Brain representations of social thoughts predict autism

Post Reply New Topic

This <click> article is interesting.

No executive summary?

Summary:
Researchers have created brain-reading techniques to use neural representations of social thoughts to predict autism diagnoses with 97 percent accuracy. This establishes the first biologically based diagnostic tool that measures a person's thoughts to detect the disorder that affects many children and adults worldwide.

Quote: For the study, Just and his colleagues scanned the brains of 17 adults with high-functioning autism and 17 neurotypical control participants. The participants were asked to think about 16 different social interactions, such as "persuade," "adore" and "hug."

How did they identify the autism group when selecting the sample? How did they select the NT controls?
What ages were they and what were the age ranges in each group? Were the subjects male, female, both and in what proportion? On what basis were the words used chosen and was this a subjective choice of the lead researcher? Based on what? What overlap was there between the two groups?

There is a great deal of really important information absent from this report. That's always a worry in "breakthrough" claims of startling new findings. MRIs differ within individuals according to the interbalance between neurotransmitters at any time. Did they control for this? If so, how? Did they collect any baseline MRIs on the subjects before doing the experiment? If not, why not?

Consider these factors above re the methodology and add more if you are a social scientist...

The Important Issue of Bias: look at this quote:
"When asked to think about persuading, hugging or adoring, the neurotypical participants put themselves into the thoughts; they were part of the interaction. For those with autism, the thought was more like considering a dictionary definition or watching a play -- without self-involvement," Just said."

The words chosen were slanted/biased toward what normcentric populations generally enjoy doing - persuading (manipulating) hugging (hello??) and adoring (fanclubs? celebrity culture?). This wasn't a level playing field at all. None of those topics are particularly likely to turn on autistic subjects, because they are less likely to have any fundamental interest in them. What would the results have been if the words were "avoiding eye contact" "turning down overloud sounds" "refusing hugs" they would have got very different results.

MRIs may offer promise in the future, but if studies are going to be as reductionist and mechanistic as this one seems to be, I can't feel any enthusiasm about it. Other things we need to know is who funded this study, was any donor money (external to the faculty) involved?

This is a study that uses new set of machine learning methods to analyze fmri data.
To put it into context, it is rare to be able to classify even a small sample of 34 people into NT and ASD groups with 97% accuracy (1 person classified wrong). This level of clarity in distinguishing NT vs. ASD is not common.
Generally, NT and ASD participants are chosen like this: ASD participants are recruited from the local area and diagnosis confirmed using ADOS (clinical observation procedure) and/or ADIR (parent interview) and/or DSM/ICD criteria. NT participants are recruited to form age-, gender-, and/or IQ-matched group with ASD participants, often through ads on craigslist. The age ranges vary from study to study, but 18-40 is common (my studies have included 20 year olds to 70 year olds). The gender balance is more males than females, perhaps 4:1 or higher ratio is common due to many more males being diagnosed with autism, esp. hfa.
fMRI protocols are usually well-eggstablished with baseline and trial blocks and data processing following standard procedures.

I once compared my feeling of social isolation to watching a play or reading a book, where you can't interact because you are not a character in the play or book.
On the outside looking in.
It's interesting that what fMRI reveals might explain that.

This is very cool. The machine learning aspect is very interesting.Now all they have to do is develop much less expensive scanning equipment. Current trends suggest this will come too.

Hmmm. I definitely view those from the aspect of seeing someone else do them. I don't like being hugged or hugging others. I tolerate it, but not happily. I can't imagine adoring someone and I doubt anyone would adore me. Maybe with persuading, I can view myself persuading others, but knowing that actually persuading others to do anything significant such as changing behavior is very difficult.

There is a story about that from a seminar given by a cardiovascular surgeon a few years ago.

The surgeon was in his office when a construction worker was shown in for a consultation. He asked the construction worker what was wrong and the construction worker said that he was experiencing leg pain. Leg pain is often a symptom of poor circulation and so the doctor told him to start walking for exercise and stop smoking.

The construction worker said something like, "Your job is to keep me healthy and my job is to make money to pay you to keep me healthy."

The surgeon was not having a particularly good day and upon hearing that snapped at him, "In that case, we might as well amputate your leg today before it gets infected and you end up with gangrene which might kill you."

The construction worker got mad and left. For the next few days the surgeon felt kind of guilty about it and wondered if the construction worker was going to file a lawsuit against him, but no lawsuit appeared.

Months later, the construction worker came back and said, "I took your advice and started exercising and now I have no more leg pain."

According to the vascular surgeon, that's the only time that he knows of when someone actually took his advice.

17 ASD participants and 17 controls. Not impressed.

I did a bit of research on Marcel Just today, because of an intuition that I've had about him that things aren't quite right.

I note that in 2011 he was a doctoral student, though he was already making grandiose claims with a view to grabbing career building headlines, and he is an avid self-promoter, financed I see by Autism Speaks:

http://www.sciencedaily.com/releases/20 ... 122956.htm

you have been warned....

How common is this "pronoun confusion" they discuss in the article? Neither my son, nor I, nor any other Aspergers or HFA person I have met has anything like this. Is this maybe only a particular type of autism they were looking at?

How does funding from Autism Speaks invalidate research or a researcher? I don't like their message, but I am glad they are funding research. Knowledge is good.

The pronoun task-related differences in the fmri study are things that you would not know through subjective eggsperience and probably not behavioral observation either, if your brain displayed the autistic pattern of activation on the pronoun task. You would not have any idea that you had this difference from most others.

Marcel Just is a professor at Carnegie Mellon and director of the brain imaging center there. He graduated from Stanford with a Ph.D. in 1972. This seems to conflict with his supposed status from earlier post as a doctoral student in 2011.

Last edited by btbnnyr on 07 Dec 2014, 12:01 am, edited 1 time in total.

in the Science Daily piece, they say autistic children have trouble "correctly using pronouns, sometimes referring to themselves as 'you' instead of 'I.'"

Is this more true of autistic children than typical children? Is it equally true of Aspergers and HFA children?

I did not have this problem as a child. My son did not have this problem as a child.

I believe it is always prudent to entertain a healthy suspicion of research funded by groups with particular investment in a political agenda for others. We know what the agenda of AS is, and it is political. Most scientists (at least, the ones I know) are concerned about these kinds of links.

Just is a great fan of Baron-Cohen's work (which I am not, though others here are), and he presents Baron-Cohen's work as uncontested fact (it isn't uncontested, and some of us dispute that it is fact). Anyway, Just's professional mission is to provide MRI evidence to back up Baron-Cohen's major claim about theory of mind.

He skirts over difficult issues such as whether his MRIs -and his theory that people with ASDs are neruologically disordered and deficient -show cause or effect (he presents his findings as cause, though that is an untested assumption, and untested assumptions cause a lot of trouble in regard to purported and reported "research findings".

He does his experimenting on "high functioning autistics" only. Whether the generalisation of his findings to all on the spectrum is his say so or the ignorance of reporters, or the influence of AS, I don't know. The generalisation seems scientifically irresponsible to me.

No doubt there will be howls of protest over the above - hey, I'm used to them by now!

I've never seen nor heard of it.

Yes, he did. I was misled by some wrong reporting, and you are quite correct. My bad!