J2014 Posted May 8, 2011 Share Posted May 8, 2011 (edited) I think SamTheSkeptic is right in that using observation exclusively is inadequate to show that a correlation is actually a causation. I think that is the main criticism, hypotheses can't be test in humans because we would have to induce the genetic lesion to see if that results in the predicted change in behaviour. If there were no ethical consideration one could of course just remove a gene for a serotonin and see if that increases the chances of aggression later in life, like it does in mice. Or we could induce the mutations in the monoamine oxidase gene,that have been shown to increase the likelihood of aggression in humans, in a normal individual and see if that leads to the same effect. We can't obviously. So how could we see if the correlations found between genes and behavious are in fact causative. One approach could be pharmacogenomics. Since we can't induce negative behavioural changes genetically, we could instead see how the effects of drugs on behaviour varies with genetics. But then again as no genetic manipulation is being induced, causation would still be difficult to prove. But if the drug directly affect the mechanism perturbed by the genetic lesion, that does bolster the argument that it was the underlying genetic mutation that induced the behaviour. Gene therapy could be another way one could get at this problem. Correct the genetic lesion and see if that changes behaviour. If eventually your view (which is currently the most popular) becomes generally accepted, then how many people will be able to get off the hook for murder because they were "genetically prone to violent behavior" ? How many sex offenders will have their sentences reduced because they had the "sex offender gene" ? The same critique could be said of determinism in general. If we don't have free will then we can blame anything for our behaviour... environment, upbringing, genes, I could even imaging a murder appealing to a judge that he did it cause "the random firing of a neuron in complex network lead to a the emergent phenomena of me swinging an axe at someones head"! The solution is that we shouldn't lock a murderer up because they are ultimately responsible for all the decisions they make, but because the decisions they make violate the rights of other people. Jail is not then an revenge, but a protection for others and a punishment to correct the offender's behaviour. Therefore, saying we should not believe that genes affect behaviour bacause it could lead to bad effects is nonsense. John Edited May 8, 2011 by JohnG Link to comment Share on other sites More sharing options...
SMF Posted May 8, 2011 Share Posted May 8, 2011 JohnG: I think SamTheSkeptic is right in that using observation exclusively is inadequate to show that a correlation is actually a causation. I think that is the main criticism, hypotheses can't be test in humans because we would have to induce the genetic lesion to see if that results in the predicted change in behaviour. If there were no ethical consideration one could of course just remove a gene for a serotonin and see if that increases the chances of aggression later in life, like it does in mice. Or we could induce the mutations in the monoamine oxidase gene,that have been shown to increase the likelihood of aggression in humans, in a normal individual and see if that leads to the same effect. We can't obviously. You seem to be confusing simple correlation with the experimental evidence provided by genetic diversity. For some traits the genetic differences (lesions?) are there for scientific observation and the twin studies, for example, are much more than simple correlations. They are natural experiments. SM Link to comment Share on other sites More sharing options...
J2014 Posted May 9, 2011 Share Posted May 9, 2011 The problem with natural experiments tho is that they are very poorly controled, and that is what prevents these corralative studies showing causation. In the twin studies for example, monozygotic twins do not just share the same DNA, they also share the same placenta. Surely any correlation between genetics and behaviour could then be due to shared developement in the womb rather than shared genetics? Similar lack of controls can be found in genome wide association studies, genes may be correlated with unknown environmental factors rather than behaviour itself, eg people with a specific mutation may just happen by chance to be more likely to be part of a specific socioeconmic group. While that specific example could be controled, what if the environmental factor with which the gene was correlated was unknown? I do not think twin studies or GWAS are worthless, they are very good at identifying possible genes that may infulence behaviour, but by themselves they do not show causation. John Link to comment Share on other sites More sharing options...
SMF Posted May 9, 2011 Share Posted May 9, 2011 (edited) So John, are you saying that in a large study which finds that if a trait is present that 100% of identical twins reared together, 100% of identical twins reared apart, 50% of nonidentical twins reared together or apart, but only 4% of adoptees reared together will share it, and 4% is the rate the trait is found in the general population as a whole, that the trait is not conclusively the result of genetic inheritance? This is an example of an extreme data set, but with large numbers of subjects the genetic component of many complex traits have been attributed, with high statistical significance, proportionally to inheritance or environment. This is what was contested in the discussion above. I don't think that anybody is claiming that simple demographic correlations with the frequency of some trait demonstrates causation. SM Edited May 9, 2011 by SMF Link to comment Share on other sites More sharing options...
J2014 Posted May 9, 2011 Share Posted May 9, 2011 So John, are you saying that in a large study which finds that if a trait is present that 100% of identical twins reared together, 100% of identical twins reared apart, 50% of nonidentical twins reared together or apart, but only 4% of adoptees reared together will share it, and 4% is the rate the trait is found in the general population as a whole, that the trait is not conclusively the result of genetic inheritance? Indeed that is exactly what i am saying, it does not conclusively show the result is due to genetic inheritance. It could also be due to sharing the exact same conditions in the womb. Non identical twins could have a lower correlation bacause although they share the same womb, they do not share the same placenta. I am not saying that this shared developement hypothesis is true, but it should not be disregarded either. Link to comment Share on other sites More sharing options...
SMF Posted May 9, 2011 Share Posted May 9, 2011 John, you are playing a distracting logical game. I don't believe that there are very many single studies, in any field, that by themselves don't have some uncertainty. This is why there is replication and extension of results by subsequent experimentation. Scientific certainty is never certain, it is always provisional and is accompanied with confidence limits. Further, like most research the twin studies are hypothesis driven and based on previous research. Your invoking of "just correlation" for this kind of research is inappropriate. Save it for the pop culture sassafras roots prevent cancer claims. If you wish to continue with this silliness, I challenge you to come up with a half dozen differences in normal individuals that can be attributed to normal variations of the placenta. Give me one example. Find me the study that shows the effects of normal placental variation and using your logic I could claim it is just correlation. How about the competence of the delivery teams? How about the disposition of the family dog? This is a game that real scientists do not play; instead criticisms are based on existing facts and contradictory evidence. What is yours for the twin studies? SM Link to comment Share on other sites More sharing options...
J2014 Posted May 9, 2011 Share Posted May 9, 2011 (edited) John, you are playing a distracting logical game. I don't believe that there are very many single studies, in any field, that by themselves don't have some uncertainty. This is why there is replication and extension of results by subsequent experimentation. Scientific certainty is never certain, it is always provisional and is accompanied with confidence limits. Further, like most research the twin studies are hypothesis driven and based on previous research. Your invoking of "just correlation" for this kind of research is inappropriate. Save it for the pop culture sassafras roots prevent cancer claims. If you wish to continue with this silliness, I challenge you to come up with a half dozen differences in normal individuals that can be attributed to normal variations of the placenta. Give me one example. Find me the study that shows the effects of normal placental variation and using your logic I could claim it is just correlation. How about the competence of the delivery teams? How about the disposition of the family dog? This is a game that real scientists do not play; instead criticisms are based on existing facts and contradictory evidence. What is yours for the twin studies? SM I think you dont understand my arguement. You're the one who claimed that twin studies showed "conclusively" the result of genetic inheritance. It sounded a bit like naive verificationism so I gave an possible alternative interpretation of the data that is not falsified by that experiement. If we are to come at this question from an unbiased perspective then all the twin studies you metioned give support to the shared placenta hypothesis. At least to the extent that they also give support to the shared DNA hypothesis. I might as well ask you if you could give me any evidence that the shared DNA of twins is not down to shared developement. The only way would be to implant the embryos of monozygotic twins into different women and then see if their behaviour still has a high degree of correlation. Or alternatively implant two genetically unrelated embryos into one person, and see if that increases the likelyhood of similar characteristics. Obviously neither experiment has or will be done. Can you give me an example of a genetic change that was induced in a human to see if it changed behaviour? Since natural experiments are very poorly controled i think you would have a hard job. Anyway as for some epidemiology on the effects of sharing a placenta, below is a study showing that approx 12% of the variance in IQ in monozygotic twins derives from sharing or not the same placentas Jacobs et al 2001, Heritability estimates of intelligence in twins: effect of chorion type. Behav Genet. 2001 Mar;31(2):209-17. http://www.ncbi.nlm....pubmed/11545537 and here is a short review with respect to schizophrenia http://www.its.calte...erspec%2007.pdf some good links in that too if you're interested. This paper ( http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2632220/?tool=pubmed ) suggests that up to "14% of schizophrenia cases would not have occurred if influenza infection during early to mid-gestation had been prevented". If that is true, then could this have been misconstrued in the twin studies as a genetic cause? What other environmental factors in the womb could also be misconstrued as genetic? I don't know. However looking for alternate explanations for what has apparently been "conclusively" shown to be true IS the job of a real scientist, and indeed you should attempt to argue that it may have been down to other factors, but you should not belittle other explanations because it doesn't fit within your narrow view of science. Edited May 9, 2011 by JohnG Link to comment Share on other sites More sharing options...
SMF Posted May 9, 2011 Share Posted May 9, 2011 (edited) OK John. The Schizophrenia article suggests that infection of the mother might be responsible for the disease. I asked you to provide evidence of phenotipic variability of normal traits, but you give me disease. In any case, because this article cautions against correlational studies regarding this relationship, saying- "Although epidemiological studies cannot establish causality," -it is basically an example, in your terms, of a bad example. I just love the the Jacobs et.al. (2001) piece because it is an excellent illustration of what I have been saying. Here is a twin study that is able to identify the contribution of the intrauterine environment component that affects intelligence, relative to the inherited and post uterine environmental component, that can be attributed to monochorionic versus dichorionic environments of monozygotic twins. But if I wished to play your game I could still ask about the effect of the competence of the delivery teams and the disposition of the family dog on the result. Instead, I think I will just accept this study as a good example of scientific investigation about which I have been talking. Just correlation eh, you make my point. SM Edited May 9, 2011 by SMF Link to comment Share on other sites More sharing options...
J2014 Posted May 9, 2011 Share Posted May 9, 2011 I asked you to provide evidence of phenotipic variability of normal traits IQ is not a normal trait? But if I wished to play your game I could still ask about the competence of the delivery teams and the disposition of the family dog Firstly you should be asking about the competence of the delivery teams because a traumatic birth can also be correlated with schizophrenia. The disposition of the family dog is not relevent because, it would, i assume, not be a shared by twins reared apart. Futhermore there is no mechanism that you can extrapolate from animal models that would give rise to such an effect. There is however for the effects of share experience in the womb (see the previous schizophrenia review paper) Just correlation eh, you make my point. SM I really dont. The fact that the epidemiolgists in the journal say "Although epidemiological studies cannot establish causality," should show you why it is not correct to imply that twin studies could show "conclusively" genetic inhevitance of behaviour traits. I used your terms to show that epidemiology is very poor at showing causation, because when controled for something else... in the schizophrenia case, infection with influenza...... a different result can be attained. This shows that epidemiolgy is by its nature poorly controled. It can show correlation, it is excellent at identifing possible genes involved in disease, but because it is poorly controled, it cannot show causation. Instead, I think I will just accept this study as a good example of scientific investigation about which I have been talking Hmmmm, i really dont think you were. I am pretty sure you were implying that twin studies could show "conclusively" genetic inhevitance of behaviour traits. They showed it is not neccessary conclusive at all. They gave one example of an environmental factor that can be misconstrued as a genetic factor, as I suggested could be the case. There could therfore be many other environmental factors also misconstrued for genetic factors. We cant know because epidemiology and natural experiments lack proper controls. Link to comment Share on other sites More sharing options...
SMF Posted May 10, 2011 Share Posted May 10, 2011 John. You are agreeing with me on everything I have said but claiming I am wrong. The Schizophrenia opinion article you offered was about disease (infection of the mother during pregnancy) and the author admitted that it was only citing correlations, while the intelligence research article was a good twin study of the type that I have been saying is good science. Strange. You don't appear to understand what you are talking about, or how this type of science progresses, so further discussion is not worth pursuing. Perhaps someone else here can clarify for me how the Minnesota and Swedish twin projects are just producing correlational epidemiology, because I think this assertion is obviously incorrect. SM Link to comment Share on other sites More sharing options...
J2014 Posted May 10, 2011 Share Posted May 10, 2011 (edited) Give me one example. Find me the study that shows the effects of normal placental variation and using your logic I could claim it is just correlation. I did the above and now you dont want to accept the logical conclusion to save face. as for the pitiful attempt to disparage my argument by saying I do not understand what I am talking about, or the progress of science, I have already explained how I think epidemiology fits into the progress of science, providing important insight in to what genes may be important that can be further explored in animal models. If you actually think that any statement in science is obviously correct, or any experiment can find conclusive evidence then you really have no idea how science works. I think it is you who has no idea what you are talking about. You certainly have very poor debating skills. You asked me for epidemiological evidence that could show, in principle, that the twin studies data could be due to developmental factors. I did and now you only respond with vague insults. As for the iq twins study, I don't think it does necessary show that placental status affects iq. There could for example be a gene, or collection of genes, that both increase the chances of two embryos sharing a placenta, and quite independently of that effect, also have an effect on iq. Edited May 10, 2011 by JohnG Link to comment Share on other sites More sharing options...
SMF Posted May 10, 2011 Share Posted May 10, 2011 John. To support your notion that twin studies are poorly controlled you found a good twin study that supports my contention that this research is experimental and not just simple correlation, and now you are disparaging your find. I conclude that you don't understand the research. SM Link to comment Share on other sites More sharing options...
J2014 Posted May 10, 2011 Share Posted May 10, 2011 (edited) John. To support your notion that twin studies are poorly controlled you found a good twin study that supports my contention that this research is experimental and not just simple correlation, and now you are disparaging your find. I conclude that you don't understand the research. SM What control in that experiment distingueishes between their hypothesis, that placental status can influence IQ, and the alternate hypothesis that genetics influenced them both? There was none. That is becasue natural experiments are limited in controls. The researchers do not randomly induce the experimental manipulation to see what the effects are, rather they find "natural" experimental manipulations. Because the induction of the experimental manipulation in not random, other factors can lead to both the assumed controled experimental manipulation being merely correlated with an effect, rather than there being a causal relationship. Surely the only way to investigate if an effect is causal or spurious is to perform a control? Since you appear not to understand what confounding factors are, or why the are so prevalent in observational studies, I conclude that you don't understand the research. This article makes some of the arguements i put forward very well Why Twin Studies Are Problematic for the Study of Political Ideology: Rethinking Are Political Orientations Genetically Transmitted? http://sites.lafayette.edu/suhaye/files/2010/01/plugin-critique_of_twin_studies_-_suhay__kalmoe__mcdermott_101007.pdf also Twin Studies of Political Behavior: Untenable Assumptions? Jon Beckwith and Corey A. Morris DOI: 10.1017/S1537592708081917 Edited May 10, 2011 by JohnG Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now