[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82845-en":3,"doc-seo-82845-105":30,"detail-sidebar-cat-0-en-105":92},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},82845,2336464648322,"Aria","https://ap-avatar.wpscdn.com/avatar/2200025388227c56fec?_k=1778556882303663488",8,"Research & Report","Do Vision-Language-Action Models Mean What They Say","Embodied Chain-of-Thought is proposed as a way to improve robot decision-making and interpretability in black-box vision-language-action models, yet whether verbalized reasoning truthfully corresponds to the policy’s actual decision process remains unclear. The work separates functional reasoning (improves task performance) from faithful reasoning (matches internal action-causing computation). It shows standard alignment can yield ungrounded, confounded steps that weaken generalization, and introduces Pinocchio to learn a dense behavioral reward for embodied faithfulness.","arXiv :2607 .0468 1v 1 [ cs .RO] 6 Jul 2026  \nDo Vision-Language-Action Models Mean What They Say? On the Role of Faithfulness in Embodied Reasoning  \nMatthew Foutter1,⋆ Matteo Cercola2,⋆ Lena Wild3 Yunshan Wang1 Michelle Li1  \nDaniele Gammelli1,4,† Marco Pavone1,5,†  \n1Stanford University 2Politecnico di Milano 3KTH Royal Institute of Technology  \n4Italian Institute of Artificial Intelligence (AI4I) 5NVIDIA Research  \n[Corresponding Author:](Corresponding Author: # mfoutter@stanford.edu)[ \\#](Corresponding Author: # mfoutter@stanford.edu)[ mfoutter@stanford.edu](Corresponding Author: # mfoutter@stanford.edu)  \nAbstract  \nEmbodied Chain-of-Thought has emerged as a promising mechanism to enhance robot decision-making and interpretability in black-box Vision-Language Action (VLA) models. However, whether this verbalized Chain-of-Thought truthfully reflects the policy’s underlying decision process remains poorly understood. We distinguish between functional reasoning, in which reasoning improves task performance, and faithful reasoning, in which reasoning truly reflects the policy’s internal decision process.  \nWe argue that SoTA alignment strategies offer a necessary but insufficient notion of faithfulness, admitting reasoning whose intermediate steps can mask the causal links in action prediction through confounding factors (e.g., reasoning that is ungrounded in the environment and internally disconnected or inconsistent), restricting policy generalization. We study this gap through a human evaluation of a SoTA reasoning model for autonomous driving, revealing an inconsistent coupling between reasoning quality and downstream trajectory improvement. We then operationalize a behavioral surrogate for embodied faithfulness through a learned critic, Pinocchio, scoring observation grounding and stepwise coherence, and use this critic as a dense reward signal in post-training an embodied policy with reinforcement learning. Across withheld driving benchmarks, our post-trained planner improves faithfulness by 4% and 18% over SoTA alignment and trajectory error post-training baselines, respectively, while maintaining competitive downstream task performance. Finally, on a synthetic out-of-distribution test set, post-training for faithfulness improves policy responsiveness to rare counterfactual scenarios by 1.6× that of a SoTA policy, suggesting that faithful reasoning traces contribute to more robust, generalizable, and interpretable embodied intelligence.  \nCogito, ergo sum.  \n—Ren Descartes, Principles of Philosophy, 1644  \n1 Introduction  \nFoundation Models (FMs) [5], particularly Vision-Language Models (VLMs) [3, 1, 13], have emerged as a general-purpose prior for learning-based decision making. Through self-supervised training on internet-scale corpora, these models inherit rich semantic grounding and broad world knowledge, enabling Vision-Language-Action (VLA) policies [35, 8] post-trained on robot demonstrations to exhibit increasingly generalizable behaviors across diverse robotic tasks and environments. More recently, the robotics  \n⋆ Equal contribution. † Shared Advisorship. Project Page: [https://mjf-su.github.io/pinocchio/](https://mjf-su.github.io/pinocchio/)  \nFigure 1: Functionality and faithfulness are distinct axes of embodied reasoning. Functionality (vertical) measures whether the CoT improves task performance; faithfulness (horizontal) measures whether it reflects the process that actually produces the action. In this work, we revisit the role of reasoning in physical intelligence through the lens of faithfulness and introduce Pinocchio, a learned critic that operationalizes faithfulness as a dense reward signal for RL.  \ncommunity has explored augmenting action generation with an intermediate Chain-of-Thought (CoT)  \n—a paradigm typically referred to as embodied reasoning [42, 6]—decomposing complex decisions into structured, intelligible steps similar to advances in language-only reasoning models [37, 14, 15]. Howev","cbCaimpD3r81C8q9","https://ap.wps.com/l/cbCaimpD3r81C8q9","pdf",5674511,5,1,30,"English","en",105,"# Abstract\n# Introduction\n## Functional vs faithful reasoning\n## Faithfulness beyond interpretability","[{\"question\":\"What is the difference between functional reasoning and faithful reasoning in embodied chain-of-thought?\",\"answer\":\"Functional reasoning means the reasoning trace improves task performance. Faithful reasoning means the trace accurately reflects the internal process that actually produces the action.\"},{\"question\":\"Why are existing alignment strategies considered insufficient for faithfulness?\",\"answer\":\"They can produce reasoning whose intermediate steps mask causal links in action prediction via confounding factors, such as ungrounded reasoning that is disconnected or inconsistent with the policy’s true computation.\"},{\"question\":\"How does Pinocchio operationalize embodied faithfulness and use it for training?\",\"answer\":\"Pinocchio learns a behavioral surrogate based on observation grounding and stepwise coherence, then provides a dense reward signal to post-train an embodied policy with reinforcement learning.\"}]",1784183390,76,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":28},"do-vision-language-action-models-mean-what-they-say","",{"@graph":36,"@context":86},[37,54,69],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/do-vision-language-action-models-mean-what-they-say/82845/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":24,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-07-24","2026-07-16",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"What is the difference between functional reasoning and faithful reasoning in embodied chain-of-thought?","Question",{"text":76,"@type":77},"Functional reasoning means the reasoning trace improves task performance. Faithful reasoning means the trace accurately reflects the internal process that actually produces the action.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"Why are existing alignment strategies considered insufficient for faithfulness?",{"text":81,"@type":77},"They can produce reasoning whose intermediate steps mask causal links in action prediction via confounding factors, such as ungrounded reasoning that is disconnected or inconsistent with the policy’s true computation.",{"name":83,"@type":74,"acceptedAnswer":84},"How does Pinocchio operationalize embodied faithfulness and use it for training?",{"text":85,"@type":77},"Pinocchio learns a behavioral surrogate based on observation grounding and stepwise coherence, then provides a dense reward signal to post-train an embodied policy with reinforcement learning.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":93},[94,98,102,106,110,115,120,122,127,130,134],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":22,"slug":121},"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":20,"slug":137},19,"General","general"]