#rewardlearning

2026-02-11

Recent findings show that hippocampal neurons shift their activity backward in time as learning occurs, enabling anticipation of rewards before they happen. Using calcium imaging in mice performing a delayed nonmatching-to-location task, the study tracked the same neurons over weeks and observed a backpropagation of tuning from reward delivery to earlier moments, including the moment of correct touchscreen choice. The results portray the hippocampus as a dynamic predictive map that reorganizes with experience to guide future behavior.

This work is of interest to psychology because it provides concrete neural evidence for predictive coding and memory-based anticipation, illustrating how learning reshapes mental representations to forecast future events. It links memory, prediction, and action in a measurable neural framework.

Article Title: Hippocampal neurons shift their activity backward in time to anticipate rewards

Link to PsyPost Article: ift dot tt/fVBpnxF

Copy and paste broken link above into your browser and replace "dot" with "." for link to work. We have to do it this way to avoid displaying copyrighted images.

#Hippocampus #PredictiveCoding #NeuralPlasticity #CalciumImaging #RewardLearning

Sumanth (@Sumanth_077)

튜토리얼을 시도해본 후 RULER(Relative Universal LLM-Elicited Rewards)이 에이전트 행동에 자동으로 보상을 할당해 수작업 보상 설계(핸드크래프트 리워드 엔지니어링)를 제거해 준다는 점을 긍정적으로 평가한 코멘트입니다. RULER 기반 자동 보상 할당의 실사용 경험을 공유합니다.

x.com/Sumanth_077/status/20165

#ruler #rewardlearning #llm #aiagents

2022-11-07

#introduction

I work as a researcher at the Fractal Analytics - Cerebral team. #cognitiveneuroscience #emotions #rewardlearning #humanbehavior #decisionmaking are some of the keywords that describe the projects done at Cerebral.

Client Info

Server: https://mastodon.social
Version: 2025.07
Repository: https://github.com/cyevgeniy/lmst