Computational framework demonstrates conceptual knowledge acquisition in autonomous agents, suggesting grounded sensory interaction pathways for artificial intelligence systems.
Key Points
To introduce Neo, an autonomous agent architecture designed to acquire abstract conceptual knowledge directly through sensorimotor interaction with an environment.
Designed an autonomous agent framework (Neo) that integrates sensory input with motor actions.
Modeled learning processes that construct higher-level conceptual representations from low-level physical observations and interactions.
Demonstrated that autonomous agents can generate grounded conceptual knowledge structures directly from environmental interactions.
Showed that sensorimotor experience provides sufficient empirical grounding to form abstract concepts without pre-programmed symbolic definitions.