182
9 Networked Minds
given the behavior of the “neighbors” they were interacting with.
14 However, the
utility function was specified in such a way that individuals could consider more
than just their own payoff from an interaction—they could also give some weight
to the payoffs of their interaction partners. This weight was proportional to the
“friendliness” of the individual and was set to zero for everyone at the beginning of
the simulation. Thus, initially, everyone was absolutely selfish, and the payoff that
others received from an interaction was given no weight (corresponding to what we
call “homo economicus”).
Additionally, we assumed that the friendliness trait could be inherited to others,
either genetically or through education. In our computer simulations, the likelihood
that someone would have offspring increased solely depending on their own payoff
(not utility). If a person cooperated but was exploited by everyone they interacted
with, they would receive no payoff. As a result, this person would also have no
offspring. Finally, if people managed to have payoffs and offspring, the friendliness
of an offspring tended toward that of the parent, but there was also a certain natural
mutation rate. The mutation rate was specified such that it did not implicitly promote
significant values of friendliness.
So, what results did our computer simulations produce? As expected, the
prevailing strategy was that of “homo economicus,” who maximizes the own payoff
in a selfish way. However, although this applied to most scenarios simulated with
our model, it did not apply to all of them (see Fig. 9.1). Offspring who lived close
to their parents (when intergenerational migration was low) tended to evolve into a
more friendly form of “homo socialis” and cared more about the payoffs of others,
i.e. had other-regarding preferences! Interestingly, this scenario corresponds to the
conditions in which humans actually raise their children.
The evolution of other-regarding preferences is quite surprising.
15 Even though
none of the assumptions of the above model promotes cooperative behavior or otherregarding preferences in isolation, they create socially favorable behavior in combination. The only possible explanation of this fact is that interaction effects between the
above rules change the overall outcome. Another interesting finding is the evolution
of “cooperation between strangers”, meaning that genetically unrelated individuals
begin to cooperate. A video of a typical run of our computer simulations illustrates
this well
16 (see Fig. 9.2).
14 This standard assumption makes our results more surprising. It can be replaced by other
assumptions such as an imitation of the best performing neighbor.
15 Note that we are talking about other-regarding preferences, not just other-regarding behavior (i.e.
cooperation), here, which has been found in many behavioral models before.
16 See http://www.youtube.com/watch?v=n6AJeIcG4zQ.
9 Networked Minds
given the behavior of the “neighbors” they were interacting with.
14 However, the
utility function was specified in such a way that individuals could consider more
than just their own payoff from an interaction—they could also give some weight
to the payoffs of their interaction partners. This weight was proportional to the
“friendliness” of the individual and was set to zero for everyone at the beginning of
the simulation. Thus, initially, everyone was absolutely selfish, and the payoff that
others received from an interaction was given no weight (corresponding to what we
call “homo economicus”).
Additionally, we assumed that the friendliness trait could be inherited to others,
either genetically or through education. In our computer simulations, the likelihood
that someone would have offspring increased solely depending on their own payoff
(not utility). If a person cooperated but was exploited by everyone they interacted
with, they would receive no payoff. As a result, this person would also have no
offspring. Finally, if people managed to have payoffs and offspring, the friendliness
of an offspring tended toward that of the parent, but there was also a certain natural
mutation rate. The mutation rate was specified such that it did not implicitly promote
significant values of friendliness.
So, what results did our computer simulations produce? As expected, the
prevailing strategy was that of “homo economicus,” who maximizes the own payoff
in a selfish way. However, although this applied to most scenarios simulated with
our model, it did not apply to all of them (see Fig. 9.1). Offspring who lived close
to their parents (when intergenerational migration was low) tended to evolve into a
more friendly form of “homo socialis” and cared more about the payoffs of others,
i.e. had other-regarding preferences! Interestingly, this scenario corresponds to the
conditions in which humans actually raise their children.
The evolution of other-regarding preferences is quite surprising.
15 Even though
none of the assumptions of the above model promotes cooperative behavior or otherregarding preferences in isolation, they create socially favorable behavior in combination. The only possible explanation of this fact is that interaction effects between the
above rules change the overall outcome. Another interesting finding is the evolution
of “cooperation between strangers”, meaning that genetically unrelated individuals
begin to cooperate. A video of a typical run of our computer simulations illustrates
this well
16 (see Fig. 9.2).
14 This standard assumption makes our results more surprising. It can be replaced by other
assumptions such as an imitation of the best performing neighbor.
15 Note that we are talking about other-regarding preferences, not just other-regarding behavior (i.e.
cooperation), here, which has been found in many behavioral models before.
16 See http://www.youtube.com/watch?v=n6AJeIcG4zQ.