Language, the Mind, and Computers: Comprehensive Cognitive Science Study Guide

The Nature of Language

• Core Concepts of Language:

  • Examples of Human Languages: English, French, Hindi, Japanese, Mandarin, Swahili, Tagalog, American Sign Language (ASL), among others.

  • Structural Elements: Sounds, Words, Phrases, Sentences, Speech, Signs, Rules, Grammar, and Hierarchical Structure.

  • Functional & Social Aspects: Talking, Speaking, Conversation, Communication, Interaction, Dialogue, Thought, Interpretation, Transmission, Reception, and Cultural Context.

  • Developmental Aspects: Learning, Development, Teaching, Practice, Cognitive Growth, Error Correction, Input Processing, Output Generation, Feedback Loops, and Fluency.

• Four Foundational Questions of Linguistics:

  1. Universality: What properties are shared across all human languages, and in what ways do languages systematically vary?

  2. Grammar: What structural principles and computational systems organize human languages?

  3. Purpose: What primary cognitive and social functions do languages perform?

  4. Origin: How does language come into existence across three distinct time scales?

     • In Individuals: Language Development / Ontogeny.

     • In Societies: Language Change / Diachronic Linguistics.

     • In Humans: Language Evolution / Phylogeny.

• Four Theoretical Dimensions to Characterize Language:

  1. External Behavior vs. Mind-Internal Organ:

     • External Verbal Behavior: Defines language as observable verbal habits and utterances. Championed by linguist Zellig Harris (1909–1992).

     • Internal Mental Organ: Defines language as internal mental knowledge and cognitive competence that enables humans to generate and comprehend speech. Championed by linguist Noam Chomsky.

2.  **Discrete Symbolic System vs. Continuous Distributed System**:
    *   *Discrete Symbolic System*: Language operates through formal, discrete symbols manipulated by rules (Noam Chomsky).
    *   *Continuous Distributed Network*: Language operates as a connectionist network of associative weights and continuous representations across distributed nodes (Psychologist David Rumelhart, 1942–2011).
3.  **Tool for Thought vs. Tool for Communication**:
    *   *Primary Tool for Thought*: Language evolved primarily as an internal representational medium for structured cognitive processing (Noam Chomsky).
    *   *Primary Tool for Communication*: Language evolved primarily to facilitate social interaction, shared intentionality, and cooperative communication (Psychologist Michael Tomasello).
4.  **Innate Capacity vs. Learned Skill**:
    *   *Innate Biological Organ*: Humans possess an genetically determined Universal Grammar or innate language faculty (Noam Chomsky; Steven Pinker in *The Language Instinct*; Robert C. Berwick & Noam Chomsky in *Why Only Us*).
    *   *Learned General-Purpose Skill*: Language is acquired through general cognitive learning mechanisms, pattern recognition, and social practice, akin to learning to drive or play the piano (Michael Tomasello in *Becoming Human*; Morten H. Christiansen & Nick Chater in *The Language Game*).

The Mind and the Mind-Body Problem

• Definition and Scope of the Mind:

  • The mind is an abstract theoretical construct rather than a tangible biological organ.

  • It encompasses mental faculties including Perception, Thought, Language, Emotion, Imagination, Memory, Desire, Decision-making, Attention, Learning, and Consciousness.

• The Mind-Body Problem:

  • Investigates which physical organ performs mental operations and how non-physical mental phenomena emerge from physical biological activity (Crane, 2025, The Mind-Body Problem, Open Encyclopedia of Cognitive Science).

• Historical Hypotheses on Mental Organization & Localization:

  • Plato & Socrates (Tripartite Theory of Psyche/Motivation):

    • Logos (Thought / Language / Reason): Located in the Head.

    • Thymos (Emotions / Spirit / Passion): Located in the Thorax / Heart.

    • Eros (Desires / Appetites): Located in the Liver / Abdomen.

  • Historical Proposals for the Anatomical Seat of Logos (Reasoning):

    • Thorax: Proposed by Parmenides (6th–5th Century BC) and Aristotle (384–322 BC), who asserted that speech emanates directly from the chest.

    • Blood in the Heart: Proposed by Empedocles (494–434 BC).

    • Temples: Proposed by Strato of Lampsacus (335–269 BC).

    • Meninges: Proposed by Erasistratus (304–250 BC).

    • Brain Ventricles: Proposed by Herophilos (335–280 BC).

• Galen's Empirical Discoveries (129–216 AD):

  • Physician, surgeon, and philosopher Galen theorized that mental functions depend on pneuma (a fluid or air-like spirit).

  • Experimental Evidence: Galen demonstrated that cutting the recurrent laryngeal nerve in living animals and humans caused an immediate loss of voice/vocal speech.

  • Conclusion: Because the recurrent laryngeal nerve branches from cranial nerves originating at the brain, Galen concluded that the brain is the physical organ controlling speech (logos) and mental activity.

• Metaphysical Approaches to Mind and Matter:

  • Substance Dualism: Asserts that mind and body consist of two fundamentally distinct substances. The body is a mechanical physical automaton, while the mind is an immaterial soul. Promoted by René Descartes (1596–1650).

*   **Monism**: Asserts that reality consists of a single unifying substance.
    *   *Idealism*: All existing phenomena are fundamentally mental.
    *   *Physicalism / Materialism*: All existing phenomena are fundamentally physical/material. Modern cognitive science adopts physicalism, viewing the nervous system as the main organ responsible for mental life.

Principles of Computation & Computational Systems

• Physical Realizability and Functionality:

  • Multiple Realizability: A single abstract function can be implemented by multiple, radically different physical systems. For instance, measuring time can be accomplished via a burning candle, an hourglass, a water clock (clepsydra), a mechanical stopwatch, or a digital quartz clock.

  • Multi-functionality: A single physical system can execute multiple distinct functions.

• Marr's Three Levels of Analysis (David Marr, 1982):

  1. Computational Level: Defines what problem the system solves and the abstract input-output mapping rules (e.g., specifying the function $f(x,y,z) = x + y + z$).

  2. Algorithmic Level: Defines how the computation is executed step-by-step, specifying representations and data structures (e.g., computing $f(x,y,z)$ via $((x+y)+z)$ versus $(x+(y+z))$).

  3. Implementational Level: Defines how the system is physically realized in hardware or biology (e.g., silicon integrated circuits, transistors, vs. biological networks of neurons, synapses, and neurotransmitters).

• Definition and Architecture of Computers:

  • A computer is any physical device that stores information and performs operations on it.

  • Information Storage: Realized as a distinct physical state of the device.

  • Operations: Realized as physical state transitions driven by mechanistic rules.

  • Historical Devices and Automata:

    • Antikythera Mechanism (2nd Century BC): Ancient mechanical analog gear system used to compute astronomical positions and predict eclipses.

    *   *Colossus* (1943): WWII vacuum-tube electronic digital computer used for cryptanalysis.
    *   *Mechanical Automata*: Self-operating physical devices designed to simulate organic behaviors:
        *   *The Digesting Duck* (1738/1764) by Jacques de Vaucanson.
        *   *The Writer*, *The Draughtsman*, and *The Musician* (1768–1774) by Pierre Jaquet-Droz and Henri-Louis Jaquet-Droz.

• Storage vs. Computation Trade-Off:

  • To process information, a system must establish a representation format and define valid operations.

  • Inverse Relationship:         $\text{More Memory/Storage} \longrightarrow \text{Less Required Real-Time Computation}$         $\text{More Real-Time Computation} \longrightarrow \text{Less Required Memory/Storage}$

  • Example: The Jaquet-Droz Writer automaton stores full character stroke trajectories on mechanical cam stacks, requiring minimal computation during execution.

  • Multiplication Example: Evaluating multi-digit arithmetic such as:         $XY \times ZW = (X \times Z) \times 10^2 + (X \times W) \times 10 + (Y \times Z) \times 10 + (Y \times W)$         Demonstrates algorithmic decomposition (high computation, low memory) versus storing a vast multiplication lookup table (high memory, zero online computation).

• Discrete vs. Continuous Systems & Digital vs. Analog:

  • Discrete Representations / Digital Computers: Store and manipulate information in distinct, segmented units (e.g., binary states $0/1$, mechanical counters, calculators, typewriters).

  • Continuous Representations / Analog Computers: Store and manipulate information along a continuous, smooth physical spectrum (e.g., vinyl records, cassette tapes, fluid dynamics, the Ishiguro Storm Surge Analog Computer 1960–1983).

• Three Classic Paradigms of Computational Models:

  1. Symbolic Rule-Based Models: Formulated by Alan Turing (1912–1954); processes discrete symbol strings via formal production rules.

  2. Parallel Distributed Processing (PDP) / Connectionist Models: Formulated by David Rumelhart (1942–2011); processes activation vectors across parallel interconnected neural networks.

  3. Information Theoretic Models: Formulated by Claude Shannon (1916–2001); quantifies communication through information entropy, signal transmission, channel capacity, and noise sources.

The Computational Theory of Mind & Cognitive Architecture

• The Computational Theory of Mind (CTM):

  • Posits that the human mind and human linguistic capabilities are formal computational systems (Rescorla, The Computational Theory of Mind, Stanford Encyclopedia of Philosophy).

• Architectural Dichotomy: Modularity vs. Interactivity:

  • Modular Architecture (Jerry Fodor, 1983, The Modularity of Mind):

    • Domain Specificity: Dedicated modules handle specific input types.

    • Mandatory Operation: Automatic, reflex-like processing.

    • Limited Central Access: Higher-level consciousness cannot inspect internal modular operations.

    • Informational Encapsulation: Modules process input insulated from central beliefs, expectations, or background knowledge.

    • Fixed Neural Architecture: Implemented in dedicated, hardwired neural circuits.

*   **Interactive Architecture**:
    *   *Domain Generality*: Shared cognitive resources across tasks.
    *   *Controllable Operations*: Executive conscious control over processing.
    *   *Full Accessibility*: Central access to internal representations.
    *   *Information Sharing*: Continuous top-down and bottom-up cross-talk.
    *   *Flexible Neural Architecture*: Dynamic, reconfigurable neural networks.

• Bottom-Up vs. Top-Down Processing:

  • Bottom-Up Processing: Processing driven strictly by incoming sensory data and physical surface features of the stimulus.

  • Top-Down Processing: Processing guided by higher-level prior knowledge, context, expectations, and mental concepts.

  • Contextual Ambiguity Example:

    • When an ambiguous visual glyph is presented between A and C, top-down reading context causes it to be perceived as the letter B (A B C).

    • When the exact same physical glyph is presented between 12 and 14, numerical context causes it to be perceived as the number 13 (12 13 14).

*   **Perceptual Encapsulation in Visual Illusions**:
    *   Visual illusions designed by experimental psychologist Akiyoshi Kitaoka (Ritsumeikan University) demonstrate informational encapsulation: even when top-down explicit knowledge confirms that an image is completely static, the visual perception module continues to perceive motion.
*   **Auditory Perception & Expectation**:
    *   The ambiguous acoustic recording "Yanny vs. Laurel" demonstrates how individual variations in frequency emphasis and top-down cognitive expectations filter identical sensory signals into distinct percepts.