While memory champions prefer mental palaces for their simplicity and capacity, accelerated leaning gurus favor mindmaps. Mindmaps are extremely versatile and can be easily modified as new information becomes available. However, mindmaps have several essential design flaws compared to yet more flexible schemes used by engineers and AI creators. We can easily add the missing elements to the mindmaps, but the resulting memory structures become significantly more complex. Here I want to navigate these treacherous waters.
Why use feedback elements?
Engineers often use feedback, both in conventional designs and in recurrent or LSTM neural networks. Our bodies also use feedback loops. We know what we want, and make small corrections as we progress. If we do not know how to do something, we perform it anyway and ask an expert for corrective instructions.
Feedback is very natural for procedural memory training and use. We know how to do something by readjusting the parameters based on past performance. It is less natural for declarative memory. We acquire the data from external sources, like articles and Wikipedia. Once acquired, the data is simply there. But what if we need to describe a procedure with feedback as a part of our declarative understanding: for example a control system of a guided rocket. Then we need to visualize complex schematics with multiple feedback loops.
It is often easier to add feedbacks to mindmaps than to visualize a complex scheme. We add to the mindmap the feedback element logical marker: it is usually a – sign in a circle with two inbound and one outbound arrow. Then we add to the second inbound arrow the hyperlink to the mental map where the feedback originated. In that particular branch, we add a marker of split arrow,
Immediately we have two guestions: how do we use hyperlinking, and how to place logical markers on the branches.
How do we place logical markers on mindmap branches?
Logical markers are usually simple geometric shapes and mathematical symbols. + or – sign with or without a circle, with or without inbound or outbound arrows can easily be used for logical marking.
Now, being purely mathematical concepts, logical markers do not have a context. So we place them as added detail on some plain surface of existing visualization.
If our visualization is a PAO, the person usually can have a t-shirt or a cape, where we can visualize the sign. If we use complex objects, they also often have branding space: notebooks, caps, cars, screens… The idea is to clean up the visualization from textures and use a clear space to embed the mathematical symbol. This also means that the relevant visualization will have fewer details than other visualizations we regularly use.
When our visualization is two-dimensional like a logo, we may need to overlap the logical marker over the logo using a semi-transparent outline. This is somewhat uncomfortable, but after three days of training becomes pretty easy.
How do we use hyperlinking?
The hyperlinking part is pretty advanced and can get messy. We kind of visualize a portal from one branch in a mindmap or location in a mental city to another. This involves a portal element – typically a logical marker, and some practice of jumping to a particular place in the mental structure and back. The connection is further improved if the source and the destination use a similar detail, like a rare color scheme. If you have a PAO, three colored scarfs will usually encode the transit. Then all you have to do is practice going through the portal.
We use hyperlinks on the web all the time. They are the links we have everywhere, and that can get us to any location. Then we press the “back” button to return. While pretty easy on any computer, hyperlinks are somewhat disorienting. Especially for me. I get motion sickness when going through them. So even with practice, I try to avoid them if possible.
Think about it. When we encounter a link on a site we will usually not press it, and if we do press it we often open another tab. There is a productivity trick for memory context swaps, but it is complex enough for a separate article.
The curse of dimensionality
You may be wondering why I am not simply adding dimensions to existing memory structures, like tunnels below mental palaces or skyscrapers with multiple layers. This is definitely an option, which I describe elsewhere. It is expensive though.
Memory structures need to be simple. There are rules on itineraries for mental palaces. Like you should not cross your own path if you can, and always follow the same itinerary creating and maintaining a memory structure.
A simple story can be remembered by an absolute novice. Two-dimensional mindmaps and mental palaces already require some training. Three-dimensional mental cities are reserved for advanced students, and even then we try to keep very simple itineraries. Adding further dimensions requires the mental wizardry of Dr Strange. So, please avoid it as much as you can.
Why use converging graphs?
The directionality of the mental itineraries brings me to another interesting question. Mental maps are diverging from a single node. What if we need to converge to one or multiple nodes? The issue is not farfetched. We use both top-down and bottom-up thinking. A neural network can use both kinds of reasoning, say in U-net or V-net. So we may need to do the same with our memory structures, combining nodes to sets, and then set to supersets.
There are several ways of doing it. The easiest way is to build a regular map, but remember that the itinerary is reversed, kind of “upside-down”. To do that you can use a different color scheme. In the series “stranger things” the authors use ambient light to show that the story is happening in the mirror dimension. This is definitely an option.
Another and very different way is using set notation. Divide graphs into subgraphs and use thematic coloring within a subgraph.
Thematic subgraph coloring
Do you remember me mentioning a few paragraphs above colored scarves? The idea is very simple. We add some sort of multicolor ribbon for each element we need to remember. Each time we introduce a new subdivision happens, a color is added to the ribbon. Typically three levels of subdivisions are more than enough for simple hierarchies. Each division per level gets a color, usually a prime color. If you need to do something fancy, you can play with a tint of the color: red vs rose or vermillion. This way no matter where you go, you remember in which subdivision you are.
Complex graphs can easily behave like labyrinths, and even with a good itinerary and color schemes, one can easily get lost. So I honestly suggest you to simplify the material you want to memorize. If the graph is sufficiently simple, with up to twelve elements you might be able to memorize it as a single atomic visualization – using the face of a clock as a mnemonic structure.
Do not try to use the tricks described in this article unless you have no other way.
Get 4 Free Sample Chapters of the Key To Study Book
Get access to advanced training, and a selection of free apps to train your reading speed and visual memory