Symposium · Philosophy of Mind · Pl. 5Andy Clark and David Chalmers · b. 1957 / b. 1966
Where Does the Mind Stop?
A man and his notebook
Clark and Chalmers compare two people trying to remember a museum's address: one recalls it from biological memory, the other looks it up in a notebook she always carries, trusts completely, and consults automatically. Functionally, the notebook plays exactly the role ordinary memory plays for the first person. Their question: why should the skull be where the mind's boundary is drawn? In the paper the notebook belongs to Otto, who has Alzheimer's disease and has learned to live by writing everything down. Inga, who remembers the address the ordinary way, is the control case. Both walk to the museum. Both, the authors argue, believed where it was before they set out.
The world it came from
The paper appeared in the journal Analysis in 1998, after several years of circulation. Clark, a British philosopher then at Washington University in St Louis, had been writing about robotics and embodied cognition. Chalmers, an Australian who had just published The Conscious Mind, was better known for arguing that consciousness resists physical explanation. Their shared background was functionalism, the view that a mental state is defined by the job it does, not by what it is made of. The 1990s had produced Edwin Hutchins's study of navigation teams on ships, where the calculation was spread over people and instruments, and Kirsh and Maglio's finding that Tetris players rotate pieces on screen because it is faster than rotating them in the head. The question was where to stop.
The idea
Where does the mind stop and the rest of the world begin? If a piece of the environment functions the way a cognitive process would if it happened entirely in the head, reliably available, automatically trusted, seamlessly integrated into action, then by their own functionalist standard, it counts as part of the cognitive process itself, not just a tool the mind happens to use from outside. The authors call this the parity principle. The point is not that tools help thinking. Everyone agrees with that. The point is that a belief can be stored in ink and still be a belief, so long as the ink is always to hand, consulted without deliberation, and accepted without a second check, just as memory is.
How the reasoning runs
Step one: state the parity principle. If part of the world does a job that we would count as cognitive were it done in the head, it is cognitive. Location is not a reason. Step two: build the case. Inga hears about an exhibition, recalls that the museum is on 53rd Street, and goes. Otto hears the same, opens his notebook, reads the address, and goes. Step three: ask what Inga believed before she recalled the address. She believed it, stored in memory. Step four: apply parity. Otto's notebook meets the same conditions: constantly available, automatically endorsed, easily accessed. So Otto believed the address before he opened the book, and the belief was in the notebook. Conclusion: some mental states extend past the skin.
Seen in life
Your phone dies in a city you know only through it. Within a minute you notice what has gone missing. You do not know your sister's number. You do not know which train takes you home, or what you agreed to do on Thursday, or the name of the restaurant your friend recommended. None of that was in your head. It was in your pocket, consulted a hundred times a day without a flicker of doubt, and it counted as knowing. Clark and Chalmers would say the feeling you have now is not inconvenience. It is a small, temporary memory loss. The information was part of your cognitive system by every test that matters, and the test the skull passes is only geography.
Why it matters
This challenges the assumption that cognition is necessarily bounded by the skull, a boundary that seems more like an accident of biology than a principled dividing line once you look closely at how thoroughly ordinary thinking already depends on notebooks, calculators, calendars, and now phones functioning as genuine extensions of memory and calculation, not mere aids sitting outside it. The stakes are practical as well as philosophical. If Otto's notebook is part of his mind, then taking it from him is closer to an injury than a theft. Designing a tool becomes a way of shaping a person. And the question of what an AI assistant is to the person who leans on it all day stops being a question about a product.
“Where does the mind stop and the rest of the world begin?”
From the source
Clark and Chalmers open their argument with the question that gives it its shape: “Where does the mind stop and the rest of the world begin?” Their answer, worked out through the notebook case, is that the traditional boundary at the skull is far less principled than it looks. The paper begins with three ways of rotating a Tetris shape: in the head, on the screen, and by a future neural implant that does what the screen does. The authors ask why only the first and third should count as thinking. From there they move to the notebook, and they end with a claim about the self: if a belief can live in a notebook, then the person whose belief it is extends that far too.
The pushback
Doesn't this stretch the concept of mind so far that it loses any useful meaning, since almost any reliably used external tool could then be redescribed as part of someone's cognition? Critics argue the notebook is still causally coupled to cognition rather than constitutive of it, a distinction the extended mind thesis is accused of blurring past usefulness. Fred Adams and Ken Aizawa gave this the name coupling-constitution fallacy: a thermostat is tightly coupled to the furnace, but it is not part of the furnace. They added a harder point. Real cognition, they argue, involves states whose meaning is not borrowed. The ink in Otto's notebook means 53rd Street only because Otto reads it that way. The notebook has no content of its own.
The reply
Clark answers the bloat worry with the conditions themselves. Most tools fail them. A library book is not always with you, and a web search is not automatically trusted. The notebook is a rare case, and the thesis was only ever about rare cases. To the coupling objection he replies that it assumes what it needs to prove: calling the notebook merely coupled already decides that cognition happens elsewhere. As for borrowed meaning, Clark asks for a test of real cognitive content that Otto's notebook fails and every part of his brain passes, and doubts there is one. That reply is strong but not decisive. Many readers still feel the gap between believing and being able to look up is exactly what the argument slides over.
Where it went next
The paper became a founding text of what is now called 4E cognition: embodied, embedded, enacted, and extended. Clark developed it into a book, Supersizing the Mind, in 2008, with a foreword in which Chalmers described his own phone as part of his mind. Adams and Aizawa, and Robert Rupert, built the standard opposition and pushed the debate toward what a mark of the cognitive would be. Psychologists tested the picture: a 2011 study by Betsy Sparrow and colleagues found that people who expect to look something up later remember where to find it rather than the thing itself. Otto and Inga are now a standard classroom example, and the usual starting point for asking what an AI assistant is to the person who uses it.
Where this leads
The Chinese Room approaches the boundaries of mind from the opposite direction, asking not how far outward cognition might extend, but whether symbol manipulation occurring entirely inside a system is even sufficient for a mind to be present there at all. The ghost in the machine is the older ally. Ryle argued in 1949 that the mind is not a hidden thing inside the body but a matter of what a person can do and is disposed to do. Clark and Chalmers take that thought one step further out. If mind is what a person does, and what a person does runs through a notebook, then the notebook is not outside the mind. It is where some of the doing happens.