“A gradual temporal shift of dopamine responses mirrors the progression of temporal difference error in machine learning”

The Systems Journal Club has spirited discussions of recent systems neuroscience papers, covering sensory, motor and cognitive issues.

Mini-series “Reinforcement Learning”

Questions to ponder over and discuss:

1. Are the two models (paper 1 vs 2) actually contentious? What kind of framework or experiment is required to resolve the contention?
2. Which of the two types of models pushes our understanding forward for how learning operates? Where does each approach fail?

January 13, 2023

Apoorva Arora (Kepecs lab, WashU Neuroscience)

“A gradual temporal shift of dopamine responses mirrors the progression of temporal difference error in machine learning” Nature Neuroscience (2022)

This paper shows that the DA transient moves backward from the time of reward to the time of cue over successive trials. This result had been missing so far in the TD-RL framework for learning via DA.

January 20, 2023

Apoorva Arora (Kepecs lab, WashU Neuroscience)

“Mesolimbic dopamine release conveys causal associations” Science 2022

This paper builds on a different model (ANCCR) to explain the association between cue and reward through DA signals. They claim that their model explains DA signals that are not explained by the TDRL framework

Venue: McDonnell Med Sci Building (455)

Time: 11:45 am to get Pizza, 12 pm for paper discussion

Full schedule and archives, Systems Journal Club

For inquiries or to be added to the journal club list, contact Camillo Padoa-Schioppa or Larry Snyder (WashU Neuroscience).