Where do morals come from

A common argument made by religious people is that if God didn't exist and we were just products of evolution by natural selection, then we wouldn't have the morals that we do. According to this argument, evolution is about survival of the fittest, so we would all be selfish and wouldn't care about right or wrong.

I'm not going to respond to this argument directly in this essay. Instead, I'm going to explain why morality is not actually that surprising to people who study evolution by natural selection, and that it's expected to arise in certain circumstances. I'll do this by providing some evolutionary groundwork and giving you thinking tools that you can use to reconstruct our behaviour. I recommend that you read everything in order, because all the points build on each other.

Teleological Language Is Ok, Sometimes

This isn't an insight into evolutionary theory, but it's a way of talking about it. Often talk about biology and evolution is told with the language of purposes and goals. This is not because they are designed or anything like that, this is just shorthand. It's easier than saying something like: "the feature and mechanism that maximised the propagation of this gene."

Chess players often say things like "the bishop is x-raying the rook" or "the knight is supporting the pawn", not because the bishop is actually firing lazer beams or because the knight is literally holding up the pawn, but because explaining things in terms of relative material value after exchanges is far less intuitive than explaining things in terms of common-sense physics. And humans are good at understanding intuitive physics but also intuitive psychology and intentions. Incidentally, this is also why chess Grandmasters say things like "what would make my knight happy?" and "what is this bishop trying to do?". They really mean something like "where would the knight have access to more squares" and "what lines is my opponent hoping to reach by bringing his bishop here?". This takes longer to say, and takes more work to understand, so they opt for the more intuitive but less accurate metaphor, and that's fine as long as you don't think the chess pieces actually have hopes and dreams.

Where this can get murky is when talking about the evolution of behaviour. This is especially murky when talking about humans since humans do have intentions and goals.

We can evolve something for a reason, without actually being conscious of that reason. We evolved to like sex because it leads to reproduction, but when people are either having sex or fantasising about it, children are the last thing on their mind.

I'll use teleological language in this article, but sometimes I will have to remind you not to take it too seriously.

Evolution Is Gene Focused

Some people will tell you that evolution is for the good of the species. Sometimes they use this to explain moral behaviour. This is Kumbaya bullshit. The first problem with this is if altruistic behaviour is an inevitable consequence of evolution, they why does it vary so much between different species? Group selection also suffers from a free rider problem. If a gene mutates and becomes better at propagating itself at the expense of the rest of the species, then that happens. There are a few mathematical scenarios where group selection can work, but for most species this is not the case. Evolution is not about the species. This is not why moral behaviour exists. Evolution is also not about the individual organism "surviving". No matter what it does, the organism will die. It's the information encoded in the genes that lives on. This is what is selected for.

Often this aligns with the prolonged survival of the organism - since the organism will make more copies of the gene, but there are conflicts. For example, genes that increase fertility during an animals youth can have deleterious effects during old age when the animal is less fertile. Motherhood is not ideal for the mother, but it does help spread her genes.

Evolution Deals With Causal Chains

Companies want to make money. But when they tell their employees what to do, they don't just ask them to make money. They tell them to do specific things that eventually result in money.

In a similar way, evolution does not encode its "goals" directly in genes, it encodes information that goes through a lot of chemical and bio-mechanical intermediates.

Imagine being inside a tank. There's no button that says "destroy your enemies". Instead there's a button that says fire. But even this button does not work by magic. Pressing the button connects a circuit which connects an electric charge which pushes a mechanism which ignites another charge which increases the pressure behind the projectile and causes it to launch in the direction the tank gun is aimed at. Aiming it at the target works by a similar sequence of mechanisms. (I have no idea how tanks work, this is just an illustration).

All that's needed is that the initial effect of the gene reliably leads to the final effect that the gene is selected for, at least in the environment that it developed in.

Humans like sex because sex reliably leads to reproduction. With contraceptives, that isn't true anymore, but we didn't evolve to only like sex if it leads to reproduction.

Evolution Deals With Tradeoffs

When we evolve body parts, we need to use resources to form them. Food, that's hard to get. There's an opportunity cost for that. Devolopment in one area is devolopment that could have been done in other areas.

Thinking of the body as an engineered system (without an engineer), there are other kinds of tradeoffs. For example, the square-cube law. This states that the surface area grows proportionately to the square of the length, but the volume is proportional to the cube.

Crudely, the volume is how much stuff you can fit in, and the surface area is how much packaging you need. This has implications for how the ability of something to fly varies with its size, and on heat transfer, and on how much oxygen it can take in.

This means that most of the horror movies that deal with animals being scaled to monstrous proportions are nothing to worry about. But it also means that humans with wings can't fly.

Biological Attributes Are Not Useful In Every Context

Something like flight or intelligence seems useful. But it's not always. You have to make a lot of changes to a body to make it fly, if it's even possible. You incur a lot of costs. And the payoff might be smaller than the cost. A similar thing is true of human like intelligence. Our brains account for 20% of our energy use, and 2% of our body weight. Our enlarged heads make childbirth a lot more risk. We have to be born early to fit at all. This means we're born relatively useless, compared to, say, horse foals that can walk a few hours after birth. Compared to other primates, human pregnancy should last between 18-21 months. Most creatures are doing fine. There's little incentive to become like us. But in certain niches, it is advantageous to evolve more intelligence or flight or whatever it is.

A squirrel that makes its living by jumping between branches to find its food can catch more food by jumping a little further and live to reproduce more. So there's an incentive to jump further and further. To become more aerodynamic. And so you have flying squirrels.

But because we don't get much advantage out of jumping between tree branches, we're not subjected to the same evolutionary pressures.

And, as I'll argue later, a similar thing is true for morality.

Behaviour Is Also Evolved

Just like the size of an animal or its teeth or its colour evolves, so too does its behaviour. The way it treats its kin, or the way they interact with members of the same species. Or how they interact with other species. Animals move, and that's behaviour.

Kin Share Genes

Following from the gene-centred view of evolution, if we evolved to care about our selves because it reproduces our genes (even though that's not what we're thinking about), then what about taking care of our children who share our genes? Or our siblings and parents?

In many species, siblings share 50% of their genes with each other (relative to the baseline for their species), parents share 50% with their children, grandparents 25%, and so on. And in general this corresponds quite well to how concern scales for family members.

JBS Haldane is quoted to have said: "I would gladly give up my life for two brothers or eight cousins".

For eusocial species like ants, something very interesting happens. You see, not all species have the 50% relatedness between siblings I was just talking about. In haplodiploid species, the males come from unfertilised eggs and the females come from fertilised eggs. Daughters share 100% of their fathers genes and 50% with their mothers (and 75% with each other). This means that sisters are more related to each other than their own children. (W. D. Hamilton calls them supersisters). The brothers often don't reproduce (the workers), but instead help their sisters.

It is often said that it is easier to understand ants if you treat them more like cells than individual organisms.

I think it's relatively clear why this should motivate some self-sacrificial behaviour and concern for kin.

But there is one unresolved question: how do they recognise kin? Remember how I said that evolution works through causal mechanisms. Animals can't psychically know that another is their relative, there has to be some sort of mechanical process that starts with an animal "looking at" another, and "knowing" whether it is a relative, or at least treating it like one. And it's possible that the assumptions made when that evolved no longer apply. For example, if we no longer live in small towns where most people you know are related to you in some way.

Kin recognition does not have to be perfect, it just has to be good enough to avoid being too altruistic, and avoid inbreeding (which leads to resessive gene issues).

There are a few different strategies that seem to be employed in nature, from imprinting to cue-based mechanisms. Humans have a mixture of strategies, who they spend time with when they are young is one of them. The army and cults often exploit these mechanisms to make people feel like brothers and prepared to sacrifice themselves for each-other.

This should account for a significant amount of altruistic behaviour, even towards non-relatives.

Cooperation Is Often Rational

Sometimes animals run into conflicts. Why? Because they're competing over limited resources.

Plants lie there and wait for their nutrients to come to them. But even plants grow taller to get sunlight and block other plants from getting it. Plants also often develop sophisticated defense mechanisms to stop animals from eating them.

Animals, on the other hand, have to go out and find food. Sometime that means plants. Sometimes that is other animals. And sometimes they have to avoid getting eaten by other animals. And one other problem they have to deal with is competition for the same food, either by other species or members of the same species.

So what should an animal do when faced with another member of the same species looking for the same food? Should it co-operate or should it defect? To a first approximation, it seems rational to defect. Just steal things. Don't be fair.

The problem is that other agents don't like that. It's in their interests to defend against that. And so animals are stuck with the age old conundrum. Do they get better at defecting, or start co-operating. The problem is that conflict is costly. Even if you get a bit stronger, a bit more aggressive, and have a good chance of winning a conflict, your opponent has to be just good enough that the cost to you outweighs the benefit.

Game theory is a branch of mathematics that deals with questions like this.

A classic example of game theory is the Prisoners Dilemma.

Two people are arrested. They're in solitary confinement so they cannot communicate with eachother. They're each given the same choice:

• If they both betray eachother, they each get two years in prison.
• If none of them snitch, they each serve one year.
• If one snitches but the other doesn't, the snitch goes free and the other goes to prison for two years.

According to Game Theory, the dominant strategy is for both of them to snitch and serve two years. This is because, no matter what the other person chooses, you can always improve your position by defecting. Even though cooperating looks better for everyone.

However, the analysis changes if you run this game more than once for the same people. This is called the iterated prisoners dilemma. Once you introduce memory, the best strategy is actually to cooperate and punish defecting - tit for tat. Out of all the programs that were entered into a competition instigated by Robert Axelrod, this is the one that won. This was the shortest and least complicated program.

The programs that did the best in the long term were the least greedy ones, judged by their self interest.

When the situation was modified to include the possibly of signal error - being wrong about who you think defected, the winning strategy was "tit for tat with forgiveness", which is where you sometimes cooperate with someone that defected. This prevents conflict escalation over a mistake.

In general, Axelrod found that the winning programs were nice (they did not defect before their opponents), but they retaliated when their opponents defected, and they tended to forgive once their oppents stopped defecting. They were also non-envious.

This indicates that ideas about justice aren't merely the dreams of impractical naive ideologues, they actually are in your own interest.

The dominant strategy is also known as the evolutionary stable strategy.

The effectiveness of a strategy depends on what strategies the rest of the population is using. A strategy might be very effective when it is the minority, but ineffective when everyone has that same strategy. And it can ossicalte between those points and settle at an equilibrium point. These strategies are not necessarily always the best no matter what, they're just ideal in certain contexts and in response to other strategies that exist in a particular environment.

Male Female Split

There's one important difference between males and females that's responsible for most other differences between them: reproduction is a lot more costly for the female. Eggs are much larger than sperm, and take more resources to make. Then there are the obvious resource costs and other issues related to pregnancy. It's also a lot easier for a male to ditch a pregnant female. Whilst a female is pregant, a male can keep reproducing with other females.

This means there is evolutionary incentive for females to be picky. There are two broad strategies for this. One way is preferring "good stock", males who will have children that are likely to successfully reproduce because they are strong and fertile. The other approach is selecting for men that contribute as fathers towards the rearing of their children. The first approach is the approach taken by "tournament species". These are the species that tend to have a few dominant males reproducing with most of the females, and males fighting for dominance. The second approach is taken by pair bonding or monogamous species.

In tournament species, there is a huge difference in size between males in females. In elephant seals, the male weights 4 times as much as the female. You can predict a lot about the social structure of a species by looking at the bones of a male and a female and comparing how different they are.

Humans are somewhere in between.

The pair bonding component gives males and females an incentive not to be cheated on. Not to be left if someone else comes along.

Nature's solution to that is love. And heartbreak. Heartbreak is an indicator that a potential partner is being genuine and won't cheat. Love isn't magic, it's a game theoretic solution to a strategic problem. But those cold strategic concerns are only for our genes. Our genes want us to mean it.

Paradoxical Strategies Are Useful

Suppose Russia has sent nukes to the USA. And the US can't stop the nukes. All they can do is send nukes in response and ensure mutual destruction. But that wouldn't help the US, they'll be dead either way. So they don't bother.

On the surface this seems rational. And indeed, if you were suddenly in that position, this would be rational. However, if you go back in time to before Russia sent the nukes, if they knew that America would react by doing nothing, then they send the nukes. But if they knew that America would "irrationally" retaliate even though it won't help them, then it would deter Russia from sending in the nukes in the first place. Thus, countries have systems to automatically retallaite with nukes and make it known. This is why many creatures have a strong, seemingly irrational desire for revenge. This doesn't mean that revenge is always rational. Suppose if someone accidentally stepped on your foot, and you dedicated your whole life, all of your savings, to getting revenge on this person. You can clearly get the message accross with far fewer resources than that. It's clearly not the optimal outcome for you.

Contracts Are Useful

This is something we all recognise when we make any sort of transaction. That buying something is better than no deal at all (although getting what you want without paying for it is best for you).

This is the case for both parties in a transaction. But neither party wants to be the sucker that gives without getting. This sets off an arms race. Getting better at cheating, whilst getting better at avoiding detection.

However, now there is a predicament. Suppose you're not a cheater. But you can't get into a contract because the person you're trying to get into a contract with doesn't know if you're a cheater. You have an incentive in signalling this in such a way that is difficult for a cheater to fake.

One way of doing this that nature has used is to involve body parts the brain has no conscious control over. Having conscious control over a heart beat, for example, is very risky. A lot of emotional displays involve things brains don't consciously control, and we're quite good at detecting them.

This is where genuine emotion comes from.

Humans Are A Social Animal

I don't mean that humans are all extroverts. This is a technical term among biologists, and refers to species that live in groups.

And just like flight, carnivorousness, and burrowing are common stragegies in certain niches that make other features more useful (e.g your diet co-evolves with your teeth and digestive system), being a social species creates a specific set of pressures related to what we might call morality.

Social animals work together to ward off predators and find food. This means that they are very attuned to other members of the same species (and occasionally other species).

This is why, as I will show, many species that exhibit moral behaviour are also social species.

Ethics In The Wild

Yawn contagion is strongly related to empathic behaviour in humans. Yawn contagion is one of many synchronisation methods among social animals.

There was an experiment conducted by Carter, Suchak, and de Waal (2011) on prosocial chimpanzee behaviour. This involved a chimpanzee choosing either a red "selfish" token, or a green "prosocial" token. The selfish token resulted in only that chimpanzee getting food, and the second resulted in a nearby chimpanzee getting food. Most of the time, the chimps chose the prosocial token. Interestingly, when the other chimpanzee spits water at them and tries to pressure them into choosing the prosocial token, the chimps chose the selfish token more frequently.

Capuchin monkeys, in another experiment involving de Waal, seem to care about fairness. I highly recommend this video. In this experiment, two Capuchin monkeys are paid cucumbers for completing a task. But once one of the monkeys gets paid in grapes, the other no longer accepts cucumbers as payment. The other experimenter, Sarah Brosnan, conducted similar experiments with chimps, and found that chimps from the same group refused the grapes until their fellow chimp was also paid in grapes.

Vampire bat mothers go on feeding trips. And when they return they often have more blood than they need, which they feed to the children of other bats. In an experiment where a vampire did not return with a lot of blood had its throat sac artificially filled with air. When other vampire bat mothers saw its inflated throat, they went on to not feed that bat's children since it didn't feed theirs because it didn't have enough blood to feed them but they didn't know that.

Hungry rhesus monkeys refuse to electrocute other monkeys for food.

Chimpanzees reconciliate after a fight, hugging and kissing. Bonobos famously reconcile with sex.

Chimps also console the losers of fights. Particularly if the loser was unfairly attacked by the dominant chimp. They tend not to console a chimp that lost a fight after provoking the other chimp.

Chimpanzees often adopt bereaved chimpanzees. Chimpanzees are dependent on their mothers for the first 8 years of their life. When a chimpanzee loses its mother, an adult male chimpanzee will often adopt it, seemingly with no benefit to himself.

Grief is widespread in the animal kingdom, to the point where “evolutionary thanatology” is being proposed as a new field of study. One famous case reported by Jane Goodal is of a male chimp who witnesses his mother die and stops eating and dies soon after.

It's not all good. Chimpanzees are also extremely xenophobic, straight up killing chimpanzees from other groups. But bonobos, our other closest relative, are much more interested in having sex with foreign bonobos.

A Recap

We have:

• Justice / Retalliation. Non-aggression
• Forgiveness
• Love
• Genuine emotion
• Concern for others

This list is by no means a full account of human morality, but I think it provides adaquate explanation for what many people think would not happen in a purely naturalistic world.

Some Implications

Really diving into the implications of this will require its own essay. But I have a few cursory ones to offer. The first thing is that moral behaviour in humans is neither an impossibility or an inveitability. It's something we're capable of but don't always do.

The second is that moral behaviour is more practical than people give it credit. If people, for example, stop believing in God and decide morality is not objective and that they should do whatever is in their best interests or whatever, that doesn't mean they will act immorally.

This might be why there seems to be a connection between responsibility/rationality and immorality. People that act in ways we consider immoral often aren't good at reasoning about the consequences of their actions, and often make their selves worse off along with everyone else.

Another implication is ethical AI. If this is the process by which we have our moral senses, then maybe there's some way of evolving that same thing in the conditions in which we eventually make our AGI systems.

Studying how we came to our morality and what our morality is seems relevent to utilitiarians. This is part of our utility functions. Whatever legal systems we end up recommending would have to take this into account.

I think we can extract some normative implications from this. Just because something is natural doesn't mean it is good. If someone has asthma for natural reasons, it is right to use "unnatural" means to live with it. But sometimes nature does have have normative implications. We should drink water rather than chlorine because that is what our body naturally needs.

If our moral sense is an evolved adaptation, then it's not clear why we should be very concerned with ethical philosophy - trying to justify morals and figure out its metaphysical nature. It would be like trying to justify why chocolate "objectively" tastes good whilst tar doesn't, perhaps in relation to its chemical structure. We don't need permission from a philosopher to eat and we shouldn't need it to act either. The role of the ethical philsopher should probably be like that of a food critic or food safety inspector. Or perhaps the ethical philosopher should try to codify our moral sense, after reading the scientific literature, and come up with a practical set of rules that can be known beforehand, and used to help us solve the problems we deal with.

Conclusion

Far from needing a supernatural origin, or being a soley human activity, morality is a perfectly understandable natural phenomenon.

I think that this sort of project should be conducted much more thoroughly - the identification of various evolutionary and game theoretic constructs that underlie morality, as well as in depth studies on humans and other animals to identify what our morality consists of. What I have done is enough to get started and begin working on implications, but it's just the start.