Thresholds of Behavioral Flexibility in Turbulent Environments for Individual and Group Success
during the next time step than one that is empty , but occasional big swings are also likely if the model proceeds in a path where many spreading agents suddenly find themselves in the same place . This kind of tipping behavior is observable in other simple models like this , including Schelling ’ s famous model of segregation ( Schelling , 1978 ).
The construction of the model also suggests the proportions of one type over another will affect outcomes in important ways . If clustering agents are a minority , their utility will be limited to however many other clustering agents there are , plus perhaps the occasional unfortunate spreading agent who is trapped near them . If clustering agents were a majority , it would benefit them , and if we allow types swapping , we should expect all agents to become clustering types .
On the other hand , if spreading agents are a minority , at first it may seem that would benefit them , but this would actually mean there are more clustering agents instead , which , depending on the distribution over the lattice and the size of the neighborhood , may impose an upper bound on how spread out they can become . In addition , the more successful spreading is , the more agents will switch to spreading , which again poses a downward pressure on the utility of being a spreading agent . Ironically , spreading agents potentially make themselves worse off as they become more successful . Spreading agents can expect high utility only under certain larger neighborhood configurations .
We have now arrived at the two complexities at the heart of this model , and which will drive our most interesting results : environmental turbulence and agent adaptability .
Environmental Turbulence and Behavioral Flexibility Environmental Turbulence
We first make the environment dynamic . Recall that agents are on a 5 × 5
lattice . To simulate a heterogeneous environment , I allow the vision ( v ) of agents to vary over three possible values : 0 , 1 , and 2 . When v = 0 , agents can only see as far as the cell that they occupy . At v = 1 , agents can see their own cell and one cell further in every direction for 9 cells total . At v = 2 , agents can see one cell further than even that in every direction . Given the nontoroidal lattice , at v = 2 agents in the middle of the 5 × 5 lattice can see all 25 cells , while those elsewhere see fewer depending on their location . This is illustrated in Figure 4 .
123