The Way Forward for Grounded Cognition - Invariant Representations in Abstract Concept Grounding DOI Open Access
Jannis Friedrich, Martin H. Fischer, Markus Raab

et al.

Published: Feb. 2, 2024

Grounded cognition states that mental representations of concepts consist experiential aspects. For example, the concept ‘cup’ consists sensorimotor experiences from interactions with cups. Typical modalities in which are grounded are: The system (incl. interoception), emotion, action, language, and social Here we argue this list should be expanded to include physical invariants (unchanging features motion; e.g., gravity, momentum, friction). Research on causal perception reasoning consistently demonstrates represented as fundamentally other grounding substrates, therefore qualify. We assess several theories representation (simulation, conceptual metaphor, spaces, predictive processing) their positions invariants. Significant problems current state become evident. outline a solution based minimalist account cognition, is epistemologically secure likely foster falsifiable empirical work. conclude that, reduced scope, by including invariants, can progress past its impasse seriously contend established theoretical frameworks, providing valuable contribution understanding human cognition.

Language: Английский

Generating meaning: active inference and the scope and limits of passive AI DOI Creative Commons
Giovanni Pezzulo, Thomas Parr, Paul Cisek

et al.

Trends in Cognitive Sciences, Journal Year: 2023, Volume and Issue: 28(2), P. 97 - 112

Published: Nov. 15, 2023

Prominent accounts of sentient behavior depict brains as generative models organismic interaction with the world, evincing intriguing similarities current advances in artificial intelligence (AI). However, because they contend control purposive, life-sustaining sensorimotor interactions, living organisms are inextricably anchored to body and world. Unlike passive learned by AI systems, must capture sensory consequences action. This allows embodied agents intervene upon their worlds ways that constantly put best test, thus providing a solid bedrock is – we argue essential development genuine understanding. We review resulting implications consider future directions for AI.

Language: Английский

Citations

36
Neural representation in active inference: Using generative models to interact with—and understand—the lived world DOI
Giovanni Pezzulo, Leo D’Amato, Francesco Mannella

et al.

Annals of the New York Academy of Sciences, Journal Year: 2024, Volume and Issue: 1534(1), P. 45 - 68

Published: March 25, 2024

Abstract This paper considers neural representation through the lens of active inference, a normative framework for understanding brain function. It delves into how living organisms employ generative models to minimize discrepancy between predictions and observations (as scored with variational free energy). The ensuing analysis suggests that learns navigate world adaptively, not (or solely) understand it. Different may possess an array models, spanning from those support action‐perception cycles underwrite planning imagination; namely, explicit entail variables predicting concurrent sensations, like objects, faces, or people—to action‐oriented predict action outcomes. then elucidates belief dynamics might link implications different types agent's cognitive capabilities in relation its ecological niche. concludes open questions regarding evolution development advanced abilities—and gradual transition pragmatic detached representations. on offer foregrounds diverse roles play processes representation.

Language: Английский

Citations

11
Deep kinematic inference affords efficient and scalable control of bodily movements DOI Creative Commons
Matteo Priorelli, Giovanni Pezzulo, Ivilin Stoianov

et al.

Proceedings of the National Academy of Sciences, Journal Year: 2023, Volume and Issue: 120(51)

Published: Dec. 12, 2023

Performing goal-directed movements requires mapping goals from extrinsic (workspace-relative) to intrinsic (body-relative) coordinates and then motor signals. Mainstream approaches based on optimal control realize the mappings by minimizing cost functions, which is computationally demanding. Instead, active inference uses generative models produce sensory predictions, allows a cheaper inversion However, devising complex kinematic chains like human body challenging. We introduce an architecture that affords simple but effective via easily scales up drive chains. Rich can be specified in both using attractive or repulsive forces. The proposed model reproduces sophisticated bodily paves way for efficient biologically plausible of actuated systems.

Language: Английский

Citations

13
Collective predictive coding hypothesis: symbol emergence as decentralized Bayesian inference DOI Creative Commons
Tadahiro Taniguchi

Frontiers in Robotics and AI, Journal Year: 2024, Volume and Issue: 11

Published: July 23, 2024

Understanding the emergence of symbol systems, especially language, requires construction a computational model that reproduces both developmental learning process in everyday life and evolutionary dynamics throughout history. This study introduces collective predictive coding (CPC) hypothesis, which emphasizes models interdependence between forming internal representations through physical interactions with environment sharing utilizing meanings social semiotic within system. The total system is theorized from perspective . hypothesis draws inspiration studies grounded probabilistic generative language games, including Metropolis–Hastings naming game. Thus, playing such games among agents distributed manner can be interpreted as decentralized Bayesian inference shared by multi-agent Moreover, this explores potential link CPC free-energy principle, positing adheres to society-wide principle. Furthermore, paper provides new explanation for why large appear possess knowledge about world based on experience, even though they have neither sensory organs nor bodies. reviews past approaches offers comprehensive survey related prior studies, presents discussion CPC-based generalizations. Future challenges cross-disciplinary research avenues are highlighted.

Language: Английский

Citations

5
Interactive Inference: A Multi-Agent Model of Cooperative Joint Actions DOI Creative Commons
Domenico Maisto, Francesco Donnarumma, Giovanni Pezzulo

et al.

IEEE Transactions on Systems Man and Cybernetics Systems, Journal Year: 2023, Volume and Issue: 54(2), P. 704 - 715

Published: Oct. 20, 2023

We advance a novel computational model of multi-agent, cooperative joint actions that is grounded in the cognitive framework active inference. The assumes to solve task, such as pressing together red or blue button, two (or more) agents engage process interactive Each agent maintains probabilistic beliefs about goal (e.g., Should we press button?) and updates them by observing other agent's movements, while turn selecting movements make his own intentions legible easy infer (i.e., sensorimotor communication). Over time, inference aligns both behavioral strategies agents, hence ensuring success action. exemplify functioning simulations. first simulation illustrates "leaderless" It shows when lack strong preference their task goal, they jointly it each other's movements. In turn, this helps alignment strategies. second "leader–follower" one ("leader") knows true uses communication help ("follower") it, even if doing requires more costly individual plan. These simulations illustrate supports successful multi-agent reproduces key dynamics observed human–human experiments. sum, provides cognitively inspired, formal realize consensus MAS.

Language: Английский

Citations

11
Learning team-based navigation: a review of deep reinforcement learning techniques for multi-agent pathfinding DOI Creative Commons
Jaehoon Chung, Jamil Fayyad, Younes Al Younes

et al.

Artificial Intelligence Review, Journal Year: 2024, Volume and Issue: 57(2)

Published: Feb. 8, 2024

Language: Английский

Citations

4
Active Inference Tree Search in Large POMDPs DOI Creative Commons
Domenico Maisto, Francesco Gregoretti, Karl Friston

et al.

Neurocomputing, Journal Year: 2025, Volume and Issue: unknown, P. 129319 - 129319

Published: Jan. 1, 2025

Language: Английский

Citations

0
Dynamic planning in hierarchical active inference DOI Creative Commons
Matteo Priorelli, Ivilin Stoianov

Neural Networks, Journal Year: 2025, Volume and Issue: 185, P. 107075 - 107075

Published: Jan. 8, 2025

By dynamic planning, we refer to the ability of human brain infer and impose motor trajectories related cognitive decisions. A recent paradigm, active inference, brings fundamental insights into adaptation biological organisms, constantly striving minimize prediction errors restrict themselves life-compatible states. Over past years, many studies have shown how animal behaviors could be explained in terms inference - either as discrete decision-making or continuous control inspiring innovative solutions robotics artificial intelligence. Still, literature lacks a comprehensive outlook on effectively planning realistic actions changing environments. Setting ourselves goal modeling complex tasks such tool use, delve topic keeping mind two crucial aspects behavior: capacity understand exploit affordances for object manipulation, learn hierarchical interactions between self environment, including other agents. We start from simple unit gradually describe more advanced structures, comparing recently proposed design choices providing basic examples. This study distances itself traditional views centered neural networks reinforcement learning, points toward yet unexplored direction inference: hybrid representations models.

Language: Английский

Citations

0
Real- Time Failure/Anomaly Prediction for Robot Motion Learning Based on Model Uncertainty Prediction DOI
Hideyuki Ichiwara, Hiroshi Ito, Kenjiro Yamamoto

et al.

2022 IEEE/SICE International Symposium on System Integration (SII), Journal Year: 2024, Volume and Issue: unknown, P. 376 - 381

Published: Jan. 8, 2024

End-to-end robot motion generation methods using deep learning have achieved various tasks. However, due to insufficient training or the occurrence of abnormal conditions, model sometimes fails tasks unexpectedly. If failures/anomalies can be predicted before occurring, irreversible task failures prevented. In this paper, we propose a method predicting uncertainty predict in real-time. For naive method, used that predicts robot's actions stochastically and also tried failure/anomaly on basis variance. it was experimentally shown variance variation data cannot distinguished. Therefore, by likelihood model, which corresponds degree discrepancy between observations, real-time treating as applied prediction failure/anomaly. The method's effectiveness demonstrated achieving high judgment accuracy rate 85% (17/20 cases) an object-picking task.

Language: Английский

Citations

3
Embodied AI for dexterity-capable construction Robots: DEXBOT framework DOI

Hengxu You,

Tianyu Zhou, Qi Zhu

et al.

Advanced Engineering Informatics, Journal Year: 2024, Volume and Issue: 62, P. 102572 - 102572

Published: May 2, 2024

Language: Английский

Citations

3