site banner

Friday Fun Thread for February 23, 2024

Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

2
Jump in the discussion.

No email address required.

https://pelorus.substack.com/p/priors-for-cognitive-enhancement

Is one of the most interesting pieces of writing I have recently come across. It introduced me to a model that I would roughly describe as

(Net X)[Or Net X complement] = Apparent X - Complement of X

I would strongly suggest reading the entire article before deeming the model too simple and obvious.

It's going to take a while for the model to fully settle in for me, but I already see applications all over the place.

I think one of the features of why so many people described Scott's early work and others from the rationalist sphere as "insight porn" was because of this property. Putting a name and formalizing something that for most people was previously only encoded in gut feelings. And I don't think there are any Academic fields other than perhaps Philosophy that are making general fuzzy models of the world (models that also work with the rules and confines of socially constructed domains).


I think I am very high focus and moderate divergence. Net is somewhat mildly unfocused. This might explain why depending on the task and context, I can vary between being perceived as having ADHD, being autistic, or being scatterbrained.

This looks like it was written by someone talking slightly too much Adderall (I Can Tell by Some of the Pixels and by Having Taken Quite a Few Adderalls in My Day) and the inverted U's being drawn as normal curves when they'd make more sense as parabolas looks like a tip-off that the author is unconsciously pushing a little too hard at that satisfying feeling of fluent-compression of concepts -- but overall this seems reasonable.

My midwit slap-another-axis-on-it extension of that model would be that (for a given task context, but maybe more broadly shared between a variety of task contexts) there's something like a y-axis of "expected reward for effort" superimposed over the x-axis of exploit-explore, or "focus-divergence" (exploit-explore only trade off at a fixed level of task effort). On that graph, I think the effect of Adderall is to push up-and-left -- to increase the expected reward for exploit-effort. This is very performance-enhancing for people who need more exploit-effort, bad for people who actually need more explore-effort, and mixed for people who already have a good explore-exploit balance but just need to put in more overall effort to improve at the task.

In this model, a depressive state is one where, for almost any available task-action context, the expected reward for effort is very low -- and that's why Adderall can sometimes be a decent antidepressant. The classical hyperactive ADHD state would be higher up the y-axis but shifted way the right, while an inattentive ADHD state might not be as far shifted to the right, but lower down below the x. Each might get some benefit from Adderall pushing up-and-left, but in different ways. And of course, some people who look like any of those phenotypes might just need different task-contexts than the ones they're presented with.