stanford medicine


Brain power - Psychiatry turns to neuroscience

Special Report

Brain power

Psychiatry turns to neuroscience

My head is clamped into a padded capsule to keep it stable as researchers measure my brain activity.


l am lying on my back inside a functional magnetic resonance imaging machine, and any movement could muddy the results. There is a strap around my chest to record my breathing rate. Sensors on my fingers monitor my sweat — and I fear that I am sweating a lot.

On a screen in front of me flash a series of strange black-and-white photographs of smiling and terrified faces. Each has the word “happy” or “fear,” in red capital letters, superimposed across it. Sometimes the word matches the expression; sometimes it doesn’t. I am supposed to push one of two buttons if the face looks fearful, and the other if it’s happy. The task sounded easy enough before the scan started. Now I find myself freezing, struggling to identify an expression that is at odds with the word. A new picture appears every few seconds, and I am beginning to feel dizzy, my eyes starting to water.

Dozens of people have done this particular facial-identification exercise with an fMRI machine measuring changes in the blood flow inside their brains. The data accumulated from their responses is revolutionizing how we define and treat anxiety, depression and other emotional disorders. It’s part of a new wave of fMRI studies, genetic research and biomolecular work that are grounding psychiatry in neuroscience, a longtime yet elusive goal for many psychiatrists, since the start of the profession in the late 19th century.

At present, the diagnosis of mental illness and its treatment are based almost entirely on clinicians’ observations and their patients’ reports. But many psychiatrists yearn to identify the biological etiology of mental illness — to pinpoint how abnormal brain function causes psychiatric symptoms — and to develop treatments targeted to fix these broken mechanisms.  It’s the difference between feeling a patient’s forehead to check for fever or doing a blood test to identify the bug that’s causing the high temperature so you can prescribe an antibiotic.

“The current approach never gets to the brain mechanisms that lead to symptoms — there’s no external validation,” says Amit Etkin, MD, PhD, who with another researcher devised the test I am taking in the fMRI. “We’re changing that.”

For roughly a decade, Etkin, 35, an assistant professor of psychiatry and behavioral sciences at the Stanford University School of Medicine, has been pioneering the use of fMRI scans to reveal how brain activity differs in healthy people compared with those suffering from mental illness. The face-expression test I am taking is one of several innovations that have allowed him and others to detail the regions of the brain that function abnormally in people with anxiety, post-traumatic stress disorder and depression.

“Amit is a poster child for how neuroscience can inform psychiatry,” says Nobel laureate Eric Kandel, MD, PhD, a neuroscientist and psychiatrist who also happens to have been Etkin’s graduate advisor at Columbia University’s medical school. “By turning to imaging early on, Amit showed that we can use it not only to study normal responses to emotion but also abnormal responses — conscious and unconscious — and start to localize where in the brain they occur.”

The effects of neuroscience on psychiatry, of course, go beyond Etkin’s work in imaging anxiety-related disorders, ranging from the improved knowledge of the neuropathology of many other mental illnesses such as addiction [see story Neuroscience of need] to the development of such new treatments as transcranial magnetic stimulation and deep-brain stimulation for depression [see story Positive charge]. Etkin’s work is also part of a broader movement to develop new ways to provide therapy to people with mental illness [see story Autism answers].

There certainly is a need for some new approaches. Mental health disorders are, by some accounts, the leading cause of disability in the United States and Canada, and the U.S. Centers for Disease Control and Prevention estimate that a suicide occurs every 15 minutes. According to one  recent federal report, about one-half of Americans will have a serious mental health condition during their lifetimes, and at present, fewer than 50 percent of those with such conditions are receiving treatment. Anxiety is the most pervasive of all mental disorders, says Etkin, and his work has been critical to redefining it in biological terms.

I wanted to understand how Etkin’s work is changing the way that psychiatry looks at mental illness, and so on this February morning I am inside the fMRI machine, hoping to grasp what exactly Etkin’s face-expression exercise elicits.

I cringe at incorrect responses that happen before I have time to think, and marvel at the many correct ones that seem to occur automatically. I am relieved once I’m done, though I’m worried about what my results will reveal. These scans are not accurate enough to be diagnostic, but still I fret: Are the images going to show a brain rife with anxiety?

Are my fMRI pictures normal?


“Why should psychiatrists care about neuroscience?”


That’s the title of a talk that Etkin gives at the start of a new neuroscience course for third-year psychiatry residents at Stanford that he has developed with the support of department chair Laura Roberts, MD. The lecture begins with a blunt discussion of the state of the profession. “Turns out we’re not terribly good at treating psychiatric diseases (worse, likely, than we have thought),” reads one PowerPoint slide. Another bullet point notes, “We also don’t understand how our current therapies work, for whom are they likely to work, and why.” Next comes one word: “Yikes!”

Over the last half-century, other fields of medicine have experienced a decline in mortality and disability rates as new devices and treatments have been developed, but psychiatry has lagged behind, says Etkin, who is also an investigator at the Mental Illness Research, Education and Clinical Center at the Veterans Affairs Palo Alto Health Care System. For instance, while the number of drug classes for cardiovascular disease has increased since the 1950s from two to 16, the number for depression and schizophrenia has barely budged, increasing from four to five. To make matters worse, current psychiatric treatments aren’t terribly effective. A definitive study on schizophrenia treatment shows that people with this disease typically go off their medications after six to nine months because they were not sufficiently effective, and the side effects are dreadful. A major investigation into the efficacy of several antidepressants revealed that only 43 percent of the study participants got well and stayed well.

‘Turns out we’re not terribly good at treating psychiatric diseases.’

The comments Etkin received from residents after the lecture indicate that it was an eye-opener for many, with at least one finding it downright discouraging. (“This is a real downer,” read the comment on the feedback form.) “I try not to paint a horribly bleak vision,” Etkin says, “but this is reality; they need to be in touch with it.”

Etkin’s talk is based on a similar presentation given by National Institute of Mental Health director Thomas Insel, MD, who also does not mince words about the present state of affairs. “Current treatments are not effective enough,” he wrote last year on his blog. “Briefly stated: in many cases patients receiving the best of current care are not recovering.”

This urgency to develop new treatments comes amid worries that psychiatry has become less appealing to many young doctors choosing a specialty. From 2000 to 2008, the number of psychiatry residency graduates declined from 1,142 to 985, according to a study last year in Academic Psychiatry. Over the last few years, the trend appears to be improving but nowhere near fast enough to keep pace with the need — or the increase in many other medical disciplines. From 2007 to 2011, for example, there was an increase of 40 first-year resident positions in psychiatry compared with 319 in emergency medicine.

The difficulties in attracting new talent can also be chalked up to other problems. For one, the financial rewards of psychiatry have diminished relative to many medical disciplines. While the field attracts young doctors excited about psychotherapy, they have fewer opportunities to be reimbursed for such work as less expensive practitioners, such as social workers and counselors, increasingly take the cases. And questions about the effectiveness of blockbuster psychotherapeutic drugs, once promoted as cures for all, have undermined the profession’s reputation and led to charges of undue influence by pharmaceutical companies.

“The profession,” says Insel on his blog, “is struggling with its identity.” Insel believes the way to resolve that is through a greater focus on neuroscience, which, he says, will draw a new generation of psychiatrists. A nationwide survey that Etkin helped conduct for the NIMH confirms its allure: Roughly nine of  every 10 residents agree that there should be more neuroscience in their training.

In response, the NIMH is offering psychiatry residents the opportunity to enroll in its annual Brain Camp, which instructs them in the most recent findings from cognitive science, neuroscience and genetics. The Yale School of Medicine revamped its curriculum to emphasize neuroscience when studying psychiatric cases.

And at Stanford, Roberts and Etkin have ushered in the neuroscience course, which is now in its second year. Beyond this, Roberts is introducing subspecialty clinical training  programs that encourage young psychiatrists to develop deep expertise in addiction, psychosomatic medicine and forensic psychiatry, along with the existing subspecialty of child and adolescent psychiatry. No matter what their specialty, she wants Stanford psychiatry residents to become literate in fMRI results, genetic tests and molecular biology, and the neuroscience class is a step in that direction. “We have to help bridge the disconnect between the latest research advances and what occurs in the everyday practice of clinical psychiatry,” Roberts says.

‘We need to hug the sciences, as that’s the key to the future.’

Roberts sees the neuroscience class as evidence of how psychiatry is moving beyond the historic split between two camps. On one hand are those espousing biological treatment (she sums up how it’s been viewed as “up-to-date and scientific” but also may be perceived as “reductionistic, impersonal and tainted”). On the other hand are those favoring psychotherapy (it’s been seen as “compassionate and valuable but also may be perceived as unproven, touchy-feely and ‘old school,’” she says). She believes that advances in genetics and neuroimaging are already bridging the gulf: Science and therapy can be used to make each other more effective, she says.  “We need to hug the sciences,” she tells me in an interview, “as that’s the key to the future — that’s where we discover new ways of understanding neuropsychiatric disease as well as its prevention and optimal treatment.” 

But by treatments, she doesn’t just mean medications. “I can’t imagine a future where psychiatry does not also involve therapy,” she says. “Psychiatry involves the underlying therapeutic relationship, as well as the insights drawn from the neurosciences; it’s the combination of the two that will lead to better outcomes in the future.”


A week before doing my fMRI, I go online to try a series of computer games that Etkin is testing for the treatment of depression and anxiety. One is an arithmetic challenge requiring me to solve equations before they disappear from the screen; some involve quickly evaluating facial expressions; another requires that I click on bubbles floating around the screen that have a word describing a positive emotion. I feel some pleasure as I hear the “pop” from selecting “jubilant” and “love,” while allowing bubbles for “fury” and “sulky” to drift away.

It takes about 40 minutes to complete these exercises, which were created by two companies. There are many more such games that have recently emerged as part of a new therapeutic approach known as cognitive bias modification, which aims to change behavior and rewire the brains of people who are depressed or anxious.

While some suggest that CBM might substitute for psychotherapy, Etkin does not believe that computers are going to replace therapists. “The utility of psychotherapy is proven without a shadow of a doubt,” he tells me, and neuroscience research is beginning to show why this is so. According to a review in Psychiatric Times in August 2011, there are at least 19 imaging studies that show psychotherapy alters brain function in patients suffering from major depressive disorder, obsessive-compulsive disorder, panic disorder, social anxiety disorder, specific phobias, post-traumatic stress disorder and borderline personality disorder.

The idea is to see if the exercises enhance activity in the regions of the brain critical for emotion regulation.

This doesn’t mean, however, that psychotherapy alone is the answer. As Etkin notes, it’s not an option for most people because insurance doesn’t cover it, and there are not nearly enough practitioners to serve all the people who might avail themselves of it.  What’s more, many people don’t want to do it.

So Etkin and others are looking for alternatives, including the computer games. In the study now under way in his lab, people with depression and anxiety undergo fMRIs before and after weeks of playing these games. The idea is to see if the exercises enhance activity in regions of the brain critical for emotion regulation, just as an athlete’s workout might be designed to build key muscles.

“If you do curls over and over, your biceps get stronger,” says Anett Gyurak, PhD, a postdoctoral scholar in Etkin’s lab who is overseeing the study. “So we’re trying to do that — train the muscle for emotion regulation.” The results so far suggest that changes are, in fact, occurring in both brain activity and how people feel; future research could help develop even more effective exercises. Indeed, Gyurak, Etkin  and psychiatry professor Alan Schatzberg, MD, have just filed a patent for a game that they believe is better designed to stimulate the neurocircuitry involved in emotion regulation.

I have no idea that I have just completed a vigorous workout, but Gyurak assures me that this lack of awareness is part of the plan, that the exercises help reframe emotion processing without your knowing it. A key aspect of Etkin’s research is that he’s focusing on “implicit” emotion regulation — a different way of thinking about what Sigmund Freud referred to as the unconscious. The point is that the neurocircuitry underlying psychiatric disorders operates without our knowing about it; these disorders typically involve deeply ingrained, dysfunctional emotional habits. One of the notable advances in the neuroscience of psychiatry, thanks in part to Etkin, is that we can now image — and identify — the neural networks behind such implicit emotions and thus see how interventions, whether they be drugs, psychotherapy or computer games, may change them.


“I came into psychiatry as a neuroscientist,” says Etkin, who has a doctorate in neuroscience as well as an MD. “It’s a perspective that serves me well.” In 2010, when he joined the Stanford faculty, he established his lab, which now has 23 people and grants totaling $4.25 million. While his job is technically 100 percent research, he sees patients every Tuesday afternoon. The son and grandson of scientists, he is also married to a psychiatrist. On a wall in their house they have his-and-her fMRI scans of their brains.

Etkin is not the sort of professor you would likely have found in an academic psychiatry department a few decades ago. Back then, the research tools were still too rudimentary to shed light on how feelings and behavior are rooted in the brain; besides, many in the psychiatric establishment viewed such research as irrelevant: The thinking was that social and environmental factors were at the heart of most mental illness.

But by the late 1990s a sea change was beginning, with the advent of new imaging techniques, developments in molecular biology and the promise of genomic sequencing. Etkin was swept up in the tide. He chose to go to Columbia so he could study with Kandel, who in 2000 was awarded his Nobel for showing how molecular changes in synapses lead to the formation of memory. Kandel has been at the forefront of calling for a rapprochement between psychiatry and biology, particularly neuroscience. In 1998, around the time that Etkin was joining the lab, Kandel published what could be considered a manifesto — “A new intellectual framework for psychiatry” in the American Journal of Psychiatry — for today’s effort to infuse psychiatry with neuroscience. Kandel called for a new generation of neuroscientists to join with psychiatry to develop a neuropathology of mental illness, a change that seems to be well under way. “I used to get MD/PhD students going into neurology, but now they’re going into psychiatry,” Kandel says. Etkin was one of his earliest recruits. He had, in fact, planned on being a pediatric neurologist, but Kandel inspired his switch.

In 2002, when Etkin began working with fMRI, scientists were still figuring out how to image brain activity relating to anxiety and mood disorders. (The first fMRIs of humans were done in 1992.) The usual approach involves having study subjects inside the fMRI perform a particular task so scientists can determine how that particular stimulus changed activity in the brain. Earlier experiments had suggested that the amygdala, a pair of almond-sized bundles of nerve fibers in the middle of the brain, is activated when people are anxious. Etkin and his colleagues, though, believed the amygdala was only one part of the process. They wanted to document in greater detail how the brain regulates — without any conscious awareness — the emotional conflict underlying anxiety. In a nutshell, he wanted to answer this question: What is different in the brains of people who can implicitly regulate their feelings of anxiety as compared with those who cannot?

Etkin was looking for a task that would elicit the unconscious brain response to emotional conflict, and the solution arose as he was riding a bus across Manhattan to meet his wife. On the seat next to him was a colleague, Tobias Egner, PhD, who had been looking at means of testing in fMRI studies how the brain deals implicitly with cognitive (not emotional) conflicts. Egner, now an assistant professor in the psychology department at Duke University, told him of a classic technique for studying non-emotional conflict — the Stroop task — first identified in Germany in the late 1920s (and later made public in the United States by the psychologist John Ridley Stroop). It involves asking a participant to identify the ink color used on cards with two words “Red” and “Green.” Sometimes the color matches the word, sometimes it does not. What researchers established is that when forced to resolve a conflict —  when the word and color don’t match, or are incongruent — study participants take longer to answer. Yet further research indicated that when participants were shown two consecutive incongruent images, the response time typically improved. This demonstrates how the brain, without our being aware of it, is implicitly primed to resolve a cognitive conflict and thus gets faster at it.

Stroop-like effects occur in a multitude of common tasks. Take a car that’s skidding on ice. The initial reaction for those new to driving in winter is to steer the car in the opposite direction. It takes mental effort to do the right thing and to turn the car into the skid. After doing it once, however, there’s less hesitation about making such a move the next time.

Etkin and Egner, under the direction of senior scientists at Columbia, wondered if they could apply the Stroop paradigm to assessing emotional conflict.

Etkin selected facial expressions and typed the words “happy” and “fearful” on top of them. What happened when he used these in an fMRI study was more than what he and Egner had hoped for: After being shown two consecutive incongruent images, the participants, who did not suffer from a psychiatric disorder, activated a select part of the prefrontal cortex, never previously associated before with emotion regulation, and they reduced amygdala activity. In other words, the activity in these regions, as well as some other regions described in the study published in 2006, appeared to be responsible for the sort of implicit emotional regulation that prevents anxiety.

Further work was needed, however, to prove that the interplay among these regions is linked to anxiety. So after moving to Stanford, Etkin and colleagues extended his Stroop testing to people who met the diagnosis for generalized anxiety disorder. The results, published two years ago, were starkly different from those of the healthy participants. When shown consecutive incongruent images, the participants with GAD had little prefrontal cortex activity and no dampening of activity in the amygdala. Moreover, the GAD participants’ response time did not speed up when shown consecutive incongruent images.

Etkin selected facial expressions and typed the words ‘happy’ and ‘Fearful’ on top of them.

“The robust group differences seen at both the behavioral and neural levels,” the researchers wrote, “suggest that the inability of patients to adapt to emotional conflict is an important aspect of the pathophysiology of generalized anxiety disorder — and potentially of other psychiatric disorders — and thus merits continued, deeper study.”

The fMRI results of the Stroop test are of more than academic interest: They have implications for treatment. In addition to being used to develop the computerized cognitive behavioral exercises, the Stroop task, for instance, may be able to serve as a marker to guide more effective use of antidepressants. Although not everyone with depression benefits from these drugs, certain drugs appear to be effective in some cases. The hope is that a set of clinical and biological markers can be developed that will make it possible to identify which drug will benefit which patients. Etkin’s lab is involved in studies that include hundreds of people with depression. It’s possible that a patient’s Stroop results can help indicate which treatment to pursue.

Along the same lines, Etkin is using the Stroop task and fMRI scans to evaluate how people with PTSD symptoms benefit from exposure therapy, in which they are exposed to the source of their fear. (It works in about half of all cases.) This is a step toward understanding how the treatment works biologically and whom it could help. As part of this research, Etkin’s lab is exploring a new treatment, transcranial magnetic stimulation, a noninvasive technique that induces electric currents in specific brain regions and thereby alters their activity. The study uses imaging to see whether TMS can change PTSD patients’ brain function in the same way as effective exposure therapy.


Five or so years ago, advances in fMRI research, genetics and molecular biology had been expected to provide a new basis for diagnoses in the forthcoming edition of the Diagnostic and Statistical Manual of Mental Disorders-5 — often referred to as psychiatry’s bible. The manual’s editors now readily admit that they were overly optimistic and that such a shift is not yet possible. Nonetheless, the new DSM is being designed so biological criteria can be added online in the future. What’s more, the NIMH last year launched an effort, known as the Research Domain Criteria Project, to develop an alternative to DSM that involves rethinking the entire classification system of psychiatric disease, based on neurobiology.

To grasp why such efforts could be turning a corner despite past disappointments requires understanding three big reasons psychiatric research has lagged behind other medical disciplines — and why recent developments suggest that these obstacles can be overcome.

1. Where do you find a schizophrenic mouse? Mouse models, the basis of much medical research, are hard to apply to psychiatric disorders.

In recent years, genetic sequencing has enabled researchers to engineer mice that have mutations similar to those discovered in certain psychiatric conditions. While pinpointing genetic causes of mental illness has been more complex than anticipated, advances in sequencing technology and genetic databases have lately yielded some significant findings, including the identification of seven  “copy number variations” — small chunks of DNA deleted or duplicated at a given spot in the genome — that increase by 10 times the likelihood of developing schizophrenia. “Never in history — the 100 years that we’ve been researching this black box — has anything been discovered that would raise your risk of schizophrenia by 10 times,” says Douglas Levinson, MD, a Stanford psychiatry professor who has contributed to the research. He is quick to make a caveat: These mutations account for no more than 1-2 percent of those with schizophrenia. Still, the discovery paves the way for creating mice with these genetic anomalies; that would, in turn, allow scientists to study how these mutations affect neuron function and develop insights about brain activity characteristic of schizophrenia.

2. How do you get a piece of someone’s brain? Biopsy samples from study participants, a staple for research in other medical disciplines, are not available to psychiatric researchers.

Included in NIMH director Insel’s top 10 research advances of 2011 is a technique called “disease in a dish.” Essentially, it’s now possible to take a skin cell from someone with a psychiatric disorder and transform it into a neuron. “For the very first time, you can make these tissues that you could not access normally,” says Ricardo Dolmetsch, PhD, associate professor of neurobiology, describing how he has used skin cells from patients with Timothy syndrome, a rare genetic disorder in which children show autism-like symptoms, “to generate little pieces of brain.” By studying these samples, he has identified abnormalities in synapse formation and levels of neurotransmitters, which offer new opportunities for developing treatments.

3. Can you test-drive a neurocircuit? The brain’s complexity has made it hard to pinpoint cellular mechanisms that cause psychiatric disease and to test therapies.

While fMRI studies such as Etkin’s are a big step forward in understanding brain function, the technology does not definitively determine cause and effect. And fMRIs measure brain function by voxel, a segment of brain about the size of a grain of rice, which is comprised of millions and millions of neurons. There is, however, a new technology, optogenetics, that can determine — on a cellular level — what causes changes in behavior symptomatic of psychiatric disorders. (It’s used only used in lab animals, as manipulating human brain cells in this fashion is seen as way too risky.) Karl Deisseroth, MD, PhD, a Stanford associate professor of psychiatry and of bioengineering, has pioneered this technology, which involves engineering a highly select group of neurons so that they can be switched on and off by pulses of light, usually delivered by fiber optic cable to the brain region of interest. Using this technique, he has discovered how to target particular neurons that cause behavior changes relevant to narcolepsy, cocaine addiction, autism and anxiety. 

Interestingly, in his study on anxiety, Deisseroth chose to focus on neurons in a subregion of the amygdala in mice that is in  roughly the same spot in humans that Etkin has shown to play a role in anxiety disorders. “Amit showed that there was altered activity in some interesting areas of the amygdala, but we didn’t know if it was causative or correlative,” Deisseroth says. The optogenetics test clearly shows the former. By stimulating particular cells with light, Deisseroth and his colleagues caused mice to behave in a markedly less anxious manner. “We are driving the neurons that inhibit amygdala output.”

And that may provide a new mechanism to target with a new class of drugs.


“Your data are really nice,” Etkin tells me a week after my fMRI, as he calls up on his computer the scans that have been composed from my performance on his Stroop task.

The scans show how my brain response differs after seeing consecutive “incongruent” images (the mismatched expressions and emotions, i.e., smiling face with the word “fear”) as compared with my seeing an incongruent image after a congruent one (i.e., smiling face with the word “happy”).

In my brain scans, there are about eight blue splotches and a few dabs of red. The former shows the regions that are less active, while the latter shows those that are more active.

Etkin points to the brightest red fleck that is just above my forehead. “There you’re activating your ventromedial prefrontal cortex,” he tells me. He then points to the blue that starts up at the top of my head and swoops toward the back of my head and down to the center of my brain. “As you engage, you also dampen the dorsal medial prefrontal cortex and the lateral prefrontal cortex, and you dampen activity in the amygdala,” he says.

Etkin calls up a diagram from one of his papers that is a composite of responses from participants who were considered healthy. It is shockingly similar to my own picture. “What we see with you is pretty close to the typical,” he says, alternating the screen between the two images. “Look at the similarities — it is a beautiful overlap.”

Etkin tells me that I made five incorrect identifications, which means that I was in the norm and that my response times were as well, with my getting increasingly fast when shown two consecutive incongruent images.

The interview with Etkin is in no way an assessment of my mental health, but it’s definitely a new way for me to think about it.

Later I print the scans and hang them on my refrigerator door. Occasionally a visitor asks me about them. I say they show a neural network we all rely upon to deal with emotional conflict. I point to the activated red spot and the dampened blue blobs and explain that they show how people cope with anxiety.

This is what it’s supposed to look like, I tell them, in a normal brain.

E-mail Jonathan Rabinovitz






©2012 Stanford University  |  Terms of Use  |  About Us