Skip to main content

Transforming the understanding
and treatment of mental illnesses.

Celebrating 75 Years! Learn More >>

 Archived Content

The National Institute of Mental Health archives materials that are over 4 years old and no longer being updated. The content on this page is provided for historical reference purposes only and may not reflect current knowledge or information.

Introduction to RDoC


DR. CUTHBERT: Why are we doing RDoC? Well, it really goes back, as Greg said, to the way we diagnose mental disorders at the current time, and our diagnostic system really depends on our Western cultural heritage in many ways that actually goes back thousands of years, if you think about the ancient Greeks, who had the concept of melancholia, which is very similar to our modern concept of depression; melancholia literally meaning black bile, their theory about what was actually going wrong and causing this.

More recently, our notions of schizophrenia and bipolar disorder stem from the work of the classic German psychiatrists, like Eugen Bleuler and Emil Kraepelin, and Kraepelin first really seriously propounded the idea that there's this distinct pair of psychotic disorders, schizophrenia and bipolar disorder, that are really distinct and separate diseases.

And that sort of notion has come down to us in the present day, and, again, it's based really just on symptom presentations and looking at the course of disorders. And as Greg said, we know that the brain is by far the most complex organ in the body, and at those times, of course, virtually nothing was known about the brain, beyond its gross anatomy, so those people had to rely on clusters of signs and symptoms that patients came in with, but unfortunately, as Dr. Insel likes to say, that's more or less equivalent to saying that you have a headache disorder or a stomach pain disorder and not really going any farther to figure out what's actually wrong. And so you are necessarily limited in how you can move forward.

And it's also the case that now, as we move into this modern era and have techniques such as Greg was describing for neuroimaging, for structural and functional analyses of the brain, we're finding that these things simply don't map on very well to our disorder categories, and there seem to be many reasons for this. So we fund all this amazing amount of brain science and behavioral science, but yet it doesn't seem to line up very well with our disorders, and as Greg said, for all of our investment, we still don't have a single test using imaging or blood measures or anything that can really provide a diagnosis that would direct the clinician to do a very precise treatment that can actually help an individual patient.

So the fact that our disorders are different is referred to as heterogeneity, meaning lots of all different things sort of given the same name. And I'll just give you a couple examples of this in terms of understanding diseases and what that means for treatment.

A few years ago, the first stem cell report for schizophrenia came out titled "Schizophrenia in a Dish," and there was a lot of excitement about that. But in one of the comments about this exciting article, Michael Owen, one of the foremost psychiatric geneticists in the world, said, "These disorders," meaning schizophrenia, "are not really disorders. There's no such thing as schizophrenia. It's a syndrome. It's a collection of things psychiatrists have grouped together."

So, again, as Greg said, it may be the case that there are five or eight different actual disease entities that we all kind of lump together and call it schizophrenia, and some people have gone so far as to say that every single patient who gets a diagnosis of schizophrenia actually has something just a little bit different wrong with them because of the complexity of the brain and all the possible ways that things could go wrong. And we hope that it's not that extreme, but this represents the problem that we have to grapple with if we really want to develop effective therapies for individual patients.

And then again, for looking at developing new treatments, recently four scientists from AstraZeneca wrote a paper called "The Drug Hunters' Perspective," and in their paper, they state, "On average, a marketed psychiatric drug is efficacious in approximately half of the patients who take it."

One reason for this low response rate is the artificial grouping of heterogeneous syndromes with different pathophysiological mechanisms into one disorder. So if you're trying to develop a new treatment, whether it's a drug treatment or a behavioral treatment or a device, if you give this treatment to a group of patients, but only, you know, say, a third or 40 percent of them actually have a problem, a particular, specific problem that's being addressed by the drug or the treatment that you've developed, well, you're not going to see an effect because all these other people will not respond at all. It's -- that treatment is completely irrelevant to what's wrong with them, but they're looped together -- grouped together so you can't actually see an effect. So these are the kinds of problems that we face.

So this just brings us to this paradox that if we really want to have a diagnostic system based on our brain science, what we know about behavioral science and genetics and so forth, we'll never get that system until we have a research literature that can tell us how we ought to do that. What should we be paying attention to? What are the appropriate tests and so forth?

But we'll never get that database as long as we are doing diagnoses and conducting our clinical studies in line with our current system that ignores all these important distinctions rather than trying to fund studies exploring what these important aspects and dimensions and elements are.

So RDoC represents essentially an attempt to move forward and say, "Let's try to look at what the brain actually does and what's important and what the major systems are and explore that and try to think of mental disorders in those terms, rather than making these symptom categories and assuming that the problems in the brain and the problems in behavior should line up one to one with these categories that were defined in the complete absence of any knowledge about the brain anywhere from 100 to 2,000 years ago.

So the statement of RDoC is in our strategic plan, and it states, "Develop for research purposes new ways of classifying mental disorders based on dimensions of observable behavior and neurobiological measures." So and there are four important parts that will help you understand sort of what that means. The first component that is specified what RDoC ought to do is to, in fact, identify these fundamental components that may span multiple disorders.

So the idea here is that we now know that there are circuits in the brain that -- you couldn't really see them really clearly in Greg's slides, but there are major circuits in the brain like the major highways that carry out certain functions. We know there are circuits for responding to threatening situations. We know there are other circuits that help us respond and get food, shelter, and things that we want.

We know there are systems for memory and storing things away to remember them later. There's a different system for what we call working memory, the sort of temporary or scratchpad memory that if someone gives you a phone number, you can remember it for a few moments while you dial the number, and then you kind of forget it because you don't need it, but it's there for a few moments.

There are other circuits that help wake us up appropriately, you know, get the system going in the morning and give us energy and then sort of help us go to sleep at night. So we understand all of these different circuits in the brain, but, again, we don't have a very good idea of how these actually relate to mental disorders because in any one disorder, like depression, many of those different circuits may be disrupted, that in any one patient, some of the circuits may be wrong, and another patient, other circuits, different circuits may be impaired. So it's a matter of saying let's not call it depression. Let's actually figure out what specifically is wrong in this patient.

Greg used the metaphor of the highways. It's sort of like saying that, you know, the problem -- that somebody calls in to say that there's a problem with traffic in the Washington, D.C. area. We have a Washington, D.C. traffic problem. Well, what you really want to know is, well, is it a problem on I-95 north to Baltimore? Is it a problem on 270, going up to Pennsylvania and Breezewood? Is it a problem of I-95 south? You know, let's say which one it is, and then we can try to fix it. But we're sort of stuck saying we have a Washington, D.C. traffic problem, and that's sort of a simple metaphor for the problem that we have, and we have to now identify what are these major circuits, and how can we best study them?

Another important point for RDoC is that the goal states, "To determine the full range of variation from normal to abnormal." So this is really important because what we're saying is that because we're studying what are essentially normal brain functions, like memory and responding adaptively to threats and going after rewards, getting food and doing enjoyable things, these are all normal functions. That's what the brain evolved to do.

And so what we're saying is that we want to look at these functions across the full range of normal behavior and then out into where things get to be a little bit off or somewhat off or really severely off and understand that whole range, because importantly now, we really understand these problems as complex traits that are continuous. It's not as though you have a cancer that you're either healthy or you have cancer or you have a flu bug, you have a virus or you don't. It's not you're well or you're sick. It's more of a continuum, like blood pressure or like cholesterol levels that they kind of shade along.

And when you get a diagnosis, say, of hypertension, it's not that you totally didn't have hypertension, and now all of a sudden you have high blood pressure, it's that there's sort of a range, and the doctor knows when you're in the normal range enough that you don't need to worry. If your blood pressure's a little elevated, well, you need to cut down on your salt, watch your diet, exercise, all those things.

If you got much higher in blood pressure, then the doctor might say, "Okay, now you need to take some anti-hypertensive medications." But even then, he or she will look at exactly the kind of blood pressure problems you have to prescribe the right medication.

So that's the idea is that here, we're really making a shift. We're not trying to figure out that somebody has a mental disorder and then say, "Okay, we're defining this disorder, and now we'll see what is wrong with you in your brain." We're turning that around and saying, "Let's think about what the brain really does and what the whole range of normal brain functioning is, just as exactly what the Human Connectome is doing, and that will help us understand disorders in terms of dysregulation and impairments in these normal processes, rather than inventing these other categories, like schizophrenia or bipolar, and expecting the brain to match up with our cultural descriptions of what appears to be wrong. So this is a very important shift in how we're going about it.

A third point is to integrate genetic, neurobiological, behavioral, environmental, and experiential components, experiential obviously meaning our experience of these disorders. Because these are complex disorders, we may never be able to simply say, "We did an imaging test, and now we know you have schizophrenia." It's really going to examine how people can function, what goes wrong in the brain, what the genetics are, how that's affected by the environment, what they feel like their experience is, and we have to look at those things all at the same time. This is not a matter of just saying, "Okay, now we understand behavior in the brain" and will tell us everything about that. It's rather understanding the relationships between the brain measures and the behavioral measures to make sense out of these very complex syndromes.

And, finally, this is such a new approach, we don't have measures of any of these things. Our -- again, our diagnostic systems are simply based on rather simple, straightforward descriptions of problems, and we don't really measure any of these constructs. We know about them, but there aren't really ways that we can measure them in clinical trials or in actual clinical use, and so we have to do a lot of measurement development and fund the research on that.

So this is the RDoC work group. It's drawn from all segments of the institute. There's a little Mac-to-PC problem there that -- so I didn't stick Kevin Quinn and Chuck Sanislow in the corner for any reason.

So exactly -- now that I've told you why we're doing it, what is RDoC, exactly? What are we doing? Well, like all of the things that we do in the extramural program, it's a research grant funding initiative directed to a very specific area, these new studies of dimensions. And it's not that we're trying to create -- you know, we've defined a new classification system, and we're testing it. It would be more accurate to say that it's research toward a new classification system. We're trying to study and validate these dimensions of function, like threat and reward and working memory, that are affected across many of our current mental disorders.

And the goal, again, is to create a research literature, to fund studies that will create a research literature that lines up the genetics, other parts of neuroscience, like neuroimaging, and behavioral science that tell us how these circuits work, what the normal range is, what seems to go wrong when things become dysregulated or impaired, and then eventually to be able to use that research literature to say how we should be diagnosing patients in the future based on the brain science.

And we hope, just as in the example from the drug hunters, that if we do that, we'll also be able to improve our treatment, which, of course, is really what we're all about is by giving us better targets that we can define a more homogenous group of patients who, in fact, do all have the exact same thing wrong with them, and then when we develop a treatment, we will be able to get a treatment that's really effective for that particular group. But all of these things are in the future because we don't know how to do that right now, and that's exactly why we have to do this research program.

So this is what the RDoC matrix looks like. And I don't want to spend too much time in this, but I think we should have a little idea of how this is laid out. So this is the end? On this thing? Yeah, that's what I -- yeah. Yeah, got it. So if you look over in the left column, you'll see this is labeled domains/constructs, so the research domain criteria refers to five major domains of functioning: negative valence systems. That means how we respond to threat and loss, you know, aversive or negative situations. Positive valence systems; as I mentioned, those circuits that help us work for reward, learn about rewards, develop habits, and so forth. The various sorts of cognitive systems, like memory and working memory. Systems for social processes and arousal/modulatory systems, like biological rhythms, arousal systems, and so forth.

So we had a workshop with experts in the field over the last few years for each of these domains in turn. We had five workshops. And at the workshops, the participants defined exactly what the dimensions are. So for cognitive systems, I've listed them here. And there are the list for arousal/modulatory systems. I haven't shown you all of the constructs, but there are these dimensions under each one of these. And you can go to our website and see the full listing of the constructs under each domain.

We use the term "construct," which is a term from psychology that means our best idea about a specific aspect of functioning in a psychological sense. And we use that to indicate that we never -- probably will never have a complete, 100 percent understanding of these systems. It's our best idea about something, but we keep refining and improving our understanding of any of these things.

So, for instance, in reward systems, the first study of the reward system was in 1954, and some scientists looking at rats simply said, "We've discovered the brain's reward system," and that was all we knew, and we knew that was really important, but that was all that was known. Now we understand that there's one system for working to get rewards, another system for experiencing that the reward is good. You know, it's one thing to want something, another thing to get it and say, "Ahh, that milkshake is just really good." And a third thing for actually learning to associate rewards, that, you know, when you see the Dairy Queen, that's when you can get a really good milkshake, and you sort of have to learn that.

So there are actually different systems for all of these things, so over the years, we've refined how we understand the brain's systems, and that's referred to as a construct. And we think of these as dimensional constructs.

So this is the list of domains and the constructs, and then, as indicated in the strategic plan, we study each of these constructs across a large number of different ways of measuring them, and we term these units of analysis. And part of the workshop process was we asked the experts to tell us, once they had defined all these dimensional constructs, what would be the elements that could be used on the basis of the literature so far to measure those dimensions.

So this one is just an example of the cognitive control construct, which is essentially an impulsivity dimension, sort of being inhibited at one end to being very impulsive at the other end, related to things like ADHD, substance abuse, gambling, fighting, and so forth.

And you can see that they define different genes that are relevant, different circuits, various ways of behavior, different kinds of reports people make of these behaviors, and, finally, the laboratory paradigms that we could use to actually study this sort of behavior.

In some cases -- you can see here there's just a gap. There's nothing there, and that indicates an area where, in fact, the science hasn't given us any leads at all and helps us target areas where we might want to particularly do further research.

So that's sort of how this is set up, and each of these constructs is sort of a promising idea, an area for study that we would encourage researchers to do. Thank you.

The idea here is that we are working towards precision medicine, which is, as you know, popular across all areas of medicine now. And in late 2011, the Institute of Medicine released a big report called "Toward Precision Medicine," and the idea is that you take classes of measurement in many different areas, people's environmental affects or their exposure to things, the exposome, their clinical signs and symptoms, which, of course, are what people want to have cured, after all. That's why they come. Genetics, epigenetics, the microbiome, other types of patient data, and together, all of these different sets of analyses, zero in on one particular patient, and presumably ideally would help to indicate the exact treatment for that individual patient. We're not quite that ambitious. Mostly now precision medicine involves small groups of patients, but that's the idea.

So similarly here, what we would say is, okay, you have problems in what we would -- used to think of schizophrenia, but for you, we're actually going to break it down into what are the specific problems that this individual has? Is it language troubles, is it hallucinations, is it delusions, is it problems in cognitive functioning or remembering things? But actually in schizophrenia, there is a lot of heterogeneity there. And so the idea is that for each patient or for small groups of patients, we study the particular genetic patterns and how that affects molecules and cells, activity in circuits, ways of measuring that activity, like heart rate or skin conductance and so forth, looking at their actual behavior, and together, these give us a good idea of disruptions in something like working memory, that that's a big problem for many patients with many disorders, where you just -- you're so stressed or so disorganized, you can't remember a phone number. If somebody says, "Okay, you know, please call (301) 443-9782," and you're like, "Sorry, I can't get that." You know, it's that kind of thing that most of us normally can do. And so we can look at that and say, "Okay, we're studying different ways of disruptions in working memory," and that, in turn, results in cognitive deficits in real-world functioning, not just with patients with schizophrenia, but patients with depression, bipolar disorder, and many different disorders. So that's the idea, is to use all of these measurements to zero in on the particular idiosyncratic constellation of problems that a particular patient or small group of patients has and then understand that.

You're looking thoughtful. Did that address your question? Okay. Okay. Good. So there are two important things that should be added here. We're not such good multidimensional artists, but I've tried to show four dimensions here. Here is the matrix sort of shrunk down. One really important aspect is development. We increasingly understand our disorders to be disorders of neurodevelopment; that is, that as the brain matures over time in this incredibly complex sequence of brain development, there are critical periods, and that interacts with things that happen to you, whether you have a really happy, secure childhood, whether you're abused or neglected, whether you have a single traumatic event, like a car accident or something. And so we really have to study development in many of our adult disorders, like schizophrenia or bipolar, we can really understand as things that may happen really early in life, including during fetal development, and set the stage and risk patterns for really getting the disorder later.

So it's critical to do this, and we hope that RDoC actually will help focus this research because it's looking at specific kinds of circuit activity and behaviors rather than these broad, heterogeneous syndromes that we have now.

Similarly, as my example shows, we want to look at the environment and the effects of the environment, both in positive ways and in negative ways, as it affects neurodevelopment and essentially events throughout the lifespan.

So that's essentially the idea, and I'll show you two examples just to give you a little better feeling for what this would look like. This is a graph showing, from some grantees that we have funded, and this shows cognitive functioning. The backs (ph) is this group's composite cognitive functioning score, and the higher the number, the better off you are in cognitive functioning. And on this X axis, we have a score called the Schizo Bipolar Score. And at this end, a zero means that you are a classic textbook case of bipolar disorder, and at this end, a 9 means you're a classic textbook case of schizophrenia, with no symptoms of bipolar disorder. And Emil Kraepelin, in his famous dichotomy, would say that you should either be bipolar or schizophrenia, with a few right in the middle with something called schizoaffective disorder, but these should all be very distinct.

Instead, in these real-world data, what you see -- and each blue dot here represents one patient -- is that there's sort of a scattering of patients all along this scale, so a 2 here would mean you're mostly the bipolar, but you show some symptoms of schizophrenia. A 4 means you have just about 50/50, systems of both bipolar and schizophrenia. A 7 would mean, well, you're looking more like bipolar, but you have some symptoms of schizophrenia.

So you can see it's not a dichotomy, as Kraepelin proposed. It's really just this continuous flow. And you can see there's this strong linear relationship across this area, with a strong correlation that the close from -- as you go from bipolar towards the schizophrenia end, on the average, your cognitive functioning tends to be worse.

And the way we've always studied this is to say, "Okay, we'll take" -- even if you grant that these patients could be called reasonably bipolar, we study bipolar versus schizophrenia, and we say, "Yeah, sure enough, we know that patients with schizophrenia have worse cognitive functioning than patients with bipolar."

This ignores the fact that there's this whole group of patients in the middle that we don't really do anything about at all, which I think is a significant public health problem, it also ignores what you can see as this massive amount of variance here, that patients aren't all the same. It's not like all patients with schizophrenia are really bad and all patients with bipolar are really good. There's this tremendous range in here that we tend to miss with our simple statement that schizophrenia has worse cognitive functioning than bipolar, even though, on the average, that's not an untrue statement.

So what we're saying in RDoC is maybe what's really important is to examine the factors that, for each of these patients, tends to account for how we understand that no matter what your diagnosis is here, your score in this range, some people have relatively preserved cognitive functioning, and other people have really bad cognitive functioning. So even if you're a pure -- person with just pure bipolar disorder, you may still have very impaired cognitive functioning, whereas some patients who get a diagnosis of schizophrenia have fairly well-intact cognitive functioning; at least, less impaired.

So what may be important is to study the factors that relate to relatively good versus rather poor cognitive functioning, independent of what diagnosis they have.

And there's another example that we worry about now, given the emphasis on our soldiers coming back from the wars and PTSD and so forth. This is a graph showing many different anxiety disorders that you can see, and the measure here is a measure of fear reactivity in the laboratory. Essentially, if you're -- you know how if you're watching a really scary movie at night and it's dark and all of a sudden, the phone rings or somebody drops something or a door slams, you go, "Ahh." You know, you just jump up and everything, whereas if you're just sitting, you know, watching the news or something, and the same door slams, you kind of look like, "Oh, okay. What was that?" You know, you're not really very upset.

So this is a laboratory measure of exactly that kind of behavior, so the higher the bar here, the higher, more intense the fear reaction is. And in this case, patients are actually imagining their own personal anxiolytic, their anxiety-producing situations.

The point I want to make is that these are patients who have had a single traumatic PTSD event, like a car accident or having one crime occur to them. These two bars show patients who have had multiple traumatic events or who have had their PTSDs for just really a long time, and it's never remitted.

You can see that, for single trauma, these patients show the biggest startle reactions of all, you know, practically jumping out of their chairs as it will, where these patients hardly show any fear reactivity at all. They're very blunted. But yet we call these patients by exactly the same disorder, and we talk about doing a treatment for PTSD, as though the treatment for these people would be the same exact same as the treatment for those people.

Clearly, you can see something has to be very different with their fear reactivity systems. The one is very excessively reactive, while the other is blunted, but yet our current diagnostic system says we should treat them all the same. To me, this is almost unethical, that we should be taking this approach, but yet to do this, we need to develop practical tests to use this in the clinic, to see if they're reliable and valid and figure out how to do this, but if we do, we should be able to develop much more precise treatments for patients with what we like to think of as a post-traumatic syndrome that is this broad range of different things, and either match our current treatments much better to the patients or develop new treatments, based on a better understanding of what's going on here.

So that's essentially all I wanted to say about RDoC. I'll just say that you may have heard about some of our new contract trials, so-called fast, for fast fail. This is another new initiative of the institute, and what's significant with respect to RDoC here is that you notice that all the trials are devoted to spectrum, psychotic spectrum, of course, autism spectrum is that way in the DSM now, and the mood anxiety spectrum.

So rather than trying to treat a specific DSM disorder, like depression or PTSD, we're looking at a broad spectrum of disorders and trying to use RDoC dimensions as the treatment targets, and that, we hope, is the first test of how we can go about this more homogenous way of treatment development. So this is sort of an experiment in how we can start doing this.

So the timeframe for RDoC is long. We are just starting with our understanding of these things, so it's really something that will take a number of years to do, to develop this database, but already, as you can see from our treatment trials, we're already, you know, trying to use RDoC as much as possible to start moving the field in the directions that it needs to go.

So just to summarize of all this, again, this is a -- very much a translational medicine approach. We're taking basic science and translating that to look and see how we can understand these clinical problems and think about our disorders in terms of basic mechanisms that get disrupted rather than defining them all on their own and then expecting the brain to match up with how we defined it. And RDoC is really a framework. All of the dimensions that I showed you aren't the only ones people can study. It's more about a different way of doing the research. And so people can study other dimensions if they want to define them and say, "Here's the brain circuit. Here's the behavior that circuit produces, and we think this is really important." That's fine too, because we want to have this grow as it goes forward.

Again, I want to emphasize that this is the idea of a dimensional approach as well, that we're not saying you're well or you're sick. There are no actual disorder boundaries at this point. Again, what we need to do is to develop a lot of research literature, and people can make those decisions later, once they have a better idea of what things look like.

It's also important to say that we're not trying to develop this as a new system. We're not going to come out with RDoC as a competitor to the ICD or the DSM. Instead, it's to produce this kind of large research database that can be used to inform how future versions of these systems would be developed.

And, again, just in conclusion, we think that in the future, 10, 20, 30 years down the line, this could lead us to a much more precise personalized medicine approach for our disorders.

>>MALE SPEAKER: Yeah, and just talking about how this influences mental health in the world, the European Union is currently conducting a planning project called ROAMER, which stands for Roadmap Mental Health Europe, to examine and propose steps for their research needs in Europe in the coming decades. And their workgroups looking at biomedical and psychological research needs have actually adopted the RDoC matrix and structure as their organizational scheme for how they propose what is needed to move forward. And I just actually had to fly over to a meeting with the European Commission a few weeks ago when they had their big meeting for ROAMER, to talk about all of these things.

So our initiative may actually have a major role in contributing to how Europe sets up its entire mental health research system in the next several years. So it's already having a lot of influence.