Cognitive and Computational Architectures for Embodied and Interactive Intelligence

Embodied intelligence arises from the dynamic coupling of brain, body, environment, and other agents. In biological systems, adaptive behaviour is grounded in sensorimotor loops through which neural processes continuously integrate perception, action, prediction, memory, and learning. Body morphology, material properties, and physical dynamics are not passive constraints, but active resources that shape computation and behaviour.

Neurorobotics translates these principles into artificial systems by studying how neural, cognitive, and computational architectures can exploit embodiment to generate robust, adaptive, and interactive capabilities. Recent advances in world models, foundation models, neuromorphic computing, and distributed agentic systems offer new opportunities to build robots that learn from experience, coordinate across multiple timescales, and adapt through interaction. Yet, many current approaches still separate neural processing, bodily dynamics, reasoning, and social interaction.

Progress requires integrated architectures in which cognition emerges from continuous reciprocal exchanges among neural computation, physical embodiment, environmental structure, humans, and other robots over time.

The Research Topic goal is to explore how cognitive principles and computational architectures can be combined to create intelligent systems that do not only process information, but actively engage with, learn from, and adapt to their environments.

Existing approaches often treat perception, action, reasoning, learning, bodily dynamics, and interaction as loosely-connected components. This fragmentation limits their capacity to exploit reciprocal dependencies among neural computation, morphology, physical experience, environmental structure, and social exchange. Cognitive and computational architectures in which embodiment and interaction are treated as constitutive elements of intelligence rather than external constraints or application domains can close this gap.

We seek contributions investigating how robots can construct and continuously update body, world, and other-agent models; coordinate processes operating across multiple spatial and temporal scales; reason and act under uncertainty; and adapt through sustained interaction with physical environments, humans, and other robots. Emphasis is placed on integrated theoretical, computational, and experimental frameworks connecting brain-inspired mechanisms with emerging approaches in foundation models, neuromorphic systems, hybrid and neurosymbolic reasoning, collective intelligence, and distributed agentic architectures.

The broader goal is to identify principles, representations, and mechanisms supporting robust, transferable, explainable, and socially situated embodied intelligence across diverse robotic platforms and real-world contexts over extended deployments.

This Research Topic welcomes theoretical, computational, and experimental contributions on cognitive and computational architectures for embodied and interactive intelligence. Areas of interest include, but are not limited to:


• brain-inspired and neurorobotic architectures;

• body, world, and other-agent models;

• sensorimotor learning, prediction, and adaptation;

• morphology-aware computation and control;

• cognitive architectures integrating perception, action, memory, and reasoning;

• foundation models and multimodal models for embodied robotics;

• neuromorphic, hybrid, and neurosymbolic approaches;

• human–robot interaction and socially situated intelligence;

• multi-robot, collective, and distributed cognitive systems;

• reasoning and decision-making under uncertainty;

• bio-inspired and nature-inspired computational architectures;

• autonomous reasoning and decision-making under uncertainty;

• lifelong, developmental, and continual learning;

• experimental validation in complex physical and social environments.

• ethical, societal and regulatory aspects of embodied intelligence.