Comment by nicoburns | Hacker News Reader

Comment by nicoburns | Hacker News Reader

Parent Full thread

> have disabling meltdowns and total inability to manage life on other days. > That describes 90% of the population.

It most definitely does not. I used to think it did too because it's my normal. But it turns out that most people manage to get on with life just fine on a consistent basis. Perhaps you have children in mind? But this applies to adults too.

> More research means scientists working towards defining physiological underpinnings, which would lead to more scientific diagnosis criteria (e.g. genomic markers, fMRI/DRI markers, etc.). And if they are found to not have those characteristics, well, what does that mean for them? Does it mean their issues aren't real? Of course not. But it certainly doesn't make sense to call them Autistic. Their disorder should be coined a different name.

I'm definitely in agreement with you that more research is called for and should be encouraged as much as possible.

My strong suspicion is that "mild" and "severe" autism (and quite possibly related conditions like ADHD) will turn out to have to have a common physiological basis. I guess this belief (as lacking in evidence as we are at this point) is partly why I think it does make sense to classify everyone under the label "Autism".

If someone labels cats and dogs and sheep and goats and a bunch of other animals as "dogs" just because they want to be called dogs, and feed those labeled samples into a classifier for training, when given novel samples of species, the pretrained classifier will recognize ("diagnose") anything on 4 legs as a dog. I imagine the same will happen if we compile a genotype database that includes dna from everyone on the "spectrum" (i.e. the "case" dataset will be highly polluted with control samples, which will no doubt turn up null results).

Don't get me wrong, I take your point. It's just that whatever is underpinning severe autism is subtle, and was already going to be difficult to tease out. Now I feel like it's perhaps impossible.

Case/control is not the only kind of analysis you can do here. You could define a numerical scale describing someone's degree of "autistic-ness" and train a classifier on that instead of a binary categorization, for example. Obviously there are challenges in doing so, just as there are with the binary case. But if a continuous scale turns out to be a better match for the underlying phenomenon, then obviously it's worth modeling in that way.

> the "case" dataset will be highly polluted with control samples, which will no doubt turn up null results

You seem to be assuming the answer to the research here. It's true that polluting the "case" dataset with control samples will make it hard to find patterns/results. But polluting the control dataset with "case" samples causes exactly the same problem.

I would suggest that we'll need lots of studies with different groupings

> a bunch of other animals as "dogs" just because they want to be called dogs

I am not suggesting that we consider people autistic because they want to be autistic. I am suggesting that a lot of the people with "mild" autism seem to have very similar symptoms and subjective experiences to those with "severe" autism (even though it presents very differently). And that it is not at all obvious that the underlying cause is different.

The analogy I would use is that of a noise in the environment. Suppose there is a high pitched whining noise, but it's fairly quiet. That would likely be annoying, but wouldn't majorly affect you. Now suppose the it's louder, akin to a loud concert. That would probably be ok for a while or some of the time, but after a while it would start to get to you and might make you irritable or more tired. Now suppose the noise was really loud, like standing next to a jet engine loud. This would be utterly intolerable and completely debilitating. The cause of the problem in each of these cases is similar, the difference is just the degree of loudness.

And different profiles of Autism could well be similar, having a similar underlying cause despite wildly different presentations. Given that there is evidence that one of the main aspects of Autism is differences in sensory processing this analogy might even be quite literal.

There is some meta humor here about the classifier not understanding absurdism.

Yes - agree here.

Father to two Autistic children, both of whom are high functioning, attend mainstream school, excel etc.

The sheer burnout my son faces after dealing with social events is palpable - masking utterly drains him.

Reply on news.ycombinator.com