Digest 2026-05-06
Papers and links from your 2026 notes that you added but may not have read yet.
Featured
Acetylcholine demixes heterogeneous dopamine signals for learning and moving
- Nature Neuroscience (2026) — Acetylcholine demixes heterogeneous dopamine signals for learning and moving
- TL;DR: ACh helps separate reward-learning signals from movement-related signals in dopamine neurons. Relevant for striatal circuitry and RL.
- 🔗 https://www.nature.com/articles/s41593-026-02227-x
Behavioral timescale synaptic plasticity: properties, elements and functions
- Nature Neuroscience (2026) — Behavioral timescale synaptic plasticity: properties, elements and functions
- TL;DR: Reviews plasticity mechanisms that operate on behavioral (not just neural) timescales. Useful for learning & memory models.
- 🔗 https://www.nature.com/articles/s41593-026-02214-2
Neuroscience Papers
| Paper | Source | Note |
|---|---|---|
| Dopamine heterogeneity (Ilana Witten) | PubMed | Heterogeneous dopamine signaling |
| Fly social behaviour and optogenetics | Current Biology | Optogenetics in fly social circuits |
| One-shot learning (Gerstner lab) | eLife | Contrasts with Collins-style framing |
| Uchida — hard-wired TD value circuit | PMC / bioRxiv | Dopamine as TD signal vs D1-MSN value |
| Value as a population code (Gerstner) | PLOS Comp Biol | Computational value representation |
| Digital twin visual cortex | Science | Large-scale visual cortex modeling |
| Neuron — visual cortex modeling | Neuron | Related to digital twin work |
→ See Interesting Papers 2026-03-19 and Visual cortex papers from Twitter
Cosyne 2026 Follow-ups
| Paper / Topic | Link | Why it matters |
|---|---|---|
| Joshua Dudman — dopamine learning signal | Nature | Dopamine as teaching signal |
| Uchida — TD value circuit (preprint) | bioRxiv | Hard-wired value computation |
| Aneta Koseska — cell working memory | eLife | Single-cell memory claims |
| Belief dynamics | arXiv | Interpretability / belief states |
→ See Cosyne 2026
Causality, Prediction & Interpretability
| Paper | Link | Key idea |
|---|---|---|
| Gershman — causality as invariant prediction | The Transmitter | Prediction ≠ understanding |
| Microprocessor interpretability | PLOS Comp Biol | Lessons from reverse-engineering a CPU |
| Shortcut learning | Nature MI | Models learn spurious correlations |
| Hypothesis testing interpretability | NeurIPS 2024 | Formal hypothesis testing for interpretability |
| Interpretability illusions | arXiv | When explanations mislead |
| Lindsay & Bau — neural system understanding | Neuroinformatics | What “understanding” means in neuroscience |
→ See causality, prediction in neuro, ai
Cellular Learning
| Paper | Link | Claim |
|---|---|---|
| What Can a Cell Remember? (Quanta) | Quanta | Overview of non-neural memory |
| Predictive behavior in microbial genetic networks | Science | Microbes show predictive behavior |
| Spaced repetition in non-neural human cells | Nature Comm | Kidney cells show spacing effect |
| Molecular memory networks | Cell | Biochemical networks as learning circuits |
| Cellular cognition (Rajan & Marshall) | PubMed | Review of single-cell cognition |
| Mechanistic model of cellular learning | Current Biology | Model for single-cell adaptation |
| Sam Gershman review | eLife | Reconsidering single-cell learning evidence |
→ See Cellular Learning
Books & Essays
- The Conquest of Happiness — Bertrand Russell — The Conquest of Happiness
- The Unbearable Lightness of Being — Milan Kundera — The Unbearable Lightness of Being
- The Vital Question — Nick Lane — The Vital Question
- AI Thoughts and Essays — your own fragments — AI Thoughts and Essays
How to use this digest
- Click any
[[note]]link to go to the original note - Papers marked with 🔗 have direct links
- Tell me “read [paper name]” and I’ll mark it off the next digest
Generated by Joi 🧠 — your Memex