IEEE Computational Intelligence Magazine, August 2007, IEEE press copyright Abstract— In this paper, we present some recent cognitive robotics studies on language and cognition integration to demonstrate how the language acquired by robotic agents can be directly grounded in action representations. These studies are characterized by the hypothesis that symbols are directly grounded into the agents’ own categori cal representations, while at the same time having logical (e.g. syntactic) relationships with other symbols. The two robotics studies are based on the combination of cognitive robotics with neural modeling methodologies, such as connectionist models and modeling field theory. Simulations demonstrate the efficacy of the mechanisms of action grounding of language and the symbol grounding transfer in agents that acquire a lexicon via imitation and linguistic instructions. The paper also discusses the scientific and technological implications of such an approach. I. INTRODUCTION Recent advances in cognitive psychology, neuroscience and linguistics support an embodied view of cognition, i.e. the fact that cognitive functions (perception, categorization, reasoning and language) are strictly intertwined with sensorimotor and emotional processes (Wilson 2002). This is particularly evident in recent studies on the grounding of language in action and perception (Pecher & Zwann 2004). For example, in psycholinguistics, Glenberg & Kaschak (2002) have demonstrated the existence of Action-sentence Compatibility Effects. In sentence comprehension tasks, participants are faster to judge the sensibility of sentences, implying motion toward the body (e.g. “Courtney gave you the notebook”) when the response requires moving toward the body (i.e. press a button nearer body). When the sentence implied movement away from the body, participants were faster to respond by literally moving away from their bodies (press a button farther from body). The data support an embodied theory of meaning that relates the meaning of sentences to human action and motor affordances. This is also consistent with neuroscientific studies on action and language, such as the involvement of the mirror neuron system for action and language learning (Rizzolatti & Arbib 1998), and brain imaging studies where words (e.g. action verbs) activate cortical areas (e.g. motor and premotor cortex) in a somatotopic fashion (Pulvermuller 1993). In linguistics, the link between the properties of language and their relationship with cognitive processes has been formalized by cognitive and constructivist linguistic theories (e.g. Talmy, 1980). This growing empirical evidence is consistent with recent advances in artificial intelligence and robotics, where the design of the capabilities of the artificial cognitive agents is based on an integrated cognitive approach (Perlovsky, this volume). For example, the design of the linguistic capabilities of interactive systems for human-robot communication are built (grounded) onto the robot’s other sensorimotor and cognitive skills (Cangelosi et al. 2005; Feldman & Narayanan 2004). Robots acquire words through direct interaction with their physical and social world, so that linguistic symbols do not Integrating Language and Cognition: A Cognitive Robotics Approach Angelo Cangelosi, Adaptive Behaviour and Cognition Research Group, University of Plymouth, Drake Circus, Plymouth PL4 8AA, UK (acangelosi@plymouth.ac.uk ) Vadim Tikhanoff, Adaptive Behaviour and Cognition Research Group, University of Plymouth, Drake Circus, Plymouth PL4 8AA, UK (acangelosi@plymouth.ac.uk ) José Fernando Fontanari, Instituto de Física de São Carlos, Universidade de São Paulo, 13560-970 São Carlos SP, Brazil Emmanouil Hourdakis, Institute of Computer Science, University of Crete, Greece