> I find the overall argument to be rather weak. The most relevant problem is that it starts on a premise that might as well be this: "Intelligence is a substance, and a nootropic is a drug that makes the brain produce more of it."
Which of course it is not. Some supplements work on a deficiency principle, but many do not.
> Of course I've intentionally phrased this as to make it seem ludicrous, but throughout the article he talks of "more" or "less" intelligence as though it were a fungible sort of thing.
Performance is a measurable thing. It is measured all the time. There didn't have to be a positive manifold to cognitive performance, we could live in a world where there are two kinds of mental performance which are zero-sum - but nevertheless, there is a _g_.
> There's no particular reason to assume that evolution would have optimized for, e.g., skill in mathematical analysis, and therefore there isn't any particular reason to assume that even the silliest of low-hanging fruit wouldn't have been picked, for instance an elderly and distinguished analyst, such as Paul Erdos, may benefit from taking amphetamine.
Of course there's no reason to expect evolution to optimize human intelligence for humans' culturally-based and idiosyncratic desires. It optimizes for reproductive fitness. Hence if we optimize ourselves for our own desires, we are probably incurring a fitness cost, and satisfying one of the loopholes.
(I swear, I am amazed at how every time this essay shows up somewhere, I can reliably count on someone to take one of the loopholes - and no matter how repeatedly, clearly, explicitly, in bulleted or enumerated lists, I have stated them - someone will proudly take a loophole and offer it as a refutation: 'ah, what if we optimize thousands of genes simultaneously? ah, what if nature optimizes for reproduction and not our desires? ah, what about the latest self-improvement fad like dual n-back which might add an IQ point or two?')
> There are also radical differences in what we measure as "IQ", which is mostly performance on a battery of strange tasks, and the sort of mental performance that leads to evolutionary success. There are, for example, arguments to the evolutionary sufficiency of ADHD[1], OCD[2].
>
> 1: http://en.wikipedia.org/wiki/Hunter_vs._farmer_hypothesis
> 2: http://en.wikipedia.org/wiki/Ocd#Causes
Those are cute examples, I'll borrow those.
> More critically, though, is that this represents yet another article on the basis of "I-think-it-oughta" evolutionary reasoning.
No, it follows from straightforward reasoning about natural selection. How could intelligence be selected for if it does not have a fitness advantage? If it has a fitness advantage, why has it not reached fixation? And so on.
> You think "intellectual low-hanging fruit" ought to be selected for, but you don't have empirical evidence. The claim is based almost entirely on hand-waving and citations of Eliezer Yudkowsky, who is not in any case an authoritative source [3], and whose claims have not been in general accepted by contemporary neuroscientists [4].
You are badly mistaken here. I am not citing Yudkowsky because his authority 'makes it so', and it's interesting that you immediately jump to thinking that. I quote him because he formulated the idea well, and I think in some respects better than Bostrom's later paper, and his formulation makes a good jumping off point for all the other material I discuss, which is heavily referenced and generally to as standard authorities as one could wish for.
> Meanwhile, experimental evidence is contradicted. For instance, amphetamine is correlated with improvements in IQ, not to the tune of 20 points, more like four[5]. In fact, things as random -- and ostensibly detrimental -- as mescaline[6] can improve certain aspects of mental functioning, albeit at the cost of others. That what are essentially shots in the dark can produce noticeable improvements in certain qualities bodes well for rational drug design, which has been a success in other fields e.g. cisplatin vs. imatinib.
There is nothing 'random' about your choice of those substances and it is dishonest to describe it as so. You chose ampehtamines and mescaline because they are some of the very few substances which can claim to improve mental functioning as opposed to be inert or poisons. (Is a random pharmaceutical drug a random drug? No, because it has been through a rigorous selection process which may have examined hundreds of thousands of millions of chemicals in the pharmacorp or university's search for new drugs.) And as you point out, the benefits are offset by costs: mescaline has some famous effects, but amphetamines in the experiments also damage performance on some tasks. Finally, you do not show how these substances are completely free lunches, and so your entire paragraph is a non sequitur.
> Another thing that is important to keep in mind is that the workings of the body can be highly counterintuitive. For instance, one might expect choline supplementation to increase the level of brain acetylcholine. It seems natural, right? But it's not true[7].
> it is now generally believed to act primarily as an allosteric modulator on ion channels linked to AMPA-sensitive glutamate receptors[9].
Could you elaborate on how your reference supports your assertion? It seems to be about discovering (half a century later) that piracetam also affects some other receptor, but I see nothing in it to support your claims about 'it is now generally believed' to act 'primarily' on this receptor.
> The most immediately difficult claim to me is the advocacy of spaced repetition, when it is not clear why the claims of the article necessarily apply to drugs but not to techniques such as this -- could we not evolve to use it instinctively?
We do use it by default. That's what it is: multiple presentations over long time periods form strong memories. The presence of the spacing effect in many differing species (http://www.gwern.net/Spaced%20repetition#generality-of-spaci...) and different kinds of memories suggests that it's a heuristic for recognizing important regularly-repeating features of one's environment, rather than expanding substantial resources memorizing noise and regularly refreshing the synapses (see the Tononi papers for more discussion of the metabolic costs of sleep & memory). Spaced repetition itself is about faking this repeating regularity by a practice involving flash cards.
> The general thrust of my post is that "evolution" is not an argument that computer scientists can throw around in order to do biology without actually studying it[12], and that evolutionary psychology is often subject to epistemological problems, cf. The Emperor's New Paradigm[13].
I am not a computer scientist. As for the rest, readers can make up their own minds.
I feel most of what you said simply contradicts me. That you have not bothered to do basic research here is evident:
>No, it follows from straightforward reasoning about natural selection. How could intelligence be selected for if it does not have a fitness advantage? If it has a fitness advantage, why has it not reached fixation? And so on.
This is why. It should be the first goddamn thing you look up. Or, you know, a systematic review of some sort. Not your intuition. If something doesn't serve the primary drivers of the evolution of intelligence --which by no means must select for general intelligence as measured by the Raven Progressive Matrices Test, then it need not be selected for! This isn't, of course something you should take from me. Read something. By someone who actually studies biology. u.u
> I feel most of what you said simply contradicts me.
I feel I provided clear argument and at least as many citations, combined with less sloppy research and supercilious arrogance. And I note that you aren't bothering to address a single one of my points, such as pointing out that your confident claims about choline supplementation seem to be totally false.
I am sure you will be glad to explain to me, beyond a bare URL.
> If something doesn't serve the primary drivers of the evolution of intelligence --which by no means must select for general intelligence as measured by the Raven Progressive Matrices Test, then it need not be selected for!
I never said evolution optimizes performance on a RAPM and most of the essay was about this.
Which of course it is not. Some supplements work on a deficiency principle, but many do not.
> Of course I've intentionally phrased this as to make it seem ludicrous, but throughout the article he talks of "more" or "less" intelligence as though it were a fungible sort of thing.
Performance is a measurable thing. It is measured all the time. There didn't have to be a positive manifold to cognitive performance, we could live in a world where there are two kinds of mental performance which are zero-sum - but nevertheless, there is a _g_.
> There's no particular reason to assume that evolution would have optimized for, e.g., skill in mathematical analysis, and therefore there isn't any particular reason to assume that even the silliest of low-hanging fruit wouldn't have been picked, for instance an elderly and distinguished analyst, such as Paul Erdos, may benefit from taking amphetamine.
Of course there's no reason to expect evolution to optimize human intelligence for humans' culturally-based and idiosyncratic desires. It optimizes for reproductive fitness. Hence if we optimize ourselves for our own desires, we are probably incurring a fitness cost, and satisfying one of the loopholes.
(I swear, I am amazed at how every time this essay shows up somewhere, I can reliably count on someone to take one of the loopholes - and no matter how repeatedly, clearly, explicitly, in bulleted or enumerated lists, I have stated them - someone will proudly take a loophole and offer it as a refutation: 'ah, what if we optimize thousands of genes simultaneously? ah, what if nature optimizes for reproduction and not our desires? ah, what about the latest self-improvement fad like dual n-back which might add an IQ point or two?')
> There are also radical differences in what we measure as "IQ", which is mostly performance on a battery of strange tasks, and the sort of mental performance that leads to evolutionary success. There are, for example, arguments to the evolutionary sufficiency of ADHD[1], OCD[2]. > > 1: http://en.wikipedia.org/wiki/Hunter_vs._farmer_hypothesis > 2: http://en.wikipedia.org/wiki/Ocd#Causes
Those are cute examples, I'll borrow those.
> More critically, though, is that this represents yet another article on the basis of "I-think-it-oughta" evolutionary reasoning.
No, it follows from straightforward reasoning about natural selection. How could intelligence be selected for if it does not have a fitness advantage? If it has a fitness advantage, why has it not reached fixation? And so on.
> You think "intellectual low-hanging fruit" ought to be selected for, but you don't have empirical evidence. The claim is based almost entirely on hand-waving and citations of Eliezer Yudkowsky, who is not in any case an authoritative source [3], and whose claims have not been in general accepted by contemporary neuroscientists [4].
You are badly mistaken here. I am not citing Yudkowsky because his authority 'makes it so', and it's interesting that you immediately jump to thinking that. I quote him because he formulated the idea well, and I think in some respects better than Bostrom's later paper, and his formulation makes a good jumping off point for all the other material I discuss, which is heavily referenced and generally to as standard authorities as one could wish for.
> Meanwhile, experimental evidence is contradicted. For instance, amphetamine is correlated with improvements in IQ, not to the tune of 20 points, more like four[5]. In fact, things as random -- and ostensibly detrimental -- as mescaline[6] can improve certain aspects of mental functioning, albeit at the cost of others. That what are essentially shots in the dark can produce noticeable improvements in certain qualities bodes well for rational drug design, which has been a success in other fields e.g. cisplatin vs. imatinib.
There is nothing 'random' about your choice of those substances and it is dishonest to describe it as so. You chose ampehtamines and mescaline because they are some of the very few substances which can claim to improve mental functioning as opposed to be inert or poisons. (Is a random pharmaceutical drug a random drug? No, because it has been through a rigorous selection process which may have examined hundreds of thousands of millions of chemicals in the pharmacorp or university's search for new drugs.) And as you point out, the benefits are offset by costs: mescaline has some famous effects, but amphetamines in the experiments also damage performance on some tasks. Finally, you do not show how these substances are completely free lunches, and so your entire paragraph is a non sequitur.
> Another thing that is important to keep in mind is that the workings of the body can be highly counterintuitive. For instance, one might expect choline supplementation to increase the level of brain acetylcholine. It seems natural, right? But it's not true[7].
https://www.sciencedirect.com/science/article/pii/S092549270... "Oral choline increases choline metabolites in human brain"
> it is now generally believed to act primarily as an allosteric modulator on ion channels linked to AMPA-sensitive glutamate receptors[9].
Could you elaborate on how your reference supports your assertion? It seems to be about discovering (half a century later) that piracetam also affects some other receptor, but I see nothing in it to support your claims about 'it is now generally believed' to act 'primarily' on this receptor.
> The most immediately difficult claim to me is the advocacy of spaced repetition, when it is not clear why the claims of the article necessarily apply to drugs but not to techniques such as this -- could we not evolve to use it instinctively?
We do use it by default. That's what it is: multiple presentations over long time periods form strong memories. The presence of the spacing effect in many differing species (http://www.gwern.net/Spaced%20repetition#generality-of-spaci...) and different kinds of memories suggests that it's a heuristic for recognizing important regularly-repeating features of one's environment, rather than expanding substantial resources memorizing noise and regularly refreshing the synapses (see the Tononi papers for more discussion of the metabolic costs of sleep & memory). Spaced repetition itself is about faking this repeating regularity by a practice involving flash cards.
> The general thrust of my post is that "evolution" is not an argument that computer scientists can throw around in order to do biology without actually studying it[12], and that evolutionary psychology is often subject to epistemological problems, cf. The Emperor's New Paradigm[13].
I am not a computer scientist. As for the rest, readers can make up their own minds.