Beyond a Concrete: Workshop on Grounding Abstract Concepts in Multimodal Interaction

Abstract concepts pose a fundamental challenge for multimodal interactive systems, as they cannot be grounded in single perceptual features or fixed motor patterns; instead, they require abstraction across perception, action, language, and affect. Examples include superordinate categories (e.g., animal, tool), generalised actions (e.g., use, make), and evaluative concepts (e.g., good, appropriate) that depend on the interaction context.

This workshop focuses on computational and interactional mechanisms for grounding abstract concepts beyond concrete sensory inputs, bringing together work on multimodal representation learning, embodied and developmental models, robotics, and hybrid cognitive architectures. The workshop aims to consolidate approaches, datasets, and evaluation strategies for abstract concept grounding, identify shared modelling assumptions, and clarify open challenges.

Expected outcomes include a comparative discussion of architectures and benchmarks and a post-workshop summary outlining research gaps and future directions, supporting multimodal interactive systems with improved generalisation and interpretability.

The workshop will be held as part of ICMI '26 in Naples, Italy.

Topics of Interest

Topics include, but are not limited to:

• Grounding abstract concepts in multimodal interaction
• Hierarchical and developmental models of abstraction
• Cross-modal abstraction and alignment
• Embodied and interactive concept learning
• Multimodal action and goal representations
• Affective and contextual abstraction
• Embodied and interactive learning of abstract language
• Hybrid cognitive architectures for abstract grounding
• Evaluation methodologies for abstract concept grounding
• Multimodal datasets and benchmarks for abstract concepts
• Applications of abstract grounding in HCI and HRI