This is about Alpha Zero.
But I think it is basically how human impulse control works. If a person chronically makes "bad" short-term-oriented choices, it may very well be because they _correctly_ don't depend on themselves to be able to execute on a long term strategy.
But I think it is basically how human impulse control works. If a person chronically makes "bad" short-term-oriented choices, it may very well be because they _correctly_ don't depend on themselves to be able to execute on a long term strategy.
https://twitter.com/humanisque/status/1421697376759296002
It's a lot less attractive to be "disciplined" and "responsible" and "prudent" if you're going to fuck it up somehow before you get the payoff.
You might as well seize some pleasure now, even if it is "self destructive" or reckless.
You might as well seize some pleasure now, even if it is "self destructive" or reckless.
If you're likely to fail anyway, there's less use in focusing on your studies. Might as well party.
If you're going to misinvest your savings, or break open the piggy bank to buy some random thing down the line, you might as well spend the money now.
If you're going to misinvest your savings, or break open the piggy bank to buy some random thing down the line, you might as well spend the money now.
If you're an alcoholic, and you're gonna fall off the sobriety wagon sometime in the next year, you might as well drink now.
Obviously, this is in part a self-fulfilling prophecy.
If you have "bad impulse control" then you can't trust your future self to execute on plans, which means it isn't worth it to have good impulse control.
If you have "bad impulse control" then you can't trust your future self to execute on plans, which means it isn't worth it to have good impulse control.
But if you have "good impulse control", you can trust your other time-slices more, so the payoff to your self control is higher, which reinforces your good impulse control.
But it isn't completely a self-fulfilling prophecy, because self-control isn't the only input to successfully executing on long term plans. You also need to make good decisions, for instance.
IQ anticorrelates with impulsivity and violence. Childhood lead exposure impacts both.
One hypothesis is that this is because these are both just direct consequences of the brain working less well.
One hypothesis is that this is because these are both just direct consequences of the brain working less well.
But another hypothesis is that relatively less self control is simply the rational policy, if you're not intelligent enough to pull off any kind of long term strategy.
You're going to make a bunch of bad decisions, and pulling off "responsible" life strategies depends on making a medium-large number of decisions well.
If your brain knows this, then it isn't going to weight the responsible strategies very highly, in the decision calculus.
If your brain knows this, then it isn't going to weight the responsible strategies very highly, in the decision calculus.
My guess is that this is not the main thing that is going on, because it would suggest that people with genetically low IQ are violent and impulsive at the same rate as people with the same IQ from lead-related cognitive deficiencies. Which I believe is not the case.
But if I found out that I was wrong about that and IQ straightforwardly tracks impulsiveness, independent of cause, this would be a crux and I would prefer the "impulsiveness is a rational meta-strategy" hypothesis.
This line of thought is also relevant for someone who wants to improve their "self-control"; to jump from the "bad" equilibrium to the "good", or even just iteratively expand the scope of places where you make the "strategic"/"long-term oriented" choice instead of the immediate g
Under this model, the core question is "does your brain believe that it is worth it to put in effort now, upfront, for benefit later? Can it trust that the hoped-for benefits are reliable?"
Which suggests that the most important thing is success spirals.
Which suggests that the most important thing is success spirals.
Make a point to succeed, and get the benefit from, iteratively more effort-requiring plans, so that your brain, correctly, epistemically, learns that those sorts of plans pay off, and that it is worth investing in them.
For many people (such as myself?), this might mean making a point to savor your successes, and soak them in, instead of just moving on to the next challenge.
@threadreaderapp unroll
• • •
Missing some Tweet in this thread? You can try to
force a refresh
