The Dimensional Turn in Psychopathology: Q&A with Robert Krueger

Robert F. Krueger, PhD, is Distinguished McKnight University Professor in the Department of Psychology at the University of Minnesota, USA. He completed his undergraduate and graduate work at the University of Wisconsin, Madison, and his clinical internship at Brown University, and he currently serves as Editor for the Journal of Personality Disorders. He has been named a Clarivate Analytics Highly Cited Researcher, and Research.com ranks him in the top 100 most impactful psychologists in the world. His current research interests center around personality and personality disorders, psychopathology, health, aging, and genetics.

Awais Aftab, MD, is a psychiatrist in Cleveland, OH, and clinical assistant professor of psychiatry at Case Western Reserve University. He is interested in conceptual and philosophical issues in psychiatry and writes online at Psychiatry at the Margins.

---

René Magritte. The False Mirror. 1929

Aftab: I’m curious about your intellectual trajectory as a psychologist and researcher. How has your thinking about the nature of psychopathology changed over the course of your career? Which experiences or mentors had the biggest impact on shaping who you are?

Krueger: I was an undergraduate at the University of Wisconsin and was mostly interested in music and computer science initially. I took a personality psychology course as an elective in my sophomore year. The course was taught by Richie Davidson and after a few lectures I was hooked. The idea of working to understand the human experience in scientific terms was thrilling to me, and it combined my interests in the humanities with my interests in formal-mathematical models. I volunteered for Richie’s lab and soon thereafter, a friend mentioned that a new professor (Terrie Moffitt) was looking for undergraduate RAs to help pilot an assessment battery. I started working with Terrie and gaining hands-on experience with assessment. Then, Avshalom Caspi moved to Wisconsin and he and Terrie started working together closely (and they were married around this time). I then completed a senior honors thesis, advised primarily by Avshalom, and ended up being recruited by them for a PhD program in clinical psychology.

Being mentored by Caspi and Moffitt (arguably the most productive and influential psychologists of their generation) was an intense experience to say the least! I enjoyed both the research and clinical work I was doing in grad school and completed my clinical internship at Brown. While at Brown I applied to some academic jobs (a long shot to be sure) and ended up getting an offer for a tenure-track assistant professor slot at Minnesota. When I showed up at Minnesota that fall, I encountered all these giants in the field. For example, I remember seeing Tellegen, Meehl, and Lykken conversing near the mailboxes and thinking I was in way over my head. Eventually, though, I got up the nerve to converse regularly with all three and found those conversations invaluable in shaping my thinking and approach.

Around the same time, I got to know Bill Iacono and Matt McGue and started working with them in the Minnesota Center for Twin and Family Research (MCTFR). In those years, I was able to publish some papers on comorbidity among mental disorders, and the connections between those comorbidity patterns and personality dimensions. In terms of my thoughts about psychopathology growing and changing, all these incredible mentors contributed to what I ended up pursuing. I would also note that the types of patterns I was writing about (based in data) matched well with my clinical experiences. I find it limiting to think about patients in terms of putatively “primary” DSM categorical diagnoses (albeit in a practical sense I appreciate the need to record a chart diagnosis). The types of things I ended up pursuing reflect a mix of clinical intuitions and formal data analysis, with the aim of articulating approaches that are based in evidence, and thereby, hopefully more accurate models for case conceptualization.

Aftab: You were involved in the creation of the DSM-5 Alternative Model for Personality Disorders (AMPD), widely considered to be one of the most important developments in the field of personality disorders. What is your understanding of why there was so much resistance to the official adoption of the model during the DSM-5 revision process? When do you suspect the model will be formally moved to the section II of the manual?

Krueger: The placement of the AMPD in the “emerging models” section of the DSM-5 was an understandable compromise at the time this occurred, which is now more than a decade ago (circa 2013, when the DSM-5 was published). In my view, the AMPD was necessary, but also, so different from the DSM-IV personality disorders approach that it led to understandable anxiety about such a dramatic change. That is, the DSM is a highly conservative document, largely by design. Extraordinary or “radical” shifts in DSM are unwelcome because the DSM aims to provide a kind of port in the storm of the very real and very serious problems we see in our patients.

Much has happened since 2013. The AMPD was very successful at attracting attention from many talented researchers and clinicians. As a result, there is now a substantial literature attesting to its validity, reliability, and clinical utility. Consistent with these developments, the ICD-11 personality disorders model (now the official WHO endorsed model of personality disorders) resembles the AMPD much more than it resembles the DSM-IV approach to personality disorders conceptualization.

The way I look at the current situation is that the DSM is compelled to follow suit. If the DSM does not shift to look more like the ICD-11 personality disorders model, it will stand in stark contrast with a substantial international consensus. Of course, American chauvinism is not unusual in the realm of international politics, so how exactly things evolve is difficult to predict. Nevertheless, to assert that the DSM must preserve invalid personality disorder categories from the late 20^th century, in contrast to the contemporary international consensus, seems generally counterproductive in the realm of psychiatric nosology.

Krueger: If the DSM does not shift to look more like the ICD-11 personality disorders model, it will stand in stark contrast with a substantial international consensus.

Aftab: While AMPD in DSM-5 retained six specific personality disorders (antisocial, avoidant, borderline, narcissistic, obsessive-compulsive, and schizotypal) as diagnoses that may be derived from the dimensional model, ICD-11 chose to discard all traditional personality diagnostic labels with the exception of the borderline pattern specifier. Was the retention of these labels in DSM-5 simply a practical compromise to ease the clinical application of the model or are there additional clinical and scientific reasons to recognize these specific diagnoses?

Krueger: I can’t articulate a rationale for recognizing those six specific categorical personality disorders in the AMPD. In fact, the goal of dimensional approaches such as the AMPD is to be able to capture the nuances of a specific patient’s personality, obviating the need for a proliferation of categories that fail to recognize the uniqueness of an individual. A system like the AMPD represents a compromise between a more “literary” or “idiographic” approach (developing a highly nuanced and specific understanding of a patient) and a more “scientific” or “nomothetic” approach (working with concepts that allow quantifiable comparisons among patients). And yes, I do consider the idiographic approach as simultaneously having high clinical utility (particularly for psychotherapy) and limited scientific utility.

Everyone is unique in some ways, but that observation provides a limited basis for achieving generalizable scientific inferences. Neurotic conflicts appear in unique and even idiosyncratic ways, but from a public health perspective, the broad neuroticism domain is the compelling target, and much good can be done in public health by targeting major maladaptive personality domains. As a clinician, the way I approach this is to start with formal personality assessment (trying to understand how this person compares with other people via normative data). This helps with initial case conceptualization, but my understanding of the patient’s needs and potentially helpful psychotherapeutic strategies becomes more detailed and nuanced because the patient’s basic dispositions end up contextualized in their current conflicts and life circumstances.

So, speaking both clinically and scientifically, an extensive literature shows that pretty much all classical personality disorder categories (not just those six) are readily reproduced via combinations of dimensions from models like the AMPD and ICD-11 model. Moreover, the multidimensional nature of case conceptualiziation is frankly indispensable, in my view. For example, when professionals use the concept “Borderline Personality Disorder,” they might mean any number of things because that category encompasses a multiplicity of potential presentations (albeit the common core tends to be severe emotional dysregulation). Consistent with a model like the AMPD, emotional dysregulation can occur accompanied by many possible configurations on other personality dimensions. A highly emotionally dysregulated patient might have rigid perfectionistic tendencies, and another might be highly impulsive. Saying both patients “meet criteria for Borderline Personality Disorder” is not particularly helpful because the dynamics will obviously differ markedly in these persons. Similarly (and this example relates directly to several of my clinical experiences), a depressed person could be highly agreeable, and a different depressed person could be highly antagonistic. The former person likely can form a therapeutic alliance more readily and probably will cooperate with a structured approach, involving things like challenging maladaptive cognitions and doing homework, and the like. The latter (antagonistic) person presents a very different challenge because they are more likely to question the therapist, to say the homework is pointless, and to bail at the first sign that difficult emotionally laden material has entered the consulting room. Both are “depressed” but the dynamics differ in ways that are somewhat predictable from their basic dispositions, necessitating a more tailored approach than the approach offered by traditional psychiatric categories.

Aftab: What are your thoughts on the recent debate that has emerged around criterion A of AMPD? (I covered the debate in the post “Either all psychopathology is personality psychopathology or there is no such thing”). How do you think criterion A can be placed on a more solid scientific footing?

Krueger: Why do we traditionally think of personality disorders as something different from psychopathology in the general sense? I suspect the source of this traditional distinction relates to the role of interpersonal dispositions in modulating the expression of psychopathology.

Krueger: It seems to me very clear that all psychopathology happens in the context of the patient’s personality. I don’t understand how it could be otherwise. The way people experience psychopathology is embedded in who they are as a person, i.e., in their personality. A vast literature illustrates this fundamental point. So, then, the question you pose is: why do we traditionally think of personality disorders as something different from psychopathology in the general sense? I suspect the source of this traditional distinction relates to the role of interpersonal dispositions in modulating the expression of psychopathology. One way to see this is to just read through DSM criteria for personality disorders vs. “other psychopathology.” The personality disorders section describes numerous interpersonal issues, whereas the other parts of the DSM are generally more silent about the interpersonal manifestations of psychopathology.

Criterion A is a way of recasting this basic issue. In the AMPD, the criterion A vs. B distinction is between self-other dynamics (A) and basic dispositions (B). Every human being (and therefore every patient) has basic dispositions (B), but only some have significant challenges with conceptualizing themselves and other people in coherent and adaptive ways (A). This makes good clinical sense to me, but I also appreciate how a literature has emerged that shows that A and B features are highly correlated, perhaps even inextricably intertwined. The difference with A features per se is they are described in more inferential terms. Criterion A describes things the clinician infers from studying the patient (e.g. lack of clarity about identity) whereas Criterion B describes things the patient (or a collateral informant) reports (e.g. grandiose assertions and fantasies). As a result, the way to study Criterion A effectively is just very difficult, practically speaking. Criterion A tends to be operationalized in questionnaires, and not surprisingly, the content of those questionnaires overlaps extensively with something like the Personality Inventory for DSM-5 (PID-5; the typical way of operationalizing Criterion B). Yet Criterion A is probably better studied as inferences clinicians make about patient object relations, as opposed to patient reports of their own personalities. This makes for a much more challenging study design where you’d want clinicians to work with a patient for some time and then report about self-other dynamics to assess Criterion A. I can see how this could be done, but I doubt I could get a grant to do it! Well maybe if I add a neuroimaging component to the study, or some genomics… but you get the idea. Criterion A seems to me very psychologically important in the clinic, but hard to study in the lab, and tractability is a major consideration in research funding. In sum, I think I can articulate how understanding Criterion A and B distinctiveness and overlap would need to be approached to yield more satisfying conclusions, but I’m not eager to try to make it happen, because it’s just not very practical, unfortunately.

Krueger: Criterion A is probably better studied as inferences clinicians make about patient object relations, as opposed to patient reports of their own personalities.

Aftab: What role do you think Hierarchical Taxonomy of Psychopathology (HiTOP) can realistically play in contemporary clinical practice and research?

Krueger: This is something I’m quite interested in these days. I spend some of my time these days working with the ARCS institute (https://www.arcsinstitute.com/). The idea driving this clinic is to bring AMPD and HiTOP types of approaches directly into assessment and treatment. Basically, AMPD and HiTOP assessment and case conceptualization is part and parcel of what the clinic is aiming to offer to patients. In addition, HiTOP has a very active clinical translation workgroup, as well as a very active clinician network (see https://www.hitop-system.org/). In fact, I would go as far as to say direct clinical application is at the forefront of current HiTOP efforts. I suspect this next decade will see a lot of activity in this area, and hopefully, a corresponding impact on the lives of numerous patients.

Aftab: What do you make of the controversy around the p-factor of psychopathology? What are your personal views about the nature of the p-factor?

Krueger: I think it’s important to be clear that “the p factor” is just the most general summary of individual differences in overall psychopathology. Because all forms of psychopathology are positively correlated in the population at large, it’s not unreasonable to create a sum score reflecting the total burden of psychopathology in an individual. That’s all the “p factor” is, in my view. It is a descriptive and not necessarily explanatory construct.

Moreover, there is much more to the structure of psychopathology than the most general dimension at the apex of the hierarchy. In that sense, I’ve been a bit surprised at all the discourse focused on the highest level, as opposed to focusing on the broader hierarchy, including its more variegated and specific components.

I think the way to move this forward is to realize that specific candidate etiologic and pathophysiologic elements are likely associated with the hierarchy at various levels of breadth vs. specificity.

The substantive challenge here is to merge data on etiological and pathophysiological data elements with more descriptive data elements throughout the hierarchy. Contemporary alternatives to the DSM such as RDoC and HiTOP can be seen as tractable and complementary approaches aimed at pursuing this merger. Some things (e.g. I suspect childhood adversity works like this) are associated with risk in a very broad sense. By contrast, other things (e.g. specific genetic polymorphisms) are associated with specific risk for specific presentations. I think efforts are better directed at studying etiological and pathophysiological data elements in the context of the overall hierarchy, as opposed to debating the meaning and nature of any specific level, including the highest and most general level (the “p factor” level).

Krueger: Because all forms of psychopathology are positively correlated in the population at large, it’s not unreasonable to create a sum score reflecting the total burden of psychopathology in an individual. That’s all the “p factor” is, in my view.

Aftab: Colin DeYoung and you have developed the cybernetic theory of psychopathology, which I have been studying recently. It’s an excellent theory, I think very highly of it, and I hope to engage with it in a more substantive manner. One question I’d like to ask you is why was it important for Colin and you to characterize the theory specifically in “cybernetic” terms? Secondly, when the theory refers to the persistent failure to move toward one’s goals, are you thinking of goals that people intentionally set for themselves (e.g. becoming a tenured professor or an Olympic athlete) or are you referring to goals in a more abstract capacity – e.g. goals related to our survival, well-being, and interpersonal functioning – that are collectively shared by humans and are products of our biological and cultural evolution?

Krueger: I appreciate your positive appraisal of the cybernetic theory of psychopathology. I can’t speak for Colin of course but would note that the cybernetic approach to psychopathology evolved from Colin’s Cybernetic Big Five Theory (CB5T), where some of what you’re asking about has been more thoroughly articulated. In my view, “cybernetic” just means the cycle of goal activation, selecting an action, doing the action, understanding the result of the action, and comparing that result with the goal. All goal directed organisms are “cybernetic” in this basic sense.

For patients, in particular, goals can exist at various levels and are often not well articulated (one might even say goals are often unconscious and highly fraught). Some goals are intentional and available to consciousness, but plenty of goals are lurking in the background and maybe not readily available to the conscious mind. In fact, I’d venture to say that these “more unconscious goals” often relate to unacceptable impulses having to do with sex and aggression (imagine that!). I suppose my psychodynamic sympathies are showing here (maybe that’s also unacceptable in contemporary discourse, but I hope not). Of course, the scientific challenges here are difficult to surmount. I’m not sure I know how to effectively assess unconscious goals, except in the sense that I do this informally as a psychotherapist. I do hope some clever graduate students decide to join us at Minnesota and help us work some of this out more formally and empirically. For the time being, I’m gratified to learn you feel the theory has merits and I hope there are people interested in trying to develop ways to further evaluate the theory empirically.

Aftab: Advocates of symptom network approaches to psychopathology eschew all talk of “latent variables,” while latent variables feature pretty prominently in HiTOP as well as the cybernetic theory of psychopathology. Where do you think the rejection of latent variables goes wrong? What are some misunderstandings in this area?

Krueger: I had the distinct privilege of visiting with Denny Boorsboom in person recently. He was in Minnesota and reached out to chat in person. I can’t speak for him directly, of course, but I think I can say that the concept of “latent variable” is unavoidable, and this is increasingly acknowledged among network enthusiasts. Basically, it’s impossible to converse productively without talking about latent constructs. It seems to me that fundamental data for developing network models (e.g. specific symptoms) are latent constructs. Consider a symptom such as insomnia. This is a much less abstract construct than e.g., the “p factor” but it’s also not lacking in latent qualities. When we say “insomnia” what is the pattern of sleeplessness? How long has it lasted? Is it highly consistent or highly variable over time? Basically, all those details of conceptualization and assessment of insomnia mean that it is wise to acknowledge that even if insomnia is more specific than depression or anxiety or internalizing, it isn’t “obviously manifest and completely lacking in latent qualities.” Moreover, network modeling approaches reveal patterns of more and less densely connected nodes, and those patterns naturally resemble latent constructs from related quantitative approaches such as those pursued in the HiTOP consortium. In short, the “thesis” of latent constructs being useful in studying psychopathology and the “antithesis” of eschewing latent constructs hopefully leads to the “synthesis” of a hierarchical approach, in which more specific and more broad variables are systematically organized in a hierarchy.

Krueger: I can say that the concept of “latent variable” is unavoidable, and this is increasingly acknowledged among network enthusiasts.

Aftab: Do you have any regrets about your academic career?

Krueger: In some ways, I regret pursuing a PhD rather than an MD, but this is generally a minor matter practically speaking because of what I am able to do professionally (i.e. research primarily albeit also some clinical practice). Indeed, I don’t envy all the challenges inherent in obtaining both basic medical training and then subsequent formal research training. That is a long and difficult road. After an MD, I wonder if I would have had the tenacity to pursue an intensive subsequent research career. So, I guess I don’t have major regrets, just some curiosity sometimes about paths I didn’t travel. More generally, I wonder sometimes about what might have become of my musical career, but seeing and knowing many first-rate musicians who find it hard to make a decent living tends to remind me of just how hard that life can be, practically speaking. Maybe I’ll retire someday and pursue music more avidly, but I love what I’m privileged to do for a living, so I’ll probably just keep doing my academic work (albeit at a slower pace as I continue to age).

Aftab: In your opinion, where has the mental health field gone wrong in its approaches to understanding and treating psychopathology? What can we do differently going forward?

Krueger: To say the field has “gone wrong” strikes me as a strong statement, given how primitive our understanding of psychopathology remains. The phenomena we encounter in the mental health clinic are the most challenging to study in a scientific sense. Straightforward strategies just don’t work in unraveling the etiology and pathophysiology of psychopathology, in comparison with many more straightforward medical problems. This is not to say medicine has it all figured out, but most of what we see in the mental health clinic can’t be explained by something like a spirochete infection. (That said, a thorough medical workup is really important, and something I try to encourage, because I’ve had experiences where there was in fact something medical lurking, like for example something an endocrinologist could help with, and the medical workup was really helpful to improving the patient’s life.)

The fundamental causes of psychopathology seem to typically be sitting in some very complicated interplay of distal and proximal etiologic factors at levels from the genome to the structure of modern societies. That’s not a cop-out, it’s just the very real and very difficult problem we face in developing a scientific understanding of psychopathology.

Krueger: Everything we try as a field ends up illustrating how murky and complex the origins of psychopathology tend to be.

When I was in grad school in the 90s, there was still optimism about things like major genetic loci or even potentially major pathogens that might explain many cases of schizophrenia, for example. But everything we try as a field (i.e. numerous types of formal scientific inquiry) ends up illustrating how murky and complex the origins of psychopathology tend to be.

With that in mind, I do think the HiTOP type of strategy is a reasonable way to proceed. The structure of manifest psychiatric problems (the things that patients present with, i.e. signs and symptoms) can be discerned using readily available assessment technologies. This provides a solid evidence base in a murky swamp. With that structure better understood, it can be linked with data from other methodologies developed in neighboring disciplines, e.g., from genetics, neurosciences, sociology, and so on. Alternative approaches (e.g. starting from animal neuroscience) are also tractable but are more distal from the human problems our patients bring us. So, I guess “what we can do differently” is to be comfortable with HiTOP type structures and not confining our thinking to DSM categories. I think this approach is already beginning to bear fruit, e.g., I would suggest that the structure of genetic risk for psychopathology resembles the HiTOP model and not hundreds of putative DSM categories. I am optimistic we can make some real progress in our lifetimes, and I’m privileged to be able to contribute to these efforts.

Aftab: Thank you!

---

This post is part of a series featuring interviews and discussions intended to foster a re-examination of philosophical and scientific debates in the psy-sciences. See prior interviews here.

---