What Is Thought?

Thought feels immediate.

We think:

• in words,
• in images,
• in plans,
• in possibilities.

But what is thought itself?

Is it inner speech?

Symbol manipulation?

Mental simulation?

Neural dynamics?

Perhaps thought is not one process at all.

Thought Is Broader Than Language

People often experience thought as inner speech.

But not all thought is verbal.

We can think:

• visually,
• spatially,
• musically,
• emotionally.

A chess player may see a position without silently narrating every move.

Inner Speech

Inner speech is a real phenomenon.

It can support:

• planning,
• self-control,
• rehearsal.

But some cognition occurs without it.

Thought should not be defined only as words in the head.

Mental Imagery

We can imagine:

• a face,
• a room,
• a route.

Mental imagery recruits some of the same brain systems used in perception.

Imagery suggests thought can simulate absent objects.

Propositional Thought

Some thoughts have a proposition-like structure.

For example:

“The door is locked.”

Such thoughts can be:

• true,
• false,
• believed,
• denied.

This invites symbolic accounts.

Symbolic Theory

Classical cognitive science often treated thought as operations over internal symbols.

A mental representation stands for something.

Rules transform representations.

This resembles computation.

Language of Thought

Jerry Fodor proposed a Language of Thought hypothesis.

The idea is that cognition may use an internal representational system with combinatorial structure.

Natural language would then express, rather than constitute, thought.

Compositionality

Thoughts appear compositional.

If we can think:

Alice likes Bob,

we can also think:

Bob likes Alice.

Complex thoughts are built from reusable components.

This supports structured representation.

Systematicity

Human thought is systematic.

Someone able to understand:

the dog chased the cat

can usually understand:

the cat chased the dog.

This suggests internal representations preserve relational structure.

Connectionist Challenge

Neural-network approaches do not require explicit symbolic tokens in the classical sense.

Representations can be distributed across many units.

Thought may emerge from patterns rather than discrete symbols.

Distributed Representation

A concept may correspond to a point or region in a high-dimensional state space.

Relations can be encoded geometrically.

This gives a different picture of thought:

not sentence-like strings,

but structured activation patterns.

Thought as State-Space Dynamics

Dynamical approaches describe cognition as movement through a state space.

A thought is not necessarily a static object.

It may be a trajectory.

Thinking becomes structured change.

Mental Simulation

One powerful view says thought simulates possibilities.

We ask:

What if I do this?

What happens next?

The mind runs internal models before acting.

Planning

Planning requires representing:

• current state,
• goals,
• possible actions,
• consequences.

Thought can therefore be understood as offline action selection.

Counterfactual Thought

Humans can imagine events that did not happen.

For example:

“If I had left earlier, I would have caught the train.”

Counterfactual reasoning extends thought beyond present reality.

Prediction

Brains constantly predict what comes next.

Some theories treat prediction as a core operation underlying thought and perception.

Thought may be model-based anticipation.

Reasoning

Reasoning is one kind of thought.

It connects:

• premises,
• intermediate steps,
• conclusions.

But much thinking is not explicit reasoning.

Intuition

A solution can appear before conscious derivation.

The mind may process structure unconsciously.

Conscious thought receives the result.

This suggests thought has multiple layers.

Unconscious Cognition

Many processes influencing decisions occur outside awareness.

Examples include:

• priming,
• automatic categorization,
• implicit learning.

Thought cannot be equated with conscious thought.

Working Memory

Working memory temporarily maintains information for manipulation.

Examples:

• remembering a number while calculating,
• holding a sentence while parsing it.

It supports deliberate thought.

Capacity Limits

Working memory is limited.

This constrains:

• reasoning,
• planning,
• problem solving.

Cognitive architecture shapes thought.

Chunking

Experts reduce cognitive load by grouping details into meaningful units.

A chess master sees:

a familiar configuration,

not:

twenty unrelated pieces.

Thought depends on learned representations.

Attention

Attention determines which information receives processing priority.

Thought is selective.

We cannot consciously manipulate everything at once.

Metacognition

We can think about thinking.

Examples:

• “I am uncertain.”
• “I should reconsider this.”

Metacognition allows monitoring and control of cognition.

Recursive Thought

Humans can embed thoughts:

“I believe that she thinks that he knows…”

This recursive structure supports:

• social reasoning,
• language,
• planning.

It can also become computationally demanding.

Conceptual Thought

Much thought operates using concepts.

To think:

“This animal is a dog”

requires categorization.

Concepts compress experience.

The next essay will examine them directly.

Perception and Thought

The boundary between perception and thought is not clean.

Expectation can influence interpretation.

Perception supplies structured representations used in thought.

Cognition is interactive.

Emotion and Thought

Emotion influences:

• what we notice,
• what we remember,
• what we value.

Thought is not independent of affect.

Emotion helps prioritize possibilities.

Motivated Reasoning

Desires can distort reasoning.

People may favor conclusions they want to be true.

Thought is not automatically rational.

Rational Thought

Normative systems such as logic and probability specify standards for reasoning.

Human thought only approximates them.

We are capable of rationality.

We are not pure logic engines.

Heuristics

Humans use shortcuts.

Heuristics reduce computational burden.

They are often effective.

They also generate systematic bias.

Embodied Thought

Some concepts may depend on bodily systems.

Thinking about action may recruit motor-related representations.

Cognition may be grounded partly in sensorimotor experience.

Extended Thought

We think with tools.

Writing externalizes memory.

Diagrams stabilize structure.

Calculators perform transformations.

A complex thought process may span brain and environment.

Notebook as Cognitive Workspace

Suppose you solve a proof by writing intermediate steps.

The page holds information your working memory cannot.

Where is the thought process?

Only in the brain?

Or in the brain–paper system?

Computers as Thought Partners

Search engines, coding tools, and AI systems can become external components in human problem solving.

They extend cognitive reach.

But using a tool does not automatically make the tool conscious or intelligent in the human sense.

Thought and Neural Activity

Every ordinary human thought depends on brain activity.

But there is no simple one-thought-one-neuron mapping.

Thoughts are distributed across dynamic networks.

Same Thought, Different Brain Pattern?

If the same abstract thought can be realized through slightly different neural states across moments or people, then mental identity may be more abstract than exact neural identity.

This supports higher-level descriptions.

Thought Without Consciousness?

Some cognitive processing clearly occurs unconsciously.

Whether all unconscious processing should count as thought depends on terminology.

The boundary is conceptual, not experimentally fixed.

Thought in Animals

Animals can:

• solve problems,
• remember,
• plan.

These capacities suggest thought need not require language.

Human language expands thought dramatically.

It may not create it from nothing.

Thought in Machines

Machines can:

• search,
• infer,
• predict,
• plan.

Should these processes count as thought?

The answer depends on whether thought is defined:

functionally,

or:

phenomenally.

Functional Thought

A functional definition might say:

If a system manipulates representations to solve problems, it thinks.

This would include many artificial systems.

Phenomenal Thought

A stronger definition requires subjective experience:

there must be something it feels like to think.

This connects thought to consciousness.

That question remains open.

Thought and Meaning

A thought appears meaningful to the thinker.

How neural or computational states acquire semantic content is one of philosophy of mind’s hardest problems.

Representation alone does not settle meaning.

Thought Is Not One Thing

The word covers many processes:

• imagery,
• reasoning,
• memory,
• prediction,
• planning,
• inner speech.

Trying to reduce all thought to one mechanism may be misguided.

The Philosophical Lesson

Thought is best understood as a family of processes that:

• represent,
• transform,
• simulate,
• evaluate

information relative to goals and context.

It can be verbal or nonverbal.

Conscious or unconscious.

Internal or extended through tools.

The Next Question

Thought depends on concepts.

Without concepts, experience would remain an unorganized stream of particulars.

What is a concept?

Is it:

• a definition,
• a prototype,
• a theory,
• a neural pattern?

That is the next topic:

What Is a Concept?