The Vector Grounding Problem
Auteurs : Dimitri Coelho Mollo, Raphaël Millière
Résumé scientifique
The remarkable performance of large language models (LLMs) on complex linguistic tasks has sparked a lively debate on the nature of their capabilities. Unlike humans, these models learn language exclusively from textual data, without direct interaction with the real world. Nevertheless, they can generate seemingly meaningful text about a wide range of topics. This impressive accomplishment has rekindled interest in the classical 'Symbol Grounding Problem,' which questioned whether the internal representations and outputs of classical symbolic AI systems could possess intrinsic meaning. Unlike these systems, modern LLMs are artificial neural networks that compute over vectors rather than symbols. However, an analogous problem arises for such systems, which we dub the Vector Grounding Problem. This paper has two primary objectives. First, we differentiate various ways in which internal representations can be grounded in biological or artificial systems, identifying five distinct notions discussed in the literature: referential, sensorimotor, relational, communicative, and epistemic grounding. Unfortunately, these notions of grounding are often conflated. We clarify the differences between them, and argue that referential grounding is the one that lies at the heart of the Vector Grounding Problem. Second, drawing on theories of representational content in philosophy and cognitive science, we propose that certain LLMs, particularly those fine-tuned with Reinforcement Learning from Human Feedback (RLHF), possess the necessary features to overcome the Vector Grounding Problem, as they stand in the requisite causal-historical relations to the world that underpin intrinsic meaning. We also argue that, perhaps unexpectedly, multimodality and embodiment are neither necessary nor sufficient conditions for referential grounding in artificial systems.
Les articles autour de lui, par date et par proximité
Plus haut, c’est plus proche de cet article ; à gauche, c’est plus ancien. Les traits relient les articles proches entre eux : les groupes de traits sont des sous-thèmes. Sélectionnez un point pour voir pourquoi il est là. Retour au résumé de cet article
Les articles autour de celui-ci
Classés selon la proximité de leur titre et de leur résumé avec ceux de cet article. Sélectionnez-en pour les lire ensemble.
Les articles les plus proches
-
"Understanding AI": Semantic Grounding in Large Language ModelsHolger Lyre · 2024 · Computation and Language · preprint, non relu par les pairs · cité 13 fois · Résumé IA disponible
Résumé scientifique
Do LLMs understand the meaning of the texts they generate? Do they possess a semantic grounding? And how could we understand whether and what they understand? I start the paper with the observation that we have recently witnessed a generative turn in AI, since generative models, including LLMs, are key for self-supervised learning. To assess the question of semantic grounding, I distinguish and discuss five methodological ways. The most promising way is to apply core assumptions of theories of meaning in philosophy of mind and language to LLMs. Grounding proves to be a gradual affair with a three-dimensional distinction between functional, social and causal grounding. LLMs show basic evidenc…
-
The Mechanistic Emergence of Symbol Grounding in Language ModelsShuyu Wu, Ziqiao Ma, Xiao-Xi Luo et al. · 2025 · arXiv.org · preprint, non relu par les pairs · cité 6 fois · Pas encore résumé
Résumé scientifique
Symbol grounding (Harnad, 1990) describes how symbols such as words acquire their meanings by connecting to real-world sensorimotor experiences. Recent work has shown preliminary evidence that grounding may emerge in (vision-)language models trained at scale without using explicit grounding objectives. Yet, the specific loci of this emergence and the mechanisms that drive it remain largely unexplored. To address this problem, we introduce a controlled evaluation framework that systematically traces how symbol grounding arises within the internal computations through mechanistic and causal analysis. Our findings show that grounding concentrates in middle-layer computations and is implemented …
-
A Categorical Analysis of Large Language Models and Why LLMs Circumvent the Symbol Grounding ProblemLuciano Floridi, Yi-Yang Jia, Fernando A. Tohmé · 2025 · arXiv.org · preprint, non relu par les pairs · cité 5 fois · Pas encore résumé
Résumé scientifique
This paper presents a formal, categorical framework for analysing how humans and large language models (LLMs) transform content into truth-evaluated propositions about a state space of possible worlds W , in order to argue that LLMs do not solve but circumvent the symbol grounding problem.
-
Will multimodal large language models ever achieve deep understanding of the world?Igor Farkaš, Michal Vavrecka, S. Wermter · 2025 · Frontiers in Systems Neuroscience · cité 6 fois · Pas encore résumé
Résumé scientifique
Despite impressive performance in various tasks, large language models (LLMs) are subject to the symbol grounding problem, so from the cognitive science perspective, one can argue that they are merely statistics-driven distributional models without a deeper understanding. Modern multimodal versions of LLMs (MLLMs) are trying to avoid this problem by linking language knowledge with other modalities such as vision (Vision Language Models called VLM) or action (Vision Language Action Models called VLA) when, for instance, a robotic agent, is acting in the world. If eventually successful, MLLMs could be taken as pathway for symbol grounding. In this work, we explore the extent to which MLLMs int…
-
Disembodied Meaning? Generative AI and UnderstandingJordi Vallverdú, I. Redondo · 2025 · Forum for Linguistic Studies · cité 3 fois · Pas encore résumé
Résumé scientifique
This study explores the cognitive and philosophical implications of Large Language Models (LLMs), focusing on their ability to generate meaning without embodiment. Grounded in the coherence-based semantics framework, the research challenges traditional views that emphasize the necessity of embodied cognition for meaningful language comprehension. Through a theoretical and comparative analysis, this paper examines the limitations of embodied cognition paradigms, such as the symbol grounding problem and critiques like Searle’s Chinese Room, and evaluates the practical capabilities of LLMs. The methodology integrates philosophical inquiry with empirical evidence, including case studies on LLM p…
-
Evaluating Large Language Models on the Frame and Symbol Grounding Problems: A Zero-shot BenchmarkShoko Oka · 2025 · arXiv.org · preprint, non relu par les pairs · Pas encore résumé
Résumé scientifique
Recent advancements in large language models (LLMs) have revitalized philosophical debates surrounding artificial intelligence. Two of the most fundamental challenges - namely, the Frame Problem and the Symbol Grounding Problem - have historically been viewed as unsolvable within traditional symbolic AI systems. This study investigates whether modern LLMs possess the cognitive capacities required to address these problems. To do so, I designed two benchmark tasks reflecting the philosophical core of each problem, administered them under zero-shot conditions to 13 prominent LLMs (both closed and open-source), and assessed the quality of the models' outputs across five trials each. Responses w…
-
Are Large Language Models Intentional? The Limits of Referential GroundingM. García-Valdecasas · 2026 · Philosophy & Technology · cité 7 fois · Pas encore résumé
Résumé scientifique
This paper reassesses Searle’s Chinese Room argument in light of Large Language Models (LLMs). While Searle criticized computational systems for being mere “syntax manipulators,” transformer models of the kind exemplified by LLMs complicate this characterization. Earlier computational systems relied on a more rigid logical foundation, based largely on pattern recognition and mechanical substitution. By tracking patterns of use, maintaining rich contextual dependencies, and exploiting implicit regularities in large datasets, LLMs show an unprecedented capacity for semantic modelling. And yet, LLMs remain limited in several respects. For instance, even the most advanced AI model suffers from i…
-
Meaning without reference in large language modelsS. Piantadosi, F. Hill · 2022 · arXiv.org · preprint, non relu par les pairs · cité 147 fois · Pas encore résumé
Résumé scientifique
The widespread success of large language models (LLMs) has been met with skepticism that they possess anything like human concepts or meanings. Contrary to claims that LLMs possess no meaning whatsoever, we argue that they likely capture important aspects of meaning, and moreover work in a way that approximates a compelling account of human cognition in which meaning arises from conceptual role. Because conceptual role is defined by the relationships between internal representational states, meaning cannot be determined from a model's architecture, training data, or objective function, but only by examination of how its internal states relate to each other. This approach may clarify why and …
Les travaux antérieurs sur lesquels il s’appuie
-
The Symbol Grounding ProblemS. Harnad · 1990 · arXiv.org · preprint, non relu par les pairs · cité 3 966 fois · Pas encore résumé
Résumé scientifique
There has been much discussion recently about the scope and limits of purely symbolic models of the mind and about the proper role of connectionism in cognitive modeling. This paper describes the "symbol grounding problem": How can the semantic interpretation of a formal symbol system be made intrinsic to the system, rather than just parasitic on the meanings in our heads? How can the meanings of the meaningless symbol tokens, manipulated solely on the basis of their (arbitrary) shapes, be grounded in anything but other meaningless symbols? The problem is analogous to trying to learn Chinese from a Chinese/Chinese dictionary alone. A candidate solution is sketched: Symbolic representations m…
-
A Philosophical Introduction to Language Models -- Part I: Continuity With Classic DebatesRaphaël Millière, Cameron Buckner · 2024 · Computation and Language · preprint, non relu par les pairs · cité 62 fois · Résumé IA disponible
Résumé scientifique
Large language models like GPT-4 have achieved remarkable proficiency in a broad spectrum of language-based tasks, some of which are traditionally associated with hallmarks of human intelligence. This has prompted ongoing disagreements about the extent to which we can meaningfully ascribe any kind of linguistic or cognitive competence to language models. Such questions have deep philosophical roots, echoing longstanding debates about the status of artificial neural networks as cognitive models. This article -- the first part of two companion papers -- serves both as a primer on language models for philosophers, and as an opinionated survey of their significance in relation to classic debates…
-
Do As I Can, Not As I Say: Grounding Language in Robotic AffordancesBrian Ichter, Anthony Brohan, Yevgen Chebotar et al. · 2022 · Conference on Robot Learning · cité 4 131 fois · Pas encore résumé
Résumé scientifique
Large language models can encode a wealth of semantic knowledge about the world. Such knowledge could be extremely useful to robots aiming to act upon high-level, temporally extended instructions expressed in natural language. However, a significant weakness of language models is that they lack real-world experience, which makes it difficult to leverage them for decision making within a given embodiment. For example, asking a language model to describe how to clean a spill might result in a reasonable narrative, but it may not be applicable to a particular agent, such as a robot, that needs to perform this task in a particular environment. We propose to provide real-world grounding by means …
-
Can structural correspondences ground real world representational content in Large Language Models?Iwan Williams · 2025 · Mind & Language · cité 11 fois · Pas encore résumé
Résumé scientifique
Basic large language models (LLMs) have no direct contact with extra‐linguistic reality: Their inputs, outputs and training data consist solely of text. Can they represent the world beyond that text, nevertheless? This paper considers whether LLMs represent real‐world domains partly thanks to structural correspondences between their internal states and those domains. I clarify the requirements for a structural correspondence to play a genuinely content‐grounding role, and argue (i) that it is a live empirical possibility that text‐based LLMs meet those requirements with respect to real‐world contents, but (ii) that this requires carefully eliminating alternative, more deflationary, content a…
-
Embers of autoregression show how large language models are shaped by the problem they are trained to solveR. Thomas McCoy, Shun-Yu Yao, Dan Friedman et al. · 2024 · Proceedings of the National Academy of Sciences of the United States of America · cité 211 fois · Pas encore résumé
Résumé scientifique
Significance ChatGPT and other large language models (LLMs) have attained unprecedented performance in AI. These systems are likely to influence a diverse range of fields, such as education, intellectual property law, and cognitive science, but they remain poorly understood. Here, we draw upon ideas in cognitive science to show that one productive way to understand these systems is by analyzing the goal that they were trained to accomplish. This perspective reveals some surprising limitations of LLMs, including difficulty on seemingly simple tasks such as counting words or reversing a list. Our empirical results have practical implications for when language models can safely be used, and the…
Les articles qui le citent
-
Dissociating language and thought in large language modelsKyle Mahowald, Anna A. Ivanova, I. Blank et al. · 2024 · Trends in Cognitive Sciences · cité 617 fois · Pas encore résumé
Résumé scientifique
Large Language Models (LLMs) have come closest among all models to date to mastering human language, yet opinions about their linguistic and cognitive capabilities remain split. Here, we evaluate LLMs using a distinction between formal linguistic competence—knowledge of linguistic rules and patterns—and functional linguistic competence—understanding and using language in the world. We ground this distinction in human neuroscience, which has shown that formal and functional competence rely on different neural mechanisms. Although LLMs are surprisingly good at formal competence, their performance on functional competence tasks remains spotty and often requires specialized fine-tuning and/or co…
-
Does Conceptual Representation Require Embodiment? Insights From Large Language ModelsQi-Hui Xu, Ying-Fang Peng, Minghua Wu et al. · 2023 · arXiv.org · preprint, non relu par les pairs · cité 13 fois · Pas encore résumé
Résumé scientifique
To what extent can language alone give rise to complex concepts, or is embodied experience essential? Recent advancements in large language models (LLMs) offer fresh perspectives on this question. Although LLMs are trained on restricted modalities, they exhibit human-like performance in diverse psychological tasks. Our study compared representations of 4,442 lexical concepts between humans and ChatGPTs (GPT-3.5 and GPT-4) across multiple dimensions, including five key domains: emotion, salience, mental visualization, sensory, and motor experience. We identify two main findings: 1) Both models strongly align with human representations in non-sensorimotor domains but lag in sensory and motor a…
-
Textual Identifiability and Anchored Semantics for Evaluating World-Model Claims in Large Language ModelsZihang Lan, Jin-Yu Yang · 2026 · IEEE Access · Pas encore résumé
Résumé scientifique
Claims that large language models (LLMs) learn world models often conflate decodability, causal employment, and external anchoring. This Theory article separates these claims and makes their evidential requirements computable. A text-only learner is represented as a Markov kernel: equal corpus laws imply equal learned-state laws, while corpus and training-procedure discrepancies yield explicit discrimination bounds on a declared finite horizon. For internal representations, a conditional localization model identifies a target-loading subspace, and a nullspace construction determines whether a target can be changed while specified associated readouts remain invariant. Rank, stability, and int…
-
Mechanistic indicators of understanding in large language modelsP. Beckmann, Matthieu Queloz · 2025 · Philosophical Studies · cité 14 fois · Pas encore résumé
Résumé scientifique
Large language models are often portrayed as merely imitating linguistic patterns without genuine understanding. We argue that recent findings in mechanistic interpretability, the emerging field probing the inner workings of LLMs, render this picture increasingly untenable—but only once those findings are integrated within a theoretical account of understanding. We propose a tiered framework for thinking about understanding in LLMs and use it to synthesize the most relevant findings to date. The framework distinguishes three hierarchical varieties of understanding, each tied to a corresponding level of computational organization: conceptual understanding emerges when a model forms “features”…
-
Beyond Hooking Onto the World: Referential Profiles and the Numerical Structure of LLM GroundingJoohwan Rhee · 2026 · arXiv.org · preprint, non relu par les pairs · Pas encore résumé
Résumé scientifique
This paper revisits the grounding problem for large language models in light of recent vector-grounding accounts. I accept the shift from classical symbol grounding to vector grounding, but argue that the current debate remains incomplete in two respects. First, reference is often treated too thinly, as if it were a fixed link between an isolated expression and an object. I argue instead that reference is profile-based, context-sensitive, discourse-level, affectively shaped, and norm-governed. Even in the human case, reference is publicly stabilized through patterns of use, correction, distinction, inference, and continuation rather than through identical private representations. Second, vec…
Aller plus loin
Comment ces articles sont-ils trouvés ?
Nous comparons le titre et le résumé de cet article avec ceux de tous les articles résumés sur ce site, et avec les articles que Semantic Scholar recommande, ceux qu’il cite et ceux qui le citent (pour un article très cité, ses citations les plus récentes et, depuis OpenAlex, les plus citées). Parmi les travaux antérieurs et les suites, les articles qui s’utilisent en profondeur, une méthode qu’il utilise et les suites très citées remontent un peu. Les niveaux (très proche, proche, apparenté) viennent de la similarité des textes, pas d’un jugement de qualité ; * signale un article sans résumé, comparé par son titre. La phrase sous un article est écrite par une IA à partir des deux résumés : vérifiez-la dans les articles. Un article jugé hors sujet par trois lecteurs quitte la liste.
Mis à jour le 6 novembre 2026.