The human memory is incredibly good at remembering physical spaces. Entire cities can easily be memorized without a conscious effort. Sadly, most matters of memory are not so trivial. Imagine you had to memorize a city, and how to get to and from any given building from any other building, without ever actually visiting the city? Not so easy anymore. What if memorizing anything was as easy as memorizing the city where you live?

I can't be the only person who can't always remember what a book said on a specific topic but can remember where on the page it was written. This is part of our obsession with highlighting: it helps us create a recognizable landscape. When I take notes for a class I remember not only the information but also where on the page I wrote it, and as a result, I have better retention.

There is an advantage to taking notes by hand as opposed to electronically. Paper material provides physical spacing through spatial-temporal signals. It is the imperfection and variation of written notes that makes each item distinguishable. Whereas, a block of typed text has no unique landmarks or features. Conceivably digital spaces could help our memory if they were intentionally designed to have landmarks and be navigable like a physical space. I could see VR mind palaces becoming a benefit of metaverse technology.

In a physical space, one hallway naturally leads to another. Perhaps this is what makes it easier to navigate brick-and-mortar hallways over ink ones. The primary temptation of the student is to memorize an equation or result rather than where the equation came from/how it connects to other concepts they already understand. Memorizing equations is more difficult in the long term than remembering derivations. If every equation is a separate idea, a piece of information on an island, random things each existing in their own vacuum must be recalled. If the path or derivation is remembered, nothing is random and each idea leads to the next, like hallways. It is easier to remember a universally applicable and highly trained algorithm of extrapolation, that requires only the most memorable landmarks to function well than it is to remember every detail of each concept or space individually.

A person who is familiar with the landscape of their discipline might be able to synthesize that knowledge and extrapolate what is around the bend. The same algorithm/learned skill that makes memorization primarily extrapolation may lead to new discoveries. The person who has memorized random disconnected facts has no such potential as synthesis is only possible between connected ideas and they do not possess an extrapolative model. Connecting new concepts to the topography of one's current knowledge should consume as much energy as learning the concept itself.