Thank you so much for making time for me today to discuss your research, impact and role as a mentor, and how you’ve navigated the waters to get to where you are today.
I’m happy to, so what would you like to discuss?
Could you tell us in your own words about your research focus and what drew you to this line of research?
My research is built around a few different themes. The first is around really trying to understand the neural substrates of emotion and how individual differences in those neural substrates would be related to individual differences in personality and risk for psychopathology. Over time there have been several different concepts that have come out of this line of research, such as the relationship between emotion and attention, and precisely how emotional stimuli grab or capture our attention.
Another theme has been the basic study of specific brain regions, in particular the amygdala, orbitofrontal cortex, and the ventral striatum. These are some of the key limbic regions that are involved in emotion and motivation, but we ask the question of “how precisely are they related?”. Then the third theme, taking a neurochemical approach, is to understand the role of dopamine in motivation by measuring individual differences. Using PET imaging we aim to measure how dopaminergic differences may impact risk for drug abuse, and expression of problems in motivation and psychopathology.
The last theme, which came about much later for me, and was spurred on by a meeting I had with Benjamin Lahey at the University of Chicago. It relates to the idea that there are large dimensions, or factors, of psychopathology that can be used to re-conceptualize how we study psychopathology. We’ve had this attitude from the psychiatric/medical model that each categorical diagnosis has it’s own unique features and neural substrates. This ignores what we know about the high comorbidity of psychopathology, and so our lab’s quest has been to ask the question: “Are there neural correlates of these large dimensions of psychopathology as opposed to the narrow symptom clusters and individual diagnoses?”. We first looked at differences in emotion, attention, and motivation for these large-scale correlates. It’s a bit funny because it’s definitely not what I feel like I was explicitly trained to do and was not really on my radar until well into my career.
That’s fantastic, do you feel that the direction that the RDoC Initiative (https://www.nimh.nih.gov/research-priorities/rdoc/index.shtml) takes aligns well with your research?
Yeah it does! I think the one difference is that the RDoC, though very sophisticated in terms of constructs, says very little about that larger structure, or meta-structure of psychopathology. A lot of what we’ve done is to meld those two things together. We were lucky enough to get one of the first round of RDoC grants, essentially making the argument that for RDoC to succeed, it also needed to pay attention to these larger factor structure issues within psychopathology.
I have noticed that reward-processing research, particularly in mental health, has had an explosion of research within the last decade or two. After reading your review paper coauthored with Michael Treadway, which was fantastic, I wondered how you see the landscape of this area of research changing in the coming years?
I think the real thing is to move away from a simplistic view of reward and to think much more in terms of the different components of reward processing and the complexity of it. It’s easy for us all to say ‘reward’ and have a decent idea of what it means, but to really talk about reward we have to be talking about reward discounting, risk, and the range of various elements that go into any computation of the immediate value of possible rewards. So to me, rather than just thinking of just ‘reward deficiency’ or ‘heightened reward sensitivity’, where we’re going to see progress is in breaking that apart into the numerous features that have to go into any aspect of reward processing. With that we have a lot better chance of understanding disorders in which there are aspects of reward processing that go awry.
In the same vein, there was one specific quote that I really liked from that paper and I was hoping to get your insight on it. It was, “Although recognizing the breadth of reward abnormalities in psychopathology is important, it would be a mistake to consider them homogenous across or even within disorders.”
Yes, I like that phrase too! And it really does suggest that not only have we sometimes treated reward in a homogenous way, but we’ve also assumed that it’s the same thing that’s going wrong even within a disorder, when you may have multiple different issues impact reward processing. For instance, when we talk about substance abuse, where you’ve got someone who’s no longer expressing much desire or positive evaluation of non-drug rewards, there are certain drug rewards that are incredibly motivating to them. This may suggest that their scaling of rewards is off rather than there being a global change in reward sensitivity. Moreover, you can have very different sorts of elements at play even within something like a substance abuse disorder. I don’t go in assuming that alterations within reward processing are identical for every person with an alcohol abuse problem or every individual who has an addiction to cocaine.
These differences could be critical to the success that we might have from a treatment standpoint. If we’re hitting the wrong thing it may be no surprise when our treatment fails. Now the trick, I think, is to go to that next level and say, “how will we quantify and operationalize those differences to the point at which we could bring them into play when we try to treat someone?”. We’re not there yet. We’re often at a level of looking at one aspect of this in isolation. For instance, Michael Treadway and I have often articulated that we have reason to be concerned about effort allocation processes in psychopathology, but arguably you wouldn’t want to come in and study effort allocation in isolation thinking that you’ve defined entirely the reward processing abnormalities of an entire disorder.
Excellent point. I’ve noticed you’ve published extensively using neuroimaging methods such as MRI and PET. As an expert what considerations should we take into account for understanding and utilizing these methods, particularly when considering affective processes?
To me, the biggest issue is the risk of taking an overly deterministic approach of looking at the results of imaging studies. Results from imaging studies are visually compelling but lend themselves to an interpretation that seems to simplify things in a reductionist way. For example, it’s sometimes assumed, “the amygdala activated and therefore I’m going to apply this reverse inference that this person was feeling anxious, or experiencing fear.” I think the visuals themselves sometimes give us this reductionist interpretation of what we’re seeing and our assumptions can be misleading. The scale is also still limited for us in terms of not being at the level of neurons or sub-nuclei. We’re at the level of populations of neurons and hemodynamic responses, and we’re often talking about signals that are not temporally resolved. These sorts of issues always need to be attended to but it’s easy to be seduced by what shows up in one of our lovely images.
Another reason I ask this question is because the SAS audience is very broad, so I want to give our readers a lens through which to see and understand some of the findings of your work.
In my own work, I’m interested in individual differences and ask questions like, “what’s the correlation between D2 dopamine receptors and this personality dimension”. I would have not been able to tackle such a question without neuroimaging. The problem which I worry about, particularly in personality research (even beyond affective neuroscience), is the risk of seeing something which is statistically significant but only describing a certain portion of the variance, but interpreting it as though we’re explaining all of the variance, or the majority of the variance. I’ve been particularly aware of this as studies come out with 200 or 300 subjects, which is novel for neuroimaging, as traditionally our sample sizes have been smaller. You see a result and there’s a correlation and you say, “Okay, can we back that up and say how much of the variance are we explaining?” You’ll find that sometimes you’re only explaining 3% of the variance, which may still be very important, but it’s not like it’s explaining 97% of the variance. Yet, when we write about it, there seems to be a temptation to portray a stronger interpretation. I don’t know that this is specific to any methodology, I think the field in general has this tendency. I’m very aware of it when reading papers. For instance, most genes only explain a very small amount of the variance in different behaviors and, to me, that always has to be kept in mind. In recent years people have been increasingly talking about being more concerned about effect sizes rather than p values. That is very salient to me, and seems like a positive development.
Yeah, I think that distinction is very important. Upon entering grad school I learned more about statistics and even that there was an approach beyond basic frequentist statistics. It’s kind of odd to first recognize this in grad school.
Well you know, it’s very much what we do. There are specific reasons we choose to teach certain aspects of statistics for our preparation in the field. However it’s clear that this knowledge on it’s own has it’s limitations. I feel like I sound a bit on the curmudgeony side, don’t over interpret this as nothing dampens my excitement every day when I come in to look at the data.
I’m definitely not getting that impression; maybe I’m just giving you the hardball questions. This actually segues nicely into my next question though! What is it that excites you the most out of your current projects?
Oh, I like this question! I think what is the most exciting at the moment is the work into the larger factor dimensions of psychopathology: this attempt to see if there are fairly non-specific correlates of psychopathology. The reason this has excited me so much is because it has caused me to reframe my thinking about how we ask research and treatment questions in the first place. As someone who works in both neuroscience and affective science, a lot of my approach has been to look for the right construct in hopes of making the most progress. So, if I want to study obsessive compulsive disorder, what are the behavioral constructs that look the most like the symptoms that we see in that disorder? That framework, since day one of me entering the field, was how I had thought about psychopathology research. There’s a lot to be gained by that, just asking, “how can we break a construct down, operationalize it, and move between the animal literature and the human literature.” But now we are stepping back and thinking about comorbidity and a larger structure of correlations between disorders. This has been the guiding direction of this newer line of work. I’ve been really excited about contemplating how to ask question differently when we think there are pervasive, pleiotrophic, non-specific influences on psychopathology.
That does sound really exciting. You’ll have to define pleiotrophic for me.
This is a term that’s often used in genetics that refers to the idea that you could have a gene that could influence a number of different things. Exactly how it comes out, or is expressed, can be quite variable. For instance, you could talk about a number of genes that don’t necessarily determine which disorder a person may have, but could determine the general risk for having a disorder.
Interesting, I’ve read some papers that refer to a p-factor.
That’s right, the p-factor term was given by Avshalom Caspi, but the first description of how to model the general factor statistically came from a paper first-authored by Ben Lahey. I am a co-author on it, but Ben was fully the leader. Caspi really caused an explosion when people read his work. Ben realized how to model this sort of general factor through a bi-factor model. People then applied this to existing data and there has been a plethora of studies on this concept in the last few years.
However, it does run up against the way we often are taught to think of psychopathology. I’ll note that when I started, I remember being a clinical student at University of Minnesota and we had this great, descriptive psychopathology course, but it seemed to me at the time, “why does anyone ever have difficulty making differential diagnoses? All you have to do is give a SCID interview to the person and go through these criteria”. I didn’t realize that there was this larger issue as discrimination between disorders is not nearly as clear as the ability to make a reliable diagnosis. With time I learned that a reliable diagnosis and a proper understanding of psychopathological symptomatology are discrete entities.
I see, changing pace a little bit, within your own research, or even personal development at any stage of your career, outside of the realm of psychology, economics, or neuroscience, from where have you drawn influence in your work?
Oh that’s an interesting question. In truth, so much of where I’ve drawn from is within those fields, but those fields are so broad to begin with. To me it’s the range of cognitive science, neuroscience, clinical science – all of those areas have so much to draw from, that I’ll admit to not spending much time outside of those areas. I think the challenge to many of us today, who are interested in the brain and behavior, is to wrap our heads around the various sub-disciplines and full-disciplines that feed in there. That to me takes up an enormous amount of weight in terms of thinking about very different perspectives that come into play. For instance, I think of some work that I have going where I talk to some people in drug discovery and pharmacology. They come at these issues from such a different starting point that just those conversations and just trying to bridge those distances ends up often being very informative. Now that doesn’t mean that I am a total, narrow geek when it comes down to it. A lot of my focus outside of science ends up being related to my great love of music, but I don’t know if that always comes back in a way that informs my science.
Yeah, I had noticed you have a CD on amazon. That’s pretty cool that you enjoy music too.
I do, however I wish it formed a larger part of my life than it does. It’s very important but it’s one of those things that’s always a challenge to balance.
So what aspects of your career have you found are the most rewarding?
It may sound corny, but it’s really been mentoring good students and post docs. I think I’ve taken more pleasure in watching students develop and launch their own careers. I also enjoy that feature of discovery. Going in and finding, “oh wait, that actually shows something! That actually worked!”, or “that’s kind of a cool paradigm…that never occurred to me.” It’s that sense of repeated surprise that has been quite delightful.
Excellent! Is there any advice that you would give graduate students working toward careers in research?
Hhmmm, so there are a few different pieces of advice I would give. The first is to understand that many of your ideas will fail, and that’s ok. One learns from those errors, and so one should not get too bent out of shape. The second would be to learn from reviews because even if you get bad reviews that are off target, they often tell you something informative. This could even have to do with how you presented the material and how you might be able to present it in the future in a way that may be easier for someone else to understand. The third is to think about what are the things you see as your potential next steps when planning into the future, such that you can draw links between the different projects you’re working on.
That’s fantastic advice! So it looks like we’re nearly up for our time today but do you have time for one more question?
I’d be happy to, this has been an enjoyable conversation.
Great! So what is your advice for students who are trying to forge their own path in research?
So let me answer this with what the challenge is first with how I see it. The challenge is how does the student define their own area of study relative to the mentor(s) with whom they work. The tension point there is very much a pragmatic question. I think most faculty are willing to encourage students to forge their own paths, but we’re limited based on our grants. You know we can’t just say, “here’s $100,000 just to go do this.”, and so the trick is often how to wed those ideas with what’s already there in the lab and what’s going on in the lab. I think that the advice is always to be thinking about how interests fit within an ongoing study. If you can figure that out, I think it can be a great stepping-stone moving forward, but it has to be done in a way that allows you to make use of the resources that are present.
Also, I’d say we should think heavily about, “what is the big idea?” Not what’s the smaller study I can do, but do I put this within this larger, long-term goal. Can I develop an elevator speech, a 30 or 60-second description of what I want to be studying in the long-term? What is the burning question that I want to answer? If you can identify that, then you may be able to start bringing it down to manageable levels. I think it’s much harder if you only focus on the narrow question for each study because we can get so into the micro-details of any given study that we can get lost. I was talking to someone the other day who works with a particular animal species and there was this fear as they were so focused on the specifics of the species that they didn’t focus on the question being asked and why using that species would be ideal to answer this big question. So my approach is to figure out what this big question is, and then you go after it in ways that are opportunistic. It may be that available data, tools, or resources will allow you to get at this bigger question and you need to be able to bring it into play.
That’s very refreshing to hear from someone who is so skilled with various methodologies. I feel like there are many ways to approach big questions and sometimes we get caught up in the methods we’re comfortable with, or that are appealing for grants.
My other piece of advice is to be totally willing to discuss those big picture ideas with people in the field. With your mentors say, ‘here’s what I’m envisioning long term’, not just sticking with, “I want to do this analysis on this dataset”. Have the discussion of how doing one analysis will give you tools or a piece of this larger picture in order to help you move forward.
The other thing is to pull from different areas of psychology and neuroscience, particularly for clinical students. I think the requirements of clinical programs make it a challenge because we spend so much time meeting other requirements and it can be hard to have enough contact points for that bigger picture.
I agree with you there. It looks like our time is up, but thank you for your thoughtful answers to my questions today.
My pleasure, thank you!