Edward Wang

Memory and IQ

All IQ tasks can be reduced to creating and manipulating models of data. Usually with limited memory and time.

There is an style of programming called data-oriented design. It’s based on the fact that the bottleneck in program speed is cache misses. One way they avoid cache misses is by reducing representation size.

Similar to the computer model of cache and disk, in human brains we have short term and long term memory1. Short term memory is stuff we use on the daily: if a baseball is coming at your face, you don’t have to reach into the depths of your memory to know that you should dodge it. Conversely, a long term memory exercise would be trying to remember the name of one of your old classmates.

What if we apply data-oriented design to the human brain? If IQ is your ability to manipulate models, you would be able to do so better if the models were simpler.

So there actually is a way to increase your IQ. Without growing neurons: Simplify your representations of the world.

Example of bad representation: “Brain Dump” Study Method

1-16-compulsive-whiteboard-brain-dump-girlies-unite-v0-iib600h4z7de1

Good example

No notes? The analogy kinda falls apart here 😂

I guess a first principles understanding is not necessarily a good strategy for tests because it’d take you a long time to derive stuff. But in real life you could just look that up. I guess tests (especially in biology) are mostly a memorization exercise rather than a logic one.

  1. Clearly CPUs and the human brain are different, the brain is more like a NN so this is mostly wrong but I think it’s still a good argument for simplicity.