Cognitive processes are realized within the organism.
External objects may cause cognition without becoming parts of cognition.
Side 115
Extended Mind
A philosophical and cognitive-science investigation of whether some tools, environments and social structures can become parts of cognitive processes rather than merely aids to them.
agent→coupling→resource→process→boundary
04lenses
16working concepts
V0content
SS-1.0standard
Where does cognition end?
The central dispute is whether cognition is defined by location inside the skull or by the organization of a functioning system.
Ask what the component does in the process.
If an external resource plays the same stable role as an internal memory process, location alone may not be decisive.
Distinguish causing a process from composing it.
A notebook can trigger recall without necessarily being part of the memory system.
Every extension thesis needs a stopping rule.
Without criteria, nearly any environmental cause could be mislabeled as cognition.
Stable reciprocal coupling is stronger than occasional tool use.
Extended systems are usually argued from reliable interaction patterns rather than one-off assistance.
The resource is readily accessible.
A resource used only rarely or with high friction is less integrated into ongoing cognition.
The agent typically accepts the resource's output.
Automatic reliance can make an external store function differently from ordinary testimony.
Agent and resource alter one another's states.
Interactive loops are stronger candidates than passive reference materials.
Integration develops through repeated use.
Learning can redistribute cognitive work across person, artifact and environment.
Extension can involve more than notebooks.
Different cases test different boundary intuitions and should not be collapsed into a single claim.
External stores preserve information.
Notes, calendars and digital systems can offload retrieval and prospective memory.
Tools restructure spatial problem-solving.
Maps, GPS and landmarks can shift which computations remain internal.
Other people can form coordinated cognitive systems.
Teams distribute expertise, memory and monitoring across members.
Interactive models can become reasoning partners.
The key question is not intelligence alone but stability, dependence, transparency and division of cognitive labor.
Extension claims face serious limiting cases.
A good theory must distinguish genuine cognitive integration from ordinary causal influence and dependency.
Too-broad criteria make the world part of the mind.
Boundary tests must exclude merely convenient or remote causes.
Control and access can be asymmetric.
A rented platform or proprietary model may be functionally integrated yet institutionally outside the agent's control.
External components can fail or change.
Dependence on unstable resources complicates claims of integrated cognition.
Functional extension need not extend experience.
What participates in cognition and what is consciously experienced are separable questions.
The boundary question is functional, not cosmetic. Calling a tool 'external' does not settle whether it participates constitutively in a cognitive process; the relation must be analyzed.