# Shaping Persistent Representations from Independent Interactions

arXiv: https://arxiv.org/abs/2609.34604 · HTML: https://mlsanity.com/paper/2609.34604/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in World models for language agents, one of 11 directions in World models.
- Builds on 16 of its 35 references that are in the map.
- First posted on arXiv 28 Sep 2026.

## Placed in

- World models / [World models for language agents](https://mlsanity.com/wm/world-models-for-language-agents/)

## Builds on (most cited first, 16 of 16)

- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [Dream to Control: Learning Behaviors by Latent Imagination](https://mlsanity.com/paper/1912.01603/) (2019)
- [VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning](https://mlsanity.com/paper/2105.04906/) (2021)
- [Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture](https://mlsanity.com/paper/2301.08243/) (2023)
- [Revisiting Feature Prediction for Learning Visual Representations from Video](https://mlsanity.com/paper/2404.08471/) (2024)
- [DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot Planning](https://mlsanity.com/paper/2411.04983/) (2024)
- [LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics](https://mlsanity.com/paper/2511.08544/) (2025)
- [In-context Reinforcement Learning with Algorithm Distillation](https://mlsanity.com/paper/2210.14215/) (2022)
- [RH20T: A Comprehensive Robotic Dataset for Learning Diverse Skills in One-Shot](https://mlsanity.com/paper/2307.00595/) (2023)
- [Generalizing Graph ODE for Learning Complex System Dynamics across Environments](https://mlsanity.com/paper/2307.04287/) (2023)
- [OvercookedV2: Rethinking Overcooked for Zero-Shot Coordination](https://mlsanity.com/paper/2503.17821/) (2025)
- [Hidden Parameter Recurrent State Space Models For Changing Dynamics Scenarios](https://mlsanity.com/paper/2206.14697/) (2022)
- [Dynamics-Aligned Latent Imagination in Contextual World Models for Zero-Shot Generalization](https://mlsanity.com/paper/2508.20294/) (2025)
- [What Can Latent World Models Know? Physical Parameter Identifiability in Multimodal Predictive Representations](https://mlsanity.com/paper/2607.27017/) (2026)
- [Benchmarking the Limits of In-Context Reinforcement Learning for Ad-Hoc Teamwork](https://mlsanity.com/paper/2605.24423/) (2026)
- [Neural operator discovery from heterogeneous trajectories](https://mlsanity.com/paper/2607.23337/) (2026)

## Abstract (arXiv)

World models learn environment dynamics from interaction experience. These dynamics depend on the current state and actions, as well as on properties that persist across interactions. Yet standard predictive training can reduce error using local evidence alone, without organizing persistent information into reusable context. We introduce SPRII, a training principle that uses relations between interactions as weak supervision for persistent context while retaining the learner's native objective. For example, different trajectories of the same system share persistent properties even when their states and actions differ. SPRII uses such relations to guide context learning without numerical property labels. Two composable components encourage contexts from related interactions to agree (Align) and use one interaction's context to predict another's future (Cross). Our analysis distinguishes three linked questions: what persistent information is accessible in the learned context (Formation), how that context influences a fixed predictor (Use), and whether it reduces task error (Value). Success at one stage does not guarantee success at the next. Controlled experiments show that more reliable relations improve representation organization, but adding a shared-property constraint can reduce access to a property that remains shared. Context substitutions change predictions at fixed model weights, while the benefit from history depends on prediction horizon and readout. Evaluations span thirteen settings, including controlled physical systems, public dynamics tasks, robotic and tactile data, and partner interaction, across multiple learner families. Relative to the corresponding baselines, SPRII yields average gains of over 10% in downstream task performance and over 15% in persistent-property readout. The project page is available at https://persistent-learning-review.netlify.app/interactive.html.