Neural and computational mechanisms of embodied learning: From brain-body interaction to translational applications.
TL;DR
The Integrative Embodied Learning Framework (IELF), a three-layer model linking computational mechanisms with overlapping neural systems and behavioral outcomes, is proposed, which generates testable predictions and extends embodied learning to developmental and evolutionary trajectories, neurological rehabilitation, and biologically inspired artificial intelligence.
Abstract
Background
Embodied learning views cognition as emerging from dynamic body-environment interactions, but evidence remains fragmented across neural, computational, developmental, and applied domains. This review aims to clarify how sensorimotor and interoceptive processes shape learning and to develop an integrative mechanistic framework.
Methods
We performed a narrative, interdisciplinary synthesis of theoretical and empirical evidence spanning systems and developmental neuroscience, computational modeling, neurorehabilitation, neurotechnology, and embodied artificial intelligence. Evidence was organized across neural substrates, learning algorithms, behavioral functions, developmental and evolutionary perspectives, and translational applications.
Results
Sensorimotor cortices and parietal-premotor networks integrate perception and action; the cerebellum implements error-driven forward models; basal ganglia and dopaminergic systems support reward-based action updating; and hippocampal and interoceptive-affective networks embed bodily experience in contextual memory and physiological regulation. Predictive coding and active inference, reinforcement learning, and Hebbian plasticity provide complementary computational accounts whose contributions vary with uncertainty, reward, and learning stage. Based on this synthesis, we propose the Integrative Embodied Learning Framework (IELF), a three-layer model linking computational mechanisms with overlapping neural systems and behavioral outcomes. IELF generates testable predictions and extends embodied learning to developmental and evolutionary trajectories, neurological rehabilitation, and biologically inspired artificial intelligence.
Conclusion
Learning emerges through recurrent brain-body-environment loops rather than isolated brain processes. IELF provides a unified mechanistic scaffold for testing embodied learning and developing personalized, closed-loop rehabilitation and adaptive artificial agents.