How Far Off Is Mind-Reading?

PopSci.com has a somewhat enthusiastic article about the possibility of using fMRI to "uncover your private thoughts"- Mind-Reading Tech May Not Be Far Off.

Neuroscientists are already able to read some basic thoughts, like whether an individual test subject is looking at a picture of a cat or an image with a specific left or right orientation. They can even read pictures that you're simply imagining in your mind's eye. Even leaders in the field are shocked by how far we've come in our ability to peer into people's minds. Will brain scans of the future be able to tell if a person is lying or telling the truth? ... While we aren't there yet, these possibilities have dramatic social, legal and ethical implications.

But what do we mean by "mind-reading"? I guess what most people mean by the term is being able to tell what someone is thinking, being able to "hear" their private thoughts. A stereotypical fictional "telepath" can get inside their targets minds and tell exactly what's going through them.

Sadly, what most fMRI "mind-reading" experiments have done is rather less impressive. they've shown that it's possible to tell whether someone is thinking about one thing as opposed to a second thing. But only if you already know what both things are, and only if you have already "read" the pattern of neural activity that corresponds to each one.

So, you could scan someone while they are thinking about, say, cats, and then again while they are thinking about dogs. From that, you could work out whether they are thinking about cats or dogs at any given point in time (here's how). If they were thinking about anything else, you'd have no idea what it was, or worse, you'd think it was either a cat or a dog. A lion, for example, would probably activate many of the same pathways that a cat does.

The great majority of "mind-reading" studies are like this. It's still pretty cool, but it's no telepathy. Is there any prospect of true "mind-reading"? In other words, could you read a mental state without knowing what you were looking for in advance?

Maybe. The parts of the brain concerned with visual processing happen to be arranged in a relatively straightforward way,which means that there are predictable relationships between visual stimuli and the areas of the brain that are activated when looking at them. Reports have claimed that it's possible to infer which picture someone is looking at out of a large set (1) and even to reconstruct the image that someone is looking at based purely on the visual cortex activity (2,3). For a good explanation of the last paper, which attracted a lot of attention, see Neurophilosophy.

Such studies come closer to true "mind-reading", but thus far the technique only works with vision. Even assuming that the same areas of the brain light up when you're thinking about something (visual imagery) compared to when you're looking at it (visual perception), the best this method could achieve would be to tell what picture was in someone's head at a given time. In ten years it might be possible to put someone in a scanner and tell, straight off the bat, that they were picturing a small white dog. But if you wanted to know what they were thinking about that dog, you'd be out of luck.

To truly read someone's mind you would need to understand how every brain state relates to every mental state. In other words, you would need to know how the brain allows us to think. At the moment, we really have no ide about that, so true mind-reading remains over the horizon.

Edit: I must be telepathic because I just saw that Mind Hacks covered a new study about mind reading a few hours ago: I know where you are secretly attending! Yet again, it involves the visual cortex.

Are 1 in 64 Kids Autistic?

Quite possibly, yes. In the last post I discussed the interesting background to a new paper about the prevalence of autism in British children, Prevalence of autism-spectrum conditions: UK school-based population study. Here's some more about the study itself.

The authors, Simon Baron-Cohen et al from the University of Cambridge, set out to assess the prevalence of “autistic spectrum conditions” in the county of Cambridgeshire, England, by sampling all of the school children aged 5 to 9 years during 2003-2004.

The most recent major study examining the prevalence of autistic spectrum conditions in Britain was Baird et al (2006), who reported a prevalence of about 1 in 86. But Baron-Cohen et al point out that this may have been too low, since Baird only looked for autism in children who were already on the government's “Special Educational Needs (SEN)” register of children with difficulties in school. If there were autistic children who were doing OK in school, or at least well enough to get by without attracting concern, they’d have been missed.

So, Baron-Cohen’s team first wrote to every school in Cambridgeshire (162 of them) and asked them to report how many of their children had been diagnosed with an autism-spectrum condition. 79 schools replied and reported 83 children diagnosed out of 8824 total, or 1 in 106 children – pretty close to Baird et al's in 2006.

But those were only the kids who had already got a diagnosis. In order to try to find undiagnosed cases, they then sent questionnaires to the parents of 11,635 children. The questionnaires included a screening form developed in Cambridge called the CAST, which asks parents about various aspects of their child’s behaviour (“Does s/he come up to you spontaneously for a chat?” “Does s/he like to do things over and over again, in the same way all the time?” Etc.)

The authors invited all of the kids who scored highly on the CAST to a face-to-face assessment to confirm whether they really had the condition. The end result was that out of 3373 kids whose questionnaires were returned, 11 were judged (in the opinion of the research team) to have an autism-spectrum condition which had never been previously diagnosed.

What does this mean? Well, good question. All it strictly means is that 11 out of 3373 children had undiagnosed autism. However, because not all of the children who scored highly on the CAST agreed to be interviewed, the authors estimate that the true figure was probably more like 22. That compares to 33 out of those 3373 whose parents reported already diagnosed autism. (Actually it was 41 reported, but only an estimated 33 were declared “confirmed”. See page 503 if you’re sceptical of this fudge, but it seems kosher to me.)

The bottom line: for every 3 children with a diagnosis, 2 others went undiagnosed. Since about 1 in 100 children have diagnosed autism, that makes 1 in 64 children with autistic spectrum conditions in total.

But this relies on some assumptions. In particular, this only works if you assume that the parents of autistic children were no more or less likely to complete the CAST questionnaire, and no more or less likely to agree to a face-to-face interview, than parents of the non-autistic kids.

However, it could well be that the parents of autistic children were already concerned that there was “something wrong” with their child and wanted to get a professional opinion, so they were keen to take part – that would mean that this study overestimated the rate of undiagnosed autism. On the other hand, it could equally well be that the autistic children were less likely to get included in the study. Maybe they just didn't want to go along to the interview with a stranger. In which case, the rate of autism would be underestimated.

Because only 29% of parents did the questionnaire and even then only about 60% of the children who scored high came up for the face-to-face, the potential for bias is great. Unfortunately, there is no way of knowing which way the bias operates. The authors acknowledge these concerns and admit that their estimates are not exact.

But this is still an important study. Even if you assume that the data were extremely biased towards finding autistic children there were still 11 cases of undiagnosed autism out of about 11,000 kids aged 5-9, compared to 83 diagnosed, which means that at an absolute minimum 1 in 9 children with autism of that age are undiagnosed. And the true figure is likely to be a lot higher, maybe 2 in 5 as the paper claims.

On this blog I've often been skeptical of claims that mental illness is extremely common. But I can easily believe that 1 in 64 children has a significant autism spectrum-condition, and that some cases go undiagnosed in primary school. While we still don't know the exact numbers, and these will always be somewhat arbitrary since they depend upon the chosen diagnostic, about 1 in 50 sounds about right. Certainly, the idea that autism is an extremely rare condition affecting more like 1 in 2000, as was believed twenty years ago, is out of date.

Baron-Cohen, S., Scott, F., Allison, C., Williams, J., Bolton, P., Matthews, F., & Brayne, C. (2009). Prevalence of autism-spectrum conditions: UK school-based population study The British Journal of Psychiatry, 194 (6), 500-509 DOI: 10.1192/bjp.bp.108.059345

Autism, the Media, and "1 in 58" - the story continues

It was about this time two years ago that my faith in the British media died. It had never been in the best of health, but up until then I believed that the (non-tabloid) newspapers were written by professionals trying to find out and communicate the truth as best they could.

Journalists might be wrong, I thought, but they did their best to ensure that they weren't. And they might have a bad habit of focussing on sensational stories that "sold papers", but such stories would at least be accurate. A career in journalism was something that strongly appealed to me.

What happened on July 8th 2007 was that The Observer printed a front-page article so demented that I’ve never been able to take that newspaper at face value since. And if you can't trust The Observer, which sat on my family's breakfast table every Sunday since before I can remember, you can't trust any of them (at least, I can’t.)

The article was about autism, and it claimed to be a report on a new research study carried out at Cambridge by the famous Professor Simon-Baron-Cohen. The upshot was that Baron-Cohen’s team had found the rate of autism to be 1 in 58 children, much higher than the previous study from a few years earlier, which found a rate of about 1 in 86. Furthermore, The Observer said, two of the team, “world experts” in autism, thought that this “dramatic rise” (i.e. a rise of 50% in a few years!) might be something to do with the much-maligned MMR vaccine.

The original article no longer exists on the Observer’s website, but the WayBack Machine has it.

It was pulled after Ben Goldacre, amongst many others (including Baron-Cohen and even the Government), criticized the piece vociferously and it was, eventually, replaced by an unsatisfactory “clarification”. Goldacre’s take on things was chronicled in a astonishing series of articles (1, 2, 3, 4, et al.) which revealed some horrific journalism, including, amongst much else, attributing opinions to people who didn’t hold them and failing to reveal conflicts of interest. If for some reason you’re not already a Goldacre fan, read those posts, and you’ll see what all of the fuss is about. But you need to know one other thing – he is literally the only journalist in the country who does what he does.

The most damning criticism at the time, however, was that the research in question could not possibly have found an increase in autism rates, let alone a “big surge”, a “dramatic rise”, or an “upward trend”. The new research, quite deliberately, used more extensive assessment criteria than previous studies. It was specifically designed to find the highest possible estimate. So only a fool would try to compare it to other sets of data and see this as evidence of rising rates. Also, even if there had been a real increase, it couldn’t have been because of MMR, because the kids in both studies will have had the MMR vaccine. The oldest kids in the Baird et al study of 9-10 year olds were born in 1991, a few years after MMR was introduced, while the youngest in the Baron-Cohen study were born in 2000, when MMR uptake was, if anything, lower, thanks to the MMR-autism scare.

Now, two years after The Observer’s “scoop”, the research is finally out in the British Journal of Psychiatry: Prevalence of autism-spectrum conditions: UK school-based population study. I’ll be examining this paper in detail in the next post, but here’s what you need to know if you remember the story from 2007:

The paper doesn’t mention the Observer affair but it’s obvious that the authors had the article in mind while they were writing up their results. They repeatedly emphasise that their prevalence estimate cannot be compared to previous ones. They make it very clear (to the point of seeming a bit stilted) that they believe that the apparent rise in prevalence of autism over time is due to better detection and diagnosis, rather than a real “epidemic”. And they do not mention MMR – not that they had any reason to, of course. The highest estimate they arrive at, which they say is probably somewhat too high but close enough, is 1 in 64 children. 1 in 58 doesn’t appear in the paper.

What’s most interesting about the paper is who wrote it. The author list includes Simon Baron-Cohen (obviously) and Fiona Scott. The Observer named Scott as an MMR-autism theorist, something she strongly denied, in 2007. However, fascinatingly, Carol Stott is not an author, although she receives a massive acknowledgement at the end of the paper – “Carol Stott was a member of the research team throughout the main phase of the study and contributed to the coordination and running of the study, data management and data collection. She also made valued contributions to team discussions.” On the basis of this, she clearly had a right to be listed as an author – but wasn’t. We can only assume that she chose not to be, because if the other authors had left her off the list without her permission they would have been guilty of a serious breach of trust.

Carol Stott, you’ll remember, played a role in the Observer autism story. Unlike Fiona Scott, she does (or at least did in 2007) believe that MMR is linked to autism, and she had very close links with Andrew Wakefield as well as displaying a, er, penchant for the scatological in some bizarre emails to journalist Brian Deer. (Goldacre does say he “genuinely warmed to her, and she regrets that many people have fallen into entrenched positions on MMR on both sides” though, so she’s not all bad!)

So, questions remain. How did the preliminary research get leaked in 2007? What happened to make 1 in 58 become 1 in 64? What’s the deal with Stott? We don’t know. But we do know a bit more about autism thanks to this paper, and in the next post I’ll be discussing that.

[BPSDB]

On Anonyminity

You'll probably have heard about the "outing" of formerly anonymous blogger Night Jack, a British policeman. Night Jack went to court in an attempt to prevent The Times newspaper from publishing his identity, but the judge ruled against him. His award-winning blog, http://nightjack.wordpress.com/, has been taken down; I don't know if this is going to be permanent.

Lawyer Jack of Kent has a typically lucid and detailed legal commentary on the case, but as a fellow anonymous blogger, I believe that this is an issue which goes beyond British law.

I write anonymously because it allows me to write things that I otherwise couldn't. I could write under my own name, as many excellent bloggers do. However, the content of Neuroskeptic would not be the same.

Broadly speaking, my "neutral" coverage of science news would remain (like this), and my criticism of journalists probably also would (like this). However, I don't feel that my more "critical" writing about science - like this - would be possible without anonymity.

I'm an academic at a junior stage of my career. Some of the targets of my (implicit and explicit) criticism are people and organizations who might well play a part in that career. Quite simply, I don't want to go on record criticizing them, for obvious reasons of self-interest.

Perhaps this just makes me a bit of a coward, but I prefer to think about it in a more philosophical light. Often when we say or write something, two things happen in parallel. We are doing something in the social world, and we are asserting a proposition.

If I were to say to someone "Your wife is having an affair", I would be doing something momentous, something that might well be very painful and damaging. This is why we don't say things like that lightly - even if they are true. We value tact. Yet at the same time, my statement is true (or false), just like any other statement of fact, and it remains true (or false) whether or not I say it.

As a society, we recognize that it's sometimes desirable to allow people to assert things without having to worry about the consequences of their words as a social act. This is why we have anonymous feedback forms, anonymous comment boxes, anonymous witnesses (in some cases). It's also why we don't regard an election as fair and open if it doesn't have a secret ballot.

And in science, we have anonymous peer review. In order for a paper to be published, it must first be subjected to the criticism of one or more experts on the topic in question, writing anonymously. The anonymonity is fundamental because it allows them to criticize the research as harshly as necessary without having to worry about the consequences. Few people want to go on record as criticizing someone else's work, especially as most scientific fields are sufficiently narrow that peer reviewers personally know the authors of most of the papers they have to critique. Yet someone has to do the dirty work of criticism.

So anonymous peer review is valued in science as a way of facilitating objectivity, something otherwise in short supply, because scientists are people with careers and reputations to uphold. At the risk of giving too much dignity to a mere blog, I see Neuroskeptic as a continuation of this review process once papers have been published. Scientific debate shouldn't be hampered by concerns about careers and reputations, although scientists being only human, it is - anonymous comment is one way of getting closer to the ideal of pure objectivity.

All of that said, anonyminity is not all roses. It's open to abuse. Someone could persue a vendetta against a rival by making apparantly objective, anonymous criticisms that were in fact motivated by nothing more than self-interest. This occasionally happens during the process of peer review - a reviewer might trash a manuscript just because they just don't like the results, or because they are planning to publish the same results and they want to do so first. And an anonymous blogger could exploit their status for similar reasons. I would like to think that I have never personally criticized anyone who is acting in good faith, which includes the vast majority of academics. I try to stick to criticising ideas, not people. But of course, I would say that.

So anonymous writing has to be seen for what it is - something that has the potential to be more objective than on-the-record statements, but with no guarantee that it in fact is. Caveat lector, as always.

[BPSDB]

Aripiprazole, Dopamine, and Well-Being - Science or Selling Point?

Suppose you were a drug company, and you've invented a new drug. It's OK, but it's no better than the competition. How do you convince people to buy it?

You need a selling point - something that sets your product apart. Fortunately, with drugs, you have plenty of options. You could look into the pharmacology - the chemistry of how your drug works in the body - and find something unique there. Then, all you need to do is to spin a nice story to explain how the pharmacological properties of your drug make it brilliant.On an entirely unrelated note, aripiprazole (Abilify) is an antipsychotic marketed in the US by Bristol Meyers-Squibb. A Cochrane meta-analysis finds that it's about as good as any other antipsychotic in terms of efficacy and side effects. As good, but no better. However, uniquely, aripiprazole is a D2 receptor partial agonist. Other antipsychotics work by blocking D2 receptors in the brain, switching them off (full antagonism). Aripiprazole also blocks D2 receptors, but it activates them slightly in the process (partial agonism).

Is that a good thing? A paper just published says yes - The relationship between subjective well-being and dopamine D2 receptors in patients treated with a dopamine partial agonist and full antagonist antipsychotics. The research in question was funded, by the way, by Bristol Meyers-Squibb. Let's see if it holds up.

The authors got 22 patients with schizophrenia who were taking an established antipsychotic, either olanzapine or risperidone. These are both D2 antagonists. Incidentally, neither of them is made by Bristol Meyers-Squibb. 11 of the patients were switched to aripiprazole, while 11 stayed on their original drug. There was no blinding, and no randomization. (The dose of aripiprazole was randomized, although still unblinded, but the assignment to aripiprazole itself wasn't).
Lo and behold, the patients who switched reported improved "well-being". Because there was no randomization and no blinding, and because the outcome was entirely subjective, this could be entirely explained as a placebo effect (or an experimental demand effect.) Especially when you consider that the patients were most likely convinced to take part in the study by being told that aripiprazole would make them feel better than their original drug.

That's not all, though. They also did some brain scanning, using PET to measure D2 receptor occupancy. On average aripiprazole blocked more D2 receptors than the other antipsychotics, which is what you'd expect, as it has a very high affinity for that receptor. But it's a partial agonist, remember - it binds to D2 receptors without switching them "off" entirely.

The paper suggests that this is a good thing because it doesn't make people feel horrible, which is what normally happens when you block almost all of someone's D2 receptors (they're rather important). By switching on the receptors as well as blocking them, it makes you feel OK.
Nice story, and scientifically it's not unreasonable. And as you can see on this plot, in the non-aripiprazole patients (triangles), D2 occupancy in the ventral striatum was negatively correlated with well-being, but in the aripiprazole patients (circles), it wasn't.

Great - except that the range of D2 occupancies in the aripiprazole group is so narrow that no correlation would be apparent even if there was one. The occupancies in the aripiprazole group are all extremely high, 80-95%. There's just no room for a correlation to appear. (Think about it this way - in children, age is strongly correlated with height, but if you only looked at a bunch of 7 year olds, you wouldn't know that.) This is high-school statistics.

This scatter-plot is in fact exactly what you'd expect assuming that a) aripiprazole strongly blocks D2 receptors, and makes people feel awful, just like any other strong D2 blocker and b) the placebo effect made some of the aripiprazole group feel (or at least say that they feel) a bit better than they otherwise would.

You'll note also that the aripiprazole group reported feeling no better than the other antipsychotic group, and that the single most miserable patient was on aripiprazole. The paper concludes on an optimistic note -

The present data suggests that aripiprazole may be associated with early and sustained improvement in subjective well-being, notwithstanding the very high D2 occupancy. This may be related to its partial agonist profile at D2 receptors.

I leave it to the reader to evaluate this claim, and to consider how likely we are to progress in our understanding of the brain when so much of the research is funded by organisations with a direct financial interest in certain theories.

[BPSDB]

Mizrahi, R., Mamo, D., Rusjan, P., Graff, A., Houle, S., & Kapur, S. (2009). The relationship between subjective well-being and dopamine D2 receptors in patients treated with a dopamine partial agonist and full antagonist antipsychotics The International Journal of Neuropsychopharmacology, 12 (05) DOI: 10.1017/S1461145709000327