Beyond “I’m Sorry”: How Apology Works
What psychology, evolution, and computers tell us about mending fences
A moment from school has remained with me for years. I’m seated near the auditorium stage, where visiting primatologist Frans de Waal stands beneath projected photos of chimpanzees in the wild. We see two furious male chimps who have just separated to avoid violence. They sit 20 feet apart, hunched on the forest floor, staring in opposite directions.
As de Waal tells us the story, in the photos we watch an older female approach one of the males and cautiously begin to groom him. Because of her age, sex, and status, she may be the only member of the troop who can safely attempt this. She sits in his space, picking at his fur, calming him, focusing on him, making him feel appreciated and safe. He resists at first but she gradually wins him over. He begins to groom her in return. She leads him to the second seated male, whom she likewise begins to groom. Grudgingly, the first male joins her in this. And as the older female withdraws, the two males begin to groom each other.
During the brief gaps in Dr. de Waal’s narration, the few dozen students around me sit quiet and still, as if holding their breath. In my memory, we are all fascinated and, in my case, very moved. How could something so human, sophisticated, and social take place among wild animals? Driven by emotion, with apparent understanding of each other’s feelings and roles, these chimps worked together to resolve a difficult impasse and save the health of the community. It actually looked a lot like wisdom. And they did it without saying a word.
Rituals like reconciliation and apology may seem like invented customs written into the social contract by Miss Manners or a kindergarten teacher. Parents make us “say you’re sorry”. Coaches teach us to shake hands. Clergy and therapists encourage us to apologize and forgive. It would be reasonable to assume humans created all these things.
But science has spotted similar behaviors in multiple animal species, and even in computer programs that interact with other programs. It’s possible that forgiveness and reconciliation are part of the nature of the universe.
In this article I’ll use the findings of psychology and game theory to sketch the architecture of apology. What are its essential elements? How do they work together? If you’ve ever had an apology rejected, or if you have a big apology waiting somewhere in your future, I hope this will help.
The Forgiveness Instinct
Evolutionary psychologists study behavior in terms of how it helped us survive over millions of years. To them, punishment and forgiveness are complementary instincts that helped protect individuals and maintain a healthy community. This was true long before we developed codes of law, or even language.
We hope that our rational mind is involved in punishment and forgiveness— for instance in the justice system. But in many cases the conscious mind simply looks for ways to justify our instinctive emotional decisions. This is why our better thinking does not always prevail, why mobs sometimes rule, and why it’s hard to “argue” someone into forgiving you. Punishment can be about tough love or righteous justice, but sometimes it’s just rage. Forgiveness can be a lofty spiritual accomplishment, a noble demonstration of reason and character, but in many cases it is simply the product of feelings like empathy, shame, loneliness, or the pain of interpersonal conflict. We tend to do what feels right. As the saying goes, “Know forgiveness, know peace; no forgiveness, no peace.”
The Evolution of Cooperation
The value of the forgiveness instinct, oddly enough, was demonstrated in a famous series of computer competitions in the late 1970s. Political scientist Robert Axelrod invited people to submit programs that would compete one-on-one in a simple game called The Prisoner’s Dilemma.
Prisoner’s Dilemma simulates 2 people who have been arrested for a crime. If neither confesses, each serves 1 year. If one confesses and testifies against the other, the one who confessed goes free and the other serves 3 years. If both confess, they both serve 2 years. The best total outcome is if both players keep quiet, but it’s risky, because the partner might betray you. In fact, in this game betrayal is probably the logical thing to do, making it very difficult to trust the other player.
The game was invented in the 1950s as a model of the nuclear standoff between the US and Russia. The best total outcome would be for both sides to disarm, preventing expense and danger. Or, if they built up weapons, the best total outcome is for neither side to ever launch. But if the other side arms itself, you will wish you had, too. And if the other side ever launches, you will wish you had launched first and disabled some of their weapons. It’s almost surprising that neither side ever launched.
Nuclear war was primarily viewed as a one-time thing, and so was the original Prisoner’s Dilemma. But in Axelrod’s contest, each program would play every other program 200 times. Being forced to continue playing with your opponent turns the 2 players into a community, and that changed things. In his book The Evolution of Cooperation, Robert Axelrod put it this way:
“What makes it possible for cooperation to emerge is the fact that the players might meet again. This possibility means that the choices made today not only determine the outcome of this move, but can also influence the later choices of the players. The future can therefore cast a shadow back upon the present and thereby affect the current strategic situation.”
If you defect on me in round one, I may defect on you in round two. But that could turn into a permanent cycle of retaliation, leading to the worst possible score for both players. As Gandhi said, “An eye for an eye leaves the whole world blind.” Without forgiveness — the willingness to trust again after a period of deserved distrust — the game would grind to a halt. So in Robert Axelrod’s contest, it turned out that the most successful program, the one with the highest total score, was the one that always started by trusting, rewarded bad behavior with punishment, but trusted again as soon as the other player behaved decently again.
This is why I wondered if cooperation is built into the universe. It may turn out that wherever life emerges across the billions of galaxies, the species that cooperates is more likely to succeed. Maybe. Who knows.
Anyway, that’s the rule on the planet where I come from. Humans (and other animals, such as dogs) have a strong instinct that allows them to forgive wrongdoing — in their young, in those close to them, and to some extent even in total strangers. Without some kind of forgiveness, there could be no society. So the good news is that forgiveness is not rare, but is actually very common. It happens around us many times a day. To err is human — and therefore, so is forgiveness.
If you will never see someone again, then from a survival standpoint it doesn’t matter whether you forgive each other. This is the logic of giving people the middle finger from a moving car, or insulting a stranger on social media. But if you know you‘ll see that person again, you might act differently. For the tribe to survive, you might have to. And somewhere along the way, evolution figured that out.
Virtually all human cultures have rituals of reconciliation, including public apology, gifts, compensation for injury, animal sacrifice, religious rituals, mediation, and third-party intervention. Not all wrongs are forgiven, and not all forgiveness leads to reconciliation. But the good news (for animals and humans) is that reconciliation often leads to relationships that are stronger than they were before the offending incident.
Reconciliation has also been observed in other species, including wolves, dogs, and at least 17 primate species. Is it wrong to use human terminology like forgiveness and reconciliation when talking about animals? Should we use terms that are less intentional, or imply less emotion? As Frans de Waal points out, we happily use human words like aggression, violence, and competition when talking about animals. To use human words for violence, but not peace, would suggest a mental bias.
The more the group needs cooperation to survive, the better they seem to be at reconciliation. One study found that wolves, which rely on each other for hunting, are better than domesticated dogs at reconciling after a conflict. Was that genetic, or did life in a wild pack simply reinforce the reconciliation instinct better than life in the suburbs? Frans de Waal tested this with a group of rhesus monkeys, which are socially hierarchical, violent, and not known for peacemaking. When put in with a group of naturally more relaxed stump-tailed macaques for a few months, the rhesus monkeys became much more tolerant and peaceful with each other, and they continued the behavior after the stump-tails were removed. So perhaps peacemaking is a genetic trait that can be brought forth with practice in a supportive environment.
One key discovery around the same time was that forgiveness is more likely when a relationship has value. Long-tailed macaques were placed in an environment where synchronized operation with a partner enabled them to get better food from a special machine. When disagreements broke out, macaques in this group showed three times more reconciliation behavior than macaques in a normal environment. Researchers called this strategic reconciliation. It’s not that the reconciliation was somehow less real than the anger it replaced. The emotions were not fake, any more than the love of a child who depends on its parents. The reconciliation was real, but might not have taken place in a normal environment where the partnership was not so valuable.
And that gives us an important clue about how to apologize.
The Decision to Forgive
Psychologist Michael McCullough tells us that forgiveness depends primarily on three factors:
- Value: Forgiveness is more common when the relationship has value — otherwise there is no motivation to forgive. We are more likely to forgive a family member, co-worker, or close friend, because the value of the relationship outweighs the risk of forgiving and being hurt again. However, if you hurt someone close to you badly enough, they may come to perceive you as permanently worthless.
- Safety: It makes more sense to forgive in cases where the harm will not be repeated. For instance, it’s easier to forgive an accidental injury than an intentional one. And it’s easier to forgive someone who has clearly changed (perhaps after many years), or a criminal who is safely behind bars, than someone who roams free and is unremorseful and likely to harm us again.
- Empathy: It also helps if we feel that the transgressor deserves our care. Perhaps we have done something similar, or see their side of things, or empathize with the suffering they’ve experienced as a result of their harmful act. In fact, punishing someone can sometimes arouse the empathy that lets us forgive them.
In some cases, simple fatigue can help people decide to forgive — because they’re just tired of the cost of hanging onto something, sometimes after many years. I’d put this in the Value category.
Now, you could say that religion provides morality based on higher law, which might lead people to forgive even when it is the hardest thing they could possibly do. That is sometimes the case. But religion has also justified many punishments and executions over the years. So we’ll leave that door closed for now, to prevent it from making the article much longer.
Since forgiveness hinges on being safe, valuable, and care-worthy, we try to show these qualities when we seek forgiveness. We use three standard signals used for this purpose:
- Gifts or compensation suggest safety and create value, and may create empathy if given directly by the offender (rather than through a third party). Studies show that victims are often satisfied with much less than the actual amount of the damage, suggesting that the payment is not as important as the gesture itself.
- Self-abasing displays and gestures suggest safety and create empathy. Dogs cower and beat their tails, apes bow low and make specific noises, and humans may tremble, blush, or weep. We often rush to forgive someone who is obviously embarrassed or ashamed. An elaborate public display of shame and remorse may let the victim enjoy an equally public display of graciousness — or may backfire, publicly.
- Apologies suggest safety, create empathy, and suggest a commitment to creating future value.
The Anatomy of an Apology
Speech is an action. It can answer a question, save a life, pay a debt, connect, wound, heal. And an apology, even though it doesn’t seem to change anything, can change everything. It helps the offended to see the offender suffer, even if it’s only by squirming, and to be reassured of future good intentions and a shared view of what happened.
Studies of apologies in different languages and cultures give us similar models, which tend to contain three elements:
1. The apology signal: “I’m sorry for X…”. This has three elements:
- Acknowledgment that an offense took place — something bad happened — without minimizing the harm or pretending that it didn’t happen. This creates safety by showing respect for the victim’s feelings, and that the offender can see things from the victim’s point of view and is not playing by a different set of rules.
- Admission of personal responsibility. The offender must not dodge the blame. “I did something wrong; I hurt you.” If a person can’t see his own fault, he can’t be trusted in the future. Some apologies include a promise to “never do it again”.
- An expression of emotion: regret, remorse, shame. A negative view of our own behavior is what distinguishes an apology from a boast. It should be humble, sincere, and at the right level — neither too cool nor too dramatic, and not overshadowing the suffering of the victim. The main thing is for the offender to show empathy and respect for the victim, and regret for his own behavior.
2. An optional explanation that does not attempt to “explain away” the offense. It may reassure the victim that the harm was not intentional, offer some detail or insight that enables empathy, or suggest a “loophole” or excuse to forgive. (“I was under a lot of pressure.”) The trick is to do this while still showing remorse and accepting responsibility for the harm that was done.
This is not the same as reframing or re-describing what happened in words or images that are more favorable to the offender. This is like negotiating the history of the event: I’ll agree that X happened, if you’ll agree that Y happened. This can work, but it’s gone beyond simple apology and into denial or renegotiation of history. In extreme form, we call it gaslighting. That’s not what I’m talking about. I’m just saying the person may have an easier time forgiving you if they know you were late because of a flat tire.
3. An optional offer of compensation, phrased as a question (“What can I do to make this up to you?”) or a proposal. Compensation graphically shows that the offender values the relationship and is willing to suffer for it. This demonstration of affection or respect creates a sense of safety, empathy, and value.
So an effective apology is not just a passive verbal request for forgiveness. By actively putting the necessary elements into place, the act of apology engages social instincts that are millions of years old, and it helps make forgiveness possible.
A Final Thought
If you’re sitting on a fence instead of mending one, I offer this idea from a 2008 interview with Michael McCullough:
“In lots and lots of cases, forgiveness is just a conversation away. I mean, there are so many people, if you ask them about the hurt that they remember from junior high or high school, what you often find is there was never any conversation back with that person who harmed them. And so the conclusion I’ve come to is, in many, many cases, if you want forgiveness — if you want to forgive or if you want to be forgiven — you need to go out there and get it for yourself. And the way you go out and get it for yourself is by trying to have the kind of conversation with the person you hurt that you want to have… So much of forgiveness comes down to interaction.”
Sources & Suggested Reading
- Good Natured: The Origins of Right and Wrong in Humans and Other Animals by Frans de Waal (the visiting professor from the start of this article) describes the evolution of personal and group morality, especially among primates.
- Beyond Revenge: The Evolution of the Forgiveness Instinct by Michael E. McCullough was so good I took notes as I read. McCullough has taught and researched social psychology at University of Miami and now UC San Diego. He made his topic scientific while also keeping it human and beautiful.
- Prisoner’s Dilemma by William Poundstone is the fun story of game theory and its roots with John von Neumann and the remarkable RAND Corporation at the start of the Cold War. Game theory has applications in economics, sociology, politics, biology, urban design, and a million other things, and leads quickly to the emerging science of complexity theory.
- The Evolution of Cooperation by Robert Axelrod is the classic book on the automated game-theory experiment that accidentally showed the competitive value of forgiveness, and perhaps the inevitability of cooperation. Two of my all-time favorite authors (Lewis Thomas and Douglas Hofstadter) wrote excited reviews when it came out, and a delighted Richard Dawkins wrote the current foreword.
- Freakonomics: A Rogue Economist Explores the Hidden Side of Everything covers a lot of topics, but in the chapter “Where Have All the Criminals Gone?” the authors explore surprising aspects of punishment as a crime deterrent.
- Why Is It So Hard to Apologize Well? by Lisa Belkin This short NY Times Magazine article on public apologies, printed on the day I originally posted these notes, is fun for its examples of the principles above.
- On the subject of why it’s sometimes hard to apologize at all, cognitive dissonance theory provides some answers. For a mind-blowing introduction that may change your relationship with your own memory, I often recommend Mistakes Were Made (But Not By Me), written by psychologists Carol Tavris and Elliot Aronson.
- Reconciling with Valuable Partners by Long‐tailed Macaques,
Marina Cords and Sylvie Thurnheer 1993 https://doi.org/10.1111/j.1439-0310.1993.tb01212.xCitations: 58
- The effect of domestication on post-conflict management: wolves reconcile while dogs avoid each other, Simona Cafazzo, Sarah Marshall-Pescini, Martina Lazzaroni, Zsófia Virányi and Friederike Range
Published:04 July 2018