 OK, why don't we get started? So a reminder that there's no class on Friday, Thanksgiving, weekend up ahead. What else is there? We're in the process, as you might imagine, of making the final now. And you, I hope, are in the process of studying for the final. And if people want office hours to go over stuff, email me. I'm gone until Saturday, but then back. So Sunday or Monday or anytime thereafter would be fine. I guess one other announcement could be if there's sufficient desire from you guys to have a review session to prepare for the final. We could do that. In that case, you need to email us with that request. And if it looks like there's a lot of people that want that, then we can do that. We've had them in the past. And then hardly anybody showed up, or people that showed up to sat there and they didn't have any questions and they hadn't studied for anything yet. And we're not going to tell you the answers to the final. So it only makes sense if, A, there's a critical mass, a number of people that want it. And B, we have it in a context where you guys have already been studying for the final and have questions that we can clarify. And we're not going to redo the whole course in a review session. But so all that said, we're happy to do a review session that would help you for the final if people want that and you have questions for that. Email us if you want that and we will use those data to make the decision. OK, so today we're talking about autism, another one in our list of diseases of the brain. And then you'll have next week on Monday, I think depression and on Wednesday schizophrenia from Henry and on next Friday, the last lecture. So we're finishing up with diseases of the nervous system. It's worth reminding us ourselves that diseases of the brain, in particular, neuropsychiatric diseases, but neurodegenerative diseases, could be up here also, are huge in terms of the economic and social and personal burden. So it's a weird metric that the World Health Organization put together here, but this plots something called disability adjusted life here. Is it something like the years of your life that you lose if you have one of these diseases listed here in terms of your contribution to society, not necessarily life itself. But you see that neuropsychiatric disorder is way up there. The main contribution is from depression, but there's many others' anxiety disorder, schizophrenia and autism, which we'll talk about today, that all contribute to this. So it's a big, it's a huge burden. And consequently, there's a lot of funding that federal funding agencies like NIH put into trying to understand psychiatric and neurodegenerative diseases and trying to cure them eventually. In terms of a big picture, you might ask yourself, so what is it that distinguishes brain diseases and might make them different than diseases of other organs, although some of these points apply to other diseases of other organs as well? So one thing is just that the brain's really complicated. So it's, in many cases, much more difficult to find the cause and end substrategic treatments for diseases of the brain than it is for cardiovascular disease or something that's run with your liver. There's often sort of a very clear mechanistic explanation for other diseases. For every psychiatric disease, there's nothing like that. And the causes are highly complex, poorly understood. The treatments are terrible. So there's a lot of work to be done. But part of it, a big part of it arises from just this broad, vague and obvious fact that the brain is a much more complex organ in terms of the number of cell types, the genes. They're differentially expressed. Any metric that you want compared to any other organ in your body. Another thing is it's relatively fragile compared to many other organs. It requires glucose and oxygen. If you are deprived of oxygen, the first organ to be damaged by that is the brain. So if you have a heart attack or you have carbon monoxide poisoning or something, parts of your brain will start dying within minutes without oxygen. And of course, with a few exceptions, but by and large, unlike, say, your skin, most neurons don't divide after birth. And so there's a limit, not none, but there's a limited opportunity for repair in the central nervous system. So if you have a lesion in your brain, that's not like having a cut in your skin. That organ can't heal itself as the skin can. And then there's two, if you look very broadly, and this sort of maps onto diseases of the brain to some extent, to a large extent. There are two big challenges, one which we heard about at the very beginning of the course, which is how to assemble the brain in the first place. So given the complexity, development is really complicated, very protracted, that goes through our adolescence. And if anything goes wrong here, you can have a disease of the brain. And that's one we'll talk about today, a pervasive developmental disorder, which is of which autism is one. And the other end of the lifespan is, of course, aging. And again, it's the fact that the brain can't repair itself. It's very complicated, the relatively fragile leads to lots of neurodegenerative diseases. So at the beginning and the end of your life is when a lot of illnesses of the brain arise. So what's autism? And we'll talk about, well, I'll show you what the diagnosis is, which isn't very principled. And then the way that people have looked at this and tried to get a handle on it scientifically. So there is a diagnosis that comes from the way that all psychiatric disorders are diagnosed, which is this thing here, the Diagnostic and Statistical Manual, that's what DSM stands for, of the American Psychiatric Association. We'll take a look at that in a minute. Pretty much every scientist, not all clinical psychiatrists, but every scientist, I think, or almost everybody, including funding agencies like NIH, agree now that what we want are diagnostic criteria that are informed by research. So you'd want some diagnosis that could use a brain scan or that would have some biological basis to diagnose, not just autism, but all psychiatric diseases, and right now we have that for none of them. So we need a lot of work on that. It first, so although presumably autism or autism spectrum disorders, as they're now called, were always around, they were only, the name was only coined as such and the concept only arose in psychiatry fairly recently. 1943, where two people, contemporaneously and unknown to one another, came up with this symptom. And Leo Kanner, whose quote from his paper is illustrated here, is often credited with being the first that came up with this. There's a couple of, and then the other person just to point out, so here's Leo Kanner, whose paper, Autistic Disturbances of Affective Contact, was here and then published a year later, although he worked contemporaneously with Kanner, was Hans Asperger, after whom Asperger syndrome is named. And they described a very small sample, I think it was just four or so boys in Asperger and 11 or so in Kanner's case. So a small, very small sample of children that were almost exclusively boys that had altered behaviors that they thought constituted a symptom that they diagnosed and then subsequently autism, that's what became known as autism and their initial observations became incorporated into the official diagnostic criteria by psychiatrists in this DSM book, which went through various iterations, it's now in his fifth. So that's the origin of it. It's worth pointing out, so a couple of facts, deriving from this initial observation and the way that autism was diagnosed here and has been diagnosed ever since then to some extent, is that it's a developmental disorder and so most of the work and most people who see psychiatrists are children, but it's pervasive and so once you have autism, the idea is that it's lifelong, so it would continue into adulthood, but most people and most of the research on autism has all been in children. Most of it is in boys, so most of it has focused on boys, but it's important to point out that there are famous people like Temple Grandin here who's written books about her autism and is very high functioning who also have autism, so it makes the important point that it's a broad spectrum and even though traditionally most of what we know about autism, both in terms of clinical presentation and in terms of research comes from young boys, there are old people and there are females who have autism as well and it's important, one very important thing is to understand the disease in these individuals and ask what ways it might be different, is it a different type of autism, et cetera. Another important thing to point out, this guy here, Stephen Wiltshire, who lives in England I think who has autism and is very high functioning, another important point to make is that most kids diagnosed with autism are at the relatively low functioning and or mentally retarded statistically, but there are also individuals who are extremely high functioning, probably several Caltech professors would meet criteria for an autism spectrum disorder and so the right way to think about it is that people with autism are atypical, they're different, they have a different profile of strengths and weaknesses and in some cases that can be a global disability if you're also mentally retarded, but in some cases not, so this guy can actually do fairly well and he can sell these works of art, what he does, you can look him up on the internet, is he can fly around a city and form an accurate memory of all the buildings and all the details and then we produce that when he sits in his studio and draws what he just saw, which most of us couldn't do. So the right way to think about it is there are disabilities and abilities, it's a different profile of strengths and weaknesses than for typically developed individuals. So what are the criteria? This is the criteria according to the fourth edition of this book, the Diagnostic and Statistical Manual and most people with autism would have this diagnosis, there's now another version, DSM-5, that has somewhat revised criteria, but I'll just show you these because the point is very similar. So people with autism have impairments in reciprocal social interactions that typically manifest early and a part of the diagnosis around age three, four or so where normal children would make a lot of eye contact, interact with other children and with their parents and engage in pretend play, a whole host of things that constitute social interactions. Children with autism don't have that, somewhat connected with that, they also don't have normal social interactions that depend on language. This used to be, in DSM-4 used to be the distinction between autism and Asperger's syndrome that people with Asperger's syndrome did have very good language functioning, people with autism didn't and then there's this category here that there tend to be unusual interests and often repetitive behaviors and in some cases, like with Stephen Wiltshire, often exceptional abilities related to this. So they're not people with autism tend to be less good and less interested in interacting with people but can be better to an obsessive extent in interacting with non-social aspects of the environment, often processing that requires a lot of focus on detail. So that's the overall, that's the sort of tryout of symptoms that constitute autism. If you look at this, you know, as a non-psychiatrist, it should look and it is, it should look very vague to you and it is very vague. Nonetheless, if you're trained as a psychiatrist and you see a whole bunch of children with autism, it turns out that there's a fairly good agreement on diagnosing autism but it's hard to capture that accurately in these sort of formal criteria that people have written down. One big question, one big factor is or issue is whether these diagnostic criteria have good sensitivity or specificity and one idea is that currently they probably have neither. So you would want to pick up if you, especially if you wanted to do research on autism, also very high functioning people with autism and a lot of the criteria currently don't pick that up. Well, they're not particularly sensitive and they're not very specific, which is why it's called autism spectrum disorder because nobody could agree on the specific categories. They used to be specific ones, autism Asperger syndrome and pervasive developmental disorder, not otherwise specified, which is a very long phrase that have now been lumped into this category of autism spectrum disorder, in part because people realize that they just didn't know where to draw the boundaries for those subtypes. Typically in all, not typically in all research studies, essentially all, what people do is compare people with autism versus typically developed, i.e. healthy individuals, which is a huge problem because it doesn't contrast autism with other psychiatric illnesses. So essentially all of the research, almost all of the research findings of brain activation differences, differences in the structure of the brain, et cetera, just contrast autism versus healthy people. And so it's unclear to what extent the findings are specific to autism. It could just be any psychiatric disorder versus typical as far as the results from any individual study are concerned. So there's a lot of work to be done there. And then in addition, as I mentioned, there's a big gradient of IQ across autism ranging from people who have a diagnosis of autism that are really mentally retarded to people that are catholic professors and would have it. And so one thing that really needs to be done is to look at how that interacts with the disease. As you might imagine in terms of just sheer disability and what difficulty you have in everyday life, IQ is a huge factor. So if you're very high functioning and have a very high IQ but have autism, you can often compensate very substantially and do quite well. Whereas if you're low functioning, you can't do well even if you didn't have autism. If you have an IQ of 40, you wouldn't be able to do very well. So IQ is a big confound often in lots of the research studies. The incidence is fairly high and higher now than it was thought to be decades ago about the same as for schizophrenia, about 1% or so. And has been rising over the years probably for a combination of reasons. One reason being that it's just diagnosed more. But another reason independently of that, many people think is also that the true incidence in fact has been going up for environmental or other biological reasons. Here's a pie chart of this. So mostly we don't know why it's gone up. There is diagnostic accretion and I can't read what these things say. Greater awareness. So these things are, it's always been there but people are just diagnosing it more. But then there are for instance some clear biological factors. So this one here, parental age, in particular for the father. Older fathers have a higher probability of having children with autism. And the age of fathers over the years has been increasing. And so there is some contribution where we know the cause that parents are just older and that contributes a little bit, a small amount to the increase in autism. But mostly we don't know which is whether it's this big question mark here. It's autism is studied and it ultimately needs to be understood at all of the different levels of analysis that you've heard about in the course. From distal, genetic and epigenetic factors to knowledge of how their protein products work or don't work properly to influence synaptic function in the brain, how that influences large scale brain function, how that ultimately influences cognition and how that eventually gives rise to behavior. So what you're diagnosing is up here, the phenotype of the child presents. What you wanna know is the stuff down here, the causal factors, often the distal causal factors, where if you knew enough, you might be able to intervene and have some kind of therapy, but that's a long distance and so that's where the challenge lies. Okay, so just a couple of other quick facts here. So as I mentioned already, the autism is a pervasive developmental disorder, which means that it arises in development, typically diagnosed around age three, age two. People are trying to find markers that would allow some prediction, if not bona fide diagnosis at earlier and earlier ages. It's thought to develop, they're thought to be changes in the brain that will predispose you or even more deterministically cause autism that are quite early, probably well before birth. So it's a developmental disorder. It's not that you are perfectly healthy and suddenly at age 30 to develop autism, that couldn't happen. So it's a developmental disorder and it's pervasive. The other point is that once you have it, this kind of, it's a lifelong condition of your brain. So that's the idea. So Ansid already mentioned it's lifelong. It runs in family. So one clear entry point for information about the biology, the causes of autism, is the fact that it has very high heritability, about one of the highest heritabilities of any psychiatric disorder. So this metric here, the maximum of which would be one, is the proportion of phenotypic variants of how people would present that you can account for by genotype, by variants in the genotype, from families, from linkage studies and families and so forth. Despite the fact that it's clearly very heritable and so presumably there's a strong genetic component, it's not, very little is known about any specific genes that contribute to autism. Another hint that there's a strong genetic basis is what I already mentioned, that it's mostly males that get autism and in high functioning people the ratio of males to females is something like 10 to one. On average it's like four to one but there's a big gender ratio that I'll talk about more in just a minute. There's a spectrum and then finally, as I mentioned, it's even though there are biological markers, it's currently only defined by just how you present and what your mother says about your childhood history when you go and see a child psychiatrist. The only reliable finding, the most reliable finding, the only reliable finding with respect to the brain is that on average children with autism have larger brains which is interesting for a number of reasons. One reason being that so do males. So males have about 10% larger brains than females and children with autism have larger brains. So again, it may be related to the fact that most people with autism are male. Okay, this just goes through the heritability. I need to know, so one is I got identical twins if your twin has autism and your identical twins, your probability of also developing autism would be much higher than if you are fraternal twins as all these numbers say. There are a lot of biomarkers that people have looked at and many of these have some prediction like head size that I mentioned, which would be related to the size of the brain, at least in childhood as you can normalize later on. And so what people are trying to find is all of these and to some extent there's some biomarkers. So there are studies now where they have sufficiently large samples where pre-people are trying to train a classifier and just use machine learning algorithms to predict from the pattern of all the different measures you could take. So you could take a blood test and a genetic test and get an MRI and do whatever you want. And you can try if you have a sufficiently large sample you can try to do some cross fold validation on that and then use that to predict whether or not somebody would go on to develop autism or not. Some of those classifiers seem to work on a limited population, none of them so far have worked completely, have convincingly worked totally out of sample. And as I mentioned already, none of them have tried to classify autism as distinct from any other psychiatric illness, it's always autism versus healthy which may not be the right classification to begin with. So what do we know about the genes that contribute to autism and what might they tell us about what goes on in the brain? There are a lot of them, hundreds that all make a small contribution to autism and very few in very rare diseases that can make a large contribution but on average any person with autism has mutations, polymorphisms across a large number of genes that each contribute a small risk to autism. Many of these are somatic mutations and copy number variants and a lot of the genes code for proteins that have to do with how the brain gets wired up. So the genes are telling us something about the mechanism of autism because a lot of the genes code for transmitter receptors or for proteins that stabilize synapses or they're involved in axonal guidance and path finding during development. So a lot of the genes seem to code for protein products involved in connectivity of the brain, either establishing that or maintaining that. So one idea is that autism has something to do with synapses and how neurons are connected to one another. Okay, let's see. Let me skip the details of this industry. This just makes the same point that I just said in passing which is that there are a number of single gene diseases that are known. So here are the diseases on the left column and the genes, if you have a mutation in that gene they cause this disease on the next column over. These are all very rare diseases but in these rare diseases you can have a very high proportion of autism which is what's shown in the middle column here. So if you have fragile X syndrome here for instance, you can have 25% of males with a fragile X syndrome would go on to develop autism or in some other diseases with this really rare one that I don't even know what it is, particularly the PC syndrome, it's 90% or something like that. So there are specific genes that will cause autism with a very high penetrance but these are all extremely rare. So if you now ask what's the actual proportion just in the population of people with autism that have this gene mutation, they're all very small numbers. So unfortunately none of these actually explain autism in the population very well because most people don't have these very rare diseases. They may tell us something of course about the mechanisms and again many of these are involved in with particular neurotransmitter receptors or synapse function in some way. Okay, so to summarize what I've told you here so you can just read this yourself. It mentions that the fact is that autism has a big genetic component but no single gene contributes to that but there are these rare diseases where you do have a strong link between particular genes and autism and that's just schematized here. So if you look at the effect size in terms of the contribution of any gene to autism and you look at the frequency of different alleles there are some very rare ones over here like fragile X syndrome, et cetera on the left that have a big effect size but most people with autism don't have that and then there are lots over here that are quite common variants and they each make a very small contribution but if you have a whole bunch of them you will have an increased risk for developing autism. Is that clear? The basic genetic scheme there. So what happens? Well, this is a very overarching picture of how people think the genes and of course unlike a disease like say Huntington's disease autism is not 100% genetic so there is also a role for environmental influences and of course genes and environment interact in very complex ways and these distal causes then eventually give rise to stuff in at this sort of endophenotype level here that gives rise to the altered behavior and cognition that we diagnose and the most proximal substrate there as I mentioned is thought to be altered connectivity. So altered connectivity in the brains of people with autism makes them process information differently that gives rise to what we see in terms of the phenotype. What caused that is changes in how synapses are set up. There's the complex interactions with the immune system that people have studied and a whole bunch of really complex things whose very distal causes are in genes and environment, most be genes. This as I mentioned arises is thought to arise quite early in development. One of the clearest pieces of evidence for this comes from mothers who took this drug for morning sickness that you probably know about philidomide which causes birth defects which is now banned so people don't take this anymore but at the time they didn't know this and so if you take this drug which prevents you from being nauseated if you're a mother and have morning sickness at particular stages of gestation this will give rise to different kinds of birth defects because the drug interferes with development and it'll interfere with different kinds of development at different ages. So I don't know if you can read this well I can't really actually but it says here what would happen if you take philidomide at 20 days at the age of an embryo or something. So here the ears are missing. Here you have small thumbs here you have stunted legs and you would get various birth defects in the child if you take philidomide in this particular window because your legs are developing etc in this time and it turns out right in here if you take philidomide there's a highly increased risk of the child developing autism. Now this is really early so this is just when the neural tube is starting to fold do you remember from your development lecture? So very early on in neural development you can already interfere with it so it's to increase risk for autism. This may not be the main mechanism by which autism is normally caused but it shows that there may will be very early events in neural development that contribute to risk for autism. Okay so what causes autism? So we just had this picture up of genes giving rise to other brain works which gives rise to the behavior and one interesting theory that this guy Simon Baron Cohen who is the cousin I think of Sasha Baron Cohen of Borat fame and is a British autism researcher one particular theory that he has pushed is looking at this gender ratio that boys have autism much more frequently than women and looking at the phenotype the way that people with autism present he has this hypothesis that people with autism their brain essentially works like an extreme male brain and this plot here shows that or is thought to show that so what he's plotted here are two dimensions which can be assessed with a questionnaire one is how much you empathize with people this empathizing quotient or EQ so if you really care about people and you empathize with them a lot you have a high score up here and so these red dots, healthy females tend to be up here females on average have a higher EQ score than do males and down here is the systematizing quotient this is how interested and good you are in sort of taking apart mechanical things and figuring out how things work and on average males are better at that than females and so the healthy males which are the blue triangles here are down here and then people with autism the green squares are also down here so his idea is that given the sex ratio that's observed in autism and given differences in cognitive abilities between males and females that map onto these dimensions of empathizing and systematizing that the autism brain is like an extreme male brain that's one hypothesis here's another way of looking at the gender difference that's one way that people think of it is that if you have a so-called threshold liability model which is quite common in psychiatry you would look at the proportion of the population that this bell-shaped curve here and the risk that they have the liability that they have from all the different genetic mutations environmental factors that would predispose them to developing autism and the idea is that if you reach a certain threshold if you have accumulated enough of these common genetic mutations you would surpass the threshold and your brain couldn't compensate anymore and you would develop autism and meet a diagnosis so there would be a threshold here and the idea is that that threshold would differ between the sexes and that there might be specific protective factors in the case of females such that you need to get to a higher threshold before you develop autism in females than you would in males so that's one model as an aside I just wanted to mention here that autism is not at all on its own in this sense so lots and lots of disorders some of which are mapped here and you could do the same for neurodegenerative diseases which are not represented here have big sex differences and so one thing that people are very interested in is looking at sex differences to try to get some insight into what causes these different illnesses so some of them probably make intuitive sense or some you've heard about and Eurexia is a lot more common females and males and so forth some probably seem counter-intuitive for instance PTSD post-traumatic stress disorder is quite a bit more common in females rather than in males even though you tend to think of this occurring in combat veterans Alzheimer's disease which isn't on here is also more common in females than in males so there's some counter-intuitive ones but anyway people are quite interested in these and of course are looking then at sex-linked genes on the X or Y chromosomes that might contribute to these various disorders as well as other factors one thing that's important to point out of course is that these sex differences are associated with lots of other things in terms of lifestyle differences that these themselves could give rise to the difference or could account for at least in part some of the differences you see in these psychiatric illnesses so it could be that so males have a higher, males tend to smoke more and so maybe sex differences in Parkinson's disease or some neurodegenerative disease might in part arise from different nicotine doses in males as opposed to females or alcohol consumption males are higher there than females and presumably that has some effect on the brain so there would be lots of indirect routes here just to sketch this out and complete this part here so the way that genes act on the brain is mostly not direct there are a few genes on the Y chromosome if you're male and a few genes on the second copy of the X chromosome that fail that escape X in activation if you're female whose protein products do act directly on the brain and do stuff in the brain but mostly the genes act by our hormones so the way that the brain as you probably know is masculinized is mostly there's a gene on the Y chromosome if you're male, SRY3 that determines that differentiates the testis the testis makes testosterone and testosterone is testosterone that masculinizes your brain and humans to some extent directly and other animals like rodents always indirectly through the conversion of testosterone to estradiol by an enzyme called aromatase in the brain but anyway, the bottom line is that it is hormones, testosterone that of course are the way that that's produced is genetically determined in large part but the effect on the brain is mostly via this route and then of course the third one keep in mind is the same story for the environment so if you are born with a penis and that initial sexual differentiation is because you're XY and had to do with testosterone right from birth your parents and everybody else will treat you as a male and that social environment will have influences on the way that your brain develops and your gender in the future so just to make the point, it should be obvious but that even though the distal source of gender in the brain is XXXY the way that that interacts with hormones and environment is very complicated and most of the effects are not directly of genes on the brain and so presumably the same thing is true in autism what's known about that people have of course tested this and tested is it the case that differential fetal testosterone gives rise to is correlated with risk for autism there's no, it's not obvious so unfortunately there's no clear data on the CX probably in part for the reason that I just mentioned which is that it's just the interaction between genes, environment and hormones is so complicated people are doing those studies some people like Simon Baron Cohen who I just mentioned would claim that they have data that shows what I just said that fetal testosterone which even if you're just male has a range some people have higher testosterone than others and so you end up being looking more masculine or less masculine as a consequence of that there's a range in males even though you're all male and to some extent that correlates with autism but it's complicated okay so that just goes through the logic of it coming to other kinds of causal models that stick more that don't go all the way back to the genes and environment and hormones but that ask about cognitive processes the most popular view is that the social cognition and social behavior that comprise the phenotype by which you diagnose autism are causes in these processes namely attention and motivation and the model goes something like this that if you're a baby born predisposed to the high risk for developing autism that you will not pay normal attention to social stimuli like the mother's face you won't make normal eye contact you won't find other people as rewarding and so you won't have normal social motivation and if that goes on for a couple of years in development that will give rise to altered social cognition the way that your brain gets trained up to have expertise in perceiving faces will be different because you haven't been paying attention to faces and you haven't been looking at faces and so that a lot of the consequences of the phenotype arise from a primary deficit that is either the attentional or motivational people have tested this and so one easy way to test it that people do in experiments nowadays is to, like in this little video is to measure where people look and so you can do this quite easily now using eye tracking and what's shown here is how you would look how you would fixate a face and you look at certain parts of the face and so you could do this in your control group you could do this in autism and you could ask other differences and there are, here's an example the details of this get more complicated quickly but to first order this is how people with autism in at least some studies, ours look at faces and here's how matched control subjects who have the same gender and IQ at age but not autism, look at faces so control subjects look a lot at the eyes people with autism look less at the eyes and more at the mouth that's one difference that seems reproducible and reliable it depends to some extent on the face and the task and the context but at least under some experimental circumstances it's a reproducible effect that shows that indeed people with autism have differential attention to faces one thing that's interesting to point out here you remember hearing about this structure the amygdala that we talked about in relation to Pavlovian fear conditioning and in relation to emotion in previous lectures the amygdala has also been studied in relation to autism and there are a number of histological studies primarily for post-mortem brains that show that neurons in the amygdala are different in people with autism whether small sample sizes because they're histological studies so there's lots more work to be done but at any rate the amygdala is one structure that's been hypothesized to be abnormal in autism and so one kind of research study that people have been doing again from our lab is to compare how people with autism look at faces to how patients that have lesions of the amygdala this one is a patient whom you heard about before that was handling the snake look at faces and there's some similarity so patients with amygdala lesions like people with autism also look less at the eyes and faces some similarity there it gets much more complicated once you look at the details but the general idea is that you could start making some comparisons between people with autism and neurological populations that have specific lesions in certain parts of the brain and to the extent that there are similarities this would give you some insight that those parts of the brain are also implicated in autism I'll just show you quickly sort of what a lot of people nowadays are doing for research on autism they're rather than taking faces and sort of whittling away at very small pieces of the question of how people with autism process social stimuli they try to use really complex more real life social stimuli like a movie and so you would show people something like this in this study that we published a little while ago several other labs have done similar studies you would use movies or sitcoms which you could think of basically the people who made these spend millions of dollars to create a sort of optimal social stimulus so you need to pay attention to people's faces you need to figure out how they're interacting these are all things processes that people with autism have difficulty with and so you could try to quantify how people with autism process this kind of a stimulus when you do that you could get data like this so these are just example data to just give you a flavor for the logic of the experiment where you would put someone in a scanner in an fMRI scanner so we did this down here in the basement of this building you would line a scanner and you would watch an episode of the office you're watching and listening to the dialogue so you're watching all this stuff unfold then you would have a group of people with autism whose co-registered brains are shown here on the right a group of controls whose co-registered brains are shown on the left and you would ask over time what do you see as the pattern of activity in these two sets of brains autism versus controls and how is it different and so you can see how it's evolving there's lots of stuff happening that just stop this so this illustrates how the properties of the stimulus it's highly attention grabbing social stimulus which is why I need to stop it otherwise you can't listen to the talk so let me just make it black for a second but the basic idea is straightforward that if you and a person with autism look at this video you could ask where in the brain will you have similar activation and where in the brain would you have different activation and so if you think about this for a second you should be able to have a hypothesis you're watching the same stimulus so you're seeing light and dark things and motion and the same visual drive at the level of the retina we just show this in a small way so you can't move your eyes around much so at the level of the retina it's the same therefore in visual cortex it should be very similar visual cortex gets the same visual input whenever there's something bright your brain and the brain of a person with autism would activate primary visual cortex but then you, unlike the person with autism would be able to figure out the social meaning of what's going on what are these people thinking what's socially awkward, what's the plot so you can figure that out people with autism would have difficulty figuring that out and so you should be able to look in the brain going from primary visual cortex to higher order visual cortices to more complex regions in the frontal lobe and ask where, at what stage in processing and where anatomically would I see a divergence in people with autism from the pattern that I see in control subjects and that would give you some insight into what processes and what parts of the brain are different you can imagine doing the same people have done the same experiment in a much more radical way with monkeys right if you have a monkey watch the office in primary visual cortex you would see the same activation as in your brain it's seeing the same thing but it doesn't know what it means and so at some point again further on in processing in the brain the monkey brain should respond differently because it has no clue as to what's going on than your brain so you should be able to pinpoint the place in processing and that would give you some clue to where in processing people with autism go awry the other thing you can do is of course use eye tracking so this just gives you a flavor for the kinds of data that you would get so if you show people a complex stimulus like this a movie and ask how they pay attention where they look if you have it on a large screen in controls and in autism this is how it would look so each colored circle is one subject direction of gaze and both groups are watching the same video and you can see that they look when there's nothing else competing for attention people with autism also look at faces quite a lot as to the controls there's some spatial dispersion some people occasionally look off to the side etc and now from this pattern which would go on for the 20 minutes of the movie of the sitcom you would want to say is there something I could train a classifier to classify accurately how people with autism look at this compared to controls is there some specific feature or something that accounts for this and you can do this if you decompose this very complex stimulus into all the different features that comprise it so for instance you could run an automated algorithm through every frame of the movie to figure out where there is high contrast and so this is fairly easy to do these days so they're like the white shirt sleeves I guess of this guy would have high contrast the lights appear and then you could just ask well is it the case that people with autism look just at brighter more high contrast regions and indeed they do so if you look at that if you just look at contrast and ask when they look on average control subjects look at contrast regions in the movie whose mean and value is down here people with autism look at regions of higher contrast so logic like this or analysis like this would allow you to identify particular factors that drive the abnormal attention in people with autism rather than following saliency that's determined by the social meaning of the stimulus they tend to be captured by low level cues like just brightness, motion and contrast and so they would miss out on a lot of the meaning because they're not allocating attention there appropriately the other point that this these data of course make is that it's not a gigantic difference between the groups even though it's statistically significant there's a lot of overlap and most of the difference in the autism group is driven by these outliers up here so again these are important points to keep in mind there's always overlap it's a big spectrum and in many cases it looks like the spectrum may not even be continuous but there might be a subgroup of people with autism that are driving most of the effects in your experiment so never mind the details of this but I'll just say it because otherwise it's gonna confuse you you can do what I just showed you here in a much more comprehensive way so you can characterize all the places in the movie not only with respect to whether they're bright or moving or something like what I just showed you but also with respect to whether there's a face whether there's something smiling whether there's something that's socially awkward you have to, that you can't do in an automated way you have to have people go through and annotate every single part of the movie but you can decompose it and then you could do something akin to just having a very big regression model that asks what are the weights of all of those different factors a face, text, a happy something happened to do with happiness contrast, et cetera, et cetera what's the relative weight on determining where you look and when you do that you get weights like what's shown in this example down here that you know color, intensity, orientation, blah blah, these all have different weights here's a face, you tend to look at faces a lot these are data from controls and so you get a profile for the weights of all these different factors and now you could do this with people with autism and you would get as it were a complex fingerprint of how all these different factors influence visual attention in people with autism so that's the direction that people are going is a more sort of multivariate approach which seems to be certainly going to be the answer so it's not that people with autism don't look at faces it's not that they only look at high contrast it's they don't look much at faces they look more at high contrast and they also look at 23 other factors in atypical ways and you want to understand that whole profile or maybe do some dimensionality reduction on it to identify factors okay let's turn the last couple of minutes to the brain for which there's not much to say and so I'm not gonna show you data for the brain as I mentioned the single most reproducible finding is that children with autism have larger brains in addition people have looked in all these different brain regions of many others but the findings have been difficult to reproduce from post-mortem histological findings the next sort of most robust finding is that indeed there are differences if you look under a microscope at post-mortem autism brains differences for instance in cell packing density in the amygdala, differences in the cerebellum differences in microglia in the brain the problem with all of these studies right now is that the sample sizes as you might imagine are very small there's not that many post-mortem autism brains most of these other studies have used MRI and there's nothing very reproducible there are a lot of findings so if you look if you do a lit search you will find a huge number of structural MRI and fMRI studies that all find various differences in people with autism compared to controls and most of those don't replicate well so we just need a lot more work on that what people are currently doing is to combine MRI with other modalities there's a lot of effort now this used to be very hard to do but people now can do it to do MRI in infants so you can do MRI in newborns you just wrap them in a towel so they're nicely swaddled and they will fall asleep and they will sleep in the scanner and you can scan them of course you can't do any cognitive activation task but you can get structural imaging and you can get functional imaging of the brain at rest that will give you information about functional connectivity so people are doing a lot of that and then finally let me just say a few words about animal models so there's a lot of work on animal models these are models nobody would claim that mice are autistic but you can have mice that show aspects of repetitive stereotype behavior that show aspects of social behavior that seem analogous in many ways to what you find in humans with autism and so there's lots of different tests you can do for instance you can look at as a sort of correlate of social communication through language you can look at ultrasonic vocalizations that they make you can look at grooming behavior or marble burrowing behavior or motor stereotypies and so forth the nice thing of course is that you can manipulate the genes and so you can find mouse homologs of genes that are known to be involved in autism in humans and you can make knockouts for those in your mouse and ask what's the phenotype that you produce and that's what's shown in this table here so there's a lot there are a lot of specific genes that people have knocked out or knocked in or done something to in the mice and asked how if I do that early on in development how do the mouse pups develop and do they show any particular components of behavior that bear some resemblance to components of the autism phenotype in humans and then that would give you some clues there's one model that's worth mentioning that was championed by the late Paul Patterson here at Caltech who was a professor in biology who passed away recently and he looked at a very important and at the time very poorly understood environmental components so almost all the work has focused on genes but it turns out there are I mean we know that there are environmental causes and one clear environmental cause that's known is the immune response of the mother and there's epidemiological data for this in humans when there were epidemics where mothers were infected with a virus it turns out that the probability of their children developing autism was statistically much higher that's over a large sample but it's known that immune activation in the mother can have an effect and people have now tried to do this in monkeys also and tried to have very specific mechanistic models the basic idea is that and so here's the experiment you would take a human virus that is thought to increase the risk of autism also in humans if the mother gets it and you infect the mouse with that so you take influenza virus and you give it to a pregnant mother female the mother female will make antibodies against that virus and the idea is that some of those antibodies against the virus may cross the placenta and get into the brain of the baby rat pup and cross react with proteins in the developing brain of the baby and impair neural development so that's sort of roughly that idea there are also other things that happen there's lots of behavior of the mother and so forth but the idea is that there would be an immune response in the mother that can influence development of the rat pup so anyway that's one model that was done here at Caltech and people are still doing a lot of work on that gonna finish up gonna skip that and finish up with these slides here there are other diseases that are much more rare but worth knowing about that are interesting with respect to autism I mentioned a few already in passing this one is fairly interesting there's a number of people looking at this including people down at the Salk Institute close to San Diego or Slabalooji and colleagues this is a genetic disorder that's also developmental and so the kids have particular facial appearance and particular behavior and it's to some extent the opposite of autism so children with Williams syndrome tend to pay abnormal attention and be abnormally motivated to interact with other people so they're extremely interested in other people they'll come up to them, look at their faces a lot they're extremely verbal and tend to talk a lot so they're very socially sort of in your face as children just the opposite of what you see in people with autism and the nice thing from a research perspective of this disease is that it results from a deletion of one set of genes on chromosome 7 and so there's a bunch of genes that are knocked out here and indeed these are trisomics and some kids differentially not... differentially lesioned so you don't have all of these regions taken out you sometimes have a portion and so you can look at different parts of the phenotype but anyway this just gives you a sense for the phenotype here this may be a little hard to see but in one experiment you would show a child this picture it shows this boy getting a cookie from this cookie jar and the mum is washing dishes and they're overflowing with a bunch of stuff happening and so if you take a person who's sort of mental age matched that has Down syndrome as a control you will find that there's a very impoverished description of not too much happening if you take a child with Williams syndrome who's mental age matched to this control you will see that there's a very elaborate response that really engages the listener and so on and so forth so they're very good at social stuff and very good at language unlike people with autism they're really bad at detailed visual spatial kinds of things so if you do another task if you ask a person with Down syndrome as a control to draw houses they will draw houses like this if you ask a person with Williams syndrome they will draw them like this and so they will be completely disorganized really disorganized and not even be able these are their drawings they're just like completely all over the place so the bottom line from this as in the case of autism what you want to understand in this disease Williams syndrome is the whole profile rather than just any particular features that are often the basis for the diagnosis is have a whole kind of a fingerprint across all these different cognitive domains which is what's shown in the x-axis so this is full scale IQ, verbal IQ, performance IQ all these different tasks that they can do and on some they're impaired on some like faces, they're very good so there are peaks and troughs here and same thing with autism to some extent this would be the mirror image of what you see in autism and so the challenge going forward is to understand take this multivariate profile ask how that's generated by changes in the brain structures that subserve each of these cognitive processes and then eventually back that out to a presumably a very complex story in development that says why some things they're very good at and other things they're very bad at but it's going to be a complex multivariate story anyway so it gives you a flavor for sort of where current research and thinking is in neuropsychiatric disorders typically a complex set of distal causes many genes that interact with the environment a pattern that is currently diagnosed in a very coarse way, psychiatrically that you want to start by breaking apart into like a more detailed cognitive profile like what's shown here and then the whole work to be done is how to connect these two sets of patterns patterns of distal causes, genes and environment patterns of in a particular cognitive profile and what connects these two of course has to be something in the brain but that's the task is to connect two sets of complex patterns alright so we'll finish there and have a happy Thanksgiving