From State Transitions to Sensory Regularity: Structuring Uninterpreted Sensory Signals from Naive Sensorimotor Experiences - Architectures et Modèles de l'Adaptation et de la Cognition Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2022

From State Transitions to Sensory Regularity: Structuring Uninterpreted Sensory Signals from Naive Sensorimotor Experiences

Résumé

How could a naive agent build some internal, subjective, notions of continuity in its sensorimotor experiences? This is a key question for all sensorimotor approaches to perception when trying to make them face realistic interactions with an environment, including noise in the perceived sensations, errors in the generation of motor trajectories, or uncertainties in the agent's internal representation of this interaction. This paper proposes a detailed formalization, but also some experimental assessments, of the structure a naive agent can leverage from its own uninterpreted sensorimotor flow to capture a subjective sensory continuity, making it able to discover some notions of closeness or regularities in its experience. The precise role of the agent's actions is also questioned w.r.t. the spatial and temporal dynamics of its exploration of the environment. On this basis, the previous authors' contribution on sensory prediction is extended to successfully handle noisy data in the agent's sensorimotor flow.
Fichier principal
Vignette du fichier
root_sensory_topology.pdf (8.18 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03537409 , version 1 (20-01-2022)
hal-03537409 , version 2 (21-06-2022)
hal-03537409 , version 3 (28-09-2022)
hal-03537409 , version 4 (10-03-2023)

Identifiants

  • HAL Id : hal-03537409 , version 3

Citer

Loïc Goasguen, Jean-Merwan Godon, Sylvain Argentieri. From State Transitions to Sensory Regularity: Structuring Uninterpreted Sensory Signals from Naive Sensorimotor Experiences. 2022. ⟨hal-03537409v3⟩
163 Consultations
96 Téléchargements

Partager

Gmail Facebook X LinkedIn More