TL;DR. “Personality disorders” are best understood as primary disturbances of self-interpersonal (aka relational) functioning, not of personality as a whole. Understood that way, they escape the charge that the notion of “personality disorder” is so broad that every mental disorder counts as a personality disorder or so non-specific that every supposed personality disorder is better described as a different category of mental disorder. With this conceptualization, the dispute between “personality disorder” and “relational disorder” becomes largely terminological for ICD and DSM, though not for psychodynamic frameworks nor for trait-based conceptions of personality.
What are personality disorders? What is the relationship between personality and personality disorders? What is the relationship to the rest of psychopathology? These questions are conceptually tricky—far more than they appear to be at first glance—and discussions are often heated because there is a general lack of clarity on the basics of what is being debated. This post is an attempt to summarize my current thinking on the topic.
It is difficult for me to overstate the influence of Orestis Zavlis on these views. Engaging with his work on the subject has forced me to look at the issues in a way I had not before and many of his conclusions seem inescapable, in particular the idea that personality disorders are relational disorders. See prior posts by Zavlis on his substack Personality Disordered (here) as well as Psychiatry at the Margins (here) and my prior discussion of personality disorders and psychopathology. See Zavlis’s academic work on Google Scholar.
Some will read this post as me defending the notion of “personality disorder”; it is a defense but in a fairly weak sense. I defend that there is a coherent way to talk about personality disorders but in descriptive terms the notion invoked in ICD and DSM is that of a self-interpersonal or relational disorder, and it is important to be clear on that. The notion invoked in the psychodynamic literature is more complicated and theoretically substantive, invoking ideas like personality organization and personality integration, but it also requires one to buy into psychodynamic theory. A trait-based notion of personality pathology tracks a different kind of concept entirely than self-interpersonal disorder.
Alright, let’s begin.
Personality is the broad totality of who a person is, their enduring characteristics that set them apart from others and remain relatively consistent over time. As ICD-11 puts it, “Personality refers to an individual’s characteristic way of behaving, experiencing life, and of perceiving and interpreting themselves, other people, events, and situations.”
There are psychologists who think of personality primarily in terms of personality traits (i.e., the Big Five: openness, conscientiousness, extraversion, agreeableness, and emotional stability/neuroticism), there are others who think of personality as encompassing additional domains, such as attachment styles, fantasy lives, self-identity, etc. According to Dan P. McAdams, for example, personality consists of 3 levels: dispositional traits (such as the Big Five), characteristic adaptations (motives, personal goals, enduring values, defense mechanisms, relational schemas and attachment patterns, etc.) and narrative identity (a story of the self that “integrates the reconstructed past, perceived present, and anticipated future to provide a life with unity, purpose, and meaning”).
According to the Psychodynamic Diagnostic Manual (PDM-3), personality “subsumes a range of internal psychological processes, including temperament, motives, fantasy life, inner conflicts, characteristic patterns of thinking and feeling, ways of experiencing self and others, ways of attaching and relating, and ways of coping and defending.”
My own personal view of personality is more congruent with McAdams and PDM-3.
Personality functioning refers to the capacities for self-functioning (identity, self-worth, self-reflection, self-direction, etc.) and interpersonal functioning (understanding others, intimacy, cooperation, managing conflict, etc.). Personality functioning is, explicitly, self-interpersonal or relational. Self-interpersonal functioning corresponds to nearly all five personality traits, but it is not clear, at least to me, that it can be completely reduced to these traits.
Personality organization is the hypothesized structural configuration underlying personality functioning. It is about the degree of integration of self- and object- representations, defensive organization, reality testing, etc. Personality organization is, partly, object-relational.
Personality disorders are primary disorders of self-interpersonal functioning, which in psychodynamic terms reflect compromised personality organization and/or rigid personality style. This definition combines the definitions from ICD-11/DSM-5 Alternative Model for Personality Disorders (AMPD) and PDM-3.
ICD-11 and DSM-5 AMPD formally define personality disorders in terms of personality functioning.
ICD-11: “Personality Disorder is a marked disturbance in personality functioning, which is nearly always associated with considerable personal and social disruption. The central manifestations of Personality Disorder are impairments in functioning of aspects of the self (e.g., identity, self-worth, capacity for self-direction) and/or problems in interpersonal functioning (e.g., developing and maintaining close and mutually satisfying relationships, understanding others’ perspectives, managing conflict in relationships). Impairments in self-functioning and/or interpersonal functioning are manifested in maladaptive (e.g., inflexible or poorly regulated) patterns of cognition, emotional experience, emotional expression, and behaviour.”
DSM-5 AMPD: “Disturbances in self and interpersonal functioning constitute the core of personality psychopathology, and… they are evaluated on a continuum. Self functioning involves identity and self-direction; interpersonal functioning involves empathy and intimacy” (Criterion B additionally requires pathological traits)
PDM-3 takes a different approach to personality psychopathology. There are no formal personality disorders; instead, there are two dimensions of personality organization and personality styles. PDM-3 also places personality organization and styles on a separate axis (P Axis) from symptom patterns of psychopathology (S Axis). Personality organization is described in terms of levels that exist on a spectrum and PDM-3 uses the terms healthy, neurotic, borderline, and psychotic to describe them.
“Personality style refers to clinically familiar personality configurations (narcissistic, obsessive–compulsive, dependent, paranoid, etc.) that practitioners have come to recognize over generations of clinical experience... There is no clear demarcation between a personality style and a personality disorder. All people have personality styles or blends of them. The term “disorder” is a linguistic convenience for clinicians, denoting a degree of extremity or rigidity that causes significant dysfunction, limitation, or suffering to self or others.” PDM-3
So according to PDM-3, we can eschew talk of personality disorders entirely (if we wanted to) and talk about the degree to which personality organization is impaired and the degree to which any particular style is extreme or rigid with negative consequences.
Mental health symptoms occur within a particular personality functioning/organization, which influences the expression, course, and response to clinical intervention. Personality functioning/organization and psychopathology are closely related. In general, the worse the personality functioning/organization, the worse the overall severity of psychopathology experienced.
Strong covariation between personality functioning impairments and psychopathology severity is therefore expected, but there are conceptual and clinical points of separation. Psychopathology is possible in someone with relatively intact personality functioning and relatively healthy personality organization (e.g., an episode of major depression or panic disorder) and, vice versa, cases of impaired personality functioning/organization can exist with only mild psychopathological symptoms outside of self-interpersonal dysfunction.
Both personality disorders and psychopathology can be described in terms of general factors, which is a statistical way of saying that different features of personality disorders correlate with each other and different features of mental disorders correlate with each other, and we can describe that shared covariation with a single statistical factor.
It is true that the general factor of personality disorders (basically a marker of self-interpersonal dysfunction severity) correlates very highly with the general psychopathology factor (an impressive r=.92, as noted by Zavlis), but there is a difference between saying that the severity of personality psychopathology goes hand in hand with general psychopathology (and vice versa) and saying that both are identical. What the association does show is that self-interpersonal dysfunction is, among other things, a marker of the severity of psychopathology, and for people who have a high burden of psychopathology, their symptoms can often be understood in the language of self-interpersonal dysfunction. This is precisely the sort of relationship that enables the PDM to have the P axis and the S axis exist in parallel.
Setting aside methodological issues such as bifactor models and self-report measures with overlapping item content, personality functioning can be differentiated from general psychopathology in terms of clinical content and defining clinical features; they need not be statistically independent. The justification does not come from variance partitioning; it comes from clinical utility. I am not claiming or trying to demonstrate that there is an ontologically or psychometrically distinct class of disorders called personality disorders. My view is simply that from a clinical perspective, there are situations where we are dealing primarily with disturbances of self-interpersonal functioning, just as there are situations where we are dealing primarily with disturbances of mood or disturbances of perception.
In addition to clinical content, I think relational disorders can also be justified in terms of hypothesized generative mechanisms. Zavlis’s computational model of personality disorder as a relational disorder, built on beliefs about self and others, mentalizing and hierarchical inference, is one example of that kind of justification. Object-relations theory and personality organization are also mechanistic hypotheses about self-other representation in the same vein, and IMO also not threatened by statistical overlap with the p-factor.
A personality disorder is not a disorder of the entirety of one’s personality or every aspect of one’s personality, just as a “movement disorder” doesn’t mean that every movement is disordered. The disturbances and impairments are primarily in the self-interpersonal domain, which is linked to the degree of personality integration.
There is a rich history in the psy-sciences of conceptualizing such difficulties as personality organization and personality functioning, but as I mentioned previously, there are people who have a very trait-focused view of personality. Their position is that personality is best described in terms of personality traits. Self-interpersonal functioning for them has no privileged relationship to personality; it’s just one aspect of personality, similar to patterns of perception and behavior. So, to characterize self-interpersonal functioning aka relational functioning as “personality functioning” is unwarranted because relational functioning is only one aspect of personality and cannot claim to represent it. One advantage for this group is that psychometrics is on their side. When we look at associations between all-things-personality, they broadly organize around personality traits (and higher-order general factors). Self-interpersonal functioning also seems to largely dissolve into this statistical structure with standard psychometric methodology.
Others have different intuitions about personality, and for them, it is not problematic at all to use the term “personality functioning” for relational functioning because their conception of personality is not trait-based; for them, personality obviously includes characteristic adaptations, narrative identity, attachment styles, fantasy lives, etc. in a central and privileged way. This is supported by developed theories of personality like McAdams’s and Kernberg’s, which are compelling at least to those who accept those theories. When I think about it, I myself share this intuition. There feels something special about the place of self-interpersonal functioning in personality. The problem for this group is that it is very difficult to tell a coherent psychometric story about this, because, as I mentioned, psychometrically, it all dissolves into personality traits and the general factor.
Psychometrics is like an acid that burns away the flesh, leaving only the skeleton behind! The skeleton is important, but so is the flesh. DSM, ICD, and PDM are not organized, primarily, around psychometric structure (HiTOP is), and the present issue is whether personality functioning can coherently exist within the logic of diagnostic manuals (and not within the logic of HiTOP). Within the logic of HiTOP, personality psychopathology simply gets distributed along spectra such as internalizing, detachment, and antagonism, and that’s that.
ICD-11 and DSM-5 AMPD invoke personality functioning, and as far as personality functioning vs relational functioning is concerned, the matter is largely terminological, as long as we don’t reify personality functioning. Call it “relational functioning” and “relational disorders” and the definition remains the same, invoking disturbances in self and interpersonal functioning.
For psychodynamic theorists, however, more is at stake, given that their approach relies on the concepts of personality organization and personality styles. While personality styles can be characterized as relational (i.e. self-interpersonal), personality organization also includes things like reality testing and defenses. (Someone could argue that even reality testing and defenses, the way they are described in the psychodynamic literature, are relational through and through. I don’t think that’s quite right because not all defenses are relational—repression, intellectualization, isolation of affect—and reality testing also concerns external reality broadly, but there are probably people out there who dispute this and believe personality organization is wholly relational.)
There are parts of the world, UK in particular, where the term “personality disorder” has become highly stigmatized and subject to institutional mistreatment; as a result, the term is anathema to many. In my own clinical and professional work in the US, I have fortunately neither encountered such drastic hostility to the term nor appaling institutional mistreatment of those diagnosed. Regardless, if “relational disorder” is more acceptable to clinicians and patients and doesn’t generate any worse misunderstandings of its own (many may misunderstand it in terms of conflict in relationships and for many the self-functioning aspect will not be obvious), I am okay with the framing. I don’t have a fundamental objection to it.
Self-interpersonal or relational is also more descriptively accurate; one need not invoke any particular relationship to personality, and hence for ICD and DSM which favor more descriptive terms without substantial theoretical baggage, relational may be more suitable. It is less suitable for PDM, which is explicitly working within the psychodynamic theoretical tradition.
A common point I hear from British critics of personality disorders is that people are being told that they “have a disordered personality.” Even according to DSM/ICD, the framing should be that they have a disorder of personality functioning. Or in psychodynamic hypothetical terms, disturbance of personality integration/organization combined with a rigid/extreme personality style.
Any psychopathology term can be given unacceptably negative connotations if you say it in a certain way. There are people out there who believe “mental disorder” is an unacceptably offensive term. “How can you tell someone their mind is disordered?!”
In my own clinical work, I increasingly utilize the self-interpersonal framing and while I use personality disorder language because currently that is the shared terminology and the only way to ensure I am being understood (most clinicians have never heard the term “relational disorder”), I explicitly am flexible about what language to use. At the same time, I don’t do so because I believe “personality disorder” is an insulting, derogatory term. If “personality disorder” has become a dirty, offensive word, it has a lot to do with the attitudes we bring to it and the perverse institutional incentives to blame and exclude some patients from care using personality disorder diagnosis as the justification. I have never been so blunt as to say to patients “your personality is disordered,” but I may say to them, “I think we should also consider a diagnosis of personality disorder, which means that the way you relate to yourself and others has become significantly disturbed, in large part due to the XYZ experiences you’ve undergone. Here are the features of [the relevant personality disorder]. How well do you think they apply to you? I want us to determine this collaboratively and if these features don’t apply to you in your best judgement, I want to understand that. I am not particularly concerned about what exact name we use, but I want us to mutually determine whether this pattern of disturbances is present or not because it influences what needs to happen for you to get better.”
Anyway, I digress.
There is no clear separation between relational disorders and other mental disorders. Symptoms of mental illness (e.g. depression, hypomania, anxiety) that become persistent and entrenched enough to be trait-like have the potential to affect self-interpersonal functioning, and if they do, they will qualify as disorders of self-interpersonal functioning. (This is very close in spirit to Zavlis’s thesis that every persistent mental disorder is a personality trait disorder, except that I also require that the mental disorder is not only persistent and pervasive but also affects self-interpersonal functioning.) Every mental disorder occurs in the context of personality functioning/organization but only some co-occur with a relational disorder.
“All mental disorders are personality disorders” is false because not all mental disorders qualify for relational disorders in the sense described above. However, a lot of chronic, persistent, severe mental disorders will enter relational disorder territory. Chronic schizophrenia, autism, social anxiety disorder, persistent depressive disorder (dysthymia), etc., can all impair self-other functioning. Are these relational disorders? They do involve relational disturbance, but secondarily; they are not primarily relational disorders because the relational disturbances are better understood as secondary to another DSM/ICD or a different psychopathological process. We can distinguish between primary and secondary relational disorders, just as we might distinguish between, say, primary and secondary anxiety symptoms or primary and secondary psychotic disorders. Secondary relational disorders are cases where self-other impairments are the direct expression of the defining features of another DSM/ICD disorder.
Another way to handle this scenario, that works only in certain situations, is to conceptualize certain aspects of chronic, persistent forms of disorders as personality styles, e.g., depressive personality, anxious-avoidant and phobic personality, somatizing personalities, and paranoid personality. This is the approach taken by the PDM.
There are many routes to relational disturbances just as there are many routes to anxiety, and just as there are primary anxious disorders, there are also primary relational disorders and those are the chief concern when it comes to classifying personality disorders.
There is a proposal led by Orestis Zavlis to reclassify the borderline pattern specifier in ICD-11 personality disorders as “relational sensitivity disorder” and to place it in a separate section elsewhere in ICD-11.
The relational sensitivity proposal is not just about the borderline pattern. Read literally, it also radically redefines “personality disorders” in the ICD in a way that might not be apparent because the nature of that change is underplayed. The proposal says:
“Personality disorder remains in ICD-11 largely unchanged in structure, but perhaps clarified to make clear what the term should denote: relatively stable and maladaptive traits that are present, to some extent, in every human being.”
“Relatively stable and maladaptive traits” is not the notion of personality disorder that is present in DSM, ICD, or PDM. That is the five-factor tradition of personality and personality disorders, one that deviates from the traditional focus on self-interpersonal functioning. I don’t actually think that personality disorders should be defined in this manner; if we are interested in the relationship between psychopathology and personality traits, we should simply have a personality trait profile axis as part of clinical assessment. In fact, Zavlis himself has criticized trait-based approaches to personality disorders in the past, arguing that they cannot define personality disorder since personality traits are related to all of psychopathology.
If self-interpersonal difficulties are no longer to be called personality disorders, then call them relational disorders, but it makes little sense to rename personality disorders as relational disorders and then create an entirely new definition of personality disorder that ICD has not used in the past (while ICD-11 incorporates trait qualifiers, it doesn’t define personality disorders using them).
The proposal states: “the correct way to frame personality pathology is in terms of degree (“how much personality pathology does each of us have”) not kind (”does a particular patient have a putative disorder of personality”).”
If we restrict ourselves to traits, sure, but I think PDM and DSM-5 AMPD show us that relational functioning can be described in terms of clinically recognizable types and styles as well. The problem is that they disappear in psychometric analyses, but then, so do many traditional DSM and ICD mental disorders.
From the proposal: ““borderline” pathology appears to be mainly about a general sensitivity in how a person feels that is particularly pronounced within the context of close relationships.”
It is actually not quite settled what borderline personality is mainly about. Borderline personality disorder is a pervasive pattern of instability in relationships, self-image, and affects, plus marked impulsivity and self-destructive behaviors. This is a polythetic construct. What is “core” to it will depend on the approach taken. Descriptively, one can emphasize emotional dysregulation (PDM-3 now calls it “emotionally dysregulated personality”; ICD previously called it emotionally unstable personality), but it is not clear whether emotional dysregulation drives self-image instability or it is a consequence of self-image instability. Some treat dysregulation as primary, while other approaches emphasize identity diffusion or mentalizing; interpersonal hypersensitivity is also among the contenders. Relational sensitivity is acceptable as such but is not obviously the winning candidate.
Another problem is that if personality disorders are defined in trait-terms with no reference to self-interpersonal disturbances and borderline pattern becomes relational sensitivity disorder, then where do narcissistic, dependent and avoidant self-other disturbances go? If only the borderline pattern becomes relational sensitivity disorder and personality disorder chapter becomes defined by traits, then other self-interpersonal patterns like narcissistic no longer have any home in the ICD. Sure, they aren’t currently named in ICD-11, but at least personality disorders are disorders of self-interpersonal functioning, which means narcissistic personality styles belongs there even if it isn’t named. The very possibility of these patterns/styles will be erased when the only relational disorder in ICD is relational sensitivity disorder.
I think the proposal by Zavlis et al. would be a lot more compelling if it:
renamed “personality disorders” as “relational disorders” in ICD-11 while continuing to define them in self-interpersonal terms, and
renamed borderline pattern specifier as relational sensitivity syndrome within the chapter on relational disorders
That is a proposal I would happily sign, primarily because it forces the field to at least be descriptively clear about the locus of pathology in personality disorders.