The Embodied Crossmodal Self Forms Language and Interaction: A Computational Cognitive Review.

Area of Science:

• Cognitive Science
• Robotics
• Computational Linguistics

Background:

• Human language relies on embodied, sensorimotor representations of the self and world.
• Ideomotor action-effect associations are crucial for language processing and interaction.
• Existing computational models often separate verbal (dialog systems) and non-verbal (robotics) interaction.

Purpose of the Study:

• To bridge the gap between verbal and non-verbal interaction in computational models.
• To investigate the role of computational models of the self in language grounding.
• To advance human-robot interaction by integrating embodied cognition principles.

Main Methods:

• Reviewing psychological and cognitive literature on ideomotor theory and language grounding.
• Analyzing existing computational methods for physical decision-making and verbal interaction.
• Synthesizing findings to propose integrated computational interaction models.

Main Results:

• Identified ideomotor theory as a key mechanism for embodied language grounding.
• Highlighted the lack of integrated computational approaches for joint verbal and physical interaction.
• Outlined a pathway for developing advanced models integrating self-representation, body schema, and language.

Conclusions:

• Computational models of the self offer significant potential for understanding language grounding.
• Integrating verbal and non-verbal communication in computational models can improve human-robot interaction.
• This review provides a foundational step towards developing such integrated models.