A connectionist approach to embodied conceptual metaphor

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Embodied Cognition

Background:

• The embodied cognition paradigm posits that cognition is fundamentally grounded in sensory-motor processes.
• This view challenges traditional cognitive science's focus on abstract computational principles.
• Existing embodied cognition theories sometimes lack the detailed explanatory power of computational models.

Purpose of the Study:

• To demonstrate how connectionist architectures can elucidate findings in embodied cognition research.
• To show that connectionist models can bridge the gap between sensory-motor grounding and abstract cognition.
• To extend a connectionist model to simulate embodied conceptual metaphor, specifically time-space relationships.

Main Methods:

• Utilizing connectionist architectures capable of learning statistical environmental structures.
• Developing overlapping, distributed representations within connectionist models.
• Extending an existing connectionist model of semantic cognition to simulate embodied conceptual metaphor findings.

Main Results:

• Connectionist models effectively learn environmental statistical structures, explaining the emergence of higher-level cognition from sensory-motor input.
• Overlapping, distributed representations in these models align with requirements of embodied cognition theories.
• Simulations demonstrated how concrete spatial experiences, including cultural cues, can structure abstract concepts like time.

Conclusions:

• Connectionist models offer a powerful framework for understanding embodied cognition, particularly how abstract concepts arise from sensory-motor grounding.
• An integrated approach combining embodied cognition research and connectionist modeling enhances understanding of cognitive processes.
• This integration can illuminate complex cognitive phenomena, such as embodied conceptual metaphors relating space and time.