Compression and Abstracts

I was thinking on how to memorize speed cards, and came up with numbers: 17 seconds for me, 13 seconds for Ben Pridmore, at best. These are theoretical limits; however, they may be wrong. The world record is 21.19 seconds (?), and that’s awfully close.

My numbers came from chunking and compression in the PAO system. I process a lot of information in my head–Joshua Foer said creativity was essentially memory, but I don’t claim to be creative–and come to a lot of odd conclusions. In this case:

  • 100-number PAO systems
  • 52 cards, so a full two digits per card (would that it were 50)
  • 7 items in Short Term Memory at once
  • 8 seconds focus to remember
  • More links means better memory

With the 100-number PAO system, you can chunk three cards–numbers 0 through 51 or whatever indexing system you care to use–into one object. Most people can track 3-5 things at once actively, and retain 5-9 things in short-term memory–you can shift your attention off two of those things and not forget them, if you don’t load other things into their slots. I’m unclear if you absolutely need attention on short-term contents to encode them into long-term memory, or even if it helps.

Let’s assume you can keep 7 objects in short-term memory, and ignore the rest of this.

7 objects. If you continuously bring in one object (3 cards) per second, that’s 7 seconds focus on each, producing 17 cards in 3 seconds with reasonable ability to remember; however, you can also compress them to improve that.

Let’s say you generate two images: Ben Franklin riding a poem and Thomas Jefferson dancing a sandwich. (I don’t have a problem confusing these, because Franklin and Jefferson are well-distinct.) You then pull in your next 15 cards (5 objects) into memory. Your STM is full.

Rather than drop that last one, you can briefly use your oldest image–Ben Franklin riding a poem–as a slot to accumulate your second-oldest image–Ben Franklin riding a poem about Thomas Jefferson dancing a sandwich. (I guess in most peoples’s cases it would be dancing on a sandwich?) You now have a free short-term memory slot (survey everything else to retain it).

So here’s a few thoughts.

First, a poem.

I don’t exactly have a visual cue for a poem. Poems are abstract concepts to me; the most visual thing I get is some idea of floating lines of text, a font unknown because I don’t have a strong idea of what that text looks like. I know there are lines, the right sides don’t exactly line up, and they are arbitrarily short; I know they break into stanzas, maybe 2-5 lines each–my impression is I have a visual idea of 3 lines; and my impression is that my image is of the top of this poem, and it fades out immediately after the second stanza begins. I can’t tell you how far that is, how that breaks out, or anything visual about it; it’s an impression, an abstract idea, and the visual representation is more of a placeholder thought than an image. The fade-out is conceptual, not visual: the ideal of the poem, the space it takes up, blurs and fades away; the image of the poem isn’t the blurry part, but it ceases to be because my mindfulness of it ends.

Let’s put that simply: I’m visualizing an image which contains an abstract component, which I have addressed as an abstract thing. The visual tag for that thing isn’t actually a picture; it’s a description of a picture whereby the common visual form is abstractly referenced, and the idea of what it is is also referenced.

Likewise, Jefferson dancing a sandwich is an abstract association. I know about dancing and sandwiches, and about dancing a jig or a waltz. The act of dancing is weakly acknowledged–there’s movement, it’s a dance–and it’s associated with the idea of a sandwich–a visual idea. These aren’t visually incorporated in my head; they’re there, the association is made, but it’s mashed together by abstract glue. It’s absurd, but it works.

To go even farther: I can turn the first visualization of Ben Franklin straddled over a semi-visual-but-not-really abstract concept of a poem into a compressed object: the poem is about Jefferson’s absurd concrete action that can only associate two concrete ideas in an abstract manner. The visual information is abstractly tied to the poem: it’s pegged to the poem, which is pegged to Franklin riding the damn thing.

This absurd mangling of abstract and concrete information in visual and non-visual forms is, itself, a cue for the two original things I had remembered: Ben Franklin riding a poem about Jefferson dancing a sandwich will, in hilarious failure, pull together Ben Franklin straddled abstractly in mid air over a thing that exists but is effectively not there–Franklin is floating straddled on nothing, but is also straddled on something very real and concrete, and these two things are the same thing–and that thing involves Jefferson and dancing and there’s a sandwich and what the hell!!!

Of course I still have those two things. And I can smash that together, hold onto it for a moment, and then shove it off. I now have at least nine things tying six cards together–more because my brain has taken an impression of the cards itself, which everyone forgets about, but you definitely saw the King of Diamonds when you made that image. I have the three PAO entries for the first three cards, the three PAO entries for the second three, then the two single images from each of those, and a single compressed image that combines both single images and their contents.

A few things stand out to me here.

First, I’m piling images together and creating more data from data. You might have your PAO system memorized, and even have it practiced for the specific 140,608 images you’re using to associate three cards together–assuming you’re not doing that on the fly. Remembering those strongly helps you simply create that impression and quickly remember it: you know thas already, and now you also have to remember this about that. You only need an impression of a well-remembered image to retrieve the image. I’ve described a system by which these well-remembered images are smashed together purely so you can keep them in your head for more time–try to get a slightly more firm impression, and maybe you can race through your images faster.

Second, I’m using abstracts freely. A lot of mnemonics books claim abstract information is hard to remember. In fact, piles of abstract information are irritatingly difficult to keep straight: abstract concepts float and muddle and mix, because a pile of abstract concepts acts as a glue to hold things together. I’ve used abstracts as glue above: I know what a thing is, I know my impression of it, and I know how to incorporate that impression into a concrete image to form an idea; I don’t need to tie it to a funny mnemonic device to turn arrogance into an arrow glance or arrowed ants.

This works for me, but I hear colors. Sometimes I hear colors that are impossible–like a blue that’s a shade of red, not purple, not a mix, just that it’s a hue that doesn’t exist and can’t be reproduced in reality because it would represent a wavelength of light that doesn’t exist. Color is rooted in numbers–the number of nanometers between peaks of light–and colors not representable in any such wavelength are represented by non-existent numbers. Obviously, I can recognize abstract concepts in a manner similar to my visual system–but, as described above, I seem to be seeing abstract information, knowing that I don’t see a thing but that I have an impression of visual information that can’t be visualized.

While it may be true that anything you can do I can do better–or maybe not–it’s been repeatedly proven that anything I can do you can pretend. The great savants, with their brilliant memories, always had some characteristic we can all imitate: Mr. S. formed images from sounds, and everyone else formed their own mnemonic system internally. We imitate all of this. If I have some weird synesthesia that can process abstracts as pseudo-visual information by sticking a concept into a visual field, it’s almost 100% likely that anyone else can, too. It looks, from inside, like I’ve done almost all of it intentionally, mostly because I never realized you can’t do that, because it sometimes happens automatically.

I’ve been trying to improve my memory lately, and this has lead me to realize a lot of strange things happen in my head. My memory for facts and figures isn’t bad, per se; I remember quite a lot, and learn quickly. My memory for facts is better than figures; but I can remember abstract facts, and I can remember one-off quotes and insignificant statements while forgetting all of the big, important stuff. Often I know more about people than they know about themselves, because I retain all the little things of no consequence and form a picture about them–and find all the things you’re hiding. I’ve likened this to playing Minesweeper in 3D: the things you say create concentrations of information around voids, outlining the unsaid in a vortex, and revealing things wholly unrelated.

What I’ve learned, most, is that I convert sounds to impressions of images–but I struggle to create images, and so I create wholly-invented images that are influenced by abstract synesthesia and associated ideas (i.e. Weird Al’s This Is the Life evokes images of rich people in penthouses with butlers, while Cable TV evokes other images; both sound the same to me because they’re both pink and have other indescribable abstract assets). I also associate everything that’s said or read–every concept, every idea, every line of dialogue–to abstract concepts, interactions, imagery. I don’t necessarily retain any of this, but I do process everything in that way; most of it’s orphaned, memorable on its own but not stuffed into a mind palace.

I’m like a room full of barrels or pots. The image is pots with colorful, round-ish things in them. Don’t ask me what. Some kind of food both liquid and solid at once. I can access any pot, I can reach in and pluck out anything you ask; if you just point me generally there, I can scoop my hand in and pour out a fount of colorful, chewy, sloshy, clattery skittle-rain. I can also dip into the nearby pots. I don’t necessarily have a good index, but I can take a look around and see if anything’s like what you’re asking.

So I seem to know everything about everything, even though I have trouble specifically remembering anything.

So there you have it: People keep telling me to somehow convert the abstract into the concrete, and I just want to integrate the abstract with the concrete. I can handle abstract concepts as abstract concepts. They’re things. I deal with things.

Does anyone have any idea wtf I’m rambling about? I’m trying to get an idea of what’s going on in my head–if anyone understands this, maybe I can improve it. Target specific things I’m doing wrong, or leverage and improve things I’m already doing that are the right thing. Or maybe someone will see something novel and can benefit, I don’t know.