Personality Functioning, Problems in Living, and Personality Traits

Abstract

The publication of the Alternative Model of Personality Disorder (AMPD) was a signpost achievement in the personality assessment. However, research on the AMPD has generally not led to either a deeper understanding of personality disorder or personality assessment or new ideas about how to provide better care for people with personality disorder diagnoses. A significant portion of research has focused on narrow issues and appears to be driven in part by ideological differences between scholars who prefer Criterion A (personality functioning) or Criterion B (maladaptive traits). I trace these issues to ambiguity about the concept of personality functioning as defined in the AMPD and its conceptual distinction from personality traits and problems in living. In this paper, I reground these concepts in coherent and distinct definitions, elaborate upon the implications of their differences, and show how these differences can help clarify and reorient AMPD research to focus on generating clinically useful models for personality pathology and personality assessment.

The concept of personality functioning has been around for a long time, but it experienced a renaissance in the form of Criterion A Levels of Personality Functioning (LPFS) in the DSM-5 Alternative Model for Personality Disorders (AMPD; American Psychiatric Association, 2013) and its close analogue in the 11th edition of the International Classification of Diseases (ICD-11). However, the publication of the DSM-5 was followed by considerable ambiguity and debate regarding what personality functioning is as defined within the AMPD, and especially how it is different than problems in living and personality traits. Such equivocality is natural because personality functioning is indeed a complex construct that has been conceptualized in different ways across time and theoretical traditions. However, this has contributed to a heavily ideological and often antagonistic flavor to personality disorder (PD) research since the publication of the AMPD, often using methods that are limited in their capacity to substantially move the field forward. The goal of this paper is to delineate differences between these concepts more clearly, with the hope of encouraging more productive, transtheoretical, self-critical, and clinically relevant research.

Personality functioning, problems in living, and personality traits

From the perspective of the AMPD, personality functioning is a unitary latent construct that reflects a person’s overall level of personological maturity, as indicated by their interpenetrating capacities to maintain a consistent sense of self, pursue and achieve prosocial goals, empathize with others, and establish mutually fulfilling close relationships. Problems in living, described as “symptoms” or “impairments” in the AMPD, are the various ways in which a person’s life can go astray, and based on which the need for clinical attention generally arises. Personality traits describe enduring differences between people that probabilistically predict the way they live their lives and the kinds of problems they are likely to have. These differences have considerable implications for how the AMPD should be studied and used clinically. In this section, I will describe and elaborate upon how these constructs are defined in the AMPD and how these definitions can be used to distinguish them more clearly from one another.

Personality functioning

The AMPD defines personality functioning using the LPFS in terms of generalized severity in self (identity and self-direction) and other (empathy and intimacy) functioning. This model of personality functioning has roots in other constructs that are familiar in various clinical and particularly psychodynamic traditions (Blatt & Lerner, 1983; Loevinger, 1976; Masterson, 1988), such as ego strength (Nunberg, 1942; Symonds, 1951) and personality organization (Kernberg, 1967) as described by Bender et al. (2011) prior to the publication of DSM-5, Skodol (2012) coincident with the DSM-5 publication, and Sharp and Wall (2021) and Morey et al. (2022) more recently. From this perspective, personality functioning amounts to the degree that a person has matured psychologically, and the degree to which that underlying maturation is expressed in one’s current ability to adapt to their current environment. Embedded within this definition is the idea that, as a social species, the intertwined abilities to understand ourselves and get along with others are core capacities that enable adaptation (Livesley, 2003; Wright et al., 2022).

AMPD personality functioning has several distinguishing properties (Table 1). First, personality functioning is a unidimensional construct (Morey et al., 2011). Whereas Criterion A in the AMPD refers to one dimension (“Moderate or greater impairment in personality [self/interpersonal] functioning”), Criterion B refers to multiple dimensions (“One or more pathological traits”) (American Psychiatric Association, 2013, p. 761). Throughout the AMPD it is described in the singular, as the core of personality psychopathology, to be evaluated on a continuum (p. 762). The LPFS is referred to as a “continuum” (p. 771) of “generalized severity” (p. 772), whose elements are “reciprocally influential and inextricably linked” (p. 772). In other words, the AMPD and supportive publications (Bender et al., 2011; Morey et al., 2022; Sharp & Wall, 2021; Skodol, 2012) are clear in conceptualizing personality functioning as a single dimension.

Table 1. Differences between personality functioning, problems in living, and personality traits.

	Personality functioning	Problems in living	Personality traits
Origins	Psychodynamic developmental theories	Clinical observations and practical concerns	Quantitative trait psychology
Nature	Latent	Manifest	Latent
Between-person structure	Unidimensional by definition	Unclear and often unimportant	Hierarchical with 5 privileged factors
Stability	Relatively low	Low and highly contextualized	Relatively high; maladaptive traits are less stable than normal range traits
Diagnostic purpose	Determine whether a person meets criteria for personality disorder	Describe which kind of personality disorder best captures a person	Describe which kind of personality disorder best captures a person
Theoretical purpose	Define personality pathology in terms of psychological maturation	Limited	Describe general variation in human behavior and links this variation to individual differences in psychopathology
Clinical purpose	Capture common features of personality pathology, predict overall extent of dysfunction	Articulate specific risks and treatment goals	Probabilistically predict the kinds of problems in living that are likely
AMPD label	Level of personality functioning	Symptoms	Maladaptive traits
Specificity to PDs	Presumed to be specific but with limited evidence	Presumed not to be specific	Often presumed specific, but evidence strongly suggests otherwise
Adaptivity	Ranges from adaptive to adaptive along a single continuum	Maladaptive by definition	Can either be maladaptive or adaptive, but these domains have different properties (e.g., polarity, extremity, stability)

Second, and related, although it can be inferred from behavioral patterns and problems in living, personality functioning is conceptualized within the AMPD as a latent variable. It is described as a “core,” inner characteristic of the individual, and specifically in reference to the degree to which a person has developed the capacity to understand themselves and relate productively to others. It therefore ranges from adaptive (i.e., age-appropriate development or even flourishing) to maladaptive (i.e., underdeveloped).

Third, personality functioning is thought to explain why different PD symptoms and maladaptive traits all correlate with one another regardless of their content (Morey et al., 2011; Sharp et al., 2015; Wright et al., 2016). Consistent with this hypothesis, removing variance associated with personality functioning makes PD symptoms (Sharp et al., 2015; Wright et al., 2016) and maladaptive personality traits (Emery et al., 2023; Morey et al., 2022) less related to each other.

Fourth, personality functioning is thought to be what is specific to PD as a class of psychopathology (Bender et al., 2011; Morey et al., 2022; Sharp & Wall, 2021; Skodol, 2012). This feature distinguishes personality functioning from personality traits, since it is well-established that associations between personality traits and non-personality psychopathology is at least as strong as associations between personality traits and PDs (Kotov et al., 2010; Livesley & Jang, 2005; Samuel & Widiger, 2008). However, as described in detail below, the specificity of personality functioning is controversial, as is the specific way it is operationalized in the AMPD.

Fifth, personality functioning is thought to provide clinicians with information about the overall severity of a person’s personality problems (Hopwood et al., 2011; Nysaeter et al., 2023). This makes it different in terms of clinical management from traits and problems in living, because whereas personality functioning provides information about overall level of care (e.g., inpatient vs. outpatient, need for a safety plan, likelihood of alliance ruptures [Hopwood, 2018; Kernberg, 1971]) and may be relevant for challenges in the application of common psychotherapy factors (Rodriguez-Seijas et al., 2019), the specific constellation of traits and problems in living provide information about the type of care that might be most helpful for a particular person (e.g., therapeutic orientation, interpersonal approach, how to manage specific risk factors) (Bach et al., 2015; Mullins-Sweatt et al., 2020).

Sixth, personality functioning is thought to be somewhat less stable than personality traits (Haehner et al., 2023; Wright et al., 2016). The word current is italicized in the AMPD (p. 762) to emphasize its centrality in the definition of personality functioning. This is why the clinical focus in most therapeutic approaches to PD is not necessarily on changing a person’s personality traits, but rather in helping them mature more generally, so that they are more capable of adapting to a wider range of situations and coping with life’s inevitable stressors. In psychodynamic theory, one’s level of personality functioning should be a relatively consistent feature, in the sense that it represents a person’s degree of socioemotional development. For instance, it should be a fairly reliable marker of individual differences in adaptability across situations (Kernberg, 2004). At the same time, people develop over time as they mature, and in that sense it is a fundamentally dynamic variable (Fonagy et al., 2003; Sharp, 2020; Uliaszek et al., 2022). But stability with regard to personality functioning is a somewhat more complicated concept than can be fully accounted for by rank-order and mean-level estimates of stability. Key to understanding this perspective is recognizing that people can regress; under certain conditions or situations, people will tend to show impairments reflective of underdeveloped personality structure (Kernberg, 1967; Loevinger, 1983; Loewald, 1981). Thus, stability of personality functioning is most clinically relevant in terms of the short-term perturbations (as evidenced by increased problems in living during stressful times; e.g., Lewis et al., 2023) whose risk increase with the failure to sustain normative development. In both the general and this more specific sense, personality functioning should be less stable than traits, but more stable than highly contextualized kinds of problems in living that will be described in the next section.

Problems in living

Problems in living are the manifest difficulties clients may present with. Examples include getting into fights, being sad, spending valuable time and resources on finding and using substances, troubled relationships with loved ones, being truant, legal problems, etc. Clients can present with all sorts of problems, and one of the most important things to figure out in a clinical assessment is what problems they are having. Unlike either personality functioning or personality traits, problems in living are not couched within specific theoretical traditions. Their assessment is a practical reality, and all major mental health traditions account for them in one way or another. From the perspective of the clinician, these are usually the features of clients’ lives that we would like to help them change (Battle et al., 1966; Horowitz, 1979). From the perspective of the researcher, problems in living are the dependent variables that we generate models of latent independent variables to try to predict (Bagby et al., 2016; DeYoung et al., 2022).

Throughout the DSM, the presence of problems in living (described as a clinically significant disturbance in an individual’s cognition, emotion regulation, or behavior) is required for diagnosis (American Psychiatric Association, 2013). Thus, the presence of problems in living is not thought to distinguish PD from other classes of psychopathology, in contrast to the level of personality functioning. Problems in living are described in the AMPD as symptoms or impairments of specific disorder types (Anderson & Sellbom, 2018; American Psychiatric Association, 2013). The AMPD uses features of personality functioning to articulate these symptoms, with the idea being that whereas personality functioning is a latent propensity to experience all kinds of problems, individuals present with specific kinds of problems and this specific configuration is useful for specifying the kind of PD diagnosis they should be given. However, the use of personality functioning features to articulate specific problems in living has arguably made it more difficult to disambiguate these two kinds of variables.

Problems in living are also described with other terms in the literature, such as dysfunctions (Semerari et al., 2014), symptoms or impairments (AMPD), harms (Wakefield, 2007), consequences (Leising & Zimmermann, 2011), or maladaptations (McCrae & Costa, 1995; see below). In contrast to the AMPD, I use problems in living here to avoid medicalizing the issues people have, which could be due to latent pathological processes, environmental context, their transaction, or a range of other factors. However, other terms, such as dysfunctions, harms, adaptations, impairments, or consequences, could also work.

Perhaps because of the similarity in terminology (i.e., the word “function”) and the use of personality functioning features to articulate problems in living in the AMPD, the distinction between problems in living and personality functioning is not always clear in the literature. This challenge echoes a longstanding problem in diagnostic criteria and psychological assessment instruments that include both latent, dispositional features and manifest, contextualized behaviors in their items or criteria (McGlashan et al., 2005; Morey et al., 2022). In the personality literature, these two kinds of features are often distinguished using the terms basic tendencies/dispositions/traits and “characteristic adaptations” (DeYoung, 2015; McAdams & Pals, 2006; McCrae & Costa, 1995), although different authors have not always articulated the distinction between dispositions and adaptations in the same ways (Henry & Mõttus, 2020). These concepts can be distinguished in terms of level of inference. Dispositions are latent aspects of the person that cannot be directly inferred. In contrast, characteristic adaptations are ways that individuals adapt their behavior (or not) to the environment they are in. Adaptations don’t need to be inferred because they are observable and/or directly experienced as they are happening, in contrast to attributions about what underlying features of the person caused them to happen (such as personality functioning or traits). Dispositions are theoretically universal and are often argued to be more stable and more strongly linked to heritable sources of variability, whereas characteristic adaptations are thought to be specific to cultural and social context and thus more dynamic (DeYoung, 2015; McCrae & Costa, 1997). In the AMPD context, personality functioning and personality traits are both latent dispositions, although they differ in other important ways (Table 1). Problems in living are manifest problems adapting that are probabilistically related to those dispositions (Widiger & Presnall, 2013).

It would be ideal for measures of dispositions to avoid assessing characteristic adaptations. However, this is challenging because one can only infer latent constructs by the way they manifest in the real world (McCrae & Sutin, 2018). In that sense, nearly every questionnaire item reflects, at least in part, a characteristic (mal)adaptation (i.e., problem in living). There are ways to get around this challenge to some degree. One approach is to try to write items that are as dispositional (e.g., I tend to mistrust people) as opposed to contextual (e.g., I am often hurt because I trust people) as possible (DeYoung et al., 2022; Kandler & Rauthmann, 2022). A second approach is to focus on the aggregate of all problems or maladaptive trait scores within or across particular domains, in the hope that whatever is contextually specific about any one particular problem or adaptation would be lost in the aggregation. Within a particular domain, this could reflect a reasonable way to capture maladaptive traits, and this is in fact very similar to the logic of the personality trait model in the AMPD (Krueger et al., 2011). The aggregation of all possible problems may also reflect a reasonable way to estimate personality functioning (McCabe et al., 2022; Oltmanns et al., 2018), which inspired the AMPD (Hopwood et al., 2011; Morey et al., 2011) and is the standard approach in circumplex measures that distinguish general severity from specific style (Wright et al., 2009). However, such an aggregate of personality functioning is potentially dissatisfying for two reasons. First, a list of all a person’s problems is likely to conflate contextual stress (e.g., presence of an impoverished environment, systemic racism, trauma experience) with internal dispositions to have certain kinds of problems. Second, both traits and personality functioning are theoretical concepts, and the content validity of those concepts would be constrained by a purely bottom-up empirical approach. Essentially, personality functioning would no longer be a latent developmental dimension, it would instead consist of the common variance among all possible clinical problems. This could be consistent with one’s theoretical preferences (e.g., Oltmanns et al., 2018), but it would not represent personality functioning as defined in the AMPD.

It follows from their manifest as opposed to latent nature that, unlike personality functioning and traits, problems in living are mostly interesting at the level of the individual person. Although different patterns of covariance might be expected among certain problems, such as the tendencies to both drink too much alcohol and be too aggressive or the tendencies to have a messy bedroom and to show up late for work, these covariances can typically be explained by the probabilistic association between traits (or disorder types) and problems (in these examples, low agreeableness and low conscientiousness; Widiger & Presnall, 2013). In the AMPD, the specific patterns of problems can be used to diagnose legacy PD types. Among individuals, it is very possible that problems in living that are correlated between people are not both present in the same person, or even that there is an opposite pattern (Bringmann, 2021; Wright & Woods, 2020). For instance, some people become particularly meek and sentimental when they drink too much, and some people have messy bedrooms because they work so much they don’t make time for anything else. The important thing about problems in living is their presence in specific individuals, not their between-person correlations with other kinds of problems. I will discuss this issue in more detail below as it pertains to factor analyses of functioning measures.

A final important distinction is that problems in living are understood as highly contextual and thus less stable than either personality functioning or personality traits. This is in part because they are thought to be strongly influenced by environmental context. They are not features of the individual, they are things the individual does or feels or thinks when they are in certain kinds of situations (Hopwood et al., 2022). The fact that problems in living are thought to be both salient and dynamic is what makes them apt treatment targets (Battle et al., 1966). Different treatment orientations vary in their approach, with some focusing on the underlying personality structure that puts people at risk for maladaptive functioning (Levy & Scala, 2012), others on environmental factors that promote dysfunction (e.g., Tyrer et al., 2003), and most somewhere in between (e.g., Heard & Linehan, 1994). Whatever the presumed mechanisms, across orientations the common hope is that problems in living will remit with effective intervention.

Personality traits

Personality traits are relatively stable patterns of thought, feeling, and behavior that distinguish people from one another. The five-factor model (FFM) emerged in the late twentieth century as the most popular way to organize traits in non-clinical personality psychology (Digman, 1990). At its canonical level, people differ in their tendency to be neurotic, extraverted, open to experience, agreeable, and conscientious. These traits are somewhat correlated, so they can be aggregated into broader traits that collapse extraversion/openness and neuroticism/agreeableness/conscientiousness (DeYoung, 2006; Digman, 1997). Their breadth can also be parsed into narrower aspects (DeYoung et al., 2007), facets (Soto & John, 2017), and item-level nuances (Mõttus et al., 2017).

One of the important innovations of the AMPD was highlighting that factor analyses of DSM-IV PD symptoms and DSM-5 maladaptive traits resulted in a structure that is very similar to the normal-range FFM (Krueger et al., 2012; Widiger & Costa, 2012; Wright et al., 2012; Wright & Simms, 2014). This corroborated the evidence and arguments FFM advocates had been supplying for decades (Samuel & Widiger, 2008; Widiger et al., 2002). This is important because it shows that individual differences in the traits associated with PD are largely the same as individual differences in the traits that describe people in general, at least in WEIRD minority world cultures (Thalmayer et al., 2022).

However, there are important differences between maladaptive and normal range traits. First, maladaptive traits are less stable than normal range traits (Bleidorn et al., 2022). One possible reason for this is that measures of maladaptive traits are more saturated with content related to problems in living and personality functioning than measures of normal range traits (Morey et al., 2022). As described above, both personality functioning and particularly problems in living should be at least somewhat dynamic, thus sprinkling these features into normal range traits would have the effect of making them less stable. Second, normal range traits are normally distributed, with meaningful variation on both tails, whereas maladaptive traits are unipolar, with maladaptive and more extreme content typically measured on only one tail. Although one can write items to capture maladaptive content on both tails of FFM traits (Widiger et al., 2012), no study that has done so has ever fit such scales to a five-factor model (Crego et al., 2018, 2020). This is likely because the maladaptive content incorporates aspects of personality functioning and problems in living. Because all features of personality functioning and problems in living will tend to correlate with one another, distributing them to both ends of a dimension will tend to make the dimension collapse (Boudreaux, 2016; Pettersson et al., 2014). The diagnostic manuals have dealt with this primarily by ignoring the tails of traits that are less likely to be related to problems (low neuroticism, high agreeableness, high extraversion, and low openness/psychoticism). The one exception is conscientiousness, which is treated differently in the two major diagnostic schemes. This trait is given two separate domain scales in the ICD-11 PD criteria (disinhibition and anankastia). In the AMPD, it includes facets from both tails within the same domain (e.g., rigid perfectionism and impulsivity), but only facets related to low conscientiousness are aggregated to achieve a domain score.

In direct contrast to personality functioning’s role in defining PDs as a class, the specific purpose of personality traits in the AMPD is to describe the ways in which personality pathology is likely to manifest (American Psychiatric Association, 2013). This is useful because understanding a person’s particular profile or pattern of traits can help link their diagnosis back to a PD type (Bornstein, 2019) and can provide reliable information about what kinds of problems in living they are likely to have (Conway et al., 2016; Hopwood et al., 2009; Morey et al., 2007; Mullins-Sweatt & Widiger, 2010).

Summary

So far, I have delineated how personality functioning, problems in living, and personality traits can be distinguished in the AMPD and in clinical personality assessment more generally. Personality functioning is a unitary developmental dimension that captures what all PDs have in common, problems in living are the specific difficulties a patient may be experiencing, and personality traits reflect enduring patterns of thought, feeling, and behavior that distinguish people from one another. I will now go on to implications of these differences for how to productively advance PD research.

AMPD research

Loevinger (1957) established the three-step framework for how to generate and test measurement tools that continues to guide assessment research today. The first step is to develop a measure of the construct whose content matches the concept the researcher intends to measure. The second is to examine its internal validity, which often includes the internal consistency of items within scales and factor structure across scales. The third is to examine its external validity, which typically involves convergent, discriminant, criterion, and incremental validity. The core message of Loevinger’s influential work was that this process is not only a way to validate measures, but also a way to generate and test psychological theories. For that reason, the match between the analyses used to test the measurement model and the underlying theory of the construct is the chief concern. A significant amount of research on the AMPD has involved each of the analytic steps outlined in Loevinger, but on occasion this research has used methods and tested hypotheses that depart from the concepts being studied in important ways. I will focus here on two common research designs that are difficult to reconcile with the definitions of personality functioning, personality traits, and problems in living articulated above.

Factor analysis and the structure of functioning

The general goal of factor analysis is to identify the between-person covariance structure of a set of indicators. This has been the backbone of personality trait research and the primary method for distinguishing individual differences dimensions from one another for several decades. Its relevance for understanding personality functioning and problems in living, however, is more tenuous, given the nature of those concepts.

Structure of the personality functioning

As described above, the LPFS is, by definition, a unidimensional construct. However, a number of studies have factor analyzed the items of LPFS measures in order to identify sub-dimensions (Zimmermann et al., 2022). This raises questions about the fit between the unidimensional theoretical model and an analysis designed to distinguish multiple dimensions. Broadly speaking, three rationales have been given for these kinds of studies.

One rationale has been to examine the psychometric properties of personality functioning measures, with the explicit goal of evaluating whether or not they fit the unidimensional theoretical model articulated in the AMPD (or its close analogue in the ICD-11; Bach et al., 2021, 2023; Bliton et al., 2022; Morey, 2017, 2018; Sellbom et al., 2023; Weekers et al., 2019; Zimmermann et al., 2015, 2023). Such studies can be construed as psychometric tests of the measure’s validity vis-à-vis the theoretical model, rather than tests of the theoretical model itself. That is, they are not intended to examine or challenge the meaning of personality functioning in the AMPD, but rather to examine the degree to which a specific measure behaves in a way that is consistent with that meaning. This is an important question for applied personality assessment.

A second rationale is that the LPFS was actually meant to be multidimensional (Gamache et al., 2019; Huprich et al., 2018; Siefert et al., 2020; Sleep et al., 2019, 2023), in contrast to what its authors have described in both the DSM-5 and separate publications. For instance, Sleep et al. (2023) suggested that the AMPD requirement of two out of four LPFS elevations for a diagnosis of personality disorder means that the LPFS should be understood as multidimensional. But this is the way unidimensional scales work: the more positive responses to items, the higher the score, and the more likely that score is above some cutoff. The fact that there are multiple items on almost all scales does not make all scales multidimensional. Looked at the other way around, the implication would be that, because you could take any subset of indicators to achieve a total score, the set must be multidimensional. But trait models are not treated this way. We do not say that, because you could add any combination of high scores on the FFM up to calculate an aggregate (overall personality score), personality must be multidimensional. In fact, it is just the opposite—personality traits are multidimensional because summarizing them with any combination of high scores makes them less useful for distinguishing people from each other and often leads to inconsistent estimates (Hopwood et al., 2011; Revelle & Wilt, 2013; Watts et al., 2022). In fact, rules in which some number of indicators must be endorsed to achieve a cutoff are common in clinical assessment. For instance, the CAGE (Mayfield et al., 1974) is one of the most popular screeners for alcohol use problems. It has four indicators. Elevations on any two of them are considered potentially clinically significant because of empirical analyses concerned with the level of CAGE severity that is most likely to indicate clinical diagnosis (Dhalla & Kopec, 2007). This is the same rationale for the 2/4 rule for the LPFS (Morey et al., 2013), but the CAGE is not conceptualized as measuring multiple dimensions. One can reasonably debate whether personality functioning is a valid or useful construct, but it is unambiguous that the AMPD authors intended for it to be understood as a single dimension.

A third rationale is that the study is concerned with other, potentially multidimensional, theoretical models of personality functioning (e.g., Berghuis et al., 2013; Livesley, 1998; Parker et al., 2004; Verheul et al., 2008). In some cases, it is difficult to tell whether such models describe latent personality functioning dimensions or manifest problems in living. For instance, the Livesley model (Berghuis et al., 2013; Livesley, 1998) defines functioning in terms of failures to adapt and achieve life’s tasks. Although this model posits latent capacities in different functional domains, the item selection strategy on measures designed to assess them focus on content that refers to specific difficulties that people have in their lives, i.e., problems in living. Another example is Kernberg’s (1967, 1971, 2004) model of personality organization, which has interpenetrating elements (reality testing, maturity of defenses, and integrated sense of self and other) that are understood to be elements of a core underlying dimension having to do with the structures within the mind that rest upon healthy maturation and development, similar to the LPFS. In Kernberg’s model, these features are based on the ability of the individual to have matured to the degree that they are capable of certain kinds of functions. Their ordering is based on maturation, not differential content in the way that personality traits differ in content. Nevertheless, there is some evidence that these features can be distinguished via factor analysis (Lenzenweger et al., 2001).

This third group of studies do not pertain directly to the unidimensional model of personality functioning described in the AMPD. It could be argued that multidimensional models of personality functioning would be preferable to the AMPD conceptualization. However, distinguishing the scales of multidimensional questionnaire measures of personality functioning from traits is a considerable challenge (Oltmanns & Widiger, 2016). When multiple dimensions of personality functioning measures are distinguished, they closely resemble traits (Dowgwillo et al., 2018; Hopwood et al., 2018; Macina et al., 2023; McCabe et al., 2021; Rossi & Diaz-Batanero, 2024; Sleep et al., 2023; Uliaszek et al., 2023). This is likely at least in part because the items that load on the scales of these measures are very similar to each other, whether they are intended to measure traits or personality functioning. There are already good models of individual differences in traits based on decades-long tradition and item development, most notably including the FFM. The value of a complementary set of multidimensional models of personality functioning seems limited for personality theory or applied assessment.

Across all of these studies, the general results of factor analyses of personality functioning measures that were inspired by the LPFS are as follows:

1. there is typically a very strong first factor (Bach et al., 2021, 2023; Bliton et al., 2022; Macina et al., 2023; Morey, 2017; Sellbom et al., 2023; Siefert et al., 2020; Uliaszek et al., 2023; Williams et al., 2018; Zimmermann et al., 2015, 2023)

2. when multiple factors are extracted they are very strongly correlated (Bach & Hutsebaut, 2018; Zimmermann et al., 2015)

3. when efforts are made to distinguish the factors, they closely resemble traits in terms of their patterns of correlation with individual differences measures (Dowgwillo et al., 2018; Hopwood et al., 2018; Macina et al., 2023; McCabe et al., 2021; Rossi & Diaz-Batanero, 2024; Sleep et al., 2023; Uliaszek et al., 2023).

A parsimonious and AMPD-consistent interpretation of these findings is that while one can naturally find multiple dimensions in any large set of questionnaire items about the different ways personality can be maladaptive, personality functioning reflects a general disposition to have problems adapting, whereas personality traits reflect the ways in which those problems are likely to manifest.

Structure of problems in living

A case could be made for generating multidimensional models of problems in living. For instance, it might be relevant to distinguish the kinds of problems people have at home or at work/school, or in certain kinds of relationships (Plutchik, 1980; Weissman & Bothwell, 1976). However, from a clinical assessment standpoint, individuals can present with many combinations of problems and the point in clinical assessment is typically to understand the individual’s particular constellation, not how that constellation might be related to overall severity or a latent multidimensional model of individual differences. In many applied assessment traditions, this is treated as an idiographic issue, such that the concern is identifying a particular patient’s unique problems, regardless of how those problems might tend to covary with other kinds of problems in between-person data (e.g., Haynes et al., 1997, 2009; Smith et al., 2015).

Within the AMPD, predictions about the classes of problems people are likely to have is the role of traits. Searching for the underlying causes of covariation in problems in living risks conflating manifest outcomes of human variability with the latent factors that predict them; the main thesis of this paper is that these domains are already too often entangled. Moreover, individual differences across areas of problems of living have discriminant associations with trait constructs (Hopwood et al., 2009). While identifying the covariance structure of problems in living may have certain kinds of utility for research, this structure is likely to resemble that of traits, similar to studies of personality functioning as described above, and make traits and problems in living more difficult to distinguish.

Summary

It may be useful to factor analyze measures of personality functioning to determine the degree to which their items can be described by a single dimension as is asserted in the AMPD; by and large evidence suggests that they do, although one can find multiple dimensions that resemble maladaptive traits as well to varying degrees across measures. There might also be practical relevance to factor analyzing lists of manifest problems in living, although this value is limited given that the clinical value of problems mostly involves characterizing individuals in a within-person context. Overall, the value of factor analyses of indicators of personality functioning and problems in living is disproportional to the attention they have received in the literature given the tenuous theoretical rationale for examining the between-person covariance structure of either.

Comparative and incremental validity of personality functioning and personality traits

Another common type of study since the publication of the AMPD pits Criterion A and B against one another regarding their ability to predict disorder constructs or problems in living (Lim et al., 2019; Martí Valls et al., 2023; Nysaeter et al., 2023; Roche & Jaweed, 2023; Sleep et al., 2019). In the AMPD model, personality functioning is intended to provide a general indicator of overall severity, and traits provide an indication of what kinds of problems a person is likely to have. By analogy, personality functioning is similar to how pleasant the weather is overall, whereas traits are similar to the different reasons the weather could be bad—because it is too cold, warm, windy, wet, or dry. It would not make sense to ask: “what predicts how unpleasant I will feel when I go outside, the quality of the weather, or the combination of temperature, wind, and precipitation?” You would expect the aggregate of unpleasant temperature, wind, and precipitation to closely approximate whatever you could say about the weather in general. However, whereas the overall quality of the weather would offer a general answer to the question of whether it is pleasant or not, details about the temperature, wind, and precipitation would provide information about why it would be unpleasant, and what you should wear if you do go outside. Both answers are useful, depending on the question. Likewise, personality functioning indicates the overall severity of personality disorder, and traits indicate the style in which that disorder tends to manifest. As with this weather question, the conclusion of these kinds of studies is typically that Criteria A and B provide relatively little incremental information. From the perspective of the AMPD, this is how the model is supposed to work (Morey et al., 2022).

There is also a technical reason for the overlap between personality functioning and traits. Questionnaire measures of personality functioning, personality traits, and the dependent variables commonly used in these studies often have very similar items (McCabe et al., 2021). Thus, the “winner” of horse-race studies often depends on methodological issues like how many items or scales specific measures have and how these issues affect measurement reliability or model over-fitting, respectively. For instance, one approach has been to compare the validity of personality functioning as assessed by the LPFS-SR (Morey, 2017) and personality traits as assessed by the PID-5 (Krueger et al., 2012) to “predict” personality disorder symptoms as measured by an instrument like the PDQ-4 (Hyler, 1994). One of the items on the LPFS-SR is “I mainly act in the moment, rather than focusing on long term goals.” One of the items on the PID-5 is “I feel like I act totally on impulse.” One of the items on the PDQ-4 is “I have done things on impulse.” All of these items capture features of personality functioning (in that more mature people are less likely to ignore the future consequences of their behavior), personality traits (in that people low in conscientiousness are more likely to be impulsive), and problems in living (in that impulsive behavior is usually problematic), regardless of what they were designed to assess. Given this overlap, questions about whether one of these variables relates more than another to the third have relatively little meaning.

There may be valid reasons to examine how personality functioning and personality traits are related to problems in living. For instance, such research could test the hypothesis that personality functioning has a more nonspecific pattern of correlation than personality traits, as would be predicted by the AMPD. It would be useful to better understand the kinds of clinically important questions for which general severity is sufficient as opposed to those in which more specific traits add useful information, and to know when it is useful for that general severity score to include content specified in the LPFS or could be a simple aggregate of all maladaptive traits and problems. These kinds of studies would be especially useful to the degree that measures of personality functioning and traits are more distinguishable, which can be achieved to some degree by using normal-range measures of personality traits (Morey et al., 2007, 2022) and by operationalizing the dependent variables of such studies as actual and clinically relevant manifest problems in living, as opposed to general tendencies with very similar indicator content as traits and personality functioning. It is also true that, given their strong overlap in cross-sectional data, it is inefficient and of limited utility to assess both personality functioning and maladaptive traits using long questionnaire measures in clinical practice. However, studies that pit personality functioning questionnaires against maladaptive trait questionnaires to predict questionnaire measures of psychopathology or life difficulties conflate personality functioning, personality traits, and problems in living, deviate from the intention and purpose of the AMPD, and are unlikely to advance knowledge about how to assess personality or help people with PD diagnoses.

Is personality functioning a useful concept?

These issues with AMPD research notwithstanding, personality functioning is clearly difficult to disentangle from personality traits using questionnaires and it is often conflated with problems in living in measures, research studies, and models of PD. This raises valid questions about whether personality functioning is actually something worth distinguishing. Its staying power, presence, and importance in the clinical literature and popularity in clinical practice is perhaps notable but only indirect evidence for the validity of the concept. Moreover, its practical utility could be largely achieved by simply averaging all maladaptive trait scores and/or problems in living (McCabe et al., 2022; Oltmanns et al., 2018; although, see Watts et al., 2022). A reasonable person could challenge the utility of personality functioning on the grounds that it has less empirical support than traits, is the most difficult to measure of the different components of the AMPD, and is most often confused with personality traits and problems in living in both the diagnostic manuals and in conceptions of PDs. Indeed, it is fair to argue on these grounds that it would be more parsimonious to do without it (Sleep et al., 2022).

There is some indirect evidence supporting the presence of personality functioning. First, the scales of maladaptive trait measures correlate more strongly with one another than the scales of normal range trait measures, and as described above, no study yet has fit a five-factor model to measures with maladaptive extremes on both tails of the five-factor traits. If personality traits were truly distinct and the increased probability of dysfunction was purely a function of trait extremity, this should not occur. One plausible explanation of these correlations is that the maladaptive content that has been added to normal range traits captures personality functioning, and the presence of personality functioning causes those traits to correlate (Emery et al., 2023; Morey et al., 2022).

Second, maladaptive traits are less stable than normal range traits (Bleidorn et al., 2022). If the only reason they are related to problems in living is because of extremity (Samuel et al., 2010; Suzuki et al., 2015), adding maladaptive extremity to trait tails should not impact their stability. A parsimonious explanation would be that this maladaptive extremity includes some combination of personality functioning and/or problems in living, both of which are less stable than normal range personality traits. Wright et al. (2016) decomposed the variance that all PDs have in common from traits that distinguish PD types, and found that the general factor (i.e., personality functioning) was substantially less stable than the traits. Haehner et al. (2023) compared the longitudinal stability of personality functioning and traits using distinct measures of each. They found that neuroticism and personality functioning were less stable than the other four traits. This could be interpreted either as evidence that personality functioning overlaps most with neuroticism (and indeed, neuroticism is generally the strongest trait correlate of low personality functioning; e.g., Sleep et al., 2019) and neuroticism is less stable than the other traits, or that neuroticism is the trait that has the most personality functioning and/or problems in living embedded within it, even within the normal range (Indeed, it is the trait that correlates most strongly with psychopathology, in general; e.g., Kotov et al., 2010).

Third, longitudinal research shows not only that traits develop over time, but they tend to do so in the direction of greater adaptivity (Bleidorn et al., 2022; Donnellan et al., 2007; Roberts & Mroczek, 2008). Particularly during young adulthood, which is also the time when prevalence of PD tends to peak and then decrease, people tend to become less neurotic, more agreeable, and more conscientious. Notably, these are the traits that are strongly related to PD symptoms on average (Samuel & Widiger, 2008). Schwaba et al. (2022) recently showed that trait changes in the direction of maturation are correlated with one another within-person, albeit imperfectly. If traits were truly distinct from one another, what would explain why they change together and in a more adaptive direction as people age? One parsimonious explanation is that as personality matures (i.e., personality functioning increases), it affects changes in all traits and particularly those most strongly related to adaptation.

From a practical standpoint, eliminating personality functioning from PD diagnosis would either require an alternative basis for distinguishing PDs from other kinds of psychopathology or lead to abandoning PDs as a concept (Hopwood, in press). To the degree that the PD concept communicates something clinically important (Wright et al., 2022), it seems useful to distinguish PDs. But given that PD phenotypes are not more stable than other kinds of psychopathology (Hopwood et al., 2022) and they are no more strongly related to personality traits than other kinds of psychopathology (Kotov et al., 2010; Samuel & Widiger, 2008), what is distinct about PDs, if not personality functioning? The absence of an answer to this question amounts to an assertion that there is no such thing as PD.

Directions for personality functioning research

Productive, knowledge-generating research depends on an alignment between methods and the hypotheses those methods are used to test (Loevinger, 1957; Watts et al., 2023). I have argued that some AMPD research has departed from the definitions of the major constructs articulated in that system. AMPD research has also made it clear that measuring personality functioning with questionnaires in cross-sectional data makes it very difficult to distinguish it from either traits or problems in living. Questionnaires designed to measure personality functioning and questionnaires designed to measure maladaptive traits are inevitably going to have pretty similar items, and these items are likely to include content related to manifest problems. Cross-sectional questionnaire studies have been the backbone of personality trait models and have helped spur research on personality functioning relatively quickly following the publication of the AMPD, but research will need to go beyond this approach to critically test hypotheses about personality functioning. Here, I describe six directions that could help move AMPD research, and in particular research on the validity and relevance of personality functioning, forward.

Clinical utility

The ultimate goal of the clinical personality assessment and diagnosis is practical (Bach & Simonsen, 2021). There are serious questions about the clinical utility of personality assessment, with some studies being supportive (Durosini & Aschieri, 2021; Waszczuk et al., 2022) but many others failing to find added clinical benefit of personality assessment, in part driven by limitations in research design (Kamphuis et al., 2021). Weekers et al. (2024) recently tested the utility of the AMPD model in a randomized control trial in which patients were randomly assigned to either an AMPD or a categorical PD assessment. Neither patients nor clinicians rated the AMPD as more useful, satisfying, or motivating than the categorical approach. Thus, in addition to testing clinically relevant predictions about the clinical utility personality functioning, traits, and problems, the field has work to do in demonstrating the value of PD assessment in general and dimensional models in particular.

Multimethod assessment

The generation of multimethod, and particularly non-verbal (i.e., questionnaire or structured interview) measures of personality functioning (e.g., Hibbard et al., 2010; Meehan, 2021; Natoli & Rodriguez, 2023; Ruchensky et al., 2022; Sinnaeve et al., 2021; Weekers et al., 2020) would perhaps allow for more precise tests of the differences between personality functioning and traits. The psychoanalytic tradition has typically promoted the assessment of personality functioning within the interactions that occur during clinical encounters, when specific maneuvers by the clinician could be used to test for the maturity of defenses, quality of object relations, etc. (Kernberg, 1981; McWilliams, 1999; Sullivan, 1954). Some of these techniques have been standardized (Clarkin et al., 2004; OPD Task Force, 2008; Perry & Cooper, 1989; Rosenbaum et al., 1997; Vaillant & Drake, 1985), however data from such measures have rarely been used to test basic hypotheses about personality functioning and how it differs from personality traits or problems in living. There are also reasons to be skeptical about the ability of clinician data to provide information above and beyond that provided by standardized instruments (Samuel et al., 2013). Moreover, questionnaires are imperfect albeit common measures of both traits and personality functioning; assessing personality functioning with one method and personality traits with another would only superficially distinguish the two sets of constructs by conflating method and construct variance. Thus, ultimately multimethod approaches would need to go beyond simply finding new methods to assess personality functioning, to leveraging different methods to test theory-guided hypotheses about the differences between these constructs (Bornstein, 2009).

Longitudinal designs

Longitudinal designs may also help discern personality functioning from personality traits. Some research suggests that there are differences in longitudinal stability between personality functioning and traits (perhaps other than neuroticism) as described above (Haehner et al., 2023; Wright et al., 2016) and some studies have suggested that different aspects of personality disambiguate more in short term stability studies than in longitudinal data (Eggermont et al., 2023; Franssens et al., 2023; Ringwald et al., 2021; Roche, 2018; Roche et al., 2023; Sinnaeve et al., 2024). Here again, however, these studies do not necessarily get at the specific temporal predictions underlying personality functioning, namely that it should explain why different features of personality develop in concert with one another (e.g., as shown by Schwaba et al., 2022) and that within this general context people who mature less are less likely to cope well with life’s stressors, as predicted by the concept of regression.

More specific tests of different theoretical models of personality functioning

Different models of personality functioning might make different kinds of predictions. For instance, there are models of personality development that don’t necessarily require the level of inference or specific mechanisms (e.g., the presence of developmental milestones, the importance of attachment relationships, the generation of a sense of self, self-awareness) asserted by psychodynamic theories, and thus make different predictions about personality development and adaptivity (e.g., Buss & Plomin, 1975; L'Abate, 1994; McCrae & Costa, 2003; Roberts & Nickel, 2017; Specht et al., 2014). There is also significant variability within psychodynamic theory about the mechanisms of personality development and the implications for functioning (e.g., Blatt & Luyten, 2009; Fonagy & Target, 2000; Loevinger, 1976; Kernberg, 1967). Furthermore, the language used to articulate many theories of development has often been loose with respect to how personality functioning differs from personality traits and problems in living. For instance, although the Livesley (1998) model is clearly aimed at measuring adaptive failures (i.e., problems in living), similar to the AMPD the model also organizes these failures into evolved capacities. In contrast to the AMPD, it distinguishes these capacities via factor analytic methods (Berghuis et al., 2013). Overall, a careful analysis of the propositions of these models could lead to a framework for critical tests of their differences (Eronen & Bringmann, 2021).

A more self-critical approach

One of the potential upsides of the AMPD is that it integrates different theoretical perspectives (Mulay et al., 2018; Waugh et al., 2017). However, it is natural that some people are more attached to one perspective than another and this creates a tendency to frame different points. Some people do not like personality dysfunction because of its psychodynamic origins, whereas others do not like traits because of its origins in quantitative psychology. While many scholars conduct relatively theory-neutral research aimed at identifying the strengths and weakness of different aspects of the model, at times alternative perspectives are miscast and even caricatured, and it seems as if studies are designed with the purpose of finding the flaws in the parts of the model that is disfavored, as opposed to learning something new about personality or helping people with PD diagnoses.

To be clear, scientific progress is often enhanced by the clash of perspectives. People from different points of view can help hone one another’s thinking and may be the most creative in generating critical tests of models they do not favor. At the same time, our field may have leaned too far into this way of conducting research in the AMPD era, as evidenced by a literature that is at times antagonistic and narrowly focused. An honest look reveals that no perspective comes close to any kind of ultimate truth about personality, PD, or personality assessment, and that the important fight is not between people who prefer different models, but between all the things we collectively do not know and the kind of knowledge that could help people with PD-related problems. In this sense, a more humble and self-critical approach, in which researchers concentrate on the limitations of their own preferred models and measures, might be a useful corrective.

For instance, proponents of the trait perspectives that would eliminate personality functioning on the grounds that it is redundant have difficult questions to answer, including: (a) why are maladaptive traits more correlated with one another than normal range traits?, (b) why are maladaptive traits less stable than normal range traits?, (c) why do traits tend to track toward maturity over the lifespan within and across people, (d) what is the difference between personality disorder and other kinds of disorder?, and (e) why is there no evidence that a five-factor model can be fit to trait data with maladaptive variance on both tails of each trait? To be clear, the conceptual coherence and measurement of personality functioning is imperfect, but the presence of personality functioning provides a plausible and straightforward answer to all these questions.

Conversely, those who assert the value of personality functioning would do well to focus on challenging questions such as (a) what specific theoretical predictions does the concept of personality functioning make about development and mental health?, (b) what evidence supports the psychodynamic account(s) of personality development and its relation with mental health problems?, (c) how can personality functioning be measured in a way that provides distinct, useful, and valid information? (d) what evidence is there for developmental maturation to explain the fact that all maladaptive traits and problems in living tend to correlate with one another across people?, and (e) do within-person temporal models of personality processes support the underlying theory of personality functioning that achieving certain developmental milestones is associated with lower likelihood of regression under stress?

Changes to journal policies to promote rigorous research

These ideological issues are amplified by the limitations of our methods to carefully examine different AMPD elements in a way that stays true to their definitions. Methodological limitations are in turn promoted by academic incentive structures (Nosek et al., 2012; Shaw & Nave, 2023; Van Dalen & Henkens, 2012). It is relatively easy to collect cross-sectional questionnaire data in convenience samples from WEIRD populations and conduct factor or regression analyses on those data. Studies like this can be done and published quickly and researchers are incentivized to publish quickly. It is not the case that cross-sectional questionnaire studies with convenience samples have no value, but they are vastly over-represented in PD research (Bornstein, 2015) and methodological ease fosters studies that disfigure constructs and ultimately impede progress. Several major journals (e.g., Journal of Psychopathology and Clinical Science, Journal of Personality and Social Psychology) have noted how the preponderance of studies with relatively low-resource methods have stilted, constrained, or distorted research and have made it explicit that such studies would have a low chance of publication; this would be one relatively easy way to improve the rigor of PD research in general and promote the kinds of studies that could provide valid tests of the distinctiveness and utility of personality functioning in particular.

Conclusion: the tripartite model of personality diagnosis

Climate change is a unitary developmental process that manifests in different ways depending on spatial considerations such as the region of Earth and the relation of the Earth to the Sun, and its manifestations are often harmful to the people who experience them. These features of climate change are closely analogous to the tripartite model of AMPD diagnosis domains personality functioning (climate change), personality traits (spatial dimensions), and problems in living (bad weather events, etc.). If a person were to measure climate change processes by what the weather was like at one particular place and time, they would miss cyclical or geographic variability and thus be unable to link their measurement back to underlying long-term changes in the earth’s temperature. This would leave room for political debates driven by ideology and associated misunderstandings and mischaracterizations, all of which would inhibit progress. We have seen ideology and methodological limitations have this effect in both climate change debates and AMPD research. In contrast, intellectual humility and a less ideological, more self-critical approach could contribute to more sophisticated designs that could advance PD assessment by determining how to distinguish personality functioning from problems in living and personality traits (Zimmermann et al., 2022).

I am on record before the publication of the AMPD supporting a tripartite model of personality disorder diagnosis that includes personality functioning, problems in living, and normal-range personality traits (Hopwood, 2011). Although my perspective has changed in certain respects (Hopwood, in press; Wright et al., 2022), I still think about this issue basically the same way because these three sets of constructs communicate distinct and clinically useful information. But I also acknowledge that:

• features of these domains are intertwined in nature,

• they overlap so much in terms of between-person covariance in cross-sectional self-report data that they are almost indistinguishable,

• there are serious empirical questions about the clinical utility of these distinct domains and of personality assessment in general,

• reasonable people could disagree about the clinical value of and empirical evidence for personality functioning as a latent developmental dimension,

• reasonable people could disagree about whether PDs demarcate a unique and clinically important class of mental health problems,

• very little is known about how to assess personality and related problems outside of WEIRD cultures,

and assert that:

• the transtheoretical nature of the tripartite model is an advantage in a field that continues to be too ideological in both research and practice,

• there are many important considerations that cannot be examined in studies of the between-person covariance of self-report questionnaires in cross-sectional data, which are overrepresented in the literature,

• the prevalence of studies that use WEIRD convenience samples, which are also over-represented, impedes progress in personality disorder research and personality assessment in general,

• these kinds of studies are probably so common mostly because they are easy to do and researchers are incentivized to publish a lot, suggesting a need to rethink incentive structures in academic psychology and journals that focus on personality assessment and PD.

Declaration of interest

No potential conflict of interest was reported by the author(s).