You will work in a team of 3 to explore modeling cognitive processes related to revising beliefs.
As we have explored during class, we may hold beliefs which underpin our understanding of how the world works, and we may hold other beliefs which represent situations within the world. Our likelihood of accepting new information and using it to refine our beliefs can be based on experience, provenance, frequency, biases, and so forth. There are many strategies for inferring beliefs, whether it be from logical reasoning with a process for revision as in the AGM model, something like Johnson-Laird’s mental models, or any number of others.
In completing this project you will do the following:
- Form a group of three people. Should there be one or two students left over, a group or two of size 2 shall be allowed.
- Propose three problems involving belief revision which you would like to research and model. These problems may be parts of larger more general problems in order to make them more approachable in the time we have remaining in the semester. The problems should have the following characteristics:
- They involve both foundational knowledge about the way the world is understood to work and situational knowledge.
- The beliefs about the world and situation are not purely based on sensory experience.
- Select one of the three problems as preferred by the group. The instructor will provide input and suggest changes as needed for time and feasibility.
- Begin to research the problem you have proposed, especially as it relates to coming to hold beliefs about a situation and revising those beliefs. What work on this problem has been done within cognitive science or its contributing fields? If not this problem, what about related or more general ones? Your goal should be to situate your work on the problem within the context of what others have done before. Produce a document of somewhere around 750 words with your (well-sourced + cited!) research.
- Model the world and a situation within it in logic. It’s not yet time to worry about computational models or revising beliefs, but this will give you deeper insight into the problem you are exploring.
- Consider the techniques we have discussed in class for representing beliefs and revising beliefs. Make decisions about how you will represent the beliefs, reasoning with the beliefs, the conditions under which you will revise beliefs, and the techniques for revising beliefs which you will use. Refer to the literature as needed and write a document (again, around 750 words) detailing your proposed methods. The instructor may push you to scale back your ambitions at this point, which is fine!
- Implement your model in Prolog. Feel free to work by analogy and steal code from class examples.
- Develop at least one more situation to use with your model of the world. This is so you can have at least two to show generality.
- Write a discussion about the situations you modeled and how well your computational model works. What does it do well? What drawbacks and limitations does it have? Is it useful as a cognitive model? What would you change? Refer back to the background section to re-enforce the context of your work. Again, around 750 words is reasonable.
- Write a short introduction and conclusion. Call the introduction – background – methods – discussion – conclusion – bibliography (+ an appendix including your code) sequence a rough draft of your final paper!
- Edit your paper given feedback and submit it!
Important Dates and Deliverables:
- October 27 – Complete element 1-3 of the project. Post the proposed problems and preferred problem on your course webpage. You may link to one page for these instead of maintaining three separate copies if you like.
- November 3 – Drafts for elements 4,5 of the project. Again post them on your webpage and prepare to present them in class.
- November 10 – Draft of 6, and a good chunk of 7. Again post them on your webpage and prepare to present them in class.
- November 17 – Complete 7, draft of 8-10. Again post them on your webpage and prepare to present them in class.
- Last week of class: Final Project Presentations. Final paper due December 3.
Some ideas which you may consider drawing inspiration from (all of which are way too broad, so you would need to narrow things significantly!):
- Understanding a crime scene
- Choosing which candidate to vote for
- Deciding the winner of a debate
- Separating “fake news” from “real news”
- Understanding situations within the field of intelligence analysis
- Deciding on a patient’s diagnosis
Final Presentation Guidelines
- Presentations will take place the last week of classes.
- Each group will prepare a 10 minute presentation of their project, including:
- An introduction to the chosen topic
- Some background about the topic as found in the literature
- The approach taken along with some justification for it as a cognitive model
- How belief revision factors into the model and the approach taken
- A brief demo of the system
- A discussion of limitations and potential future improvements
- There will be time for 1 or 2 questions as the next group is getting ready to present.
Submission Guidelines
When done, post your final presentation on your course webpage. Also post your final paper as a pdf. Be sure your paper is broken up into Introduction / Methods / Results / Discussion / Conclusion. The results section is a great place to present a demo of the scenarios you discuss in the methods section!