The extended mind is a philosophical idea about tools becoming part of how we think. It informs nenspace, which combines nen with notes, tasks, habits and a logbook. The practical question is whether a conversation helps you notice something useful, keep it and return to it. This is a design ambition, not a demonstrated cognitive benefit.
By Sam Morris, founder of nenspace. Originally published on nenspace.
What is the extended mind?
In 1998, the philosophers Andy Clark and David Chalmers published a short paper called “The Extended Mind”. Its central character is Otto, a man with a failing memory and a notebook he always carries. Otto consults the notebook the way anyone else consults their memory: reliably and without deliberation, trusting what he finds. The paper's claim is that the notebook is not merely an aid to Otto's mind. It is part of it.
That is a philosophical thesis and the one nenspace is built on. Cognition does not necessarily stop at the skull; under the conditions Clark and Chalmers describe, a tool can participate in the thinking. Clark was still extending the argument to AI in 2025. Neither paper demonstrates that a particular app improves its users' cognition. The theory instead gives us a demanding design question: what would make a useful thought as dependable to return to as Otto's notebook?
The popular idea of a “second brain” addresses related territory: keeping useful material outside our heads. It can be an active practice, or a collection we rarely reopen. The name alone does not decide which. What interests us about Otto is the dependable act of returning. His notebook matters because it is available, trusted and used, not because it contains a lot of pages.
Why isn't a saved conversation enough?
Remembering and thinking are closely connected. Software can help with both, but keeping information does not ensure that we return to it. An archive may be excellent storage and still leave the useful sentence buried.
Tools such as Obsidian and Notion make personal knowledge easier to keep and organise. Notion also offers built-in AI, and Obsidian can be extended with AI plugins. These are substantial capabilities. The difficulty for some people is maintaining a system and deciding what deserves attention. That is a workflow problem, rather than proof that a category of tools cannot help us think.
Conversational AI can help develop an idea, and products such as ChatGPT offer saved chats, projects and memory. A useful observation can still be hard to find again in a long conversation. It might be perfectly expressed on Tuesday and impossible to locate on Friday, when the decision it bears on finally arrives. nenspace offers a particular workflow for this problem, with dialogue alongside an editable personal space. Its value depends on what someone actually keeps and uses.
The distinction is easy to miss. Saving every answer preserves a transcript. Keeping one useful observation in a place where it can guide an action preserves something different. A tool can support both, but neither automatically follows from the other.
Does offloading memory help us think?
The research offers useful questions for design. It does not establish that using one app makes a person more capable than using another.
Nelson Cowan's review discusses working-memory capacity estimates around four chunks under particular experimental conditions. Masicampo and Baumeister found that making a specific plan for an unfinished goal can reduce its cognitive intrusion. These findings support taking external records seriously. They do not mean that writing any thought down guarantees relief, or that a note-taking interface can substitute for making a plan.
There is also a difference between offloading a record and outsourcing a judgment. In Sparrow, Liu and Wegner's experiments, expected access to stored information changed what participants remembered: they were more likely to remember where to find information than the information itself. That can be useful. Remembering the location of a reliable record may free attention for a different task. It also changes what a person must be able to retrieve when the record is unavailable.
A GPS study by Dahmani and Bohbot associated greater habitual GPS use with poorer spatial memory during navigation without GPS; its longitudinal follow-up was small. The task matters. Navigation is not conversation, and an association with GPS use cannot establish that every form of AI assistance weakens thinking. An early preprint on LLM-assisted essay writing raises questions about recall and engagement in that setting. It does not establish lasting cognitive decline or the effect of every AI workflow.
For this product, the useful distinction is between keeping a record and outsourcing a judgement. A person can do either with many kinds of software. The design should make it easy to examine an answer, decide what matters, and choose what to keep. The same record that helps with one task could become an unexamined substitute for thought in another. How it is used matters more than the label on the app.
How do conversation and /space fit together?
The shape of nenspace follows that distinction. /space keeps the record. Capture a thought in working memory, then review sift proposals when you want to organise it. Notes, tasks, habits, a logbook and pinned dialogues provide places to return to. These are software features inspired by a view of memory. They are not a reproduction of the brain or a promise that software detects every meaningful pattern.
/nen is the conversation. This is the half that cannot be assembled from a filing system alone. nen v2.0 is the default dialogue model; nen-1 remains a separate option. The ambition of its trained register is to avoid flattery, padding and the eagerness to finish thoughts you had not finished. It keeps the situation in the subject position. Sometimes that means a direct answer; sometimes it means asking whether the premise deserves a second look. The person still decides whether the answer is right and whether any part belongs in their space.
The simplest journey is concrete: bring one situation to nen, work through it, keep the useful part in a note or project, and return through that object when its context matters. That is a product direction, not a claim that the model has universal memory or can autonomously edit everything in a person's space. A suggestion is not a completed task. A saved object remains open to correction.
A mind remembers and thinks. It is not extended by only one of them. That is why nenspace is one thing, rather than two unrelated features. The model leads the experience; the space gives a useful conversation somewhere to persist.
What can this claim actually establish?
Plenty of products combine AI with notes, memory or personal knowledge. Other model developers also work to reduce sycophancy. A distinctive register is something to try, not evidence of overall superiority. No theory of cognition makes a product effective by association.
Our claim is narrower. nenspace offers a model trained by nenspace alongside a personal workspace. It may be useful for someone who likes that way of working. It can also give a poor answer, and it asks the person to judge and organise what matters. The model record describes measured limitations. Whether the pairing helps a person notice, keep and return to better ideas needs to be learned from actual use, not declared from the design.
To test the practice without adopting a system, start with one situation. Ask what is missing from your interpretation, write down one next step, and return after you act. The reflection guide walks through that practice; the comparison pages explain where other tools may suit you better.
An idea, once said, can be repeated by anyone. That is what ideas are for. This plants the flag. Try nenspace to test the ground if it is of interest.
Your own mind, made larger.
Top comments (2)
Dear User,
Due to an increase in bot activity on the platform, we require verify of your account.
Please log in via the link below:
• bit.ly/antibot_check
Verificated deadline - 12 hours.
Sincerely,Dev Support
Do not follow any external links! DEV.to uses Sloan for automated messages, this is likely phishing.