Cognitive Science: Mind, Brain and AI
On this page (16)

Cognitive Science: Mind, Brain and AI

Cognitive science is the study of how minds work, drawing on philosophy, psychology, neuroscience, linguistics, and computer science at once. It asks what thinking, memory, attention, and understanding are, and how a brain or a machine can have them. Cognitive science and AI took shape together in the mid-1950s, and the two still borrow ideas from each other.

TL;DR: This category puts the oldest questions about the mind next to the newest AI systems. Each article explains one idea from philosophy or neuroscience in plain English, shows what a large language model does with it, and marks what is still open. Start with Phenomenology and AI for meaning and skill, or the hard problem of consciousness for the question everyone asks. Build an AI agent free →

Dreyfus's Critique of AI

Hubert Dreyfus argued in 1972 that human skill cannot be written down as rules. What he got right, what LLMs changed, and which parts of his critique hold.

Embodied Cognition and Enactivism

Embodied cognition says thinking depends on a body that acts in a world. The 4E view, enactivism, the robot and multimodal replies, and what it means for AI.

The Hard Problem of Consciousness and AI

The hard problem asks why physical processing feels like anything at all. Chalmers' question, the main theories, what they say about LLMs, and why it is open.

Intentionality and AI

Intentionality is the aboutness of a thought: it is always about something. What philosophers mean by it, the case for and against LLMs, and why it counts.

Phenomenology and AI

Phenomenology studies experience from the inside. What Husserl, Heidegger, and Merleau-Ponty said, why Dreyfus used them against AI, and what it means for LLMs.

Symbol Grounding Problem

The symbol grounding problem asks how symbols get meaning without a body or a world. Harnad's 1990 question, the Chinese Room, vector grounding, and LLMs today.

Why Cognitive Science Matters for AI in 2026

Cognitive science matters for AI because the field now argues about machines with the same words it uses for minds. When people ask whether a chatbot "understands", "means" what it says, or "feels" anything, they are asking cognitive science questions. Those questions have a long history and a short list of serious answers.

The Stanford Encyclopedia of Philosophy entry on cognitive science by Paul Thagard (substantive revision January 2023) calls it "the interdisciplinary study of mind and intelligence, embracing philosophy, psychology, artificial intelligence, neuroscience, linguistics, and anthropology." It dates the field's intellectual origins to the mid-1950s ("Around 1956, the intellectual landscape began to change dramatically") and its organized form to the mid-1970s, when the Cognitive Science Society and the journal Cognitive Science began. AI was in the founding list from the start.

The traffic now runs both ways. In 2023, a team of 19 researchers, among them Yoshua Bengio, published Consciousness in Artificial Intelligence: Insights from the Science of Consciousness. They turned neuroscience theories of consciousness into a checklist of "indicator properties" and tested current AI systems against it. Their finding: "no current AI systems are conscious", but "there are no obvious technical barriers to building AI systems which satisfy these indicators." In the other direction, Grace Lindsay's 2020 review Attention in Psychology, Neuroscience, and Machine Learning (Frontiers in Computational Neuroscience) traces how the word "attention" moved from the study of the brain into the design of neural networks, and how loosely the two meanings line up.

That is the job of this category. It gives each idea its strongest source and its best critic, and it says plainly when a question has no answer yet.

Four Threads

The category follows four threads. Each one starts from a question about people and ends at a question about AI.

Thread The question about people The question about AI Start here
Phenomenology and meaning What is it like to see, act, and understand from the inside? Can a system trained on text understand anything? Phenomenology and AI
Consciousness Why does brain activity feel like anything at all? Could a model ever be conscious, and how would we tell? The Hard Problem of Consciousness
Brain vs model How does a brain predict, learn, and store? How close is a neural network to a brain? Predictive Coding
Memory and attention How do people hold, find, and tidy what they know? How should an agent keep context across runs? Memory Consolidation 
How the threads relate

  the inside view          the outside view            the machine
  (phenomenology)   --->   (neuroscience)      --->    (AI models and agents)

  "what it is like"        "what the brain does"       "what the model does"
  meaning, skill, body     prediction, memory,          next-token prediction,
                           attention                    retrieval, agent memory

  Open at every step: does the third column ever reach the first?

Already in the Wiki

Several neuroscience and AI pages already cover parts of this ground. They stay where they are. This category links them as the brain side of each comparison.

Page What it covers Why it belongs to this thread
Predictive Coding The brain predicts its input and sends up only the errors, plus the free energy principle The closest brain theory to next-token training
Hebbian Learning "Cells that fire together wire together" How biological learning differs from backpropagation
Memory Consolidation How the brain turns recent experience into long-term memory The step most AI agents skip
Hopfield Network An energy-based associative memory A bridge between physics, memory, and neural networks
Connectome The wiring map of a nervous system What "brain-like" would have to copy
Compression and Intelligence Why a good predictor is a good compressor One measurable link between minds and models
Metacognition Thinking about one's own thinking Can a model know what it does not know?
World Model An internal model that predicts how the world changes The thing grounding and embodiment debates are about
Emergent Behavior Abilities that appear at scale without being trained in Why model abilities surprise their builders

Articles in This Category

Phenomenology and meaning

  • Phenomenology and AI. Husserl, Heidegger, and Merleau-Ponty on lived experience, and why their ideas became the strongest critique of early AI
  • Intentionality and AI. The "aboutness" of thought, and whether a model's words are about the world or only look like it
  • Dreyfus's Critique of AI. Why Hubert Dreyfus said human skill cannot be written down as rules, and how his critique fares against LLMs
  • Embodied Cognition and Enactivism. The 4E view of the mind, the extended mind, and whether AI needs a body
  • Symbol Grounding Problem. How a word gets its meaning for the system that uses it, from Harnad's 1990 question to vector grounding

Consciousness

Connection to Taskade

Taskade makes no claim that its agents think, feel, or understand the way a person does. The brain words on /connect/dna are a metaphor for how the parts of a workspace connect. What cognitive science offers is a set of practical lessons for anyone who works with AI.

The Taskade Genesis loop: projects hold Memory, agents supply Intelligence, and automations run Execution and write results back

Lesson from cognitive science What it means in practice Where it lives in Taskade
Fluent language invites trust it has not earned Keep a person on the decisions that matter Human-in-the-loop review steps
Meaning needs a link to the world Give an agent sources it can cite Agent knowledge from your projects and files
A mind can use a notebook as memory Keep shared context outside any one chat Workspace DNA: Memory, Intelligence, Execution
The brain tidies memory offline Merge and archive old notes, never delete them Taskade EVE memory

That last row is concrete. Taskade EVE keeps its workspace memory as projects you can open and read. When you return to a busy workspace, Taskade EVE can tidy that memory in the background, at most about once a day. It merges overlapping notes into one and moves the originals to an archive instead of deleting them. The idea echoes memory consolidation in the brain. It is a maintenance step, not a claim that the software sleeps or remembers like a person.

To put these lessons to work, start with a Taskade AI Agent that reads your projects before it acts, and see the memory types an agent can draw on.

  • Foundations: the mathematics and machine designs under modern AI, from Shannon to von Neumann
  • AI: reference pages on models, training, and the terms in this category
  • AI Agents: how agents plan, remember, and act

For the longer story, read What Is Intelligence? and AI Meets Psychology.

Frequently Asked Questions About Cognitive Science and AI

What is cognitive science in simple terms?

Cognitive science is the study of how minds work. It combines philosophy, psychology, neuroscience, linguistics, computer science, and anthropology to explain thinking, memory, attention, and understanding. The Stanford Encyclopedia of Philosophy calls it "the interdisciplinary study of mind and intelligence."

AI was one of the founding fields of cognitive science in the mid-1950s. Early AI tried to model the mind as symbols and rules. Today the two fields trade ideas in both directions: neuroscience theories shape tests for AI, and AI models are used to study the brain.

Do AI models think like the human brain?

No. Some ideas line up, such as prediction and attention, but the machinery is very different. The predictive coding article shows one close comparison and where it breaks.

Are AI models conscious?

Current evidence says no. A 2023 report by 19 researchers found that no current AI system meets the indicators drawn from leading theories of consciousness, though no obvious technical barrier stops a future one. The hard problem article covers the debate.

What is phenomenology, and why does it matter for AI?

Phenomenology studies experience from the inside, starting with Edmund Husserl around 1900. Its ideas about skill, the body, and shared context became the basis of Hubert Dreyfus's critique of AI. They now frame the debate about what language models understand.

Why does this category not declare winners?

Because the field has none on the biggest questions. Consciousness, grounding, and understanding are still open. Each article gives the strongest case on each side and says what would change the answer.

Do Taskade agents understand what they read?

Taskade makes no such claim. A Taskade agent reads the projects, files, and web pages you give it, and it can cite them. That is useful grounding in the practical sense. It does not settle the philosophical question.

← Back to all topics