New Mutations - New Eugenics?

True or false: you inherit your genes from your parents.





Mostly true, but not quite. In theory, you do indeed get half of your DNA from your mother and half from your father; but in practice, there's sometimes a third parent as well, random chance. Genes don't always get transmitted as they should: mutations occur.



As a result, it's not true that "genetic" always implies "inherited". A disease, for example, could be entirely genetic, and almost never inherited. Down's syndrome is the textbook example, but it's something of a special case and until recently, it was widely assumed that most disease risk genes were inherited.



Yet recent evidence suggests that many cases of neurological and psychiatric disorders are caused by uninherited, de novo mutation events. Here are two papers from the last few weeks about schizophrenia(1,2) - but the story looks similar for autism, intellectual disabilities, some forms of epilepsy, ADHD, and others. Indeed they're often the same mutations.



Biologically, a given mutation is what it is, whether it's de novo or inherited. But on a social and a psychological level, I think there are crucial differences, and in particular I think that if it turns out that de novo mutations are important in disease, we're going to see attempts to take these variants out of circulation - far more so than in the case of the very same genes, were they inherited.



The old eugenics movement was based on the idea that if we stop people with bad genes from breeding - by sterilization, voluntary or otherwise, say - we'll be able to eliminate diseases and other undesirable traits. This idea is now generally regarded as extremely unethical, but many of its opponents have shared with the eugenicists the belief that it could work.



But if de novo mutations are what cause the majority of disease, then this approach would be pointless. Sterilizing certain people, or encouraging the healthy ones to have more children, would never be able to eliminate the 'bad genes' because new ones are being created every generation, pretty much at random.



So the de novo paradigm ought to be welcomed by opponents of eugenics. It wasn't just morally wrong - it was biologically misguided too.



But hang on. This is the 21st century. We have in vitro fertilization (IVF), and you can analyze the genes of an IVF embryo before you decide to make it into a child. In the near future, we might be able to routinely sequence the genome of any unborn child shortly after conception.



From there, it would be a small step to allowing parents to decide not to have children with de novo mutations.



This would be, in its effects, a form of eugenics - in the sense that it would produce the effect that the old eugenicists wanted. No more 'bad' genes, or not nearly as many. Opinions will differ as to whether it's morally different. But I would have said that politically, it's a lot more likely to happen.



I can't see forced sterilization returning any time soon. But if you were expecting a baby and you knew that it was not just carrying your and your partner's DNA, but had also suffered a mutation - might you not want to avoid that?



Psychologically, it matters that it did not inherit the gene. It would be a big step to decide that your child should not inherit one of your own genes. Of course, some genes are obviously harmful, like one that raises the risk of cancer, but think about the grey areas - a gene for social anxiety, mild autistic symptoms, obesity, a personality trait.



You might well feel that carrying that gene is what makes you, you; and so it would be natural for your child to have it. You might decide that if it was good enough for you (and all your ancestors), it's good enough for your children. You might well resent the very idea that it's a 'bad' gene at all, as an attack on your own self-worth.



But none of that applies if it's a de novo mutation. Indeed, quite the opposite - all those same considerations would probably lead you to want your children to carry as close as possible to a carbon copy of your DNA, with no random changes. It was good enough for you.



My point is that I think there will be much more support for the idea of genetic screening against de novo mutations than against inherited genes. More people will want it, it will be more socially acceptable, and more widely used. I'm not saying this would be a good or a bad thing, just making a prediction. In the future, diseases and traits that are primarily caused by de novo mutations will increasingly selected against.

New Mutations - New Eugenics?

True or false: you inherit your genes from your parents.





Mostly true, but not quite. In theory, you do indeed get half of your DNA from your mother and half from your father; but in practice, there's sometimes a third parent as well, random chance. Genes don't always get transmitted as they should: mutations occur.



As a result, it's not true that "genetic" always implies "inherited". A disease, for example, could be entirely genetic, and almost never inherited. Down's syndrome is the textbook example, but it's something of a special case and until recently, it was widely assumed that most disease risk genes were inherited.



Yet recent evidence suggests that many cases of neurological and psychiatric disorders are caused by uninherited, de novo mutation events. Here are two papers from the last few weeks about schizophrenia(1,2) - but the story looks similar for autism, intellectual disabilities, some forms of epilepsy, ADHD, and others. Indeed they're often the same mutations.



Biologically, a given mutation is what it is, whether it's de novo or inherited. But on a social and a psychological level, I think there are crucial differences, and in particular I think that if it turns out that de novo mutations are important in disease, we're going to see attempts to take these variants out of circulation - far more so than in the case of the very same genes, were they inherited.



The old eugenics movement was based on the idea that if we stop people with bad genes from breeding - by sterilization, voluntary or otherwise, say - we'll be able to eliminate diseases and other undesirable traits. This idea is now generally regarded as extremely unethical, but many of its opponents have shared with the eugenicists the belief that it could work.



But if de novo mutations are what cause the majority of disease, then this approach would be pointless. Sterilizing certain people, or encouraging the healthy ones to have more children, would never be able to eliminate the 'bad genes' because new ones are being created every generation, pretty much at random.



So the de novo paradigm ought to be welcomed by opponents of eugenics. It wasn't just morally wrong - it was biologically misguided too.



But hang on. This is the 21st century. We have in vitro fertilization (IVF), and you can analyze the genes of an IVF embryo before you decide to make it into a child. In the near future, we might be able to routinely sequence the genome of any unborn child shortly after conception.



From there, it would be a small step to allowing parents to decide not to have children with de novo mutations.



This would be, in its effects, a form of eugenics - in the sense that it would produce the effect that the old eugenicists wanted. No more 'bad' genes, or not nearly as many. Opinions will differ as to whether it's morally different. But I would have said that politically, it's a lot more likely to happen.



I can't see forced sterilization returning any time soon. But if you were expecting a baby and you knew that it was not just carrying your and your partner's DNA, but had also suffered a mutation - might you not want to avoid that?



Psychologically, it matters that it did not inherit the gene. It would be a big step to decide that your child should not inherit one of your own genes. Of course, some genes are obviously harmful, like one that raises the risk of cancer, but think about the grey areas - a gene for social anxiety, mild autistic symptoms, obesity, a personality trait.



You might well feel that carrying that gene is what makes you, you; and so it would be natural for your child to have it. You might decide that if it was good enough for you (and all your ancestors), it's good enough for your children. You might well resent the very idea that it's a 'bad' gene at all, as an attack on your own self-worth.



But none of that applies if it's a de novo mutation. Indeed, quite the opposite - all those same considerations would probably lead you to want your children to carry as close as possible to a carbon copy of your DNA, with no random changes. It was good enough for you.



My point is that I think there will be much more support for the idea of genetic screening against de novo mutations than against inherited genes. More people will want it, it will be more socially acceptable, and more widely used. I'm not saying this would be a good or a bad thing, just making a prediction. In the future, diseases and traits that are primarily caused by de novo mutations will increasingly selected against.

Antipsychotics - The New Valium?

Antipsychotics, originally designed to control the hallucinations and delusions seen in schizophrenia, have been expanding their domain in recent years.

Nowadays, they're widely used in bipolar disorder, depression, and as a new paper reveals, increasingly in anxiety disorders as well.

The authors, Comer et al, looked at the NAMCS survey, which provides yearly data on the use of medications in visits to office-based doctors across the USA.

Back in 1996, just 10% of visits in which an anxiety disorder was diagnosed ended in a prescription for an antipsychotic. By 2007 it was over 20%. No atypical is licensed for use in anxiety disorders in the USA, so all of these prescriptions are off-label.

Not all of these prescriptions will have been for anxiety. They may have been prescribed to treat psychosis, in people who also happened to be anxious. However, the increase was accounted for by the rise in non-psychotic patients, and there was a rise in the rate of people with only anxiety disorders.

The increase was driven by the newer, "atypical" antipsychotics.

Whether the modern trend for prescribing antipsychotics for anxiety is a good or a bad thing, is not for us to say. The authors discuss various concerns ranging from the side effects (obesity, diabetes and more), to the fact that there have only been a few clinical trials of these drugs in anxiety.

But what's really disturbing about these results, to me, is how fast the change happened. Between 2000 and 2004, use doubled from 10% to 20% of anxiety visits. That's an astonishingly fast change in medical practice.

Why? It wasn't because that period saw the publication of a load of large, well-designed clinical trials demonstrating that these drugs work wonders in anxiety disorders. It didn't.

But as Comer et al put it:

An increasing number of office-based psychiatrists are specializing in pharmacotherapy to the exclusion of psychotherapy. Limitations in the availability of psychosocial interventions may place heavy clinical demands on the pharmacological dimensions of mental health care for anxiety disorder patients.

In other words, antipsychotics may have become popular because they're the treatment for people who can't afford anything better.

These data show that antipsychotics were over twice as likely to be prescribed to African American patients; the poor i.e. patients with public health insurance; and children under 18.

Comer JS, Mojtabai R, & Olfson M (2011). National Trends in the Antipsychotic Treatment of Psychiatric Outpatients With Anxiety Disorders. The American journal of psychiatry PMID: 21799067

Antipsychotics - The New Valium?

Antipsychotics, originally designed to control the hallucinations and delusions seen in schizophrenia, have been expanding their domain in recent years.

Nowadays, they're widely used in bipolar disorder, depression, and as a new paper reveals, increasingly in anxiety disorders as well.

The authors, Comer et al, looked at the NAMCS survey, which provides yearly data on the use of medications in visits to office-based doctors across the USA.

Back in 1996, just 10% of visits in which an anxiety disorder was diagnosed ended in a prescription for an antipsychotic. By 2007 it was over 20%. No atypical is licensed for use in anxiety disorders in the USA, so all of these prescriptions are off-label.

Not all of these prescriptions will have been for anxiety. They may have been prescribed to treat psychosis, in people who also happened to be anxious. However, the increase was accounted for by the rise in non-psychotic patients, and there was a rise in the rate of people with only anxiety disorders.

The increase was driven by the newer, "atypical" antipsychotics.

Whether the modern trend for prescribing antipsychotics for anxiety is a good or a bad thing, is not for us to say. The authors discuss various concerns ranging from the side effects (obesity, diabetes and more), to the fact that there have only been a few clinical trials of these drugs in anxiety.

But what's really disturbing about these results, to me, is how fast the change happened. Between 2000 and 2004, use doubled from 10% to 20% of anxiety visits. That's an astonishingly fast change in medical practice.

Why? It wasn't because that period saw the publication of a load of large, well-designed clinical trials demonstrating that these drugs work wonders in anxiety disorders. It didn't.

But as Comer et al put it:

An increasing number of office-based psychiatrists are specializing in pharmacotherapy to the exclusion of psychotherapy. Limitations in the availability of psychosocial interventions may place heavy clinical demands on the pharmacological dimensions of mental health care for anxiety disorder patients.

In other words, antipsychotics may have become popular because they're the treatment for people who can't afford anything better.

These data show that antipsychotics were over twice as likely to be prescribed to African American patients; the poor i.e. patients with public health insurance; and children under 18.

Comer JS, Mojtabai R, & Olfson M (2011). National Trends in the Antipsychotic Treatment of Psychiatric Outpatients With Anxiety Disorders. The American journal of psychiatry PMID: 21799067

Gamma Waves: The Brain's Clock, Or Neural Noise?

Gamma waves are very hot at the moment.


Gamma band activity is a term for electrical oscillations recorded from the brain that have a frequency of over 25 Hz. In most brains, a peak frequency of about 40 Hz is seen. This makes gamma waves the fastest brain waves.

If you believe some recent claims, gamma waves are the answer to all the mysteries of life and the universe. They're said to underlie the symptoms of schizophrenia and autism, and they've been invoked to answer deep questions such as the binding problem and maybe conciousness itself. You can even buy a Nintendo game that promises to boost them.

A new paper from Burns et al casts doubt on all of these grand claims. Gamma-based theories of brain function all assume that gamma waves act a bit like a clock, with a consistent rhythm of about 40 Hz. Activity of about 40 Hz is indeed observed in brain recordings but is that just because the brain is randomly generating all kinds of signals, and only the 40 Hz ones "get through"?

To put it another way, imagine that you got a letter in the mail at 9 am every morning. That could be because someone is sending you one letter each day like clockwork. But it could also be that loads of people are sending you letters at random times, and your mailman only has room in his sack to deliver one each morning.

Here's the key data, recorded using electrodes implanted into the brains of two male macaque monkeys:


This shows that the monkey data closely resemble what you'd expect if gamma activity were filtered noise, and are not what you'd see if it were a more meaningful "clock". The "triangle" on the graph shows the number of bursts of a given frequency and duration.
The data also show that the phase of the gamma activity isn't consistent, which it would be if it were clocklike. In fact, the phases change entirely randomly.

So if gamma is just "filtered noise", what's the "filter"? Why 40 Hz, not 80 or 4000? Probably because this is just the maximum frequency at which neurons can fire. It takes a certain finite amount of time for cells to communicate with each other: a silicon chip can get a clock speed of many billions of hertz, but a cell just physically can't.

There's a catch, though. These monkeys were asleep, anaesthetized with the powerful opiate sufentanil. This is a good choice of drug: unlike most other sedatives and anaesthetics, you wouldn't expect an opiate to directly affect gamma oscillations. But still. If you believe that coherent gamma waves are the key to high-level concious experience, as many do, you might not expect to see much of that in the primary visual cortex in asleep animals.

However, this is clearly a very important issue, and it's not the first gamma-skeptic paper. In 2008, Yuval-Greenberg et al reported that many attempts to measure gamma activity using EEG were contaminated by electrical activity from scalp muscles. Rather than coming from the brain, the "gamma" activity reflected nothing more than tiny eye movements. The implications are still being debated.

This paper attacks the gamma hypothesis from a completely different angle, saying that even the "real" gamma in the brain, may be nothing more interesting than filtered noise.

Burns SP, Xing D, & Shapley RM (2011). Is gamma-band activity in the local field potential of v1 cortex a "clock" or filtered noise? The Journal of neuroscience : the official journal of the Society for Neuroscience, 31 (26), 9658-64 PMID: 21715631

Gamma Waves: The Brain's Clock, Or Neural Noise?

Gamma waves are very hot at the moment.


Gamma band activity is a term for electrical oscillations recorded from the brain that have a frequency of over 25 Hz. In most brains, a peak frequency of about 40 Hz is seen. This makes gamma waves the fastest brain waves.

If you believe some recent claims, gamma waves are the answer to all the mysteries of life and the universe. They're said to underlie the symptoms of schizophrenia and autism, and they've been invoked to answer deep questions such as the binding problem and maybe conciousness itself. You can even buy a Nintendo game that promises to boost them.

A new paper from Burns et al casts doubt on all of these grand claims. Gamma-based theories of brain function all assume that gamma waves act a bit like a clock, with a consistent rhythm of about 40 Hz. Activity of about 40 Hz is indeed observed in brain recordings but is that just because the brain is randomly generating all kinds of signals, and only the 40 Hz ones "get through"?

To put it another way, imagine that you got a letter in the mail at 9 am every morning. That could be because someone is sending you one letter each day like clockwork. But it could also be that loads of people are sending you letters at random times, and your mailman only has room in his sack to deliver one each morning.

Here's the key data, recorded using electrodes implanted into the brains of two male macaque monkeys:


This shows that the monkey data closely resemble what you'd expect if gamma activity were filtered noise, and are not what you'd see if it were a more meaningful "clock". The "triangle" on the graph shows the number of bursts of a given frequency and duration.
The data also show that the phase of the gamma activity isn't consistent, which it would be if it were clocklike. In fact, the phases change entirely randomly.

So if gamma is just "filtered noise", what's the "filter"? Why 40 Hz, not 80 or 4000? Probably because this is just the maximum frequency at which neurons can fire. It takes a certain finite amount of time for cells to communicate with each other: a silicon chip can get a clock speed of many billions of hertz, but a cell just physically can't.

There's a catch, though. These monkeys were asleep, anaesthetized with the powerful opiate sufentanil. This is a good choice of drug: unlike most other sedatives and anaesthetics, you wouldn't expect an opiate to directly affect gamma oscillations. But still. If you believe that coherent gamma waves are the key to high-level concious experience, as many do, you might not expect to see much of that in the primary visual cortex in asleep animals.

However, this is clearly a very important issue, and it's not the first gamma-skeptic paper. In 2008, Yuval-Greenberg et al reported that many attempts to measure gamma activity using EEG were contaminated by electrical activity from scalp muscles. Rather than coming from the brain, the "gamma" activity reflected nothing more than tiny eye movements. The implications are still being debated.

This paper attacks the gamma hypothesis from a completely different angle, saying that even the "real" gamma in the brain, may be nothing more interesting than filtered noise.

Burns SP, Xing D, & Shapley RM (2011). Is gamma-band activity in the local field potential of v1 cortex a "clock" or filtered noise? The Journal of neuroscience : the official journal of the Society for Neuroscience, 31 (26), 9658-64 PMID: 21715631

Pharmaceuticals And Violence

A French study reveals which medications are most often associated with violence and aggression: Prescribed drugs and violence.


The authors trawled the French records of drug side effects from 1985 to 2008. By law, doctors in France must report any adverse event which is either serious, or unexpected, to the authorities.

They found a total of 540 reports mentioning "violence", but only 56 of these were clear-cut incidents of physical aggression towards others. Suicide and self-harm were not included, unless they also involved violence to other people.

There were 76 suspect drugs (because some reports included one or more). Here's the Hall of Shame:

16 reports involved benzodiazepines (Valium) or similar drugs.
13 implicated dopamine-boosting drugs used to treat Parkinson's disease.
4 were caused by serotonin-based antidepressants like Prozac. Older antidepressants were not associated.

Antipsychotics and anti-epileptics were also high on the list.

There were also reports involving and the antiviral drugs interferon (3), ribavarin(2), and efavirenz (3); the stop-smoking aid varenicline (4); anti-acne drug isotretinoin (4); and the banned weight-loss drug rimonabant (2). All of these can also cause depression, and I've blogged about some of them before for that reason.

Of the perpetrators, 86% were men. Nearly half had a prior psychiatric history, but that's not surprising because many of these drugs are prescribed to people with mental illness.

In terms of the number of reports of violence relative to the total number of adverse events for each drug, Parkinson's drugs were "worst". However, this doesn't mean much, because it might just mean that these drugs are generally mild in terms of side effects.

So it's an interesting dataset, but it's impossible to come to any firm conclusions as to how common these effects really are. Cases might go unreported if they're thought to be "normal" violence; and regular violence could also get wrongly blamed on a drug - criminals get sick too.

Finally, we ought to remember while these effects are inherently attention-grabbing (and Parkinson's drugs in particular have given rise to some tabloid-friendly stories), the overall rate was tiny - less than 3 cases per year, for all prescribed drugs, in a nation of over 60 million people.

Rouve N, Bagheri H, Telmon N, Pathak A, Franchitto N, Schmitt L, Rougé D, Lapeyre-Mestre M, Montastruc JL, & the French Association of Regional PharmacoVigilance Centres (2011). Prescribed drugs and violence: a case/noncase study in the French PharmacoVigilance Database. European journal of clinical pharmacology PMID: 21655992

Pharmaceuticals And Violence

A French study reveals which medications are most often associated with violence and aggression: Prescribed drugs and violence.


The authors trawled the French records of drug side effects from 1985 to 2008. By law, doctors in France must report any adverse event which is either serious, or unexpected, to the authorities.

They found a total of 540 reports mentioning "violence", but only 56 of these were clear-cut incidents of physical aggression towards others. Suicide and self-harm were not included, unless they also involved violence to other people.

There were 76 suspect drugs (because some reports included one or more). Here's the Hall of Shame:

16 reports involved benzodiazepines (Valium) or similar drugs.
13 implicated dopamine-boosting drugs used to treat Parkinson's disease.
4 were caused by serotonin-based antidepressants like Prozac. Older antidepressants were not associated.

Antipsychotics and anti-epileptics were also high on the list.

There were also reports involving and the antiviral drugs interferon (3), ribavarin(2), and efavirenz (3); the stop-smoking aid varenicline (4); anti-acne drug isotretinoin (4); and the banned weight-loss drug rimonabant (2). All of these can also cause depression, and I've blogged about some of them before for that reason.

Of the perpetrators, 86% were men. Nearly half had a prior psychiatric history, but that's not surprising because many of these drugs are prescribed to people with mental illness.

In terms of the number of reports of violence relative to the total number of adverse events for each drug, Parkinson's drugs were "worst". However, this doesn't mean much, because it might just mean that these drugs are generally mild in terms of side effects.

So it's an interesting dataset, but it's impossible to come to any firm conclusions as to how common these effects really are. Cases might go unreported if they're thought to be "normal" violence; and regular violence could also get wrongly blamed on a drug - criminals get sick too.

Finally, we ought to remember while these effects are inherently attention-grabbing (and Parkinson's drugs in particular have given rise to some tabloid-friendly stories), the overall rate was tiny - less than 3 cases per year, for all prescribed drugs, in a nation of over 60 million people.

Rouve N, Bagheri H, Telmon N, Pathak A, Franchitto N, Schmitt L, Rougé D, Lapeyre-Mestre M, Montastruc JL, & the French Association of Regional PharmacoVigilance Centres (2011). Prescribed drugs and violence: a case/noncase study in the French PharmacoVigilance Database. European journal of clinical pharmacology PMID: 21655992

Psychiatry and Phrenology

The notorious John P. "Most Published Research Findings Are False" Ioannidis has turned his baleful statistical gaze upon the literature on brain volume abnormalities in psychiatric disorders.


Reports of regional volume differences in the brains of people with mental illness compared to healthy people have appeared in increasing numbers in recent years. Such studies have given plenty of positive results. People with depression have smaller hippocampi. The amygdala is bigger in people with autism. And so on.

Last month, Ioannidis took a comprehensive look at this literature and he argues that it suffers from a fairly serious case of "excess significance bias" - essentially, that scientists are somehow biased towards reporting differences between patients and controls, and are not telling people about the times when there wasn't a difference. This could be because of publication bias, p-value fishing or other scientific sins.

Scientists tend to call a difference between two groups significant if it has a p value of less than 0.05. This means that if there were no real difference, just some random noise, this result would be less than 5% likely to occur.

However, there's many ways you could end up with a low (i.e. good) p value. You would get a significant result, even if the true difference was very small, if you do a big enough study. Even a small difference will be detected if you study enough people. On the other hand, when the true difference is huge, you might only need a small study to get the same p value.

A power calculation is a way of specifying how likely a given study would be to detect a difference of a given size, based on the size of the study. These are usually used ahead of time to work out how big your upcoming study needs to be, assuming you can guess roughly how big the real effect you're interested in is going to be.

Ioannidis turned this on its head and asked: assuming that the true difference in the brain volume is what the average of all the published studies says it is, how many of the published studies were big enough that they ought to have succesfully detected it?

He found 41 seperate meta-analyses for different brain regions in various disorders. These were published in 8 papers - because each paper reported on multiple regions. He only looked at meta-analyses published in the past 4 years, but these analyses will themselves have included older work. This means that this paper is a kind of meta-meta-analysis. He didn't directly consider the raw brain scans at all.

The meta-analyses found many significant volume differences - but in 29 of those 41, there was an excess of significant papers. In other words, the papers were too small to have a good chance to detect the effect that they themselves found - suggesting that something funny was going on. Although, strangely, in 10/41 there were too few, and only in 2 were there the "right" number.


For what it's worth, studies on schizophrenia and on relatives-of-people-with-schizophrenia showed the least evidence of this problem, while autism was terrible, with 4 times as many significant papers as expected by chance. I'm not sure this is worth much, though. We don't know if this tells us more about schizophrenia vs autism, or more about the researchers that study them.

Anyway, this is an important study, and the inverse power calculation approach is certainly a useful one. It's not new, but it's not used as widely as it ought to be. It does make the assumption that the meta-analyses are "right" about the effect size, and then paradoxically concludes that they are biased. However, this means that the true bias is probably even bigger than this suggests (because if the analyses as biased, the true effect size is smaller than assumed, and the studies should have been even less likely to find it.)

Unfortunately, this doesn't tell us which of the studies are wrong, so it's not directly useful for people researching mental illness. It tells us that there is something wrong with scientific publishing, however. Truth be told, I suspect that a similar picture would emerge if you did this kind of thing in many other fields of science. The only real solution, in my book, would be to require the pre-registration of scientific studies. Ioannidis actually advocates this at the end of the paper.

Ioannidis JP (2011). Excess Significance Bias in the Literature on Brain Volume Abnormalities. Archives of general psychiatry PMID: 21464342

Psychiatry and Phrenology

The notorious John P. "Most Published Research Findings Are False" Ioannidis has turned his baleful statistical gaze upon the literature on brain volume abnormalities in psychiatric disorders.


Reports of regional volume differences in the brains of people with mental illness compared to healthy people have appeared in increasing numbers in recent years. Such studies have given plenty of positive results. People with depression have smaller hippocampi. The amygdala is bigger in people with autism. And so on.

Last month, Ioannidis took a comprehensive look at this literature and he argues that it suffers from a fairly serious case of "excess significance bias" - essentially, that scientists are somehow biased towards reporting differences between patients and controls, and are not telling people about the times when there wasn't a difference. This could be because of publication bias, p-value fishing or other scientific sins.

Scientists tend to call a difference between two groups significant if it has a p value of less than 0.05. This means that if there were no real difference, just some random noise, this result would be less than 5% likely to occur.

However, there's many ways you could end up with a low (i.e. good) p value. You would get a significant result, even if the true difference was very small, if you do a big enough study. Even a small difference will be detected if you study enough people. On the other hand, when the true difference is huge, you might only need a small study to get the same p value.

A power calculation is a way of specifying how likely a given study would be to detect a difference of a given size, based on the size of the study. These are usually used ahead of time to work out how big your upcoming study needs to be, assuming you can guess roughly how big the real effect you're interested in is going to be.

Ioannidis turned this on its head and asked: assuming that the true difference in the brain volume is what the average of all the published studies says it is, how many of the published studies were big enough that they ought to have succesfully detected it?

He found 41 seperate meta-analyses for different brain regions in various disorders. These were published in 8 papers - because each paper reported on multiple regions. He only looked at meta-analyses published in the past 4 years, but these analyses will themselves have included older work. This means that this paper is a kind of meta-meta-analysis. He didn't directly consider the raw brain scans at all.

The meta-analyses found many significant volume differences - but in 29 of those 41, there was an excess of significant papers. In other words, the papers were too small to have a good chance to detect the effect that they themselves found - suggesting that something funny was going on. Although, strangely, in 10/41 there were too few, and only in 2 were there the "right" number.


For what it's worth, studies on schizophrenia and on relatives-of-people-with-schizophrenia showed the least evidence of this problem, while autism was terrible, with 4 times as many significant papers as expected by chance. I'm not sure this is worth much, though. We don't know if this tells us more about schizophrenia vs autism, or more about the researchers that study them.

Anyway, this is an important study, and the inverse power calculation approach is certainly a useful one. It's not new, but it's not used as widely as it ought to be. It does make the assumption that the meta-analyses are "right" about the effect size, and then paradoxically concludes that they are biased. However, this means that the true bias is probably even bigger than this suggests (because if the analyses as biased, the true effect size is smaller than assumed, and the studies should have been even less likely to find it.)

Unfortunately, this doesn't tell us which of the studies are wrong, so it's not directly useful for people researching mental illness. It tells us that there is something wrong with scientific publishing, however. Truth be told, I suspect that a similar picture would emerge if you did this kind of thing in many other fields of science. The only real solution, in my book, would be to require the pre-registration of scientific studies. Ioannidis actually advocates this at the end of the paper.

Ioannidis JP (2011). Excess Significance Bias in the Literature on Brain Volume Abnormalities. Archives of general psychiatry PMID: 21464342

The Schizophrenic Computer

All over the world, inanimate objects are getting schizophrenia. Last week, it was a dish (full of neurons).

Before that, it was a computer program. That's according to a paper, which appeared in Biological Psychiatry last month, although it involved no biology, called Using Computational Patients to Evaluate Illness Mechanisms in Schizophrenia.

The authors set up a neural network model, called DISCERN, and trained it to "read" stories. The nuts and bolts are, we're reassured, not something that readers of Biological Psychiatry need to worry about: "Its details, many of which are not essential in understanding this study..."

Anyway, it's basically a series of connectionist models. These are computer simulations of a large number of simple units, or nodes, which can have "activations" of varying strengths, and which have "connections" to other nodes. The model "learns" by modifying the strength of these connections according to some kind of simple learning rule.

Connectionist models are a bit like brains, in other words. A bit. They're several orders of magnitude simpler than a real brain, in several different respects. Still, they can "learn" to do some quite complicated things. You can train them to recognise faces and stuff, which is not trivial.


Anyway, DISCERN is a connectionist model of language, but it's not necessary a model of how the human brain actually learns language. Because we just have no idea how the human brain does that. We don't even know if our brain acts as a connectionist network at all, above the cellular level. Some cognitive scientists think it is, but others think that those guys are talking out of an orifice connected to their mouth, but not their mouth. Not in so many words you understand.

So they set up this system and got it to learn 28 stories, each of which consisted of multiple sentences. Some of the stories were the autobiography of a doctor - "I was a doctor. I worked in New York. I liked my job. I was good doctor" - he was not a great communicator, clearly. Others were a story about gangster ("Tony was a gangster. Tony worked in Chicago..." etc.) The network had to read these stories and then recall them.

The core of the study was that they tested to see what happened when they interfered with the program by introducing certain bugs - interfering with the activations or connections of nodes in particular parts of the model. They tried 8.

They compared the computer's performance to that of 37 actual patients with schizophrenia (or the related schizoaffective disorder) who were tested on a similar task, compared to 20 healthy controls. When the human patients came to recall the stories they'd read, they tended to make more errors of particular kinds: mixing up who did what ("agent switching"), and adding stuff that wasn't in the story ("derailment").

What they found was that DISCERN made the same kinds of errors when it was given 2 particular deficits, "working memory disconnection" and "hyperlearning". The other 6 deficits didn't cause the same pattern of findings. Hyperlearning was the best match.

They comment that

A majority of three-parameter best-fit hyperlearning simulations also recurrently confused specific agents in personal stories (including the self-representation) with specific agents in crime stories (and vice versa) in a highly nonrandom fashion.

Noteworthy was the high frequency of agent-slotting exchanges between the hospital boss, Joe, and the Mafia boss, Vito, and parallel confusions between the “I” self-reference and underling Mafia members, suggesting generalization of boss/underling relationships.

Insofar as story scripts provide templates for assigning intentions to agents, a consequence of recurrent agent-slotting confusions could be assignment of intentions and roles to autobiographical characters (possibly including the self) that borrow from impersonal stories derived from culture or the media.

Confusion between agent representations in autobiographical stories and those in culturally determined narratives could account for the bizarreness of fixed, self-referential delusions, e.g., a patient insisting that her father-in-law is Saddam Hussein or that she herself is the Virgin Mary.

So if you believe it, they've just made a program that experiences schizophrenic-type paranoid delusions.

It's fair to say that this is speculative. On the other hand, it's an interesting approach, and at least it's theory-based, rather than just an attempt to use ever more powerful genetic, neuroimaging and biological techniques to find differences between a patient group and a control group.

Hoffman RE, Grasemann U, Gueorguieva R, Quinlan D, Lane D, & Miikkulainen R (2011). Using computational patients to evaluate illness mechanisms in schizophrenia. Biological psychiatry, 69 (10), 997-1005 PMID: 21397213

The Schizophrenic Computer

All over the world, inanimate objects are getting schizophrenia. Last week, it was a dish (full of neurons).

Before that, it was a computer program. That's according to a paper, which appeared in Biological Psychiatry last month, although it involved no biology, called Using Computational Patients to Evaluate Illness Mechanisms in Schizophrenia.

The authors set up a neural network model, called DISCERN, and trained it to "read" stories. The nuts and bolts are, we're reassured, not something that readers of Biological Psychiatry need to worry about: "Its details, many of which are not essential in understanding this study..."

Anyway, it's basically a series of connectionist models. These are computer simulations of a large number of simple units, or nodes, which can have "activations" of varying strengths, and which have "connections" to other nodes. The model "learns" by modifying the strength of these connections according to some kind of simple learning rule.

Connectionist models are a bit like brains, in other words. A bit. They're several orders of magnitude simpler than a real brain, in several different respects. Still, they can "learn" to do some quite complicated things. You can train them to recognise faces and stuff, which is not trivial.


Anyway, DISCERN is a connectionist model of language, but it's not necessary a model of how the human brain actually learns language. Because we just have no idea how the human brain does that. We don't even know if our brain acts as a connectionist network at all, above the cellular level. Some cognitive scientists think it is, but others think that those guys are talking out of an orifice connected to their mouth, but not their mouth. Not in so many words you understand.

So they set up this system and got it to learn 28 stories, each of which consisted of multiple sentences. Some of the stories were the autobiography of a doctor - "I was a doctor. I worked in New York. I liked my job. I was good doctor" - he was not a great communicator, clearly. Others were a story about gangster ("Tony was a gangster. Tony worked in Chicago..." etc.) The network had to read these stories and then recall them.

The core of the study was that they tested to see what happened when they interfered with the program by introducing certain bugs - interfering with the activations or connections of nodes in particular parts of the model. They tried 8.

They compared the computer's performance to that of 37 actual patients with schizophrenia (or the related schizoaffective disorder) who were tested on a similar task, compared to 20 healthy controls. When the human patients came to recall the stories they'd read, they tended to make more errors of particular kinds: mixing up who did what ("agent switching"), and adding stuff that wasn't in the story ("derailment").

What they found was that DISCERN made the same kinds of errors when it was given 2 particular deficits, "working memory disconnection" and "hyperlearning". The other 6 deficits didn't cause the same pattern of findings. Hyperlearning was the best match.

They comment that

A majority of three-parameter best-fit hyperlearning simulations also recurrently confused specific agents in personal stories (including the self-representation) with specific agents in crime stories (and vice versa) in a highly nonrandom fashion.

Noteworthy was the high frequency of agent-slotting exchanges between the hospital boss, Joe, and the Mafia boss, Vito, and parallel confusions between the “I” self-reference and underling Mafia members, suggesting generalization of boss/underling relationships.

Insofar as story scripts provide templates for assigning intentions to agents, a consequence of recurrent agent-slotting confusions could be assignment of intentions and roles to autobiographical characters (possibly including the self) that borrow from impersonal stories derived from culture or the media.

Confusion between agent representations in autobiographical stories and those in culturally determined narratives could account for the bizarreness of fixed, self-referential delusions, e.g., a patient insisting that her father-in-law is Saddam Hussein or that she herself is the Virgin Mary.

So if you believe it, they've just made a program that experiences schizophrenic-type paranoid delusions.

It's fair to say that this is speculative. On the other hand, it's an interesting approach, and at least it's theory-based, rather than just an attempt to use ever more powerful genetic, neuroimaging and biological techniques to find differences between a patient group and a control group.

Hoffman RE, Grasemann U, Gueorguieva R, Quinlan D, Lane D, & Miikkulainen R (2011). Using computational patients to evaluate illness mechanisms in schizophrenia. Biological psychiatry, 69 (10), 997-1005 PMID: 21397213

Schizophrenia In A Dish...?

...or a storm in a teacup?


According to a new paper just out in Nature from the prestigious Salk Institute, schizophrenia may be associated with differences in neural wiring which can be observed in cells grown in the lab, thus offering a window into the normally inaccessible development of the human brain.

The paper is here, and here's an open-access Nature news bit discussing it: Schizophrenia 'in a dish'. It's certainly an incredible piece of biology. They took fibroblasts, a cell found in the skin, from 4 patients with schizophrenia and 6 healthy controls.

Using genetically modified viruses, they turned these cells into human induced pluripotent stem cells (hiPSCs), which have the ability to become any other type of cell in the human body. Then, they made those hiPSCs turn into neurons by putting them in a dish with various brain-related chemicals and culturing them for three months. Not entirely unlike those brains-in-a-vat that philosophers like to talk about...

To test the connectivity of these cells, they then infected them with a modified rabies virus, after first infecting them yet another modified virus to make that work. Rabies can only spread from cell to cell via synapses between cells; they could spot the infected cells because the rabies was modified to carry a special fluorescent protein. So they could tell how many connections the neurons made.

What they found was that cultures derived from schizophrenia patients made fewer connections:


The distinct lack of red in the schizophrenia patient's dish shows that the rabies virus was less able to travel from cell to cell; the normal amount of green, yellow and blue shows that this wasn't just because it couldn't get into the cells in the first place.

OK, that's extremely cool. But then it gets a bit tricky. They tried adding five different antipsychotic drugs to the dishes for 3 weeks. Four did nothing; one, loxapine, made the cells form more connections. But it's odd that it was loxapine, a drug with unremarkable efficacy, which did this; they also tried clozapine, the only antipsychotic which is verifiably more effective than any others, and it didn't.

Loxapine is similar to (and metabolized to) amoxapine, an antidepressant; that's an issue, I would say, because we already know that antidepressants cause cells to sprout new connections. It would have been good to have used some antidepressants and some other medications as a control.

They did a lot of other work, but the data are hard to interpret. The cells "mis-expressed" about 600 genes, but we're not hold how many genes they tested. 25% of them had been previously linked to schizophrenia, but you could say that of lots of genes: is that more than would be expected by chance alone?

The patients were also unusual. Patient 1 suffered an onset of schizophrenia at age 6, and died by suicide aged 22; childhood-onset schizophrenia is extremely rare. Patients 2 and 3 were brother and sister; this means their data may not be independent, so there are (being conservative) only really 3 patients here.

Overall it's a great idea, a technical tour-de-force, and I'm sure we'll be seeing much more work along these lines on schizophrenia and other neurological and psychiatric disorders. However, as it stands, schizophrenia remains mysterious.

Brennand KJ, Simone A, Jou J, Gelboin-Burkhart C, Tran N, Sangar S, Li Y, Mu Y, Chen G, Yu D, McCarthy S, Sebat J, & Gage FH (2011). Modelling schizophrenia using human induced pluripotent stem cells. Nature PMID: 21490598

Callaway, E. (2011). Schizophrenia 'in a dish' Nature DOI: 10.1038/news.2011.232

Schizophrenia In A Dish...?

...or a storm in a teacup?


According to a new paper just out in Nature from the prestigious Salk Institute, schizophrenia may be associated with differences in neural wiring which can be observed in cells grown in the lab, thus offering a window into the normally inaccessible development of the human brain.

The paper is here, and here's an open-access Nature news bit discussing it: Schizophrenia 'in a dish'. It's certainly an incredible piece of biology. They took fibroblasts, a cell found in the skin, from 4 patients with schizophrenia and 6 healthy controls.

Using genetically modified viruses, they turned these cells into human induced pluripotent stem cells (hiPSCs), which have the ability to become any other type of cell in the human body. Then, they made those hiPSCs turn into neurons by putting them in a dish with various brain-related chemicals and culturing them for three months. Not entirely unlike those brains-in-a-vat that philosophers like to talk about...

To test the connectivity of these cells, they then infected them with a modified rabies virus, after first infecting them yet another modified virus to make that work. Rabies can only spread from cell to cell via synapses between cells; they could spot the infected cells because the rabies was modified to carry a special fluorescent protein. So they could tell how many connections the neurons made.

What they found was that cultures derived from schizophrenia patients made fewer connections:


The distinct lack of red in the schizophrenia patient's dish shows that the rabies virus was less able to travel from cell to cell; the normal amount of green, yellow and blue shows that this wasn't just because it couldn't get into the cells in the first place.

OK, that's extremely cool. But then it gets a bit tricky. They tried adding five different antipsychotic drugs to the dishes for 3 weeks. Four did nothing; one, loxapine, made the cells form more connections. But it's odd that it was loxapine, a drug with unremarkable efficacy, which did this; they also tried clozapine, the only antipsychotic which is verifiably more effective than any others, and it didn't.

Loxapine is similar to (and metabolized to) amoxapine, an antidepressant; that's an issue, I would say, because we already know that antidepressants cause cells to sprout new connections. It would have been good to have used some antidepressants and some other medications as a control.

They did a lot of other work, but the data are hard to interpret. The cells "mis-expressed" about 600 genes, but we're not hold how many genes they tested. 25% of them had been previously linked to schizophrenia, but you could say that of lots of genes: is that more than would be expected by chance alone?

The patients were also unusual. Patient 1 suffered an onset of schizophrenia at age 6, and died by suicide aged 22; childhood-onset schizophrenia is extremely rare. Patients 2 and 3 were brother and sister; this means their data may not be independent, so there are (being conservative) only really 3 patients here.

Overall it's a great idea, a technical tour-de-force, and I'm sure we'll be seeing much more work along these lines on schizophrenia and other neurological and psychiatric disorders. However, as it stands, schizophrenia remains mysterious.

Brennand KJ, Simone A, Jou J, Gelboin-Burkhart C, Tran N, Sangar S, Li Y, Mu Y, Chen G, Yu D, McCarthy S, Sebat J, & Gage FH (2011). Modelling schizophrenia using human induced pluripotent stem cells. Nature PMID: 21490598

Callaway, E. (2011). Schizophrenia 'in a dish' Nature DOI: 10.1038/news.2011.232

Neurology vs Psychiatry

Neurology and psychiatry are related fields - if for no other reason, because neurological disorders can often manifest as, and get misdiagnosed as, psychiatric ones.

But what's the borderline between neurology and psychiatry? What makes one disease "neurological" and another "mental"? Are some psychiatric disorders more "neurological" than others?

It's a rather philosophical question and you could discuss it for as long as you wanted. Rather than doing that I thought I'd have a look to see which disorders are, at the moment, considered to fall into each category.

To do this I did a quick search the archives of two journals, Neurology which the world's leading journal of... well, guess, and the American Journal of Psychiatry. I looked to see how many papers from the past 20 years had either a Title or an Abstract which referred to various different diseases. You can see the results above. Note that the total number of papers varied, obviously, and I've only plotted the proportion.

Some interesting results. Schizophrenia, which is probably considered "the most neurological" psychiatric disorder, is in fact the least talked about in Neurology. Depression is top amongst the "core" psychiatric ones.

Autism occupies a middle ground, discussed by psychiatrists at 70% and neurologists at 30%. That didn't surprise me, but what did was that ADHD is almost as neurological as autism. Mental retardation is also intermediate, though it's 30:70 in favour of neurology. Whether autism is really less neurological than mental retardation, is a good question.

Then out of the disorders with a known neuropathology, Alzheimer's disease, Huntington's disease and "dementia" (which overlaps with Alzheimer's) are a bit psychiatric while stuff like headache and epilepsy is almost 100% neurological. Why this is, is not entirely clear, since both dementia and epilepsy are caused by neurological damage, and they can both cause "psychiatric" symptoms.

I suspect the difference is that it's just much harder to treat Alzheimer's, Huntington's and dementia. With epilepsy or meningitis, neurologists have a very good chance of controlling the symptoms and few patients will be left with ongoing psychiatric problems. But with the neurodegenerative disorders, neurologists can't really do much, leaving a large pool of people for psychiatrists to study.

Someone once said that neurologists take all of the curable diseases and leave psychiatrists with the ones they can't help. These figures suggest that there may be some truth in this.