The document discusses the value of creating agent-based models that incorporate human values to better represent human behavior. It describes an experiment that tested whether agents with values were able to reproduce empirical data on human behavior from a psychological experiment. The document raises questions about the differences between value-based agents and rational economic agents, why modeling humans with values is interesting, what types of models best explain human behavior, and whether models need to be realistic representations of the systems they aim to study.
4. We tested the agents ability to reproduce
empirical data on human behavior in a
psychological experiment
5. Is the question interesting even if the results
are negative?
6. Why would be interesting to know if agents with
values do not reproduce human behavior?
7. Some questions for discussion
• What are fundamental differences between value-based agents and a
homo economicus agent?
• Why are we interested in humans with values in the first place?
• Would we be interested in any model that reproduces human
behavior?
• What kind of models allow for the best explanations?
• What kind of models can we trust?
8. Should models be realistic?
• Models always differ from the system they are meant to represent
• The relation between reality and the model implies how they create
knowledge about a system
• Broero et al.:
• Case-Based Models
• Typifcations
• Theoretical Abstractions
• We can make infinite models that reproduce some output. So how do
we differ between them?