3 - On using ABMs in Cognitive and Computational Social Science

Question 1

Under what circumstances are cognitive models developed?

Answer 1

Often developed based on cognition in isolation (lab experiments, surveys, settings with high control)

Question 2

When does human cognition occur?

Answer 2

• In dynamic social interactions
• Others influence our attention, learning, decision making etc.

Question 3

What makes environments dynamic?

Answer 3

They create feedback loops that are often hard to predict

Question 4

What is the methodological challenge in CogSci?

Answer 4

If individual cognitive models are able to be used on a larger scale

Question 5

What are advantages of ABMs in CogSci?

Answer 5

• how multiple cognitive entities interact over time and space
• Calibrate (Parameterize - express in parameters) and validate (test predictions) cognitive models in complex systems
• Encourage model development

Question 6

When is an ABM a useful methodological tool?

Answer 6

• When cognitive function depends on the actions of other agents/environments
• When there is heterogeneity (diversity) within and between groups
• When time is a factor (change over time)
• When there is a spatial distribution

Question 7

What are the uses of CogSci in ABMs?

Answer 7

• Calibration (provides parameterization)
• Validation (test predictions)
• Longitudinal question (e.g. evolution - culture - social norms)
• Ethical questions (scenarios that are not “ethical” to manipulate in real-life)

Question 8

What varies between agents?

Answer 8

• Degree of autonomy
• Self interest
• Sociability
• Learning
• Complexity

Question 9

What is multi - realizability in ABM?

Answer 9

When multi-level (subgroups) multi-agent systems can be implemented in different ways at lower levels and still generate the same/similar macro-level phenomena
(get to the same outcome with many different implementations)

Question 10

What are some current opportunities of AMBs?

Answer 10

• how to scale and incorporate real-time simulation with massive amounts of data
• model equivalence (similarities between models)

Question 11

What is the minimality procedure?

Answer 11

The idea that you should have the simplest set of rules in your model to generate macroscopic effects

Question 12

What are the shortcomings of the minimality procedure?

Answer 12

• Reduces validity
• Isn’t driven by theory
• Can create non-plausible models

Question 13

What is a generative model?

Answer 13

Generate a theory of behaviour by understanding the mechanisms that generates behaviour

Question 14

What do generative models require?

Answer 14

• External mechanisms (environmental and social)
• Internal mechanisms (behavioural/ cognitive mechanisms)

Question 15

What are the 3 different approaches in Computational Social Science?

Answer 15

• Deductive (explain social science phenomena with math/computer science/logic-game theory)
• Generative (Generate theory and test it with data)
• Complex (combine complex systems methods with data science, machine learning etc.)