Research map / Reinforcement learning / Visual and representation RL
Of Mice and Machines: A Comparison of Learning Between Real World Mice and RL Agents
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
The paper did what is described above.
Problem
x
Approach
x
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add TISB and variance-penalized TD learning | TDMPC-2 · Mouse visitation pattern overlap | visitation pattern overlap: 20.9% -> 86.1% | Helped |
| Add TISB and variance-penalized TD learning | TDMPC-2 · Initial six steps | waiting incidence: 0% -> about 32.4% | Helped |
| Add TISB and variance-penalized TD learning | TDMPC-2 · Episode length | mean episode length: 7.09 steps -> 13.89 steps | Helped |
| Use 50% negative experiences in training batches (ablation) | TISB · Training experiments over 5000 steps | performance: 50% | Helped |
| Amplify negative-experience rewards by a factor of 200 during sampling (ablation) | TISB · Cellworld predator-avoidance task | path diversity and wall-following behavior: 200 | Helped |
| Add Prioritized Experience Replay (ablation) | TDMPC-2 · Cellworld predator-avoidance task | behavioral policy | No clear effect |
| Add SMiRL without TISB (ablation) | TDMPC-2 · Cellworld predator-avoidance task | waiting behavior | No clear effect |
| Combine SMiRL with TISB (ablation) | TDMPC-2 with TISB · Cellworld predator-avoidance task | path-finding behavior | Hurt |
| Use Q-value variance penalties in the TD target with TISB | TDMPC-2 · Before predator encounter | change in goal-agent distance: -0.2 to -0.5 -> -0.1 to -0.2 | Helped |
| Use Q-value variance penalties in the TD target with TISB | TDMPC-2 · Initial environment entry | distance traveled: approximately 45% less | Helped |
| Add SMiRL (ablation) | TDMPC-2 · Cellworld predator-avoidance task | trajectory diversity | Mixed |
| Train offline IQL on mouse trajectories instead of CQL or COMBO (ablation) | Offline RL models · Mouse trajectory reproduction | waiting and trajectory fidelity | Helped |
Closest papers by meaning
- Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments (NeurIPS 2025 · similarity 0.89)
- Intrinsic Goals for Autonomous Agents: Model-Based Exploration in Virtual Zebrafish Predicts Ethological Behavior and Whole-Brain Dynamics (NeurIPS 2025 · similarity 0.86)
- Inverse Reinforcement Learning with Switching Rewards and History Dependency for Characterizing Animal Behaviors (ICML 2025 · similarity 0.85)
- Predicting Future Actions of Reinforcement Learning Agents (NeurIPS 2024 · similarity 0.85)
- RAT: Adversarial Attacks on Deep Reinforcement Agents for Targeted Behaviors (AAAI 2025 · similarity 0.84)
- SEEA-R1: Tree-Structured Reinforcement Fine-Tuning for Self-Evolving Embodied Agents (NeurIPS 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.