Interesting Papers 2026-03-19
-
Dopamine heterogeneity — Ilana Witten
-
Fly social behaviour and optogenetics
-
One-shot learning — Wulfram Gerstner lab
- Shows something opposite to Anne Collins style results / framing.
- https://elifesciences.org/articles/47463
-
Uchida — hard-wired circuit
- https://pmc.ncbi.nlm.nih.gov/articles/PMC12458361/#S2
- Notes from 2026-04-14 TD scratch notes:
- Working hypothesis: dopamine in lNAcc is closer to a TD / teaching signal, while D1-MSN population activity may look more like value / reward prediction.
- D1-MSN activity should not be read like fast glutamatergic excitation; D1 receptors are modulatory rather than directly spike-driving.
- Open question: why visual cue-related D1 activity reflects upcoming reward prediction, but direct reward / opto-triggered dopamine does not simply produce a matching increase in D1 activity.
- Possible answer noted: dopamine changes intracellular state / gain / plasticity, while local inhibition and network dynamics shape whether somatic activity actually rises.
- Circuit sketch: D1 MSNs in NAcc are GABAergic and may carry value-like signals back toward VTA, while dopamine returning to NAcc may carry TD-error-like information.
- Method note from the paper: somatic rather than axonal calcium was recorded because muscimol may directly inhibit dopamine axons, which express GABA receptors.
-
Value as a population code — Gerstner computational paper
-
Key-value memory in the brain — Sam Gershman
- Link to add later