Your Memory Was Never Only in Your Head

People assume memory is whatever sits inside their skull. That is a common scientific mistake.

The extended mind

In 1998, two philosophers of mind, Andy Clark and David Chalmers, published a paper that is now one of the most cited texts in the philosophy of cognition: the extended mind thesis. They proposed a simple thought experiment. Imagine two people. The first, Inga, has an intact biological memory and recalls the address of a museum from inside her mind. The second, Otto, has Alzheimer's and writes every address in a notebook he keeps with him at all times. When Otto opens his notebook and finds the address, functionally the same thing happens as what occurred in Inga's mind. Clark and Chalmers concluded that Otto's notebook is, cognitively speaking, part of his mind rather than an external accessory.

That idea is the scientific foundation of what people now call a second brain.

But a second brain is not an Instagram trend. It is a cognitive principle that existed long before anything went digital.

Knowledge needs a resilient network

Another scientific layer belongs here, this one from network science.

Albert-László Barabási and his colleagues showed that systems with a connected network structure are more resistant to damage and loss than systems whose knowledge is concentrated at a single point. The human brain, on its own, is a single point of failure. One blow, one illness, or even just the passage of time and a change in life context can cut off access to that knowledge. But when the same knowledge is stored across a network of connected nodes outside the head, losing one node does not bring down the system.

This is exactly the logic that holds in structural design. A structure with many parallel connections resists local collapse better than one where the entire load sits on a single column.

The fully documented historical case is Niklas Luhmann, the twentieth-century German sociologist. Over his career Luhmann built a system called a Zettelkasten, a card box holding close to ninety thousand interlinked notes. Each card carried one atomic idea, a unique number, and references to other related cards. Luhmann always insisted that his scientific output, seventy books and more than four hundred papers, was not the product of individual genius. It was the product of a conversation he held with that system every day. In his own words, he thought with his card box, not only with his brain.

Why a biological memory alone is not enough

Human working memory is limited. In 1956, George Miller showed that the human mind typically holds around seven units of information at once, and later research, including Nelson Cowan's work, brought that number closer to four for most people. Which means the human brain was fundamentally not designed to hold large volumes of information in short-term memory. And if knowledge is not reviewed and retrieved from long-term memory either, it decays quickly, following the forgetting curve Ebbinghaus documented in the nineteenth century.

Which means knowledge that lives only in your head is an eroding asset, not a durable one.

What context change taught me

Now let me put that scientific framework next to my own story.

I was in Iran until 2021. For seven years I was a faculty member at a university while practising architecture professionally at my own firm. Architectural work is cognitively nothing like the common assumptions about it. An architect has to hold thousands of small decisions in mind simultaneously: the site, the geography, the culture and subcultures around the building, structural loads and their effect on the design, the properties and strength of materials, urban planning regulations, the client's budget, and the aesthetics of the space. If those decisions do not hold together in one coherent system, the building physically collapses. For years, without giving it a name, I was managing a complex knowledge system. That system just happened to live on drawing sheets and in university meetings, not in a retrievable archive.

In 2020 another layer joined it. I decided to make a podcast. No grand plan, just an attempt to say things I wasn't saying anywhere else. Some of my longtime followers may remember Arch-Gap, thirty-five live conversations with prominent contemporary Iranian architects. That podcast grew, reaching more than fifteen thousand regular listeners and hundreds of thousands of plays. Every episode was a small archive of experience and analysis, and still scattered: in my head, in audio files, and in the memory of listeners.

Then came the move to France. And here exactly what Clark and Chalmers had predicted happened to me, without my knowing its name at the time. When the context of my life changed, every piece of knowledge that existed only in my head, documented in no retrievable external system, effectively became inaccessible. New language, new market, new professional network. Seven years of academic experience and the experience of building a podcast turned, in a single moment, into two separate islands with no bridge between them.

At the same time, AI entered everyone's working life. At first I thought it was just a tool for producing text and images quickly. Then I understood that producing text was never my problem. My problem was that the knowledge I had gathered across different fields over more than a decade was connected nowhere. Even my attempt to write a book in English for architects, aimed at connecting the knowledge of image creation with tools like Midjourney, could not create semantic continuity for me.

Knowledge becomes capital when it can be retrieved

This is where the writing of Dan Koe, a writer and thinker working on the digital economy, took on a precise meaning for me. In an essay titled You Have $100,000 of Knowledge Trapped in Your Brain, he argues that most intelligent, experienced people carry a large volume of combinatorial knowledge, a mix of several fields that nobody else holds in exactly that combination. But because that knowledge was never turned into a retrievable, transferable system, it stays effectively dead capital. Something that exists but produces no value, exactly like money kept under a mattress.

In another essay, The Synthesizer, Koe describes a different career path. A synthesizer, in his account, is someone who follows a unique trajectory, uses curiosity as their compass, connects ideas from apparently unrelated fields, and out of that connection builds something none of the source fields would have produced alone. He stresses that writing is the most important skill on that path. Not because the finished piece of writing is valuable, but because the process of writing is itself a way of thinking and connecting ideas.

At exactly that point, I decided to re-architect my thinking system. Not metaphorically. Literally, like an architecture project.

A second brain with two layers

My current system has two layers.

The first layer is Obsidian. In its internal logic, Obsidian is Luhmann's Zettelkasten, only digital. Every note is a node, and the nodes carry bidirectional links. A note about structural principles in architecture can connect to a note about narrative structure in podcasting, and that to a note about content architecture for European clients. This is exactly what cognitive science calls associative memory. Knowledge becomes powerful when its nodes are connected, not when it is held in separate, unrelated files.

The second layer is Claude. This is where the large difference between my system and Luhmann's appears. Luhmann had to find the patterns among ninety thousand cards himself, by hand. Claude does that automatically. When a new note from a client meeting, or an idea from my old architectural practice, enters the system, Claude compares it against the whole archive, finds hidden patterns and unexpected connections, and builds a first draft of content or strategy out of those connections. I keep the human in the loop. The final decision and the quality check are mine. But the search and the initial connection no longer rest on my limited memory.

Scientifically, this is precisely what the research on distributed cognition describes. The process of thinking does not necessarily happen within the boundaries of the skull. It happens in the interaction between the brain and reliable external tools. And when that external tool, like Claude, has processing and connecting capability of its own, you are dealing with an augmented cognitive system rather than a passive archive.

One more scientific effect enters here: the spacing effect. In that same nineteenth-century research, Hermann Ebbinghaus showed that spaced review of information slows the forgetting curve and moves knowledge into long-term memory. The practical problem has always been that nobody has time to sit down and manually work out which old note has become relevant again. That is exactly what Claude does in my system on my behalf. When a new topic comes up, it puts my old architecture or university teaching archive back on the table, even when I have forgotten the connection myself. Which means spaced review, which normally demands severe personal discipline, happens automatically in this system.

From habit to cognitive infrastructure

At forty-two, in the middle of this rebuild, I decided to study business management formally. Not to collect a master's degree from a French university, but to learn how to turn this logic of information architecture into a system that is scalable, repeatable, and transferable to other people. I am forty-three now, and what I am building is no longer a daily habit or a burst of motivation. It is a cognitive infrastructure that works independently of my mood.

The lesson this experience taught me is entirely scientific and has nothing to do with motivation. The difference between someone whose knowledge is lost with every change of context, and someone who builds every new experience directly on top of the previous ones, has almost nothing to do with intelligence or innate talent. The real difference is whether that person built a retrievable system for storing, connecting, and regenerating knowledge. Luhmann did it with paper and manual numbering. I do the same with Obsidian and Claude, at a larger scale and greater speed.

That difference becomes critical when someone has moved between several very different fields, as I have. From architecture to academia, from academia to podcasting, from podcasting to migration, and from migration to AI. Without a reliable external system, every move means losing part of your previous intellectual capital. With a properly designed system, every move means adding a new layer to the same architecture without losing the layers underneath.

What Compacio builds for brands

This is what we implement for other businesses at Compacio. Most brands have scattered content. Every post, every article, every campaign is an island separate from the rest. When the content manager changes or a campaign ends, the same thing that happened to me during migration happens to the brand. Knowledge accumulated in one person's head, with no retrievable external system, is simply lost.

What we do is build a knowledge architecture for the brand. A connected archive of market research, customer experience, and produced content, made retrievable, connectable, and continuously regenerable with the help of AI. Every client interview, every performance report, every strategic decision becomes a node in that same network rather than a forgotten file in a folder. The result is a system that gets stronger with every new project instead of starting from zero each time, and that works regardless of who is sitting at the desk that day.

That work connects naturally to a growth system of record, evidence-led content and SEO, and measurement that leads to decisions.

If you want to see how this architecture would work for your own business, start a conversation with the Compacio team.