Thomas Insel on Science, Zip Code, and Future-Proofing Psychotherapy

Thomas Insel: It’s a pleasure to be here, and I’m glad that we’ll have a chance to talk about some of the nonmedical aspects of mental health care, which have not received enough attention.

TI: There are two parts to that. I think the first part is that we have bought into a medical model for how we think about mental disorders broadly. And the second part is that the medical model is part of a large healthcare industry, at least in the United States. I don’t know if this is true in other places, but in the United States, healthcare is a massive business, a $3.5 trillion business.

A lot of that business is driven by a particular model which says that illness is due to a singular, often simple cause, whether that’s a bug or a gene or a particular endocrine factor, and that the solution is a relatively simple intervention, often a drug. And that has proven to be really a good business model for the pharmaceutical industry and, to some extent, the medical industry, which has done pretty well over the last four or five decades.

And I must say that for a lot of people with medical problems, this has worked pretty well. I think if you had gotten HIV in the 90s, you certainly were better off than if you got it in the 80s. And if you have cardiovascular disease today, you’re certainly much better off than you would have been 30 years ago. And that’s true now, fortunately, for some forms of cancer as well, where we’re seeing remarkable progress with new diagnostics and new treatments.

I just don’t see the same sort of breakthroughs and the same opportunities yet for people who have PTSD, depression, OCD, a range of mental disorders. It feels to me like that medical model has helped some but not enough in the mental health field. Part of why I wrote the book was to try to understand why we haven’t made more progress. And part of that “why” goes right to that issue that the simple bug-simple drug model that has been so effective in the world of infectious disease is really not so effective for the millions of people who have a mental disorder.

TI: In some ways! It’s a mixed bag. I think that we learned an enormous amount, but I would say that it’s still very much in process. I don’t think we’ve fully gotten the return on the investment. I think we will, and that science is going to be really critical for us in trying to go deeper into understanding these disorders.

The problem for me was that—and this is just a personal reflection and is not in any way an indictment of the NIMH—but when I look at this state of care and what’s happening for most people, particularly those with severe mental illness, with schizophrenia, bipolar illness, severe depression, severe PTSD, it’s not a scientific problem these people face.

They face incarceration. They face homelessness. They face this massive injustice in a kind of crisis-driven system that actually leads them out of the care system and into these other pathways that are often deadly and certainly unfair, generally punitive, and not compassionate. So, that’s not a NIMH problem.

For me, so much of the sorts of public health problems that we’re facing aren’t really about genes or neuroimaging or the science. It’s more of an almost, and I loath to use the term, but really a social justice issue. And what we are seeing is this egregious set of policies that lead to people with mental illness going everywhere except into compassionate care because there’s no capacity for that.

So it actually had nothing to do with NIMH. I left NIMH and kind of never went back because if you want to address those issues, you’ve got to go someplace else with a very different army. And it’s not the army of neuroscientists and those who are brilliant in the fields of genomics and data sciences. It’s an army that is really willing to take on those big social problems and begin to deal with them.

And I think we know what to do. I think we know how to do that, and that’s beginning to happen. But my goodness, it’s not going to happen through NIMH funding. It’s just not their job. That’s something very different from the world that they’re focused on.

TI: That’s right. A crisis of care. It’s not really a crisis of science. It’s not because we don’t have good research or that we’re not spending the research dollars correctly. I argue, actually, that we probably need more research, more science, more funding for NIMH.

You know, we always need better treatments; we always need new diagnostics. But let’s get real here. We haven’t been implementing the things that we discovered 30 years ago. NIMH spent a huge amount of money in the 80s and 90s on the Nurse Home Visitation Program. I write about this a lot in my book because I think it was just a brilliant investment.

But it’s not a research question anymore. We don’t need to put a lot more NIMH dollars into that. We need to implement this for millions and millions of families who are disadvantaged and who need that kind of support, because we know it works.

I don’t want to see us get caught up in this academic cycle of “let’s keep studying this problem.” At some point, you have to try to solve the problem and not just study it, and that was what led me moving from this kind of research career to a career that was much more about advocacy, policy change, about making sure that we were starting to invest in the kinds of services and broad social supports that we need and sadly lack in this country.

TI: Yeah, exactly. I think where I ended up, and it’s so interesting when you write a book like this; you think you know what you’re doing, but you have no idea. You usually end up someplace very far away from where you started, and that was exactly the case here.

I started this book when I was working at Google, where I was trying to develop really interesting ways of digital phenotyping. I was convinced that technology was really going to transform mental health care, and I still think that’s probably true. But I ended the book by realizing that the problems that we’re focusing on are really problems of mental health. That’s very different from mental health care. And I have to say, I don’t think I understood that.

When I started the book, every conversation I had about health or mental health was about health care or mental health care. And it wasn’t until I was two-thirds of the way through this, and in this odyssey that I took around the state of California to try to understand why we hadn’t seen more improvements in public health measures like morbidity and mortality, that I began to realize, like, wait a minute, this is not a health care problem.

All this stuff, incarceration, homelessness, poverty, health disparities, is happening way outside of healthcare. It’s actually something very different. We could probably fix healthcare. We could probably do so much better on health care, but barely move the needle for morbidity and mortality.

As an example, I was just looking at this over the weekend: the chances of turning 70 years old or living to 70 in terms of life expectancy are at about 82% for White females and about 54% for Black males in the United States. That 82% to 54% disparity is not really a function of what medications they’re on or how many clinic visits they have, or even what health insurance they have. That contributes a little bit, we think it accounts for maybe 10% or 20% of that disparity. But most of the disparity in race- and gender-based mortality in this instance is really about your zip code. It’s about your lifestyle, your exposure, your environment. It’s about a lot of other stuff that’s not really in the healthcare system.

I guess the really hard question to ask, and the one that I’ve been thinking a lot about lately since the book came out is, do we need to rethink what we mean by health care? And specifically, do we need to rethink what we mean by mental health care? Is it really just about medication and psychological treatments and maybe some rehabilitative care? Or is there something more essential that has to do with recovery, has to do with thriving, has to do with wellness? Does that need to come into focus, and does that need to be within the scope of what we mean by healthcare?

TI: I struggled with that in the book. I start from a perspective that psychotherapy is a really powerful intervention and that we have specific, skill-based therapies that have been demonstrated to work. I also understand that outcomes may depend more on the therapist and the therapy, and that’s always a challenge in any kind of randomized clinical trial that one does on these interventions. But the evidence is pretty compelling for both the safety and ultimately the effectiveness, which is quite different from the efficacy of psychotherapy.

So the question is, with a treatment that’s so powerful, why have we seen this gap, and why has it become so difficult to actually get it delivered in the way that it should be? I think there are a couple of things. One is, we need to look closely at the training of psychotherapists, how we do it, where we do it, and also when we do it. We’ve had this notion that you train, and then you have supervision for a period after graduate school, and then you’re kind of on your own until your next licensure comes up.

I think we want to look more carefully at how we make sure people get the kinds of skills and the feedback to get better and better. I’ve been fascinated by a company with which I have no connection but am really intrigued by, called IESO. It’s not in the United States, it’s just in the UK, but they’ve really focused on, how do we help our therapists who are online to get better and better?

They’ve built this natural language processing engine so that every interaction between therapist and client is captured. It goes through this engine, and they have a dashboard that shows them levels of therapeutic rapport, levels of effectiveness of their comments, and also the state of play for the client; better, worse, what’s the emotional tone in the interaction? It’s really fascinating to watch.

But what’s amazing about it is that by getting this kind of real-time feedback, therapists have gotten better and better. And when you look at outcomes, they went from 49% recovery to 67% recovery just by providing this real-time feedback, not just to patients and clients, but to therapists themselves. It was actually more useful for the therapist than the client. But ultimately, the clients enjoyed that impact.

So I think part of what we need to do is to think about how we help our therapists to navigate and to improve what they do. The other part is we have to ask, what do we pay for? Are we paying for a number of hours spent, or are we paying for outcomes? Basically, are providers being rewarded for how long somebody stays in treatment, or for getting people out of treatment and getting them well? We need to begin to look at the incentives that are built into the system and ask, are we incentivizing for the right things?

TI: No, it’s neither. There’s nothing biological here. It’s really taking language and decoding it. If you think about what we do in psychotherapy, it’s listening, it’s observing, it’s communicating. And through that, we hope that there’s understanding and trust and change ultimately through the relationship.

That process of using language to communicate is a process which has really been revolutionized by artificial intelligence and very good data science through this thing called natural language processing, which was created to try to understand how words got glued together and what coherence looks like in language.

But over time, it’s been used to measure sentiment, like mood, and is now being used to measure how well people are connecting and if they’re communicating effectively. This is a multi-billion dollar industry that’s been taken over largely from the call centers. Call centers are now far better than they were five years ago because of the ability in real-time to decode the communication between two people.

What IESO has done is to take that same kind of effort and said, “Let’s provide objective evidence about how a therapist and client are communicating and relating and actually literally measure things like trust and measure therapeutic alliance.” And they found ways to define that, which I think are really interesting.

It may not be for everybody, but it is fascinating to me that by capturing that kind of data objectively, they have been able to provide a source of feedback that actually helps people do what they’re trying to do, which is create trust, create the therapeutic alliance, build that rapport. Who would have thought that you would actually do that through technology?

And yet, they’ve demonstrated that this can work without any burden on either the provider or the client. It doesn’t take any extra time. It’s kind of like the speedometer in your car, you know, it’s a part of the dashboard, it tells you as you go how fast you’re going and how you’re driving.

TI:  It is. I guess I would just push back with the word “just,” because I think we need both. We need both that subjective experience, like, how was this for you? And then, you know, the objective readout of what does the algorithm say? And it may be in the gap between those two that there’s a lot we can learn.

There’s this really interesting new science that is just beginning to shine a light on our behavior, and particularly on our language in a way that I think will revolutionize psychotherapy; it will revolutionize the study of mood, behavior, and cognition. I really think we’re just beginning to see that happen.

One kind of untapped example of this, which I’ve been so intrigued by but haven’t yet seen really developed, is that you can use this natural language processing approach to measure the coherence of speech, because every two words have a vector that attaches them. So if I use the word “dog,” it’s not unlikely that the word “bone,” or the word “cat,” or the word “food” would come up in the same phrase, right?

But the word “algorithm” or the word “church” may not be as easily associated as that. And so by measuring what we call semantic coherence, the likelihood that words could come together or maybe wouldn’t be found together, you get a sense of how people are thinking and how things get put together. In contrast, great poetry often has longer vectors, less coherence.

But as people become psychotic, for example, this is a very sensitive way of picking up thought disorder. And you could say, “Well, yeah, but you could just listen to them and know that’s happening.” Maybe, but how helpful would it be to be able to say, “Well, their coherence moved from 0.6 to 0.74.” Or to be able to provide a tool so that a nurse in an emergency room in a rural community, who really isn’t trained to do a lot of the assessment of thought disorder, would be able to say, “Well, according to this tool, this person’s semantic coherence is about 0.68.”

In understanding thought disorder and psychosis, for example, it provides an objectivity that we’ve come to expect for assessing diabetes or hypertension. It gives us a number which is reproducible and which ties back to something that’s truly actionable because based on that number, you might decide “this person is, in fact, currently psychotic and needs to be treated along this pathway,” versus “this person is a very good poet who tends to put ideas together that are very creative and that are different, but this is not necessarily pathological.” So I think we’re at the beginning of a revolution in our ability to add objective measures to what we are currently and have traditionally done just subjectively.

TI: Actually, the subjective experience may be what really counts or is far more important. But that’s why I brought up the IESO example, because I think there is an opportunity for technology to improve the quality of what we provide in the psychotherapeutic relationship.

It may turn out that we don’t need that. But I think the data would suggest that there’s room for improvement. And, to be fair, there are people who are just naturally gifted as clinicians and who just have the ability to do this without a huge amount of training and without needing many years of experience and probably won’t need that kind of a tool.

But there are a lot of us whom I think would benefit from getting that continual feedback in a way that’s passive and ecological, because it’s done within the hour. It’s not, you know, in a supervisory hour. And it gives you a sense of something that is probably fundamental to the treatment process, which is the development of a therapeutic alliance.

TI: I think that’s a really key question, and it’s one that I also struggled with in working on the book. I’ve spent four decades making the argument that these emotional and behavioral problems are medical problems. And I ended up in the book saying, yeah, these are medical problems, these are brain problems, and they deserve the same reimbursement, the same rigor, the same science that we would expect for any other medical problem.

But the solutions are much broader and much different. The solutions are relational, they’re environmental, they’re political. We have to really widen the lens here if we want to begin to have the impact that I think all of us care about, particularly at a population level, and the medical model just isn’t really built for that.

I talk a lot in the book about—and to be fair, you’re right, this is more about serious mental illness—but I talk a lot about recovery. And I have to say, I was not the person pushing the recovery model. I sort of see there’s a medical model and a more recovery relational model. I think we need them both, but the recovery model, to me, is really defined by these three P’s that I talk a lot about in the book: people, place, and purpose.

If we really want to think beyond just symptom relief and we want to see people thrive, we want to see them recover, we want to see them have a life, then we have to be thinking about more than the medical model. We have to be thinking about, how does someone with a mental illness have a shot at getting the things that all of us want? Social support—that’s the people, a safe environment—that’s the place, and a purpose—a reason to recover, something that they wake up for, something that they see as a mission.

We don’t do that in the medical model. That is not what we mean by mental health care in 2022. And what I’m arguing for in the book and in trying to start this kind of new social movement around mental health is that we just take on a broader perspective that says, actually, we should reframe what we mean by care, and the care should include the three P’s, that providers ought to be able to write a prescription for housing, and we ought to expect Medicaid to pay for a clubhouse which provides the three P’s every day for people with serious mental illness.

We need to think about how we get beyond this simple idea that there’s a magic bullet intervention, that if we get just the right pill to just the right molecular target in just the right patient, we’ll solve this problem, because that’s probably not ultimately the way we solve this problem. It’s going to be actually from multiplexing the problem or thinking about people, place, and purpose and providing a much broader range of care, not a more narrow focus on medication.

TI: I guess the argument is we need both. I think about psychotherapy as learning to play the violin. You’re learning a skill. It takes time, it takes practice, and it often usually takes a really good teacher. But that’s really hard to do if you have a bad tremor. So, I’d start by treating the tremor so somebody has a decent opportunity to be able to actually learn how to play the violin, but I wouldn’t stop with treating the tremor. I think that is a part of it. You need both, and you need to be able to do both over a long period of time.

And I guess what I feel really strongly about is two things. One is that our field has been, unfortunately, very fragmented between medical approaches and psychological approaches. The science says that the two of them together are better than either one alone. And yet in practice, we rarely see them combined in a way that’s most effective for patients or clients. I think that’s something we need to fix.

But the second part of that is, we often don’t pay for this in a way that it merits. There’s a tendency, I think, by both public and private payers to undervalue the treatments. It often is easier to pay for the medication because, by the way, they’re almost all generic, super cheap, it’s easy to write a prescription, and payers are very comfortable with that. It’s harder to require the combination and to be able to pay for the combination.

It’s so funny, I was just in a conversation about the use of psychedelics. And if there’s one area today where everybody is thinking, “Oh, this is the new…” you know, it’s very hyped. “This is the new magic bullet,” that psychedelics are really going to matter. Again, it’s just one more pill that you can take, and you’ll be able to play the violin.

And yet, what’s so interesting is when you talk to people in that space, they talk about psychedelic-assisted psychotherapy. It’s so refreshing. It’s the first time in 40 years I’ve heard people committed to combining medical and psychological approaches in a way that’s really thoughtful and potentially very impactful. It’s such a paradox, with all the hype around taking the magic pill. That is actually the place where we may find and understand the importance of combining the two therapies.

TI: I don’t remember saying that, but one of the things that I tried to convey in the book is that the process of change is also a process of neuroplasticity. And the idea that there are medical treatments that affect the brain, and then there’s psychotherapy that affects behavior, is really probably grossly simplifying. It’s very likely that the change that occurs with medical treatments partly relates to opening people up to behaving in different ways and exposing them in new ways.

TI: Which changes the brain. And likewise, that going at this from a psychological perspective also changes the way people think, changes the way they behave, which also changes the brain.

To go back to my violin analogy, when you learn to play the violin, you wire your temporal cortex. There’s no way around that. We have to begin to think a little more mechanistically about what actually happens with behavior change and to realize that behavior and the brain are inextricably linked together in ways that we can’t often see, but we have to accept.

TI: I like that idea, Lawrence. We have to get out of our sort of tribal approach to this. It’s so frustrating, and I kind of understand it, you know, it’s where people come from, it’s their identity, but what if we flip the narrative and say, “What’s most helpful?” What actually helps a 14-year-old with anxiety or a 24-year-old with psychosis? It’s not about our role. It’s not about our skill set, necessarily. I mean, we have to think much more broadly about putting all of the tools in the toolkit together in a way that serves that person in a way they will want and accept it.

We haven’t been very good at that. I mean, even the very fact that we built a care system that’s really built for payers, to some extent, for providers, but not for the consumer. And it’s one of the reasons why I think we get very low engagement. Only about 50% of people who should be in care or could be in care and would benefit are actually buying what we sell.

Bridging the Divide

I think the next decade is an opportunity to say, “Can we meet them where they are?” Particularly for young people. They’re not likely to show up at a brick-and-mortar office. They are likely to be on TikTok or Discord, or now maybe even Twitch. I mean, there are lots of places where you find them. Is there a way to meet them there? Should we rethink the mental health care that we want to deliver so that it’s much more person-centered, more culturally sensitive and adapted, and begin to understand that what we’ve been doing hasn’t really worked for a lot of what we had hoped it would? Yeah, we have great treatments, we have great skills, we have something that really is useful, but it’s not getting the people in the way they want it. Particularly, I would say, for communities of color, LGBTQ communities, I mean, there are just lots of people who feel on the outside and who see mental health care as we built it as not friendly and not matched to what they’re looking for.

This is a place where I think technology can make a big difference. It can help us to democratize care and give people choices that they haven’t had, particularly people who are in rural areas and underserved communities. People who feel that, for whatever reason, they’re part of a small niche in society that’s been underserved. I think now is the time we can say, can we create a different platform, meet people where they are in the ways that they would want to be engaged, and give them something useful?

I guess in some ways, helping people with mental illness is a little bit like what we learned with COVID, where there was this gap between creating vaccines and delivering vaccinations. I think NIMH and others have done a spectacular job of creating the equivalent of vaccines for psychological treatments, for medical treatments, and for people who struggle with emotional and psychological issues. We haven’t been so good at delivering the vaccination part, actually delivering these in a way that people want them and can use them and can benefit. I think that is the challenge for the next decade.

TI:  I think it’s already happening. In my career, I’ve never seen the kinds of transformations we’re now witnessing—and I don’t think that’s too strong of a word, it really is a transformation of this workforce and care system. You have the aggregation of large numbers of private practice psychotherapists into these massive groups, and there are companies that have gotten very wealthy through doing this. Lifestance and Uplift Health are doing a piece of this in several states. It’s very interesting. It’s changing the culture of how people practice. It ultimately will provide them with resources, as they get in group practices that will make their jobs in some ways more effective and hopefully easier.

You also have the advent of teletherapy on a big scale. Last year $5.1 billion was being invested in mental health startups. How amazing is that? You’ve got hundreds of new companies starting off. Eight of them are already unicorns, meaning they’re valued at over $1 billion. You have a company that I find really interesting, Cerebral, that’s a little more than two years old. It started at the beginning of the pandemic. It’s arguably one of the largest mental health care providers in the United States today. They have many, many thousands of providers. They talk about having served 350,000 clients in the last two years.

So, we’re going through this massive change. I don’t know where it’s going to end up, but I would imagine many of the people who are listening, who are in private practice, are thinking about, should I (and maybe they already do) work for Talkspace or Cerebral or Lyra or Ginger or Modern or Better Help. I mean, there’s so many of them that are hiring. In a way, it’s sort of an invitation to a new economy, a gig economy, just like we saw for Uber. People are having opportunities. They have a lot more possibilities of what they can do and how they can spend their time and work.

I don’t know how this is going to end up, but I guess the question I’m asking myself, again, going back to what does this mean for the 14-year-old with anxiety or the 24-year-old—

TI: So, are they better off or worse off at the end of this? Or is there no change? I do know that there are now startups that are just for African American male therapists so that African American male clients who are looking for that can find it.

So I think it’s early. I always say this, Lawrence, this whole transformation of mental healthcare delivery is probably a play in five acts, and we’re in Act 1. In Act 1, we’re getting to see who the main characters are; we’re trying to solve the problem of access. And by the way, we’re starting to address some of the conflicts and some of the problems that are coming up.

I think Act 2 is going to be really interesting. I think it’s going to be more about improving quality and starting to find ways of measuring outcomes and all of that. We’re not there yet. It’ll be really interesting to see how that works out.

But what a fascinating time to be in this field! It’s all changing very quickly. In 2027, you know, five years from now, I think we’ll be having a really different conversation. I think the access issue may be largely fixed through the democratization of care and through the fact that it doesn’t matter where you live or what your race or ethnicity or zip code might be, you’ll be able to find someone who can help or someone who has at least signed on to help who looks and talks and maybe even understands you in a way that might be hard to do today. The question will be, can they teach you to play the violin? Do they have the skills and the experience to be able to do this well?

TI: Well, the NIMH in 1970 or 1980 would have done that. But in 1990 or 1991, there was a fissure and the federal government created SAMHSA, the Substance Abuse Mental Health Services Agency, and they said to NIMH, “Going forward, you’re like any other NIH Institute. You’re just like NIAID or NINDS. Your job is science. You’re a research agency. We don’t want you to get involved in service delivery. You shouldn’t be thinking about that. That’s SAMHSA’s job.”

The reality is that SAMHSA is still a fairly small agency. The federal government still, it’s changing a little bit, but largely has delegated to states and counties the provision of mental health services. So what you get for mental health care is going to be very different depending on where you live, what state, which county—

TI: Yeah, but there’s still a large investment. I look at what we’re doing here in California, with $4.4 billion now dedicated to youth mental health, the transformation of the Medicaid system, the development of the Mental Health Services Act—it’s this millionaire’s tax that pays for mental health care. This year that will generate about $3.7 billion for mental health care in the public sector. There’s a lot of stuff you can do and a lot of stuff that’s happening.

I wouldn’t lay this on NIMH. Really none of this is their job. On top of all that government spending, last year we had $5.1 billion coming from the venture capital industry invested in startups. That’s two and a half times the size of the NIMH budget.

So there’s a lot of investment, a lot of money being pushed into the system right now. We just need to make sure it’s going to the right things and that we’re holding funders and beneficiaries accountable for results. So that it’s not just pouring money in and not actually seeing changes in outcomes, which, at the end of the day, that’s what we care about. We want to make sure that, in fact, the rate of suicide is coming down, the rate of employment is going up, kids are finishing their education. It’s not just measuring PHQ-9s [a depression questionnaire]. It’s actually knowing that people are beginning to recover and function in a way that we haven’t been measuring and we certainly haven’t seen over the last 30 years.

TI: Gosh, I have to think about this for a moment. It probably should say something like, “He Served in the Golden Age,” because this was just an extraordinary moment to be leading this research effort and to see where the science could take us in terms of understanding the brain and health and disease.