Notes for October 22, 2003

Fallacies: Fake Precision and the Genetic Fallacy

**We drew numbers for our presentations that begin on November 12. About 7 people will present each day.

-Process validity is when processes are the same in the model as in modeled thing. Outcome validity is what the model predicts and how it behaves when observed.

-It is a good idea to rely on statistical measures to measure an agent’s performance because you want an agent to be as similar to a natural agent as possible.

-A reasonable explanation of behavior is the way an agent interacts with the environment.

-The majority of experiments dealing with the synthetic approach lack evaluation because they come from an engineering perspective and do not need a theoretical explanation.

-The use of a robot is an advantage to study intelligence because a robot is a model used for understanding.

-We can determine what a model is really measuring through its validity.

-Depending on the goal, a model that is functionally equivalent to something can be really accurate if it is of a different construction.