3 ms·

How does it predict the opposite for extreme intelligence, or any other extreme trait? It's not always the case; but sometimes what we think of as a single qual

How does it predict the opposite for extreme intelligence, or any other extreme trait? It's not always the case; but sometimes what we think of as a single quality is actually the result of several genetic factors. When you say intelligence is "under mutation-selection balance", how does that relate to extreme intelligence?

gwern 7y ago
To be under mutation-selection balance, mutations affecting a trait must be selected at a sufficient rate to maintain the trait, because it affects fitness. Those with the highest number of mutations affecting the trait must be selected against. The easiest way to affect fitness is to damage health. As it happens, mental illnesses damage fitness enormously. So, for a mutation-selection balance trait, those with the least mutations damaging health and fitness will have the highest trait values ie. extreme intelligence.
- solidsnack9000 7y ago
  In the long run, yes. But it doesn’t mean that on a case-by-case basis we won’t see people with trait enhancing mutations that are maladaptive in other ways — like high intelligence in combination with mental illness.
  - gwern 7y ago
    But the theory does not predict those should be common, and predicts they should be rare or nonexistent. So you can't invoke it as support for these findings: it is in fact a bullet which must be bitten (similar to the bullet they must bite when they admit that yes, IQ does correlate with mental health up to the Mensa level in ordinary samples like Swedish population registry or the Scottish population sample, but then postulate that it magically mysteriously reverses abruptly and mental illness rates shoot up orders of magnitude somewhere within the top percentile where conveniently those ordinary samples just can't spot any reversal).
    - solidsnack9000 7y ago
      It does predict they should be rare but extremes are rare. What is your theory for how to understand these results?
      - gwern 7y ago
        Mensa-level extremes are not rare at all. They're something like 2% of the population (probably even more when you consider they allow multiple retests and other non-IQ tests). > What is your theory for how to understand these results? There is nothing to understand. It is purely an artifact of their sampling process. There is no reversal and no mystery. IQ simply correlates with better performance and health as far up as we can measure it into samples like TIP or SMPY, period, and 100% consistent with mutation-selection balance. Mensa attracts losers, and those losers self-select even further into a survey asking about self-diagnosed problems, and these researchers were too lazy to do a good study and too dishonest to discuss forthrightly all the reasons why one knows a priori their results only demonstrate that sampling from Mensa is a really bad idea (I say 'dishonest' because they do not mention any of the very well known contrary results and I am also told these points were all brought up to them before publication and they refused to address any of them).

gwern 7y ago

To be under mutation-selection balance, mutations affecting a trait must be selected at a sufficient rate to maintain the trait, because it affects fitness. Those with the highest number of mutations affecting the trait must be selected against. The easiest way to affect fitness is to damage health. As it happens, mental illnesses damage fitness enormously. So, for a mutation-selection balance trait, those with the least mutations damaging health and fitness will have the highest trait values ie. extreme intelligence.

solidsnack9000 7y ago
In the long run, yes. But it doesn’t mean that on a case-by-case basis we won’t see people with trait enhancing mutations that are maladaptive in other ways — like high intelligence in combination with mental illness.
- gwern 7y ago
  But the theory does not predict those should be common, and predicts they should be rare or nonexistent. So you can't invoke it as support for these findings: it is in fact a bullet which must be bitten (similar to the bullet they must bite when they admit that yes, IQ does correlate with mental health up to the Mensa level in ordinary samples like Swedish population registry or the Scottish population sample, but then postulate that it magically mysteriously reverses abruptly and mental illness rates shoot up orders of magnitude somewhere within the top percentile where conveniently those ordinary samples just can't spot any reversal).
  - solidsnack9000 7y ago
    It does predict they should be rare but extremes are rare. What is your theory for how to understand these results?
    - gwern 7y ago
      Mensa-level extremes are not rare at all. They're something like 2% of the population (probably even more when you consider they allow multiple retests and other non-IQ tests). > What is your theory for how to understand these results? There is nothing to understand. It is purely an artifact of their sampling process. There is no reversal and no mystery. IQ simply correlates with better performance and health as far up as we can measure it into samples like TIP or SMPY, period, and 100% consistent with mutation-selection balance. Mensa attracts losers, and those losers self-select even further into a survey asking about self-diagnosed problems, and these researchers were too lazy to do a good study and too dishonest to discuss forthrightly all the reasons why one knows a priori their results only demonstrate that sampling from Mensa is a really bad idea (I say 'dishonest' because they do not mention any of the very well known contrary results and I am also told these points were all brought up to them before publication and they refused to address any of them).