The CogSci Track: Bayesian Cognitive Science

For readers who want to use probabilistic modeling to build and test theories of minds — human or machine.

The roadmap

graph TB
    A[I. Foundations<br>counting → Bayes] --> B[II. The Tools<br>GenJAX]
    B --> C[III. Continuous Probability<br>& Bayesian Learning]
    C --> D[IV. Structure — in full<br>generalization, Shepard, causality]
    D --> E[V. Chains, Walks & Sampling — in full<br>memory search, sampling the mind]
    E --> F[VI. Decisions & RL — in full<br>inverse RL, theory of mind, POMDPs]
    F --> G[VII. Model Complexity<br>the DPMM chapter]
    G --> H[VIII. Deep Networks<br>LLMs & in-context learning]
    H --> I[IX. Ethics<br>fairness, bias, alignment]

    style D fill:#27ae60,color:#fff
    style E fill:#27ae60,color:#fff
    style F fill:#27ae60,color:#fff

Reading order

  1. Part I — Foundations: all chapters
  2. Part II — The Tools (GenJAX): all chapters
  3. Part III — Continuous Probability & Bayesian Learning: all chapters
  4. Part IV — Structure, in full — the generalization chapter (all four parts, including Shepard’s law and No Free Lunch) is a centerpiece of this track
  5. Part V — Chains, Walks & Sampling, in full — including Memory Search as a Random Walk and Sampling the Mind
  6. Part VI — Decisions & RL, in full — including Inverse RL (goal inference and theory of mind), POMDPs, and the modern-RL frontier
  7. Part VII — Model Complexity: the Discrete Bayesian Nonparametrics chapter (Anderson’s rational model of categorization lives here); bias-variance and GPs are 🤖-badged extras
  8. Part VIII — Deep Networks: LLMs & In-Context Learning (in-context learning as implicit hierarchical Bayes) and World Models
  9. Part IX — Ethics, Fairness & Safety: fairness formalisms, bias in data, alignment

This project is generously funded by the Japanese Probabilistic Computing Consortium Association (JPCCA).