http://www.sciencedaily.com/releases/2009/04/090406132056.htm

Considering evolutionary game theory, in class we have discussed two strategies of Dove and Hawk, symbolizing passive, cooperative behavior and aggressive behavior, respectively.  We found that even though Doves do well against one another (share food), Hawks do so well against Doves that Dove cannot be an evolutionarily stable strategy, even though Hawks do poorly against one another (fight over food).

This implies that all Doves would die off in a community if a group of Hawks were to invade.  It is worth noting, however, that a community of all Doves would fare better than a community of all Hawks, as two players cooperating benefit more than two being aggressive (from our model, payoffs of (1, 1) vs (1/2, 1/2)).  Is it possible that in some situations cooperative behavior can survive?  Or does “survival of the fittest” imply that any dominated species will always be eliminated?

The above article tries to answer these questions through observing cooperative behavior of yeast.  As seen in class, evolutionary stability is meant to suppose no intelligence or coordination on the part of the players, and instead strategies are to be viewed as hardwired into the players, such as through encoding in their genes.  This makes studying yeast a particularly good choice – with no thoughts or emotions, their choices of action are direct results of encoded behavior.

To test such behavior, an experiment was designed where sucrose was introduced as the yeast’s only food source.  Yeast prefer to eat glucose, but if none is available, can secrete an enzyme which breaks down the sucrose into smaller sugars they can absorb.  However, in this process a majority of the absorbable sugars are diffused away where they become freely available to other yeast cells – only one percent remains for the original cell.  This means that some yeast cells can simply absorb the sugars broken down by others and do not need to exert work to do the decomposition themselves, which implies that their behavior would result in a higher chance of evolutionary success.  It seems rational then to suppose that the cooperative decomposition behavior would not be evolutionarily stable, and those cooperating would die out.  However, it turns out that this is not the case.

The guaranteed one percent for the cooperative yeast turns out to be enough to keep the behavior alive in the network – mainly because the game being played is what the article calls a “snowdrift game.”  The game gets its name from a fictitious situation in which two people are stuck behind a snowdrift and have the option to wait for the other person to shovel or shovel themselves.  While it is best for each player to wait for the other person to shovel, if both wait, they will be stuck forever.  While waiting, or in the actual case of yeast, cheating is the best option, if nobody is willing to shovel, or cooperate, everyone will die out.  As a result, some yeast will always cooperate to keep themselves alive, and as more cooperate, it becomes more advantageous for others to cheat.  But if too many cheat, there is not enough food, and some cheaters will need to switch to cooperative behavior.  Thus an equilibrium is always reached in which some fraction of the yeast population is willing to cooperate and the remaining fraction cheats.  Cooperative behavior is, at least in this case, evolutionarily stable.

The behaviors in this case differ from the pure strategies discussed in class - while cheating is the best strategy, it is only effective if some yeast cells cooperate, and if none cooperate, the whole population will die out.  The situation is an example of mixed strategies.  Playing cooperative is the best response to everyone cheating, and cheating is the best response to everyone playing cooperative, so there will always be a mix of both behaviors.