Final Project: Open-Ended Exploration of Memory Models#

Overview#

For your final project, you will design and conduct an original investigation into a memory model of your choice. Your goal is to do something substantive and interesting with the model and dataset. This project is open-ended: you are encouraged to explore novel ideas, extensions, and applications. You may work with up to 2 other students on this assignment. If you choose to work in a group, all group members will receive the same grade for the assignment, regardless of individual member contributions, group dynamics, etc.

Choosing Your Model and Dataset#

You may select:

  • A model from a previous assignment (e.g., Hopfield Networks, SAM, CMR, Laplace TCM).

  • A different model from the literature (e.g., a model from a published paper of your choice).

  • A new model that you design yourself.

Similarly, you may select:

  • A dataset from a previous assignment (e.g., Murdock, 1962 free recall data).

  • A different experimental dataset from sources such as the Penn Memory Lab Data Archive.

  • A synthetic dataset that you generate to test specific properties of a model.

Project Scope and Expectations#

Your project should explore the model and dataset in a meaningful, non-trivial way. Some possible directions include:

  • Examining the model’s strengths, limitations, or failure modes
    (e.g., testing whether a model’s predictions break down in edge cases).

  • Extending or modifying the model
    (e.g., adding new parameters, changing the retrieval dynamics, incorporating neural constraints).

  • Applying the model to a new type of dataset
    (e.g., fitting a free recall model to recognition memory data).

  • Comparing multiple models
    (e.g., testing whether SAM or CMR better predicts a given dataset).

Deliverables#

Your submission should include:

  1. Code Implementation

    • Submit a Google Colaboratory notebook (or similar format) with well-commented code.

    • Ensure that all figures are generated within the notebook.

    • The notebook should run without errors.

  2. Project Report (2–5 pages, PDF format)

    • Introduction: Describe the model, dataset, and the research question you are addressing.

    • Methods: Explain how you implemented the model, the experiments you ran, and any modifications you made.

    • Results: Present key findings with clear figures and tables.

    • Discussion: Interpret the results. What insights did you gain? Were there unexpected outcomes? What could be improved?

    • References: Cite any papers, datasets, or external sources you used.

Evaluation Criteria#

Your project will be graded based on the following criteria:

  • Creativity & Interestingness (30%)

    • Is the project novel, insightful, or thought-provoking?

    • Does it go beyond trivial or obvious analyses?

  • Correctness of Implementation (30%)

    • Is the model correctly implemented and well-documented?

    • Do the results make sense given the model and data?

  • Logic & Clarity (20%)

    • Is the report well-organized and easy to understand?

    • Are the arguments and conclusions well-reasoned?

  • Depth & Detail (20%)

    • Does the project explore the model in sufficient depth?

    • Are analyses thorough and well-supported by results?

Submission Instructions#

  • Submit a Google Colaboratory notebook (or equivalent) containing your implementation.

  • Submit a PDF report (2–5 pages) with figures and results.

  • Ensure that all files are well-organized and clearly named.

Good luck, and enjoy the exploration!