Psychiatric neuroimaging and computational psychiatry hold great promise. With computational models and brain imaging, we will solve psychiatry and understand what causes disorders. We put out great expectations and hopes, but will we ever be able to fulfill even a small part of these promises? Or are these just empty words to convince funding bodies?
When I left high school, I was very interested in understanding how the human mind works. I wanted to understand what drives us, what makes us human. Back then, I was fed up with all of maths, physics and biology because I could not see how these subjects could help us understand how the human mind works and what we as human mankind have to do on this planet. I was much more attracted by philosophy and Freud, which seemed to give us answers to these very deep questions. This is why I naively decided to study psychology – if any of the potential subjects, I was sure that this would give me better insight into the human being. The first years of studying were quite disappointing. I learned about t-tests, the big five personality traits and some efficacies of CBT treatments. But nothing enlightening about the human mind. Well, maybe I was expecting too much, or maybe we still knew very little. When I then heard about neuroscience and the brain I thought that this might be the direction that could help us understand these issues better. These techniques opened up completely new perspectives – seeing where in the brain we think and being able to understand what humans think. This was very exciting. So I did my major in neuropsychology. And yes, this excitement carried on. I loved studying the brain and I felt like for the first time we were able to finally understand how the human mind and brain works. I wanted to understand more how this kind of research works and also how the brain could go awry and lead to psychiatric disorders. So I decided to do several internships, in neuroimaging as well as in psychiatry. I was sure that these two fields must meet at some point and provide the answers that I had spent so long looking for. And indeed, it was fascinating to learn how we can get insights about the brain mechanisms by just recording magnetic field inhomogeneities during mental processing. But maybe even more important were my experiences as intern in psychiatry and neurorehabilitation. Yes, during adolescence I came across people with psychiatric problems. I had friends that always made impulsive decisions, to their own disadvantage. And another friend seemed to separate himself more and more from everyone else, until he ultimately left school; and as I learned later, had to be treated on a ward for psychosis. But when I worked in psychiatry, it opened up a whole new world of human struggling. Working with adolescents that do not want to live anymore. Adolescents that had traumatic experiences, who were deprived during childhood, or that suddenly suffered from hallucinations. And I was working with them day after day on the ward. It was fascinating and frightening at the same time to observe how some of them did not want to live anymore. And how the psychotherapists and nurses fought so hard to bring them back ‘on track’. And it was great to see how over the months on ward you build up relationships and you are able to support these kids and help them get better. I was so impressed how good clinicians were able to help those adolescents in their despair and supported them throughout their darkest times. It was great to see how some of them got better and were able to leave the ward and to become independent. But the reality also was that some did not get better, and the therapists ran out of interventions. And some of them went on to other therapies, and some of them did not go on and ended their lives. This experience taught me several things. First, I was able to observe the abyss of the human mind, standing at a safe distance and observing what havoc the mind can wreak on an individual. Second, I was amazed about how good and caring therapists were able to help those kids. And this was not because they used a particular form of CBT. No, it was much more their clinical experience, it was because they cared about the patients and because they were able to connect to them even when they were at their worst. Third, I also learned how helpless we are and how often we are unable to help. How come that we still do not understand why someone suddenly has hallucinations? How come that we have no proper intervention for anorectics? Why are such massive impairments understood so poorly? Are we really that ignorant and stupid? We are able to understand quantum states, but not why an IQ drops by 20 points after a psychotic episode.
I wanted to change that. And I thought that maybe through brain imaging we could solve some of these problems. So I went on to do a PhD in child and adolescent psychiatry. And I was able to find a good position where we studied ADHD and OCD. The former, I had a lot of experience with from other internships, the latter I had little knowledge but it was a fascinating disorder. So we worked eagerly to understand how their brains may work differently and how this could be related to their impairments. And I realised that only through computational models we might be able to go slightly beyond the purely descriptive nature of classical psychiatric neuroimaging studies. One thing I learned during that time is that it is actually bloody difficult to do proper psychiatric neuroimaging research. It is not easy to build proper hypotheses because the literature is so self-contradictory. The psychological theories are so removed from actual brain mechanisms that a translation is almost impossible. One of the biggest challenges, however, was to recruit patients. And this is not only because we demand quite a lot from them (3 hours of clinical interviews, 5 hours of tasks and scanning, etc). It also is because these patients suffer! Their lives are not fun. They have some serious issues. And they are often not able to live independently. So how should they actually join a study and master these additional demands? If you don’t leave the house because it causes you a lot of distress, how in the world should you be able to come to a new place (called ‘laboratory’) and lie in a scary machine and let people you don’t know do tests on you? This certainly is NOT fun!
So I was very impressed by the patients that were actually willing to join our studies. They were willing to take all these burdens and additional hardships, just to help us with our studies. They were investing so much of their energy because they believed in our research. They truly believed that we might find out something new and that we might eventually be able to help to cure these disorders. This noble desire to help others, even in the light of their own suffering, is simply amazing.
But: Can our kind of research actually help these patients? Are we really addressing the right questions that will give the right answers? Are these patients going through all these struggles for a reason? Sometimes I was and I am not sure about whether our findings really advance our understanding of psychiatry and eventually help my patients from the ward.
It often feels like psychiatry is a strange field. We are always chasing the newest developments in cognitive (neuro-) science, hoping that we find the holy grain which leads us to a panacea. A few decades ago, we believed that neuropsychological tests like an Iowa Gambling task would show us which psychiatric disorders suffer from impairments in the OFC. In the 90s, we thought that conflict and reward tasks could reveal the regions in the brain that cause impairments in disorders. And now, we hope that by augmenting our tasks with computational models, and by mining our data with machine learning algorithms, we will be able to understand psychiatry. But is there really any substance to it? How can I justify to my patients that they have to go through all these struggles? Can we really make a difference?
Frankly, I don’t know. I wish I would. I wish I could say, yes, eventually, our work makes the world a better place. But I am not sure I can.
I am doing science now for almost 6 years, and I sometimes really struggle with disillusionment. There are hundreds of papers that justify their basic research with a potential application for psychiatric and neurological patients. But not many of them seem to have materialised for patients. So are these claims just empty words that help us convince journal editors and grant agencies? Do we actually care enough about our patients? And how could our gambling tasks in the scanner have a relation to the problems of our patients in psychiatric hospitals?
A difficulty I sense is that most of us researchers have little experience and understanding of psychiatric disorders. If you want to be successful in science – one of the most competitive fields one could potentially imagine – it is almost impossible to spend years and years on the ward to become an excellent therapist too. So how can we scientists understand our patients and develop novel hypotheses? And likewise, many clinicians lack knowledge and experience to be fully on top of the recent developments in brain research, and how our models work, and thus struggle to understand our ideas and methods.
So, where will the endeavour ‘computational psychiatry’ lead us to? Will we actually find remediation? I think there are some major problems that we are failing to address so far. One is that we use highly artificial environments. And although certain processes might be impaired in real life, we might not find them in our artificial tasks. Secondly, we completely ignore the social environment and their influences. We believe that the brain of a patient is a secluded entity completely remote from everything else. But we should know that the social environment has massive effects on disorders. It can prevent a person from becoming ill, or vice versa. So how should we predict future disease states if we ignore this important variable? Lastly, I feel that omitting our current diagnostic classifications and replacing them by other (equally arbitrary) measures somehow feels like throwing the baby out with the bathwater. The former were built over centuries and work pretty ok in everyday clinics. Yes, they are not perfect, but maybe we should appreciate the past and not try to completely reinvent the wheel.
But I can also see some potential real ground-breaking advances. I think that using machine learning approaches we might actually be able to separate different subgroups with distinct etiologies, but common symptom expression. However, to achieve this, we need data from many, many patients, the right tasks to provide useful data, and we need to join our forces across labs. We need to overcome the narrow boundaries of labs and share data and tasks. A second advance of our field, I believe, is that by using computational models, we have to sharpen our minds and hypotheses. We have to make concrete mathematical predictions and rigorously evaluate them. For this, however, we need to try to understand our patients. We need to be exposed to their problems. We need to be on the ward, we have to closely follow what they experience. Only through that we will be able to generate the right hypotheses and gather the right data.
So is it worth to spend all that money on computational psychiatry? And is it worth to make our patients going through the additional burden of participating in our studies? I hope so. Maybe one of the most exciting things about science is that we never know where we will end up. Maybe a breakthrough is just around the corner. But I also believe that we as scientist must constantly doubt ourselves. We have to ask ourselves whether it is the right thing we are doing. Whether I spend all the trust (and money) that I receive from the public wisely or not.