From seeing to remembering: Images with harder-to-reconstruct
representations leave stronger memory traces (paper)
https://arxiv.org/pdf/2302.10392
3 Likes
It makes sense to me that there’d be a meaningful correlation between somewhat puzzling imagery and memorability.
We try to make sense out of chaos. We’re pattern finders.
Although, if everything you see is nonsense, then all the nonsense starts to blend together.
“Meaningful madness” is probably an okay guiding principle.
3 Likes
I wonder if it would be useful to turn Major system objects into anthropomorphic characters with ChatGPT-Dalle3 for the purposes of these absurd, memorable and novel visuals
3 Likes
That’s what Dominic O’Brien did when he created his number system (The Dominic System). Each two-digit number is a character with an action and a prop.
Some people have done the same with the Major System. For example, my character for 09 is an actor from the 70’s TV series Soap. He’s washing his hands with a white bar of SoaP.
For 59, I imagine my cousin - a tennis player - LoBbing a tennis ball.
61: My brother-in-law, a retired airline pilot, is flying a JeT.
This allows me to use PAO with numbers.
1 Like
This has some interesting implications.
For the story method specifically. We tend to remember specific types of narratives better than others, so a balance between story coherence but imagine absurdity is optimal.
1 Like
One of my issues with these kinds of studies is that they are too reductionist to apply to real world trade-offs
I’d much rather see what top memory athletes converge on for optimizing performance
Memory competitions aren’t the real world though. They’re artificial tasks asking for specifically tailored memory skills.
But if that’s what interests you, well you have this forum with several memory champions describing what they do and what works for them… Lots of them are active and very generous with their time and explanations. 
1 Like
I never claimed they are. What they do provide are real world trade-offs, such as learning per unit of time. That part holds for exams or work-related memorization too.
Well more or less. They can give a rough idea of how fast you could memorize real-world stuff, but take it with a big grain of salt.
Mnemonic systems like that are very explicit about what and how to remember it. Real-world application is way less clear cut and has more room for ambiguity and since there is no clear conversion process, making mnemonic images regardless of methods used, takes longer than you would think and way longer than with plain numbers. At best you could use word list memorization as a rogh guide - coincidentally studies very often use words as to-be-rememberd data.
1 Like
I noticed one thing as well: the more images have verisimilitude, the better they will be. This does not mean the images need to be realistic or boring; it just means that they should have meaning and internal logic.
Besides sharpness being very important, images become even better as they become more detailed. I should note that I haven’t read the post yet, but I’d like to share my experiences.
2 Likes