Resident | ||
Last Move | Last Opponent | Strategy |
---|---|---|
Cooperate | Cooperate | |
Cooperate | Defect | |
Defect | Cooperate | |
Defect | Defect | |
Initial |
Invader | ||
Last Move | Last Opponent | Strategy |
---|---|---|
Cooperate | Cooperate | |
Cooperate | Defect | |
Defect | Cooperate | |
Defect | Defect | |
Initial |
Background: This model restricts payoffs to the standard Prisoner's Dilemma values (i.e. two cooperators get 3 points each, two defectors get 1 point each, one defector and one cooperator get 5 and 0 points respectively). In models of single interactions defectors should win and completely eliminate cooperators. In this case, however, when pairs of individuals meet they can interact multiple times (not necessarily in succession, although this has implications about the animals' biology). This is important because they can remember what happened on the last interaction which leads to a wealth of new adaptive strategies. For example, one individual may not want to cooperate with another animal that defected last time. It is assumed that the individuals do not know ahead of time what the number of interactions will be. Unlike the last model, one type (the resident) starts off occupying 90% of the environment, while the other type (the invader) occupies the remaining 10%. The purpose behind this is to set up experiments that test whether a new type can invade some other type that is already established in the environment. This is particularly relevant in an evolutionary context in which some novel type may be present at low levels, and can only establish itself if can displace the type that currently represents the majority.
Instructions: Specify the different strategies for the resident (top) and invader (bottom). In this version, the strategy may be based on what the opponent and the player did in the last round. For each type (resident and invader) you must specify five different actions, one for each of the combinations of what they did and their opponent did last time, and one for an initial meeting. You may choose from cooperate, defect and random (flip a coin to decide whether to cooperate). Read each strategy by going across the row, so that the very first strategy (if you haven't changed anything yet) for the resident is "if I cooperated last time and my opponent cooperated last time than this time I will cooperate." Therefore, the default for both types is always cooperate. You also must specify the number of times (# meeting) that each pair interacts before they part ways. Meeting one time should be the same as the Model 1. Meeting more than once would mean that multiple flower locations are conveyed between a given pair of individuals. It is assumed that they have time to check the information to determine whether their partner gave them the correct information between exchanges. You may not enter more than 25 for the number of interactions, and it is suggested that you keep this number below 10. Hit RUN when you are ready to observe what happens.
Interpretation: The graph on the right depicts what happens to the resident and invading types over time. Note that there is a good deal of fluctuation and that you will probably want to observe the dynamics for each set of parameters more than once. When individuals only meet once, then defecting is the best strategy under these parameters. Can you find other strategies that outcompete pure defection (defect for all possible interactions) when individuals meet multiple times? If you think you have found a good strategy, pit it against another colleague's strategy and see how they fare. Are there cases in which being the resident or invader type makes a difference? What would this mean?