even in Drosophila, no one is on the record being able to engineer in, say, a 10% increase in leg length without other effects. Or make the Ubx flies that look like butterflies. Of course they can screen for these things, but that's not at all the same as engineering. When I see a fruit fly with tiny colored spots on its wings, then I think we'll have entered the age of bioengineering.

Randall Parker said at June 9, 2003 10:21 PM:


If you want to insert genes to replace the specific spots where the original genes are located and you want to do that for many genes then that seems much harder than putting the new genes in at some standard location chosen as the place where all new genes would get added. I'm thinking of the problem of how to go around and replace literally dozens of different genes with other variations of those same genes. How do you test for the desired result when so many genes are going to be substituted? What are the odds that many different changes will all take?

I'm thinking that people will want to change multiple sites in order to get some desired change in intellect or personality since I expect multiple sites would need to be changed to, say, raise IQ by 20 points or make changes to guarantee getting an extrovert rather than an introvert.

Intact fetus: well, growing up large numbers of ES cells, doing gene therapy to them, and then trying to select among them to eventually choose one to try to make become an embryo has its own problems. The cells are no longer really back at the epigenetic state they were right after the fertilized egg first started dividing. It seems to me there is an epigenetic state advantage of doing the gene therapy to the fertilized egg. But then you need a really reliable mechanism for getting the genes to replace the genes you want to replace. IOW, I'm not thinking that the way it is currently done is the optimal approach in the long run.

I actually expect people will want to change literally hundreds of genetic variations in their offspring once the meaning of enough variations become known. People will want to eliminate purely harmful variants. They will want to avoid variants that predispose for various diseases. They will want to go for improvements in looks, intelligence, coordination, and other qualities.

I'll fix my comment about your 20 year period estimate.

Your drosophila example with GC: people will known how to increase leg length in their offspring or other traits because scientists will be able to compare short and tall people to identify which genetic variations contribute to height changes. The human race has such a large number of existing variations that many people will be able to find what they want for their children by using existing variations.

Edward A.Greenhalgh said at June 16, 2004 9:35 PM:

the real danger the public over looks with GE is the failure of government to trtuhfully test the products and to have good standards. Please see my website www.cancerfraudbadbiotech.com as it outlines just how far government will go to cover up fraud . There are dangers the public just does not consider. Please note the fmr US surgeon general regretted my treatment and Noam Chomsky commented it may be of phenomenal importance. Thank you. E.A.Greenhalgh

Edward A.Greenhalgh said at June 16, 2004 9:35 PM:

the real danger the public over looks with GE is the failure of government to trtuhfully test the products and to have good standards. Please see my website www.cancerfraudbadbiotech.com as it outlines just how far government will go to cover up fraud . There are dangers the public just does not consider. Please note the fmr US surgeon general regretted my treatment and Noam Chomsky commented it may be of phenomenal importance. Thank you. E.A.Greenhalgh

g. salter said at June 28, 2004 6:12 PM:

It does not have to be a communist collective that rules over the GE masses, it could easily be a right wing govenment or a dictatorship that controls the obedient masses and makes money off their labours etc....or why couldn't it be a group that leads by consensus (no leaders), or the "leaders" could be freindly AI's that are evolved or programmed to serve for the better good and therefore have no greed or self-centered biases?

Post a comment
Name (not anon or anonymous):
Email Address:
Remember info?

Go Read More Posts On FuturePundit
Site Traffic Info
The contents of this site are copyright ©