Sunday, January 1, 2012

Morality Part 3: Seeds of Morality

This is a multiple part blog on morality: [Previous]    [Next]

“Survival of the fittest” makes the natural world sound like an anarchistic, competitive, vicious, Machiavellian stab in the back.  This is born out by a glance outside.  Fierce competition, no doubt; animals kill and eat one another.  But is there also goodness in nature? 

There are so many ways organisms do seem to cooperate.  Ecologists refer to mutualism and symbiosis when both organisms benefit from interaction, co-operation when this occurs within a species, commensualism, where one benefits and the other is unaffected.  Phoresy is the use of one organism by another for transportation, inquilinusm for housing.  Interdependence is assisting others to reap indirect effects (such as donating anonymously to a political candidate).  All of these relationships, found throughout nature, might be construed by an anthropomorphic observer to be “moral” because in some sense they can be seen as good.   

But why do creatures do this?  One motivation is pretty straightforward: If a genetic mutation stimulates an organism to help its young that trait will likely persist in the genome, just as genes for indiscriminate infanticide will not survive for long.  Since it’s the gene, and not the organism, that matters, instincts to help siblings and cousins will be rewarded as well. Helping distant relatives is one step closer to a selflessness, as there may be many cousins about, but when a simpleminded (or no-minded) creature can't tell who is related to whom may discover that neighbors are often enough relatives too, and use nearness as a good proxy for kinship.  If this is so, the “nice-to-neighbor” gene combination may spread as well. In sedentary populations, this could happen.  It would look to an outsider as just plain niceness.

The kinship impulse is powerful, as any human parent can attest.  But acting friendly may improve an organism’s fitness another way: by status or reputation.  Whether it's achieved by bestowing benevolence or by demonstrating superior resources, status is a powerful currency that is paid back in both survival and procreation.  So one would expect status-seeking genes to survive as well.

Finally, one creature might help another by arranging (or expecting) reciprocal aid.    That is worth looking at more closely. 

The classic model of the Prisoner’s Dilemma goes like this.  Two prisoners are interrogated separately, and their sentences depend only on what they tell the authorities.  Each can accuse the other or not.  Each player knows the other one faces the same dilemma, but will not know of his choice in advance.    If both accuse, the punishment is harsh, and if neither accuses there is a light punishment for each.  But if one accuses and the other is silent, the accuser gets off free (while the accused is punished severely). 

Each can see that the short answer for self-preservation is this: No matter what the other does, it’s best to accuse.  So they both accuse the other and both avoid the most severe punishment, but they are each punished harshly, both forfeit the minor slap-on-the-wrist, and neither walks free.  This outcome is not optimal -- it actually seems like a recipe for back-stabbing. 

It's analogous to one organism coming up against another in nature where each must react immediately, either positive (including offering assistance, resisting impulses to attack, sharing resources, greeting, grooming) or negative (including abandonment, attack, thievery, threat, and hoarding).   If both are hostile, there is a standoff.  If both are cooperative they have a team.  If one is hostile and one is friendly, the first may injure or kill the other and steals his stuff.  Often there is a celebratory meal.   So the cost of extending friendship in this situation is too high --again, we might expect an instinct for hostility. Teamwork would be a rare event.

And this is the way it probably is for the lower creatures – nice behavior to kin and neighbors, and maybe an effort shown to gain status, when status matters.  Other than that it’s every creature for itself.

But when there evolves the ability to distinguish individuals from the teeming mass, and to notice and remember whether they are hostile or cooperative and when players come up against each other repeatedly -- in other words, in a more realistic situation -- the Dilemma becomes a lot more interesting.

Axelrod and Hamilton famously gathered 14 strategies for the Prisoner's Dilemma from experts in game theory and threw them all together in a computer program and ran it 200 times.  Surprise: the eight “nice” strategies (those not striking first) all came out ahead.  The best one was impressively simple: Start out friendly and if you find a friend be friendly back.  But whenever your partner strikes, strike them back immediately.  This “tit for tat” strategy was the winner in most situations. 

Axelrod ran it again with 63 strategies and 1,000 iterations, but this time the payoffs were in offspring, so that the balance in populations changed over time.   This was a more accurate model of evolution, in which the fittest code replicates,and the less fit expires.  Notice there were just simple principles and rudimentary reactions to local stimulus at play.  No consciousness.  The same thing happened – tit for tat came out ahead.  Nature, this suggests, is shot through with niceness.  

There was another interesting result; the advantage of being nice is compounded when the nice players were allow to cluster – to identify and play mainly amongst theselves.  Not only niceness, but a germ of community too!  Incidentally, this was not true for defectors – cheats only lost more when they played against cheats.  Oddly, the “suckers,” who were always nice to all opponents, actually worked against the system by self-sacrificing to “cheats” who then took advantage of everyone.

So "tit for tat" seems the best on paper.  But in the real world mistakes and misunderstandings occur frequently, so when two nice players interact they risk accidentally flipping into mutual retaliation.  Therefore, a random element of forgiveness within a tit-for-tat strategy, would improve the outcome in the long run.  Oh man,  forgiveness too! 

So here's the joyous statement I have been working up to: We are hard-wired for cooperation.  These may be the precious seeds of moral consciousness.


  1. Do simulations show that random forgiveness helps?

  2. Yes, Martin Nowak's Generous Tit-for-Tat forgave 1/3 of single defections randomly. It beats Always Defect, but is pushed back by Always Cooperate.

    So it goes Tit for Tat, shifting to Generous, then to Cooperate, which is beaten out by Always Defect, which loses to Tit for Tat and Generous. There is no equilibrium. When Generous plays amongst Tit for Tat (when TT makes mistakes) Generous wins.

    This is from Ridley's The Origins of Virtue.

  3. I'm not sure I was very clear on that, and I can't see how to edit a post.

    "Tit for Tats" play standard Tit for Tat, but in this model they make occasional mistakes, which can throw a cooperative arrangement into perpetual mutual defection.

    "Generous" plays Tit for Tat but cooperates 33% of the time someone defects against him (i.e., forgiveness)

    "Cooperate" are the ones who are always cooperating, regardless.

    And "Defect" are the opposite of "cooperate;" they always defect.

    The progression is as I explained, with no stable state and no extinctions.


I've allowed comments without login.