So today we're going to talk about one
more other element that you should

consider when designing the agents for
your agent based model.

We've eluded to this before but you have
to choose the level at which you're going

to actually create the agents themselves.

Now it doesn't even have to be the same
level for every agent type in the model

but we're get into that in a little bit.

As I mentioned before, an agent doesn't
necessarily have to be an individual or

an employee or a person on the street or
something like that.

It could be a corporation. It could be an
organization. It could be a cell.

It could be whatever, right?

So here, just as an illustration, I have
two different models created in the

medical field. So on the left, we have
this tumor model. A model where you're

looking at cancer. As you can see, maybe
it's not completely obvious, but each of

these little dots represents a cell.
So the agents in this model are cells.

Over here, we have an AIDS model. A model
of AIDS transmission in a community or a

society. In this case, each of the little
agents is represented by a human being,

a person. It's interesting to me. We have
two different models, both kind of disease

related phenomenon. But in one case we
have the model being done at the person

level and in the other case at the
cellular level.

Why would you choose these
two different levels?

Well the main reason is because of the
question they were interested in answering.

In the tumor model, they're essentially
looking at how different treatments can

affect tumor growth. In the HIV model, the
AIDS model, they're looking at how HIV

propagation happens throughout an entire
community.

They're very different kinds of goals and
in each case they chose the level of an

agent to correspond to the research
question they were interested in asking.

So that's why you get these two different
level choices.

By the way, both of these models are
available in the NetLogo models library.

So one other thing we should talk about
is...we've talked about agents but there

are a couple of special words that are
often thrown around in the agent based

modelling literature that relate to
different types of agents that aren't your

traditional formal agents.

One word that is thrown around is a meta-
agent. A meta-agent is an agent that is

composed of other agents. For instance,
you could imagine you have a business and

you have the business represented as an
agent interacting with other businesses.

But then maybe within that business you
have different departments and those

different departments take different
actions and those different actions are

part of the business organization but the
business is not just the composition of

those different departments.

You could even go lower than that.
Within the departments you could have

employees and those employees could
be agents.