WHAT IS IT?
This is an evolutionary biology model. It models population genetics with
respect to the fitness of traits that are affected by social and environmental
conditions. The model has two breeds of patch agents: altruistic agents and
selfish agents.
The basic premise of the model is that the selfish agents and the altruistic
agents are competing for each spot in the world by entering into a genetic
lottery. You can imagine these agents as plants who "seed" for a spot, and the
dominant seed generally wins. The details of the lottery are explained below in
'Understanding the Model'
Under normal (non-interfering) environmental conditions, the selfish agents
win, and the altruistic population is driven to extinction. However, as outlined
in 'HOW TO USE IT: Environmental Variables', when the environmental conditions
are made more harsh, the altruistic population is able to survive, and even
dominate the selfish population.
Understanding the Model
1.Each patch is a cell that has a fitness. Each patch is the locus of a
lottery for its space. It and the four surrounding patches put in "seeds" to try
to get the patch turned to their type of patch, altruist or selfish. Being
successful in the lottery is getting patches to turn to your type of breed.
We're assuming here that the breed (altruistic or selfish) is the important
genetic trait.
2.Patches live in five-cell, plus-shaped neighborhoods. Whenever a patch is
calculating something about its fitness, it is the center of the neighborhood.
For another patch, when that patch is calculating, it becomes merely one of the
neighbors.
3.Each patch calculates its own fitness using equation:
if it is an A: 1 - cost + (NumberAltruists in Neighborhood/5 * benefit from
Altruists)
if it is an S: 1 + (NumberAltruists in Neighborhood/5 * benefit from Altruists)
Thus, the fitness of the S patch will be higher than the fitness of the A's.
If the cost is .2 and benefit is .5, for an A surrounded by two S's and two
A's, then the fitness of this spot is 1 - .2 + (3/5 * .5) = 1.1
4.Each patch has calculated its fitness this way. Now, A patch looks to its
four neighbors. Each of the five patches, including itself, puts a weighted seed
into a genetic lottery for this center spot. So, for example, if the
neighborhood is ASASA, each of the three A's register their fitness value, and
each of the two S's put in their fitness. The A's are added, and the S's are
added. Let us assume that the A's add up to 3.2 (this includes the A in the
center spot), and the S's add up to 2.6. These two numbers are the
altruistweight and selfishweight respectively, in the lottery for the center
spot. Now, the larger number, whichever it is, is called the Major seed; it is
divided by the sum of all the fitnesses. Thus, 3.2/(3.2 + 2.6) = .552 This
number is the Altruism seed in the lottery. The minor seed is 2.6/(3.2 + 2.6) =
.448. (Notice that the Altruism seed of the parent is 3/5 = .600, while the
child's is .552. Even though altruism is dominating, it is losing ground.)
5.There are a number of ways of doing the lottery itself. Currently, we
choose a random number between 0 and the sum of fitness. In the above example,
this is a number between 0 and .999. Now, if the Number is below the Minor seed,
the minor weight gets the spot, and if it is above the major seed, the major seed
gets the spot. So, in the example, if the random number is anywhere from .449 to
.999, then the Major seed gets it. If it is between 0 and .448, the minor seed
gets it.
HOW TO USE IT
Basic Controls
1.The SETUP button sets up the model by creating the agents.
2.The GO button runs the model.
3.The DEN%_ALT slider lets you determine the percentage of overall altruists
in the model. This value is weighed against the DEN%_SELFISH value, which
determines the percentage of overall selfish agents.
4.The ALTRUISM_COST slider determines the value of cost in the above fitness
equations.
5.The BENEFIT_FROM_A slider determines the value of benefit in the above
fitenss equations.
Environmental Variables
There are two environmental variables: Harshness and Disease.
1.The HARSHNESS slider sets the value for the resistance of empty patch spots
to being populated by agents. The value for this slider determines a
corresponding value in the lottery for each empty (black) spot on the grid; the
higher this value, the more difficult it is to populate.
2.The DISEASE slider sets the value for the possibility that the agents in
occupied spots will die. The value for this slider is factored into the genetic
lottery, and determines the percentage chance that each agent will die out from
their spot.
THINGS TO TRY
1.At first, run the model with Harshness and Disease both at 0. Notice that
the selfish population quickly dominates the world, driving the alrtuistic
population to extinction. How do respective population sizes affect the outcome?
2.Play with the values of cost and benefit. What are realistic values for
actual genetic competition? How does initial population size effect the
significance of these values?
3.Increase the Harshness and Disease values, independently, and with respect
to one another. What are the effects of the Harshness Model? of Disease? How
are the values dependent on one another? At what values does the altruistic
population begin to have greater success?
4.Consider why the introduction of Harshness and Disease conditions affects
the success of the altruistic population. How does each population, run alone,
respond to the Harshness and Disease conditions? If you imagine the black spots
as Voids (a third type of competing agent), what is the fitness relationship
between Altruists and Voids? Selfish agents and Voids?
EXTENDING THE MODEL
This model is based on a paper by Mittledorf and Wilson, in press, and has
thus been developed under its auspices. However, the model can be extended in a
number of interesting directions, including adding new environmental variables,
adding different types of agents, and changing the altruistic and selfish
weighting under different environmental conditions. This model does not address
the behaviors of individuals, but only the relative weights of genetic traits. A
next step in considering the evolution of altruism is to model altruistic
behaviors. See the related model: Cooperation (Altruistic Behavior).
STARLOGOT FEATURES
This model uses only patch agents. The fitness of the agents is determined by
setting a 'fitness' variable. Each agent then compares its type of fitness
(altruistic or selfish) to the total value of fitness in its neighborhood.
Because each patch can talk to one another, they can find out the fitness of each
member of their neighborhood, and use that value in calculating the total weights
of altruism and selfishness in the neighborhood.
The use of commas in the code allows all of the agents to synchronize in
their fitness-checking, and thus to keep the neighborhood values accurate.
RELATED MODELS
Cooperation (Altruism Behavior) Model
Divide-the-Cake Model