Tuesday, April 12, 2016 ... /////

Humans are the smartest animals

Differences between species aren't precise but may be way too obvious

P.Z. Myers has continued his tirade against the suggestions that the intelligence of humans could increase due to genetic modification. In his new text, On the absurdity of g, he attacks the very notion of the IQ or g factor.

Animals may be smart. I've actually seen cooler videos on IQ tests that some apes, monkeys, and even rodents, could have passed, but I can't find them. Despite the cuteness of these smart animals, we know that most humans would beat them in these – and especially other – tasks. When I say that some human is stupider than some average animals, I am always exaggerating.

Myers argues that the intelligence is multi-dimensional and I agree with that. But this fact doesn't make the concept of the IQ useless or inapplicable.

Why? Because if there are many "types" of intelligence, quantities that may be labeled $IQ_1,IQ_2,IQ_3,\dots ,IQ_n$, and let's assume that all of these quantities are distributed around 100 with the standard deviation 15, the natural selection typically punishes individuals according to the weakest link, the minimum from the values $IQ_i$. And needless to say, many non-cognitive skills and variables affect the survival in Nature and in the society; those are probably more consequential than any of the $IQ_i$ variables.

Savants have some extreme skills – they differ from each other because there are many disciplines in which they may dominate – but those extreme isolated skills are typically "useless" because a balanced mixture of skills is typically better for the survival and on the job market. Myers wrote a similar thing but his wording made me laugh:

But the whole thing about being neurotypical is that we’re in a state of a kind of balanced mediocrity…all the bits and pieces of our minds are working together in a low-stress fashion, and in a similar way to how most other people’s minds are working. You just can’t make mediocrity extreme.
I laughed out loud because I called him a typical "genius of mediocrity" in the previous blog post and he has proudly embraced this label. He may be a genius of mediocrity but as he must have overlooked, aside from him and his likes, there also exist people who are actually intelligent as well as people who are incredibly intelligent.

Because of the evolution pressure that tries to punish those with a very low $IQ_i$, one of the values (tribes with no memory are likely to die away even if they are able to calculate 1,000 digits of $\pi$ using marbles), individuals tend to have similar values of all those $IQ_i$ variables. If you allow me to replace this assertion with just a slightly different formulation of a similar fact, when someone is good at mathematics at school, she is likely to be good at physics as well, and so on.

Despite the multi-dimensionality, the skills at all mental tasks tend to be positively correlated. This correlation is a consequence of the fact that various mental activities depend on "similar circuits" in the brain etc. And, as I mentioned, the "balanced" character of all the "dimensions" of IQ is pushed even closer towards the same universal values by the natural selection that tries to disfavor the lowest values of the lowest $IQ_i$.

It means that despite the multidimensionality, there exists a vaguely defined direction in the space of cognitive skills or an inaccurately defined variable that will be very useful for predicting many aspects of the individual or group's cognitive skills. If you check the Wikipedia's definition of the g factor, you will see:
The g factor (also known as general intelligence, general mental ability or general intelligence factor) is a construct developed in psychometric investigations of cognitive abilities and human intelligence. It is a variable that summarizes positive correlations among different cognitive tasks, reflecting the fact that an individual's performance on one type of cognitive task tends to be comparable to that person's performance on other kinds of cognitive tasks. The g factor typically accounts for 40 to 50 percent of the between-individual performance differences on a given cognitive test, and composite scores ("IQ scores") based on many tests are frequently regarded as estimates of individuals' standing on the g factor...
So the very basic experimental observation (which may also be justified theoretically by the natural selection and the weakest link, as I did) that makes it useful to think about things like the g factor or the IQ is that the results in various kinds of cognitive skills tests are positively correlated with each other. So one may construct some kind of a weighted average of all the "kinds of IQ" to produce one that is most helpful for the predictions of all types of cognitive tests, one that is rather strongly correlated with all of them.

The ironic fact is that this "positive correlation between different kinds of IQ scores" of an individual – between his "different kinds of cognitive skills" – is what underlies the usefulness of the IQ, and this positive correlation is actually totally acknowledged by Myers as well – because he actually says that the humans are so great because their different cognitive strengths are so balanced, that we're good at everything. But for some reason, he uses this balance as an argument against the IQ. This is bizarre, isn't it? At least one side must be making a serious mistake.

Be sure that it's Myers who is thinking irrationally. The correlation between the different kinds of the IQ does make one IQ useful, and Myers' skepticism is just wrong.

People may be smarter or stupider and the differences may often look striking. But those are differences corresponding to the "few dozens" of difference in the people's IQs. I am confident that if you measured the animals' cognitive skills using any sort of an IQ test that they would be capable of mastering at all, the animals would generally get vastly smaller scores.

The animals, although some of them may really be smart and many of them (and most of pets) are considered cute, are just "obviously" stupider than even the stupidest humans. I haven't thought about the problem too deeply but it seems obvious to me that when you preserve the "additive linear IQ scale", most mammals would almost certainly have negative IQs – more than 6 human standard deviations lower than the average humans – and non-mammals would be much less intelligent still, perhaps negative hundreds or so.

Dogs, chimps, and dolphins could be the "strongest contenders" in the search for the smartest animals who are not men (assuming that women aren't running, of course). But if you introduced any particular IQ test (probably one not demanding a pen), the differences could be really radical.

My main attitude to the IQ is the following: The IQ doesn't measure "everything" but it clearly measures "something". One may see that certain groups of people get considerably and repeatedly higher IQ scores than other groups, pretty much regardless of the detailed character of the IQ tests. This doesn't mean that they're "superior" in all respects or that the precise difference is the most important number in the world. But the IQs are meaningful, not purely random, they contain some information, and they're guaranteed to be correlated with (if I avoid the verb "imply") lots of other things.

So I am sure that lots of you have disliked the IQ measurements – I mostly did. You were worried that you may get a score below 160 and maybe even below 150. You didn't like that sensible observers around you could have been misled by this result because your "actual" IQ is clearly some 170 or so. Some classmates who have seen the "typical tasks of IQ tests" before could have gotten a higher score than you did, although you know that they're not really smarter. On the other hand, and that's important for the usefulness of the IQ, you didn't get the IQ below 110 like many classmates of yours. The correlation between the IQ and your hypothetical "better ways" to estimate the people's intelligence was imperfect but self-evidently positive. Even the most childish IQ tests with colorful squares and triangles were rather useful e.g. to eliminate folks who couldn't become geneticists or particle physicists.

It would be hard to measure some birds' or sharks' IQ and the dependence on the methodology could be dramatic. At the same moment, I am pretty much certain that the actual difference between the IQ of a rodent and the IQ of a frog is even (much) higher. A frog's or bird's IQ could differ by 50 points, depending on the methodology, but the difference between their IQs is probably equal to several hundreds or more. I am confident that predators (animals) generally "feel" and find it obvious that their prey is stupid in comparison, just like we feel that a bunny is cute but not too clever. So even in the absence of some "high precision science", there is a useful concept that may be exploited to produce valid statements and successful predictions about lots of cognitive skills of the animals.

The same comments apply in the opposite direction – IQs higher than the present humans. I don't mean demons, angels, and God – as in the diagram above, because I don't really think that they "exist" in some tangible way – but I do mean the future species or modified humans etc. Their intelligence may be (and, very likely, will be) vastly higher than the intelligence of the present humans.

It is not possible to quantify "the" precise IQ of these future hypothetical genetically modified humans. There is no "the" privileged definition of the IQ that could be used even in extreme circumstances. On the other hand, the fact that their IQ – whatever the exact definition or test or measurement protocol is – will be higher than the contemporary people's IQ will be as self-evident as the fact that even loop quantum gravity physicists are smarter than the average squirrels let alone earthworms.

The noise may exist and in the case of the IQ, it surely does. But the existence of noise and various systematic and statistical errors does not imply that the signal cannot exist. The signal exists – it always exists on top of noise. And P.Z. Myers is just denying basic facts about the reality, e.g. that some animal species are vastly more intelligent than others.

IQ and error margins

The climate skeptic activist Tom Nelson recently criticized the concept of the IQ because no one ever quantifies the error margins and this makes the concept unscientific.

It's true that the error margins are rarely quantified, if ever. But this fact has good reasons. "The thing" simply cannot be defined by one overwhelming definition. You can operationally define the IQ as the result of some particular IQ test and you try to make the IQ test "good". But there's no way to quantify the deviation of the IQ measured by your test from the IQs measured by other tests and methods – because there's no objective way to decide "which tests are still good enough". The differences between the IQs will depend on your tolerance towards "not really professionally composed" IQ tests. And this adjective has an unavoidable subjective aspect, too.

So "the IQ" (which would be a single quantity everyone could agree upon) isn't sufficiently sharply defined to allow you to quantify the error margin. However, many things are sufficiently sharply defined so that you may estimate the error. If you find that the IQ resulting from a particular test is correlated with the grade in a subject, you may quantify the correlation and the average error margin – the difference between the actual grade and the grade predicted from the IQ. For big enough statistics, it works to a certain extent that may be quantified and if you consider a new person that looks like a "part of the same ensemble", you can predict the grade from the IQ including the error margin because the error margin was measured.

In other words, the typical absence of error margins when people talk about the IQ is a symptom of the IQ's not being quite a precision science. But it is not enough to show that it is a totally illegitimate or unscientific or vacuous concept. The repeatedly verifiable existence of positive correlations between various cognitive skills – and IQ tests – is what rules out the hypothesis that the IQ is total bogus.

The claim isn't that there exists some canonical definition of the IQ or a measurement protocol to get it. There almost certainly isn't such a thing. The claim is that if you take one of these sensible enough protocols or definitions, you will obtain a variable that is imprecise but precise enough to allow you to make lots of valid statements about individuals and groups whose IQ ended up being sufficiently different.

Off-topic: 10 AGW consensus-skeptical papers a week in 2016

It's off-topic but I was highly intrigued by it. Despite John Kerry's and some other left-wing aßholes' attempts to prosecute people who have made sane conclusions about the topic of the alleged climate threats, The Washington Times building on Pierre Gosselin's monitoring of the situation pointed out that so far in 2016, about 10 papers that disagree with what many climate alarmists (if not most of the IPCC) call the "consensus on climate change" are being published in real-world journals every week.

That's an increased rate from 5-or-so per week in the recent 2 years. Imagine, every day, more than one AGW skeptical paper appears – while some people would almost like to outlaw any disagreement with the climate alarmist propaganda. The disconnect between the global warming activism (and global warming fascism, which is more accurate once it involves some powerful politicians) and the actual science has grown to astronomical proportions.

Fukushima tritium water is OK to dump

Japan is getting ready to pour some water with an increased fraction of tritium (hydrogen-3) from the retired Fukushima power plant to the ocean. I hope it works OK but there is obviously a controversy. I wouldn't like to drink the water, although I do believe that it would probably be safe as well, but I would be willing to take the responsibility for releasing it.

Other isotopes have been removed and the tritium that remains in the water is 57 milliliters. Tritium decays to helium-3 by beta-decay with lifetime of 12 years. The volume of the ocean is over $10^{24}$ liters and $10^{-18}$ of its hydrogen atoms are tritium, so it's about one million liters. Well, slightly less than that because the tritium is more rare in deeper ocean. 57 milliliters get diluted really quickly, anyway. I find it plausible that not even one fish would be lethally harmed by the operation.