Language and Intelligence: Embodied Cognition Perspective

Embodied Cognition Hypothesis

Cognitive processes are shaped by the activation of the agent's sensorimotor systems, influencing concepts, words, and thought.

Classical Cognitive Science Models (Rejected)

Sandwich Model

Views perception, cognition, and action as sequential, lacking feedback. The body is a tool for detection and execution.

Computer Model

The mind is a computational system transforming inputs into amodal symbols. The body is an input receiver, the brain is hardware, and mental representations are software.

Embodied Cognition Model

Cognition is embodied action, emphasizing the body's role. Agency and interaction with the environment are central. Higher cognitive processes depend on sensory and motor processes.

Mind, Body, and Environment

Emphasizes the sensorimotor coupling of living beings and the environment. Rejects the separation of mind from the mind–body–environment system. The mind includes external pathways for information.

Enactivism

Views humans as self-organizing dynamic systems that maintain identity by adjusting behavior to environmental changes. Cognitive processes emerge from recurrent sensorimotor interactions.

Embodiment of Concepts and Language

Concepts are abilities to interact, not representations of objects. Concepts are the reactivation of neural patterns from experienced objects. Language is grounded in sensorimotor experience; understanding language involves predicting action effects.

Action-Based Language Theory

Concepts are models of action. Understanding language involves simulating related sensorimotor experiences. There's a link between concepts, words, and action.

Embodied Intelligence

Intelligence is a dynamic adaptation to the sociocultural context, not a constant component. It develops through embodiment and interaction with the environment. Intelligent behavior results from the agent's body constraints and the environment.

Referential Competence

Referential competence is the degree of integration between symbolic and iconic representation systems. It reflects the connection between abstract levels and concrete, embodied experiences.
Referential competence is defined as the "degree of integration of symbolic and iconic representation systems, i.e., the strength of the symbol-referent links in semantic memory" p.594p. 594

Measuring Intelligence and Referential Competence

WAIS-IV

Measures intelligence using sub-indices like Verbal Comprehension, Perceptual Reasoning, Working Memory, and Processing Speed. It focuses on abstract, decontextualized cognitive skills.

Linguistic Analysis

Analyzes language content and style to measure referential competence. Measures include dictionaries of referential activity (WRAD), reflection (RefD), affect (AffD), and somatic sensations (SenSD).

Study Findings

Intelligence scores correlate with the use of words related to somatic and sensory sensations in narratives (SenSD) but not with other referential competence measures.
There is a positive correlation between sensory somatic dictionary (ISenSD) positively correlated with the overall scores of intelligence (FSIQ: r=.558r = .558, p=.002p = .002), the working memory scores (WMI: r=.531r = .531, p=.003p = .003), and the subtests of Digit Span (r=.487r = .487, p=.007p = .007) and Letter–Number Sequencing (r=.429r = .429, p=.020p = .020).
The study supports the embodied cognition hypothesis: sensorimotor experiences are basic to sense-making processes, concepts, and language.