What is the difference between internal and external language?

Internal language is the innate system within our minds that determines the sound and meaning of infinite expressions, localized in the brain. External language is its manifestation in sounds or symbols used for communication.

Is language the most fundamental aspect of human cognition?

Language is considered fundamental, like vision, arising from our genetic endowment. It's seen as the core of human cognitive nature, the mode for constructing thoughts, and the basis for creative capacities.

Are there limits to human intelligence and cognition?

Yes, Chomsky argues that as biological organisms, humans have inherent limits to their cognitive capacities, similar to how genetic endowment enables certain traits while precluding others. This challenges the belief that intelligence can answer any question in principle.

Can technology like Neuralink expand human cognitive abilities?

Technology can expand cognitive capacity in limited ways, similar to how a book does, by augmenting existing abilities. However, Chomsky believes it cannot fundamentally alter our native cognitive capacities or allow us to understand things beyond our inherent scope, like the unintelligible aspects of quantum mechanics.

What are the limitations of deep learning in machine learning?

Chomsky distinguishes between engineering utility and scientific understanding. Deep learning, by processing vast examples to find patterns, is seen as engineering (like a Google translator) but not science, as it doesn't explain the fundamental nature of language and can perform well on non-grammatical structures.

Does societal evil stem from human nature or institutions?

Chomsky posits that evil and societal problems arise primarily from institutions, not inherent human nature. While human nature expresses itself through institutions, it can accommodate various types, and particular negative institutions are products of historical contingency.

What gives meaning to life?

The meaning of life is not a general answer but something we determine through our own activities and actions. The significance of one's life is created, not discovered.

Key Moments

Noam Chomsky: Language, Cognition, and Deep Learning | Lex Fridman Podcast #53

Q: What are the most fascinating aspects of language from a linguistic perspective?

A key fascinating property is 'structure dependence,' where grammatical interpretation relies on the structural relationships between words, not just their linear order, even when it seems computationally more complex and goes against immediate auditory input.

Lex Fridman

Science & Technology4 min read36 min video

Nov 29, 2019|556,762 views|15,932|848

Save to Pod

Key Moments

TL;DR

Noam Chomsky discusses language as a core cognitive faculty, limits of human cognition, and critiques deep learning.

Key Insights

Language is a fundamental, innate human cognitive faculty, not just an external communication system.

Human cognitive capacities, like biological traits, have inherent limits, influencing what we can understand.

Deep learning offers engineering utility but lacks scientific rigor, failing to uncover the underlying principles of language.

Institutions, not inherent nature, are the primary source of societal evil; human nature accommodates diverse social structures.

Scientific progress involves questioning the obvious and seeking intelligible explanations, even when they challenge established views.

The meaning of existence is not predetermined but created through individual actions and significance.

LANGUAGE AS A FUNDAMENTAL COGNITIVE FACULTY

Noam Chomsky posits that language is an innate, internal cognitive system, a unique human trait determined by our genetic endowment. It is not merely an external tool for communication but the very faculty that structures thought and enables creative, unbounded reasoning. Unlike a biological system like vision, which varies across species, the human language faculty is a universal, defining characteristic of our species, forming the basis for our higher cognitive abilities and unique achievements.

THE STRUCTURE AND NATURE OF LANGUAGE

Language operates on two distinct levels: the internal generative system within the mind and the external expression of sounds and meanings. The internal system dictates which language an individual speaks, based on genetic predispositions, and is localized within specific neural configurations. While language is often used for externalizing thoughts, its primary role is internal thought construction. This internal system possesses properties that, in their minimal form, resemble arithmetic operations, suggesting a potential basis for universal communication.

LIMITS OF HUMAN COGNITION AND UNDERSTANDING

Contrary to the belief that human intelligence can solve any problem, Chomsky argues that our cognitive capacities, as biological organisms, inherently possess scope and limits. Just as genetic endowments enable certain developments while precluding others, our minds are structured in ways that both grant richness and impose bounds. This is exemplified by the historical shift from seeking mechanical explanations of the universe to accepting mysteries, indicating a potential limit in our pursuit of an entirely intelligible world.

INSTINCTIVE CONCEPTIONS AND COGNITIVE BIASES

Humans possess instinctive conceptions of how the world works, such as a preference for physical contact and a tendency to impose regular geometric figures onto observations. These deeply ingrained cognitive biases, evident even in infants, suggest fundamental constraints on our perception and understanding. The mind actively constructs reality based on these predispositions, shaping our interpretation of sensory input and influencing what we can readily comprehend.

THE CRITIQUE OF DEEP LEARNING AND AI

Chomsky views deep learning as an engineering pursuit, not a scientific one. While useful for pattern recognition and practical applications like translation, it fails to uncover the underlying scientific principles of language. By processing vast amounts of data without theoretical grounding, deep learning systems can succeed even when violating fundamental linguistic rules. This approach is contrasted with science, which seeks critical experiments to test theoretical questions, highlighting the limitations of data-driven methods in achieving true understanding.

THE ROLE OF INSTITUTIONS IN HUMAN BEHAVIOR

Chomsky contends that societal evils and the development of specific institutions, like market systems, arise from historical contingency and choices, rather than inherent human nature. Human nature is adaptable and can accommodate a wide range of social structures, both cooperative and dominating. The expression of human nature is shaped by the cultural and historical context, suggesting that the prevalence of certain behaviors or institutions is not predetermined but a result of societal development and choices.

LINGUISTIC INSIGHTS AND STRUCTURE DEPENDENCE

A profoundly fascinating aspect of language is structure dependence: the principle that grammatical rules operate on the structural relationships within a sentence, not just linear proximity of words. This complex computation, which ignores linear cues, is universal across languages. Understanding this principle reveals a deep insight into the surprising nature of language and its underlying neural basis, demonstrating a sophisticated cognitive operation that goes beyond simple sequential processing.

THE QUEST FOR SCIENTIFIC INTELLIGIBILITY

The history of science reveals a persistent drive for intelligibility, the desire to understand the world through comprehensible mechanisms. However, certain scientific advancements, like Newton's discovery of action at a distance and the unintelligible nature of quantum mechanics, challenge this goal. This suggests that while science can describe phenomena, it may not always provide explanations that align with our innate conceptual frameworks, pointing to potential limits in what we can truly make sense of.

BRAIN-MACHINE INTERFACES AND COGNITIVE EXPANSION

While technologies like brain-machine interfaces aim to expand cognitive capacity, Chomsky suggests such advancements offer expansion in a known sense, akin to the utility of books, rather than a fundamental alteration of cognitive abilities. These tools may increase bandwidth or access to information but are unlikely to enable understanding of concepts fundamentally beyond our native cognitive architecture, just as they cannot impose radically different biological systems like an insect's vision.

THE MEANING OF EXISTENCE AND HUMAN AGENCY

Chomsky believes that the meaning of human existence is not a preordained universal answer but is actively created by individuals through their actions and pursuits. It is a matter of personal significance, not a fixed definition. This perspective emphasizes human agency in shaping their lives and contributing to the world, suggesting that meaning is derived from what we do and the impact we have, rather than from an external source or predetermined purpose.

Mentioned in This Episode

●Software & Apps

●Companies

●Organizations

●Books

●Concepts

●People Referenced

Common Questions

It's conceivable that communication with aliens could be established based on universal properties like arithmetic and core principles of human language, as suggested by experiments with simple Turing machines.

Topics

Artificial Intelligence Neuroscience & the Brain Mindset & Self-Improvement Technology & Innovation Science & Mathematics Society & Philosophy Philosophy Of Mind Human Nature Cognitive Science Limits Of Cognition

Mentioned in this video

People

Noam Chomsky

One of the most cited scholars in history, known for his work in linguistics, cognitive science, philosophy, and political thought. He spent over 60 years at MIT and recently joined the University of Arizona.

Woody Allen

A filmmaker whose comedic work, including a character worried about the universe expanding, is referenced by Chomsky to illustrate overcoming existential anxieties.

Marvin Minsky

A cognitive scientist and AI pioneer who conducted an experiment with Turing machines, suggesting a basis for universal communication with aliens through arithmetic.

Ernest Becker

An author who explored the human fear of mortality, a topic Chomsky discussed in relation to his own childhood worries and the meaning of existence.

David Hume

A prominent philosopher who commented on Newton's work, noting that by showing the imperfections of the mechanical philosophy, Newton left mysteries remaining for science.

Companies

Neuralink

A company working on brain-computer interfaces, aiming to read and stimulate neurons to expand cognitive capacity.

Software & Apps

Cash App

A finance app presented as a sponsor, offering features for sending money, buying Bitcoin, and investing in stocks, with a promotion for the FIRST organization.

Google parser

A tool for analyzing grammatical structure in text, used as an example by Chomsky to discuss the difference between engineering utility and scientific understanding in deep learning.

Turing Machines

Theoretical machines used in computing, which Minsky experimented with to suggest a potential basis for universal communication with intelligent alien species through arithmetic.

Organizations

FIRST

A STEM education non-profit that inspires students through robotics and Lego competitions, supported by Cash App and Lex Fridman.

MIT

Massachusetts Institute of Technology, where Noam Chomsky worked for over 60 years and where Lex Fridman first met him.

University of Arizona

University where Noam Chomsky recently joined and where the conversation with Lex Fridman took place.

Locations

Stata Center

A building at MIT where Lex Fridman had a memorable encounter with Noam Chomsky in an elevator.

Concepts

Schrödinger's cat

A thought experiment in quantum mechanics that illustrates the paradoxical nature of quantum superposition, which Chomsky uses to exemplify an unintelligible world.

Deep Learning

A subfield of machine learning based on neural networks that is progressing rapidly, but Chomsky questions its scientific value, contrasting it with engineering applications.

Tagalog

A language contrasted with English to illustrate the fact that our genetic endowment determines which language we speak.

English

One of the languages spoken by humans, contrasted with Tagalog to illustrate the internal system of language tied to genetic endowment.

corpus linguistics

A method of linguistic analysis that examines large collections of text (corpora) to identify patterns, similar to how deep learning operates, but Chomsky notes it's less effective than critical experiments.

Skinnerian behaviorism

A school of thought in psychology focusing on observable behavior and its environmental determinants, which Chomsky argues is falsely vindicated by some in deep learning.

structure dependence

A deep and puzzling property of language where grammatical rules depend on the structural relationships between words, not just their linear order, which Chomsky highlights as a fascinating aspect of linguistics.

paleoanthropology

The study of human prehistory through fossil and archaeological records, which Chomsky compares to corpus linguistics in its reliance on available data when critical experiments are impossible.

Media

Wall Street Journal

A publication whose corpus is used as an example for deep learning systems, which Chomsky argues focuses on coverage of random experiments rather than critical scientific inquiry.