What information reaches the system?
Sensation, language, memory, social cues?
Cognition begins with structured information rather than raw behavior alone.
Side 81
Cognitive
Science
A study of mind as a problem shared across psychology, neuroscience, linguistics, artificial intelligence, philosophy and anthropology. Cognitive science asks what representations and processes make perception, memory, language, reasoning and action possible.
representation→process→behavior→model→integration
06disciplinary lenses
05model questions
05integration problems
81Side
Cognitive science studies the architecture behind intelligent behavior.
Its distinctive move is to combine levels of explanation rather than treating mind as the property of one discipline.
How is information encoded?
Symbol, feature, map, probability?
Different theories disagree about whether cognition requires explicit internal representations.
What transforms one state into another?
Rule, association, inference, dynamics?
Processing models explain how representations change over time.
What behavior or judgment follows?
Action, report, decision?
Behavior constrains theories but rarely identifies one mechanism uniquely.
Which explanatory level matters?
Neural, computational, psychological, social?
Good explanations connect levels without confusing them.
Representation is one of cognitive science’s central disputes.
Some models treat cognition as manipulation of symbols; others emphasize distributed patterns, prediction, action or embodied interaction.
Explicit structures and rules.
Concepts and propositions can be manipulated according to formal relations.
Information across patterns.
Connectionist models represent content through activity spread over many units.
Represent uncertainty.
Beliefs can be modeled as distributions updated by evidence.
Body and action matter.
Representation may depend on sensorimotor structure rather than abstract symbols alone.
Environment carries part of the task.
Cognition can exploit external structure instead of encoding everything internally.
Models anticipate input.
Perception and action can be framed as reducing prediction error or uncertainty.
Cognitive processes compete for explanation.
The same behavior can be modeled as rule-following, association, search, inference or dynamic interaction.
Serial
One stage follows another.
Useful when processing order matters and intermediate states can be distinguished.
Parallel
Multiple processes operate together.
Parallel models explain rapid integration across features or sources.
Associative
Experience changes connection strength.
Repeated co-occurrence can alter later expectations and responses.
Inferential
Evidence updates hypotheses.
Reasoning and perception can be treated as uncertainty reduction.
Dynamic
State evolves continuously.
Attractors and coupled variables can explain cognition without discrete symbolic steps.
No single method reveals the mind directly.
Cognitive science triangulates across behavior, brains, models, development, language and culture.
| Method | What it reveals | Main limitation |
|---|---|---|
| Behavioral experiment | Performance, errors, response time | Mechanism remains underdetermined |
| Neuroscience | Neural implementation | Activation does not equal cognitive explanation |
| Computational model | Explicit mechanism | Fit may not imply biological realism |
| Developmental study | How capacities emerge | Age-related changes have many causes |
| Comparative/cultural study | Which capacities generalize | Contexts differ in many dimensions |
Different cognitive frameworks disagree about what kind of machine a mind is.
The disagreement is productive because each framework exposes assumptions the others hide.
Mind as symbolic computation.
Explains structured reasoning well but can struggle with flexibility and grounding.
Mind as learned distributed network.
Explains pattern learning and graceful degradation while making symbolic structure less explicit.
Mind as probabilistic inference.
Models uncertainty elegantly, though the source of priors and representations remains debated.
Mind as body–environment activity.
Emphasizes action and context over detached internal computation.
Mind as hierarchical prediction.
Unifies perception and action around prediction, but broad formulations can become difficult to falsify.
Mind as evolving state.
Focuses on trajectories, coupling and attractors rather than discrete representations.
The hard problem is not collecting disciplines; it is making their explanations meet.
Cognitive science is strongest when evidence at one level constrains models at another.
Behavior
Describe what the organism can do and under which conditions it fails.
Computation
Specify the problem being solved and the information required.
Algorithm
Specify the process that could transform input into output.
Implementation
Ask how the process is realized in biological or artificial machinery.
Environment
Ask which parts of the apparent computation are supplied by body, culture or external structure.
Cognitive Scienceinterdisciplinary study of mind
VisionDavid Marr · levels of analysis
MindwareAndy Clark · cognition, embodiment and prediction
Unified Theories of Cognitionarchitecture and integration