Computation models

Hello everyone! In this presentation, I will discuss the concept of modelling and how computational
models help us to understand complex systems in social science. I will also discuss Agent-Based
Simulation which is a tool of sociological understanding. At first, let's go briefly to the concept of
modelling.

Modelling focuses on simple entities and interactions within a system. Their implications result in
robust changes in the underlying structure of the complex system. They also normally produce
surprising results which can motivate the new predictions about the system, and they can also be
easily communicated to others.

Let's discuss the concept of modelling with an example of a road map. Road maps are a very good
example of one of the best models we use in our daily life. Road maps help us to navigate and
manipulate the world. In addition to providing the navigation, road maps also provide additional
details such as a useful pattern of population. In this example, we can see the road maps of the area
surrounding the University of Sydney. Another image on the right shows the original view of the
same area without road maps. The road map leaves unnecessary details and provides us with an
easy way to navigate the road. A good map provides enough details while isolating unnecessary
information. We can place all the details of the world such as main highways, roadside restaurants,
hospitals, schools, etc on a separate layer and we can combine these layers depending on our needs.
Different people want different information such as a person who is sick wants to know the details
nearby hospital and how to reach there. Similarly, a parent wants to find the best school in the area
and the time it will require to travel to and from school. If we keep adding more layers of
information to the map, it will start to become complex and confusing which results in abandoning
the map and use the real world itself. If we look at the map on the third image with more layers of
information such as terrain and traffic, the map becomes more complex. Good modelling requires
having enough and the right information on the map depending on the needs of the user.

Now, let's discuss a more formal modelling approach which can be referred to as the “model of
modelling. This concept is based on the mental modelling ideas developed by Holland for creating
artificial learning systems. As we can see, the top half of the diagram represents the real world that
we want to model. In the lower half, we represent our model of the real world. In this diagram,

S represents many states of the real world.

F(S) represents the transition function

T represents the time

The transition function maps the real-world states at the time ‘t’ to a new state at the time t+1. The
modeller will not know the value of function F(S). Even though the modeller would like to know the
value of F(S), it will normally be impossible because of the dimension of state space and also because
of the complexity of the function F(S).

If we take an example of forecasting the weather, the actual state of this system S is enormous
which consist of positions and characteristics of different atoms. The function F(S), which is a law of
physics, tells us about the change of state of atmos every moment. The modeller can also use the
alternate approach in which the size of the state is reduced to find a simpler transition function. In
this weather example, we can consider using equivalent class E(S) which will map the configuration
of atoms into measures of barometric pressure and humidity. After this, we need to find a function
f(s) that will help to predict the transformation of pressure and humidity over time.

These concepts and principles of modelling are important to understand the concept of a simple or
complex system. A reductionist hypothesis is one view of modelling the complex system. According
to this hypothesis, if we can simplify the model in the right way, we will be able to understand
everything around us. In this way, the social world around is “just particle physics”. Once we
understand the concept of particle physics, we can use this knowledge to unpack any higher-level
system such as sociology, psychology physics and so on.

When the model becomes more complex in its structure, many useful features are lost and we shift
from modelling to simulation. One example of simulation is Agent-Based Simulations. In the
diagram, we can see an example of bees. The first image shows the real world image of bees. The
second image on the bottom left shows an example of modelling and the third image shows a
simulation. Simulation provides more general features of the system without going into too much
detail on the complexity of the structure of the system.

One example of simulation is an agent-based simulation. Agent-based simulation has become one of
the popular tools for understanding social phenomena. It consists of agents having goals and beliefs.
These agents in Agent-Based Modelling interact with each other and they have the resources that
are required to achieve their goals. ABS has many advantages over traditional modelling tools. Using
ABS, we can study the dynamics of the system as we don’t have to limit our attention to equilibrium
outcomes. Because of this, it is possible to study different processes, path dependence and so on.
There is also no limit on the number of agents which makes it a more realistic tool for modelling
social processes.

Agents in ABS models has cognitively and socially interesting properties. In the view of cognition,
agents are heuristic in which they follow simple behaviours such as habits, routines, etc. Agents are
also adaptive as they respond to the feedback from their surroundings through the process of
learning and evolution.

The Checkerboard model of segregation is one of the best-known examples of ABS. This model
addresses the origin of residential segregation by race or ethnicity. In the original model, the world
consists of agents which are occupying the squares of the checkerboard. The agents are divided
into two classes which represents the binary social division and also affect the distribution of agents
in the space. This model assumes that each agent has a threshold for tolerating members of the
other groups available in its neighbourhood. At first, the agents are randomly distributed across the
board and some of the squares are left unoccupied. By observing their neighbourhood, some agents
can change their location when other types of agents exceed a certain threshold. Because of this,
any dissatisfied agent randomly relocates to the new position on the board. The result of this model
shows that there is still a chance of segregation even when the agents are highly tolerant.

As we have seen, ABS is an important and useful tool to understand the different aspects of society.
ABS and other modelling tools help us to understand the complex system of society in a better way.