Tucker-Drob and colleagues demonstrated that general intelligence (g) has a genetic basis, showing that it is not merely a statistical artifact. Their findings are detailed in the study titled "Genetic 'General Intelligence,' Objectively Determined and Measured," - https://www.biorxiv.org/content/10.1101/766600v3.full
James Lee (behavioral geneticist):
Roderick McDonald criticized the naming fallacy. Just saying that all correlated scores on a test are caused by a common factor is circular reasoning.
Instead, it is a much more important scientific discovery to identify an outside variable that causes observed scores to correlate.
An alternative to the factor model is the network model. It claims that there is no "mental property" that can be affected in a manner propagated to all of its observable indicators. These connections or "bonds" were hypothetical in the early 20th century. Today, they could be considered being genes that are inherited together, or brain regions that are connected.
In the factor model, if a SNP has a causal impact on a trait, then its relationship with observed scores are in proportional to the effect of the SNP, multiplied by the factor loading.
In contrast, network models make no prediction like that. There is no particular expected relationship among SNP effect, observed scores, and factor loadings.
We re-analyzed GWAS data to test these competing theories. "We found a very large number of SNPs whose pattern of associations with items matched the factor models. . . . We had a lot of SNPs that had effects that were almost exactly as predicted."
(Everything Lee is saying is in relation to personality traits, especially neuroticism.)
It is reasonable to expect these results to apply to intelligence because factor models tend to fit better with intelligence data than personality data. A study from de la Fuente et al. (2021) indicates that this may be so, but there is caution because there may be some ascertainment bias in that study.
So, what kind of thing is a psychological trait? The question remains open. But the common factors of factor analysis are real things that are subjected to physical causes of genetics and the brain
on the other hand - if you are interested in my theoretical position on this :-) ...well you can look at the tutorials and other ideas here: https://osf.io/preprints/psyarxiv/65gzh
We need to be clear what our goals are. Can education help individuals function better cognitively at work and in their personal lives? Of course it can. Can targeted interventions for strategic problem solving and decision making help with cognitive performance? Yes, ample evidence for that too. Relational reasoning training improves fluid reasoning test scores (e.g. Ravens' matrices) and has real world benefits for instance. Can brain training games improve cognitive functioning - yes, this has been demonstrated for working memory and attention and cognitive resilience - to the extent that such training is used by professionals in practice. Focusing on such goals is my interest, much like any educator would be interested in them. Network theories understand 'general intelligence' in this embedded, real world performance way, more than 'g' theorists do.
This is an ongoing debate and yes, interesting to follow. But it is clear to me from an intervention perspective that of course IQ levels change with interventions - education alone increases IQ by 2-3 points per year. Individuals go in and out of college with considerable IQ gains (10+ points), and so on. The Flynn effect has been considerable and this cannot be explained by genetics.
1. Education has no effect on g. IQ and g are not the same thing. There is a world of difference between effect on g and on broad abilities - The reason it matters a lot is because g has high predictive validity, compared to non-g abilities. So even if you think education is useful, it is incredibly intellectually dishonest to say that something increases IQ, without differentiating between effect on g and non-g abilities. It matters a lot!
2. I have examined studies on relational frame theory - which you, and the others claim increases fluid reasoning and in 5 studies with control groups, examined in meta analysis, i have found that 1 or 2 has any meaningful impact on fludi reasoning, but i remember that the effect on matrix reasoning was almost non - existent in 3 or 4 studies. Even if it does have some impact on fluid reasoning, ones performance in real life will still be limited by working memory and processing speed. We know that working memory training not only does not have a far transfer, but it also does not have a near transfer - to other types of working memory - for example dual n back does not translate into other types of working memory. this has been well documented.
3. "Can targeted interventions for strategic problem solving and decision making help with cognitive performance?" They cannot be implemented efficiently in real life. They cannot compensate for g.
I appreciate what you are doing, I don't want to come across as cynical, but there is a mountain of fancy words and fancy ideas that you have produced, without any valuable practical implementations.
It would be much more helpful if you found some biochemical pathways that influence cognition.
Sorry for spelling and grammar, I am in a hurry and am not a native english speaker.
Tucker-Drob and colleagues demonstrated that general intelligence (g) has a genetic basis, showing that it is not merely a statistical artifact. Their findings are detailed in the study titled "Genetic 'General Intelligence,' Objectively Determined and Measured," - https://www.biorxiv.org/content/10.1101/766600v3.full
James Lee (behavioral geneticist):
Roderick McDonald criticized the naming fallacy. Just saying that all correlated scores on a test are caused by a common factor is circular reasoning.
Instead, it is a much more important scientific discovery to identify an outside variable that causes observed scores to correlate.
An alternative to the factor model is the network model. It claims that there is no "mental property" that can be affected in a manner propagated to all of its observable indicators. These connections or "bonds" were hypothetical in the early 20th century. Today, they could be considered being genes that are inherited together, or brain regions that are connected.
In the factor model, if a SNP has a causal impact on a trait, then its relationship with observed scores are in proportional to the effect of the SNP, multiplied by the factor loading.
In contrast, network models make no prediction like that. There is no particular expected relationship among SNP effect, observed scores, and factor loadings.
We re-analyzed GWAS data to test these competing theories. "We found a very large number of SNPs whose pattern of associations with items matched the factor models. . . . We had a lot of SNPs that had effects that were almost exactly as predicted."
(Everything Lee is saying is in relation to personality traits, especially neuroticism.)
It is reasonable to expect these results to apply to intelligence because factor models tend to fit better with intelligence data than personality data. A study from de la Fuente et al. (2021) indicates that this may be so, but there is caution because there may be some ascertainment bias in that study.
So, what kind of thing is a psychological trait? The question remains open. But the common factors of factor analysis are real things that are subjected to physical causes of genetics and the brain
Here you go... thre is no definitive 'answer' to this ongoing debate. It is an ongoing theory construction and discovery process - that's the nature of science (making it distinct from dogma, etc. https://open.substack.com/pub/iqmindware/p/major-theories-of-intelligence-a?r=284rjh&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
on the other hand - if you are interested in my theoretical position on this :-) ...well you can look at the tutorials and other ideas here: https://osf.io/preprints/psyarxiv/65gzh
We need to be clear what our goals are. Can education help individuals function better cognitively at work and in their personal lives? Of course it can. Can targeted interventions for strategic problem solving and decision making help with cognitive performance? Yes, ample evidence for that too. Relational reasoning training improves fluid reasoning test scores (e.g. Ravens' matrices) and has real world benefits for instance. Can brain training games improve cognitive functioning - yes, this has been demonstrated for working memory and attention and cognitive resilience - to the extent that such training is used by professionals in practice. Focusing on such goals is my interest, much like any educator would be interested in them. Network theories understand 'general intelligence' in this embedded, real world performance way, more than 'g' theorists do.
As I said - I recommend a close reading of: https://www.mdpi.com/2079-3200/11/1/19
This is an ongoing debate and yes, interesting to follow. But it is clear to me from an intervention perspective that of course IQ levels change with interventions - education alone increases IQ by 2-3 points per year. Individuals go in and out of college with considerable IQ gains (10+ points), and so on. The Flynn effect has been considerable and this cannot be explained by genetics.
1. Education has no effect on g. IQ and g are not the same thing. There is a world of difference between effect on g and on broad abilities - The reason it matters a lot is because g has high predictive validity, compared to non-g abilities. So even if you think education is useful, it is incredibly intellectually dishonest to say that something increases IQ, without differentiating between effect on g and non-g abilities. It matters a lot!
2. I have examined studies on relational frame theory - which you, and the others claim increases fluid reasoning and in 5 studies with control groups, examined in meta analysis, i have found that 1 or 2 has any meaningful impact on fludi reasoning, but i remember that the effect on matrix reasoning was almost non - existent in 3 or 4 studies. Even if it does have some impact on fluid reasoning, ones performance in real life will still be limited by working memory and processing speed. We know that working memory training not only does not have a far transfer, but it also does not have a near transfer - to other types of working memory - for example dual n back does not translate into other types of working memory. this has been well documented.
3. "Can targeted interventions for strategic problem solving and decision making help with cognitive performance?" They cannot be implemented efficiently in real life. They cannot compensate for g.
4. The Flynn effect - see answers listed here: https://qr.ae/pYeX0b
I appreciate what you are doing, I don't want to come across as cynical, but there is a mountain of fancy words and fancy ideas that you have produced, without any valuable practical implementations.
It would be much more helpful if you found some biochemical pathways that influence cognition.
Sorry for spelling and grammar, I am in a hurry and am not a native english speaker.
As they say, we will have to agree to disagree!
More ‘incredible intellectual dishonesty’ 😅 https://www.sciencedirect.com/science/article/abs/pii/S0022096509002021
Rather than personal insults I think we should leave it in good scientific spirit as your namesake - ‘in my opinion’ (IMO) 🙏