Monday, February 27, 2006

Autism Hysteria - An Unusual "Recovery" Anecdote

A controversial blog ( documents the claimed recovery of Steve Colville's daughter from autism. Steve makes the following claims:

1) The definition of autism has broadened over time to the point where it's no longer meaningful. This explains, in part, the "autism epidemic".

2) Many children labeled autistic are not autistic (and Thomas Sowell, author of The Einstein Syndrome, is right about a similar assessment).

3) Even though his daughter showed behaviors consistent with some of the DSM-IV diagnostic criteria for autism, she is not autistic, and never actually was autistic. (She was actually diagnosed PDD-NOS).

4) Autistic behavior may be caused by a lack of bonding between a child and his/her mother. This also explains, in part, the "autism epidemic".

5) There exists real autism which is truly life-long (apparently meaning that no development of speech or social skills is possible when afflicted by it).

Claim #1 above overlaps with some of my own views. (See Does Autism Exist?). I mostly object to the labeling of an increasing number of children as brain disordered. I don't mind the recognition that a good number of people are autistic - people who in the past would've been labeled weird, stupid, retarded, nutty or even gifted. When used as an identity label, it can be very useful.

Claim #2 is meaningless, as autism is a social construct. There is no objective test for autism, so it's possible to claim that a given person is autistic or not really autistic, or to say that she used to be autistic, etc. The only formal convention is the DSM-IV, and that is still pretty subjective. This is why it's possible for researchers to speculate and disagree as to whether Einstein was autistic.

The same applies to claim #3, i.e. whether his daughter was really autistic or not. This reminds me of Raun Kaufman, who is claimed to have gone from being autistic to being someone "without a trace" of autism. Given the common but unsubstantiated belief that autistic children don't develop fully, this has become quite controversial. Naturally there are those who say Raun Kaufman was not really autistic, and those who say he's now a very high-functioning autistic. The key here is that Raun Kaufman does not believe he's autistic, and that's really the assessment that matters. I should note that some evidence presented in favor of the view that Raun Kaufman is recovered is based on stereotypes. For example, it is often mentioned that he has a University degree, as if autistics could never have a University degree. (I have a Master's degree myself, BTW).

A diminishing number of researchers still believe that autistics never develop speech, and those who do develop speech are described as 'not truly autistic'. This clearly makes no sense. Some autistics who develop speech still have substantial deficits. At the very least psychiatrists would have to come up with a new label for those who truly never develop speech, but the big problem would be that it is impossible to predict which young child falls into which category. Again, medical tests for autism or kinds of autism don't exist.

Refrigerator Mother

Claim #4 could be seen as a variation on the refrigerator mother theory. This has been the most controversial claim of all. For example, Xian Diaz said:

I'm curious to hear how your wife feels about this whole situation, and how she felt when she was blamed for her daughter's affliction.

It is clear mothers feel offended when such theories are proposed. My wife gets pretty rattled. The recognition of my autism and its strong genetic component are seen as a vindication of her parenting skills. But frankly, being offended by a theory has no bearing on the theory's scientific merit. Furthermore, it would be foolish to claim that parenting is irrelevant with autistic children, or any children for that matter.

I do believe Political Correctness was one of the reasons why the Refrigerator Mother theory was no longer studied by researchers. (There's very little research on it in the last couple of decades). When autism began to be recognized as a highly heritable condition, Refrigerator Mother began to be phased out. But if you think about it, today many researchers continue to look for other environmental triggers. Except, this time, they aren't looking for environmental triggers that have to do with childhood experiences or psychological influence. They are only looking for environmental accidents, i.e. physical causes. There is no good reason for this focus, except fashionable belief systems, and the unpopularity of certain views.

Autism is highly heritable, but it is not 100% heritable. Note, however, that nothing that is evaluated subjectively is ever 100% heritable. Personality and intelligence are probably even less heritable than autism. Homosexuality is less heritable than autism. There must be something that makes one identical twin have one personality type and the other twin have a different personality type. The cause of this is not likely going to be level of mercury in the brain. It's probably something a lot more subtle that has to do with the unique experiences of each twin. (Or perhaps it's something that occurs in the womb - we just don't know).

There are a couple case reports that indicated a higher proportion of autistic behavior in children who had experienced profound institutional privation at an orphanage. This was not nutritional privation, BTW. This seems to suggest, in principle, that some psychological factors could amplify the perceived severity of someone's autism.

The Anecdote

At first I was inclined to dismiss Steve's anecdote as yet another anecdote of sudden development that coincided with the latest treatment attempted by desperate parents who are always trying something new. After reading his account, however, it was clear that this was unusual.

Then suddenly [the headmaster] leaned over, picked up my daughter and held her tightly to his chest... And screamed for the best part of an hour as the headmaster walked about... Then when she'd calmed down a little he put her down, placed a chair in her hand and pushed her into a semi-circle of children having a music lesson.
My daughter sat down and looked about for a few seconds as though waking from a sleep. Then her eyes focused on the teacher. And then my daughter (and remember at this point she was making no eye contact and was almost completely non-verbal) simply began to copy the teacher's hand actions and mouth the words to the song...
And then three months later she began to talk... The best moment came when, during one of the hugging sessions, she tried to escape and said: "Daddy, you shouldn't be doing things now that you should have done when I was a baby!"

This is just a summary. The whole description is pretty lengthy. But essentially, it seems far-fetched to say such an improvement is coincidental. The part about beginning to speak in sentences all of the sudden is not surprising. Speech develops that way - it's not a gradual process but an explosion. But the sudden improvement seen right at the beginning of this therapy appears notable.

Some people dismiss this as a kind of ABA. This is nonsense. ABA involves intensive training up to 40 hours per week for several years. It's not just hugging for 3 months.

Some key information to understand what went on here is found at the beginning of the account:

She began to say single words at about 12 months, and by 18 months had begun to make sentences. At 18 months she had a development test and seemed fine.

It is clear this little girl understood spoken language and could also speak in sentences before 18 months. Then suddenly she stopped speaking. This is what most parents refer to as a "regression". It appears that true regression can occur due to conditions such as Landau-Kleffner Syndrome. In other cases, it's not clear if the child forgets some skills or simply stops using certain skills.

Some additional information seems relevant:

Then when she was two years old we moved house... At about this time our second child was born... my wife went back to her own country for a while, during which time Denise's behavior became intolerable.

So it seems she experienced considerable stress around the time of her "regressions". It is possible she simply withdrew and had constant anxiety, not so much that she forgot to speak in sentences.

Social withdrawal due to depression and stress has been reported. See Autistic devices in small children in mourning (2000). It should be noted that not all children are affected this way. Clearly, there needs to be some kind of genetic predisposition. There is some evidence that autistic children react differently to stress.

Incidentally, my mother says I was the happiest child before my brother (who is 4 years younger) was born. I believe I experienced a sort of regression at the time. I have seen how stress has affected my son, particularly when we first enrolled him in a preschool, when I've been out of town and when my wife has been out of town. But my son has never really spoken in sentences. A regression per se has not been obvious with him. Additionally, my son is quite loving and likes hugging. These are some of the reasons I don't believe this approach or outcome could apply to my son, even though the profile of hyperactivity and so on appear to be very similar.

In trying to understand the sudden improvement observed at the beginning of the therapy, I'm reminded of some anecdotes retold by Thomas Sowell (who Steve has read as well). He tells, for example, of a family sitting in a living room, very worried, discussing how to deal with their late talking child. Realizing this, the child approaches his relatives and says "I'm sorry". He also tells of children who don't speak in public but who are caught practicing speech when they are alone.

Steve's daughter did know how to speak and so on. When the headmaster forcefully hugged her for half an hour, I'd hypothesize she realized her parents were very worried for her, that this was going to become a major problem for her, and much like the child who said "I'm sorry", started to behave in a way more in line with the expectations of the parents. The hugging, in fact, might have worked like an aversive. I could be wrong, but without evidence for this or the opposing hypothesis, who's to say which is right.

Holding Therapy

It's true what they say: There's nothing new under the sun. After a bit of research, it's clear Steve's miracle cure is exactly the same as Holding Therapy. This is an old therapy invented by Dr. Martha Welch in the 1970s. She hypothesized that autism was caused by the lack of bonding between mother and child. The therapy involves forcefully hugging a child despite any distress experienced by the child.

As with ABA, holding therapy has been described by people on the spectrum as a form of abuse [ref].

Others are not so forceful. This is what Temple Grandin has had to say about Holding Therapy:

During my travels to many autism conferences several parents have reported to me that holding therapy was beneficial. It is not the "cure" that some of its proponents tout, but it had a beneficial effect on some children. In my opinion, the benefits of holding therapy could be obtained by less stressful methods. I cringed when I watched the BBC show The Visit and I am glad I did not have to endure forced holding. Fisher (1989) describes a gentler approach to holding that worked with her daughter.

I should note that Temple Grandin is one of the few people on the spectrum who still sees autism as a disorder that needs to be defeated through various means, including drugs.

Proponents of Holding Therapy contend that it forges a bond between mother and child. But there is no scientific evidence whatsoever to support this assertion. As I noted, my own hypothesis is that it works as a kind of aversive, i.e. the child is forced to act more in accordance with expectations in order to avoid further hugging. This will probably appear to work well with children such as Steve's daughter, who have the skills necessary to pretend to be NT (which, BTW, requires considerable energy expenditure). With children who are simply incapable of emulating expected behavior, I'm thinking forceful hugging could easily lead to PTSD.

I have learned that Donna Williams has criticized Holding/Hug Therapy in her book Autism - An Inside-Out Approach on much the same grounds.

Don't get me wrong. I do think it is advisable to hug children as much as possible. I not only hug my son constantly. I tell him how special and unique he is. I just wouldn't recommend hugging children against their will, and particularly not until they are screaming with pain.

Further Issues

I have some problems with what Steve has to say. For example, at one point he says something like "she was bad, bad bad... now she's bad, but in a good way if you know what I mean." I'm not sure if he realizes this insults all autistics, including my son and myself. He's basically saying "autistics are bad, bad, bad, and in a bad way." He has also implied that autistics don't get excited and so on - perpetuating estereotypes that should not exist anymore. Steve doesn't like the word "neurotypical" and prefers to speak plainly by using the word "normal". I wonder if he thinks his daughter's development and reaction to psychological stress was "abnormal", and whether he thinks it will never be "abnormal" again. I'd obviously prefer if Steve did not imply that my son or myself are "abnormal".

Friday, February 24, 2006

Does Autism Exist?

Certainly, autism exists, otherwise why write about it and why label myself autistic or Aspie? But the question is: does it actually exist as a distinct natural entity or only as a cultural/social/psychiatric construct?

In a paper titled ADHD is best understood as a cultural construct (2004), Sami Timimi explains that there's no objective test that may be used to diagnose ADHD and that reported prevalence rates of the disorder vary widely. This theory (remarkable as you will see) can actually be applied to any spectrum disorder in the field of psychiatry and it fits autism quite nicely.

What the theory points out is that the characteristic behaviors of a spectrum disorder occur to a lesser degree in the general population. So the question arises as to how you distinguish someone who is disordered from someone who isn't. More specifically, how do you distinguish those who are "only a little autistic" vs. those who are "not quite autistic"? Psychiatrists have rules of thumb that guide them in making the distinction; usually this simply means that the affected person or a relative is looking for help. But evaluations by affected persons, relatives and psychiatrists are all subjective. There is no medical test that will determine that someone is autistic, nor is there one that will rule out a psychiatric diagnosis of autism. Furthermore, different psychiatrists will invariably have different opinions as to where the boundary lies.

But surely some people who are severely autistic are truly afflicted by something objective, right? Social construct theory is not concerned with severity. The reasoning continues to apply regardless of where the boundary between normal and abnormal is situated.

Many conditions which are not social constructs themselves are often diagnosed as autism. Such is the case of Fragile-X Syndrome and Rett Syndrome, both objectively diagnosable conditions. And we should also expect that many other objective conditions yet to be identified may also look a lot like autism. In this sense, autism should be seen as both a blanket term and a social construct.

A possible objection is to point out that autism could simply be a blanket term for many objective conditions, and not a social construct. Some would even propose those objective conditions are: Asperger syndrome, classic autism, and so on. The problem with this reasoning is that since the characteristic symptoms of autism blend into behaviors of the general population, the boundary between normal and abnormal is always inherently arbitrary, and the same may apply to the boundary between arbitrarily defined sub-conditions. (The boundary between autism and Asperger's is notoriously problematic as it is [ref]).

Some readers might be thinking that a genetic test for autism will render this theory useless. But this is actually where the predictive beauty of this theory can be appreciated. Scientists are currently expecting to find 3 or 4 or 15 genes that will define a person as autistic. The more genes you have, the lower in the spectrum you are. What social construct theory predicts is that no set of genes that accurately mirror a diagnosis will be found. Genes will only increase the likelihood of autism, but not cause it. And an absence of genes will not rule out a diagnosis. Genome scans to date are quite consistent with this prediction (in addition to being consistent with a blanket term view).

Based on current knowledge about the heritability of autism, it could be argued that the best possible genetic test will only be able to make a prediction of classic autism with 60% accuracy and a very broad spectrum with 90% accuracy – and this would be a extremely sophisticated and complex test.


Social construct theory is clearly not simply an ideology or a view. It is a scientific theory that could be falsified in principle (with an objective medical test), is useful and makes predictions.

A common misconception is that finding differences between autistics and non-autistics falsifies this theory. For example, pointing out the differences in grey and white matter volumes which have been documented might be thought by some readers to undermine this theory. But this is not sufficient, because (1) The differences observed are 'in average', i.e. the findings cannot be generalized to all autistics, and (2) Such differences would be expected in any behavioral spectrum regardless of where the comparison boundary is situated. Comparing any two arbitrarily selected groups would probably yield neurological differences in average. For example, you could compare classic autistics vs. non-classic autistics, or Aspies vs. those with ADHD, or men vs. women – finding neurological differences is not surprising but generalizing them is basically a stereotype.


At this point some readers might point out that the theory that autism is a social construct is, in turn, a useless intellectual artifact.

The usefulness of this theory lies in the kinds of problems it explains and in that it provides a new framework to think about issues related to autism. For example:

1) Social construct theory explains the "autism epidemic". Any subjective boundary can shift through time. This should also be true of differences in regional prevalence.

2) Social construct theory explains the emergence of Asperger's syndrome as a form of autism.

3) This theory suggests that the psychiatry profession is, perhaps unintentionally, gradually broadening the conception of brain disorder. There appear to be no controls on how much longer this trend will be allowed to continue.

4) Based on this theory it can be inferred that many individuals labeled as having a brain disorder (and labeling perhaps has consequences of its own) in fact do not.

Thursday, February 23, 2006

Problems with the Mercury Hypothesis

I'll start this blog with some criticism of the hypothesis that mercury poisoning is responsible for autism across the board. Some parents who propose this hypothesis appear to be quite belligerent and eager to demonstrate their hate for autism and autistics (see Anyone who disagrees is labeled "mercury poisoned" as well, and presumed to be inferior (just as I can only assume they see their own children). Disagreements are thus explained away as a lack of reasoning skills. As appalling as this is, it is not the basis of my criticism.

I could analyze in detail the studies that demonstrate a link and those that don't, comparing their methodology. But this has been done [ref][ref][ref].

What I intend to do is point out a number of well known facts about autism and mercury toxicity which are totally inconsistent with the hypothesis. At the very least, there are no mercury toxicity-based theories that attempt to explain them.

  1. There's no mechanism by which mercury could cause savant skills to emerge.
  2. The most severe symptoms of mercury poisoning include such things as kidney damage, burning mouth, and perspiration. These are not symptoms of autism, even in a minority of cases.
  3. Mercury poisoning cannot account for neuroanatomical differences observed in early childhood, such as white and grey matter volume differences, increased head circumference, and differences in neuron size and density.
  4. Mercury poisoning cannot account for increased false memory discrimination, information processing advantages, and other cognitive differences that have been documented.
  5. It has been hypothesized that autistic children have a genetic liability in their body's mercury metabolism, having to do with the APO-E4 protein. This, however, would mean that a good portion of the siblings and parents of autistic children who had been themselves vaccinated with thimerosal should also be autistic. This is not the case. In fact, twin and sibling studies suggest that a single genetic variation is not sufficient to cause autism.
  6. Additionally, several different alleles have been linked to autism, and these are prevalent in the general population. None of the alleles that have come up in genome-wide scans are related to mercury disposal in any way.
  7. Objectively diagnosable conditions which look like autism have a purely genetic etiology. These include Fragile-X, Tuberous Sclerosis, and Rett Syndrome.
  8. There are no reports of adult autistics running to get chelated, and some of them are OK with their autism, which would seem to be a strange way to experience being poisoned.

This, of course, does not mean that the symptoms of autism couldn't be amplified by a small amount of mercury poisoning, or that some diagnosed cases may be caused by mercury poisoning. But it is unethical and scientifically ludicrous to say that autism and mercury poisoning are the same thing. It is also silly to suggest that chelation is an autism treatment, as it does nothing other than treat heavy metal toxicity. And it is just wrong to diagnose and label a child or an adult as mercury poisoned without any medical tests to back up the assertion.