Empiricism, Rationalism, and Language: Lecture Notes on Knowledge, Perception, and Innate Structure

Empiricism: Core Idea and Historical Context

• Core claim: Knowledge originates from experience; external objects cause perceptual input, and the mind forms ideas from those impressions.
• Locke (17th century) on the source of ideas:
  • Senses convey distinct perceptions of sensible qualities (e.g., yellow, white, heat, cold, soft, hard, bitter, sweet).
  • The things in our mind are in fact conveyed to it by our experiences with objects in the world.
  • Source of knowledge is experience; a loop from world to mind.
  • Phrase to remember: "The source of knowledge is experience"; our beliefs are not built in but arise from experience with things out there.
• Early lineage and echoes:
  • Philosophical tradition stretching back 18 centuries, with ideas echoed by earlier North African, Greek, Hindu philosophies.
  • Tabula rasa (blank slate) concept: we start blank and are shaped by experience; this is a kernel in many empiricist narratives.
• David Hume (18th century) and the problem of foundations:
  • All our simple ideas, at their first appearance, come from simple impressions; ideas derive from impressions.
  • Emphasizes that knowledge begins with sensory input but leaves open how the mind structures those impressions into reliable knowledge.
• Problems for empiricism identified in the lecture:
  • Problem 1: Epistemic reliability of the senses
  • Sensory illusions show that seeing does not guarantee truthful information about the world (e.g., visual color/gray-scale illusion, size illusion with circles, motion or brightness cues).
  • Classic examples shown in class: two gray boxes that appear different but are the same color; middle circles that appear different sizes but are the same size; moon illusion on the horizon vs. in the sky.
  • Takeaway: Senses alone do not always convey truthful information; perception can be tricked by contextual cues (lines, shadows, surrounding stimuli).
  • Problem 2: Knowledge of non-sensory concepts (a priori ideas)
  • Geometry and arithmetic: triangles, circles, parallel lines, and the ratio π remain true even though they are not directly given by sensory experience.
  • Example discussions: triangles and the Pythagorean theorem; circles and the ratio of circumference to diameter; infinite digits of π; parallel lines having no end in reality.
  • Frogs and species knowledge: knowledge about species-wide properties (e.g., frogs’ birth in water, tadpole metamorphosis, general properties of frog species) extends beyond any single sensory instance.
  • The existence and truth of mathematical and abstract entities (like prime numbers) seem to hold regardless of concrete sensory experience.
• Core tension illustrated by famous thought experiments:
  • Descartes’ skeptical possibility (the evil demon or deceiving senses): if senses can mislead, how can knowledge be grounded in them?
  • Plato’s cave analogy: potential misperception by relying on shadows—what we think we know could be misled by appearances.
  • The Matrix-like modern retelling: the possibility that sensory input might not reflect the world as it is.
• The synthesis and takeaway from the empiricist critique:
  • Knowledge seems to require more than sensory input; the mind actively reflects on and organizes sensory data.
  • The role of reflection ( Locke's term) in forming new ideas from sensory input by turning raw experience into structured knowledge.
  • Hume’s emphasis on impressions as the starting point is compatible with the sense that there is more to knowledge than raw data, but empiricism struggles to explain how non-sensory truths arise.
• Philosophical implications and non-binary stance:
  • Empiricism is not necessarily opposed to built-in cognitive resources; a non-binary view allows domain-specific balance (e.g., mathematics may require more rationalist structure, while social facts might be learned from experience).
  • The lecture suggests that knowledge involves both impressions and mental organization; a strict dichotomy between empiricism and rationalism is insufficient.

Rationalism and the Idea of Built-in Mental Structure

• Core claim: Knowledge arises from a combination of sensory input and structure imposed by the mind (innate or evolutionarily evolved priors).
• Locke’s idea of reflection and the rise of the rationalist response:
  • The mind imposes organization on sensory experiences; knowledge is not just raw data but structured processing.
• Rationalism vs empiricism is not a strict binary; a nuanced view is more accurate:
  • Some domains may rely more on built-in structure (e.g., grammar-like rules in language) while others may rely more heavily on experience.
  • Evolution provides a mechanism for built-in biases and organize processing that shape how we learn from experience.
• Mechanisms proposed for built-in structure:
  • Bias to generalize: from a few observed instances, we infer broader general properties about the world (e.g., frogs and their species-level traits, generalization beyond observed frogs).
  • Spatial and dimensional constraints: our language and semantics may spring from a world that is three-dimensional in space and time; thus space-related words (on, under) may be constrained by environmental structure.
• An important caveat: built-in knowledge or biases are not guaranteed true simply because they are in the mind; they can be wrong and must be tested.
• The non-binary stance extended to language:
  • Language learning involves both innately structured knowledge and patterns learned from input; one cannot rely solely on one or the other.
• The role of evolution in shaping cognition:
  • Innate or evolved biases provide efficient priors for learning under uncertainty; they help generalize from sparse data and structure linguistic input.

Language as a Key Test Case: Rules, Grammar, and Evidence

• Focus on grammar, not lexicon, to test empiricism vs. rationalism in language:
  • Lexicon (words) like the word for cat vary across languages; not the main burden of the argument.
  • Grammar (rules for combining words) is where innate structure might reside.
• Two core grammatical rules explored in the lecture: 1) Question formation rule (the "recipe"):
  • Concept: to form certain questions, you take the element you want to ask about, move it to the front, and replace it with a question word (who, what, where).
  • Informal recipe (illustrative): take the thing you want to ask about, place it at sentence front, replace with a question word, yielding well-formed questions like:
    • When did Moira go to the store?
    • How do I get this darn thing to turn on?
    • Where did David put the purple socks?
      2) The word that in English: after verbs like think/believe/heard/say, the complement clause can be preceded by that or not, and meanings remain the same in many cases.
  • Observed that in most sentences, thought/believe/etc. can be followed by either with or without that, e.g., Anita thought David would win vs Anita thought that David would win.
  • However, when embedding and combining with other rules, the combination of two rules yields a puzzle: the presence or absence of that becomes sensitive to context.
  • Example: Who do you think that would win? vs Who do you think would win?
• Experimental investigation with language learners:
  • Psycholinguistic study or classroom experiment: test intuitions about sentence acceptability using iClicker polls and natural language data.
  • Key experiment setup: create sentences that combine the two rules (question formation and the optional that) and test acceptability.
  • Observations from the poll prompts:
  • In live polls, participants show clear preferences that reflect a natural language intuition (e.g., Who do you think would win? preferred over Who do you think that would win? in certain contexts).
• Empirical data from child language corpora:
  • Corpus study: 25 children, over 1,000,000 words, about 30,000 utterances containing questions.
  • Finding: Only about 1% of the observed questions had the more complex embedded-structure forms, indicating the most complex combinations are rare in everyday child-directed speech.
  • Crucially, researchers found zero examples of exactly the complex patterns that would demonstrate how the two rules interact in the precise way that adult speakers judge as natural.
  • Yet, human learners nonetheless acquire correct sentence formation and the rules governing these interactions. Children learn how to form acceptable questions with embedded clauses beyond what their direct exposure to such sentences would predict.
• The “irrationalism” (innate knowledge) argument, as presented in class:
  • Empiricist view would claim linguistic knowledge is learned from exposure to examples; seeing actual examples should suffice to learn grammar rules.
  • The observed data show a gap: learners converge on correct generalizations about rule interactions despite sparse or absent direct exemplars in natural input.
  • This discrepancy suggests that language knowledge may rely on built-in or innately specified constraints on how grammar can interact, rather than being derived solely from exposure to examples.
• Takeaway and implications:
  • The language data pattern supports a rationalist or innatist component for certain aspects of grammar, but the overall picture is nuanced: knowledge of language likely arises from a combination of innate structure and experiential input.
  • The debate about language innateness will continue; the lecture emphasizes moving beyond binary positions and exploring how evolved cognitive architecture could enable rapid and robust language learning.
• Preview of ongoing themes:
  • The next topics will further explore built-in knowledge, instinct, and how they interact with learning and culture in language.

Key Mathematical and Logical Concepts Mentioned

• Circle and triangle geometry relates to non-sensory knowledge:
  • π is the ratio of circumference to diameter: rac{C}{D} = rac{C}{2r} = rac{A}{r^2} ext{ (contextualized values)}
  • Alternative expressions for circumference and area:
  • $C = 2\pi r$
  • $C = \pi D$
  • $A = \pi r^2$
• Pythagorean geometry: the relation in a right triangle is $a^2 + b^2 = c^2$
• The broader philosophical symbols and terms:
  • Tabula rasa: innate ignorance at birth that must be shaped by experience.
  • Impressions vs ideas (Hume): impressions are the raw sensory data; ideas are the brain's reflections on those impressions.
• These equations and concepts illustrate how abstract mathematical truths can be known even when not directly observed through the senses.

Connections to Broader Themes and Real-World Relevance

• AI and learning systems:
  • The lecture opens with a nod to state-of-the-art AI (e.g., ChatGPT) learning from patterns in data, drawing a parallel to human empiricist concerns about data-driven knowledge.
  • The idea that learning from experience is powerful, but the presence of innate or structured priors can greatly accelerate learning and generalization.
• Educational and cognitive implications:
  • Understanding that language and knowledge involve both data-driven input and internal structure can inform teaching approaches and curriculum design.
  • Acknowledges that humans may rely on built-in biases to generalize from sparse data, which has implications for language acquisition and cognitive development.
• Philosophical implications:
  • Challenges to strict empiricism by showing necessary roles for mental organization and structure in generating knowledge.
  • The discussion of non-binary positions invites careful consideration of where innate structures may exist and how they interact with learning from experience.
• Ethical and practical considerations:
  • If knowledge is a mix of built-in structure and experience, how should we approach education, bias, and artificial intelligence design to ensure robust, fair, and explainable learning?

Summary Takeaways

• Empiricism claims knowledge derives from sensory experience, but sensory data can be deceptive and some knowledge (geometry, numbers) seems not reducible to direct experience.
• Rationalism argues that the mind imposes structure on experience; there may be built-in cognitive priors that guide learning and knowledge acquisition.
• Language provides a critical test case where traditional empiricist accounts struggle to explain learners’ knowledge of rules and their interactions, suggesting innate grammatical constraints.
• A nuanced, non-binary view acknowledges both experiential data and built-in structures (evolved biases, language-specific priors) shaping knowledge across domains.
• The discussion foreshadows deeper exploration of built-in knowledge and instinct in language and cognition in upcoming sessions.

Glossary of Key Terms

• Empiricism: Knowledge is primarily grounded in sensory experience.
• Rationalism: Knowledge relies on built-in mental structures or innate ideas.
• Tabula rasa: The mind starts as a blank slate, shaped by experience.
• Impressions vs ideas: In Humean terms, impressions are vivid sensory inputs; ideas are reflections or concepts derived from those impressions.
• Reflection: The mind’s act of organizing and structuring sensory experience to form knowledge.
• Innateness / Built-in priors: Cognitive structures that are present in the mind or evolved in the species to facilitate learning.
• Embedded clause / embedded question: A complex sentence structure where one clause sits inside another (e.g., who did you think [that] would win?).
• That-trace effect / that-optional rule: The interaction of the word that with other grammar rules can become obligatory in certain contexts.
• Non-binary approach: A stance that allows both empiricist and rationalist elements to contribute to knowledge depending on the domain.

References to Historical Figures Mentioned

• John Locke: Senses convey into the mind distinct perceptions of sensible qualities; knowledge arises from experience.
• David Hume: All simple ideas originate from simple impressions; knowledge begins with impressions.
• Plato: The cave allegory, awareness of being misled by appearances.
• Descartes: Skepticism about knowledge grounded in the senses; possible deception by the senses.
• Philosophical lineage includes echoes from premodern thought and non-European philosophical traditions.

Quick Study Prompts (for exam prep)

• Explain the two main problems empiricism faces according to the lecture and give an example for each.
• Describe how the lecture portrays the relationship between impressions and ideas in Hume’s view, and how this relates to the claim that knowledge requires more than sensory input.
• Summarize the two grammatical rules used to test language knowledge and explain why their interaction poses a challenge for strict empiricism.
• What is the argument for innately built language knowledge based on the child corpus study? What did the study find about the presence of certain complex sentences in everyday speech?
• How does the non-binary view reconcile rationalist and empiricist perspectives when it comes to different domains like mathematics versus social facts or language?

Important Formulas to Remember

• Pi and circles:
  • $rac{C}{D} = \pi$
  • $C = \pi D$
  • $C = 2\pi r$
• Geometry basics:
  • $a^2 + b^2 = c^2\quad\text{(Pythagorean theorem)}$
• Area of a circle:
  • $A = \pi r^2$