Paper
Causally Robust Reward Learning from Reason-Augmented Preference Feedback
Authors
Minjune Hwang, Yigit Korkmaz, Daniel Seita, Erdem Bıyık
Abstract
Preference-based reward learning is widely used for shaping agent behavior to match a user's preference, yet its sparse binary feedback makes it especially vulnerable to causal confusion. The learned reward often latches onto spurious features that merely co-occur with preferred trajectories during training, collapsing when those correlations disappear or reverse at test time. We introduce ReCouPLe, a lightweight framework that uses natural language rationales to provide the missing causal signal. Each rationale is treated as a guiding projection axis in an embedding space, training the model to score trajectories based on features aligned with that axis while de-emphasizing context that is unrelated to the stated reason. Because the same rationales (e.g., "avoids collisions", "completes the task faster") can appear across multiple tasks, ReCouPLe naturally reuses the same causal direction whenever tasks share semantics, and transfers preference knowledge to novel tasks without extra data or language-model fine-tuning. Our learned reward model can ground preferences on the articulated reason, aligning better with user intent and generalizing beyond spurious features. ReCouPLe outperforms baselines by up to 1.5x in reward accuracy under distribution shifts, and 2x in downstream policy performance in novel tasks. We have released our code at https://github.com/mj-hwang/ReCouPLe
Metadata
Related papers
Cosmic Shear in Effective Field Theory at Two-Loop Order: Revisiting $S_8$ in Dark Energy Survey Data
Shi-Fan Chen, Joseph DeRose, Mikhail M. Ivanov, Oliver H. E. Philcox • 2026-03-30
Stop Probing, Start Coding: Why Linear Probes and Sparse Autoencoders Fail at Compositional Generalisation
Vitória Barin Pacela, Shruti Joshi, Isabela Camacho, Simon Lacoste-Julien, Da... • 2026-03-30
SNID-SAGE: A Modern Framework for Interactive Supernova Classification and Spectral Analysis
Fiorenzo Stoppa, Stephen J. Smartt • 2026-03-30
Acoustic-to-articulatory Inversion of the Complete Vocal Tract from RT-MRI with Various Audio Embeddings and Dataset Sizes
Sofiane Azzouz, Pierre-André Vuissoz, Yves Laprie • 2026-03-30
Rotating black hole shadows in metric-affine bumblebee gravity
Jose R. Nascimento, Ana R. M. Oliveira, Albert Yu. Petrov, Paulo J. Porfírio,... • 2026-03-30
Raw Data (Debug)
{
"raw_xml": "<entry>\n <id>http://arxiv.org/abs/2603.04861v1</id>\n <title>Causally Robust Reward Learning from Reason-Augmented Preference Feedback</title>\n <updated>2026-03-05T06:35:21Z</updated>\n <link href='https://arxiv.org/abs/2603.04861v1' rel='alternate' type='text/html'/>\n <link href='https://arxiv.org/pdf/2603.04861v1' rel='related' title='pdf' type='application/pdf'/>\n <summary>Preference-based reward learning is widely used for shaping agent behavior to match a user's preference, yet its sparse binary feedback makes it especially vulnerable to causal confusion. The learned reward often latches onto spurious features that merely co-occur with preferred trajectories during training, collapsing when those correlations disappear or reverse at test time. We introduce ReCouPLe, a lightweight framework that uses natural language rationales to provide the missing causal signal. Each rationale is treated as a guiding projection axis in an embedding space, training the model to score trajectories based on features aligned with that axis while de-emphasizing context that is unrelated to the stated reason. Because the same rationales (e.g., \"avoids collisions\", \"completes the task faster\") can appear across multiple tasks, ReCouPLe naturally reuses the same causal direction whenever tasks share semantics, and transfers preference knowledge to novel tasks without extra data or language-model fine-tuning. Our learned reward model can ground preferences on the articulated reason, aligning better with user intent and generalizing beyond spurious features. ReCouPLe outperforms baselines by up to 1.5x in reward accuracy under distribution shifts, and 2x in downstream policy performance in novel tasks. We have released our code at https://github.com/mj-hwang/ReCouPLe</summary>\n <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n <category scheme='http://arxiv.org/schemas/atom' term='cs.RO'/>\n <published>2026-03-05T06:35:21Z</published>\n <arxiv:comment>Published in International Conference on Learning Representations (ICLR) 2026</arxiv:comment>\n <arxiv:primary_category term='cs.AI'/>\n <author>\n <name>Minjune Hwang</name>\n </author>\n <author>\n <name>Yigit Korkmaz</name>\n </author>\n <author>\n <name>Daniel Seita</name>\n </author>\n <author>\n <name>Erdem Bıyık</name>\n </author>\n </entry>"
}