When the DSM gets it wrong: vulnerable narcissism and autism
Am I becoming a DSM truther? (Let's hope not!)
Hello! I hope your summer is kicking off well. I’m mostly doing my best to dodge the rising East Coast heat. Summer is my least favorite season (I tell uncomprehending sun-lovers to think of me as a friendly vampire).
I’ve been writing recently about book proposal-related topics (e.g., problems with biography, the root of creative insight), but this latest post marks a return to form with a research deep dive. I hope you enjoy!
The more I read, the more I become convinced our current approach to describing and defining psychiatric disorders is hopelessly flawed.
It’s not possible to cover all my research learnings here. (My Reader app says I’ve annotated over 100 research studies on neurodivergence… oh god).
So I’ll illustrate what I’m talking about with an example: problems with the DSM’s approach to narcissistic personality disorder (shorthand: NPD or narcissism) and how it obscured a link to autism.
Two tracks: clinical practice v. research
When it comes to our understanding of psychiatric conditions, there are separate tracks: clinical, and research. (This is… mostly relevant, so bear with me.)
The clinical track is represented by the DSM (and its international equivalent the ICD), which are used by clinicians as diagnostic tools. Clinicians are the ones who diagnose and treat patients. They adhere to official manuals for a number of reasons: credibility, consistency, practicality.
The DSM is published by the American Psychiatric Association, and any revision goes through a multi-year process involving expert working groups, public comment, internal review panels, and a vote by the APA’s Board of Trustees. This is diagnostic definition-by-committee; a process that is biased toward conventional thinking and slow to evolve.
The research track, on the other hand, is comprised of the decentralized patchwork of scientific research. Whereas clinical approaches are mostly dictated by the singular authority of the DSM, research teams are free to pose whatever hypotheses they like and see what the evidence says. They build on past ideas, but they’re not bound by them.
Research findings sometimes conflict, which can be a good thing. Conflicts surface methodological flaws, as we’ll see in a study below, leading to breakthroughs.
Changes to the DSM require years of vetting and layers of process. On the research side, well-evidenced and persuasive findings can rapidly float to the top and springboard further investigation. That’s not to say research is perfect. But it’s the more innovative and up-to-date of the two tracks.
Which is all to say: If you want to truly understand psychiatric disorders, dogmatic adherence to the DSM is not the way. We’ll see that in action in a bit.
Rethinking the DSM’s approach to narcissism
More relevant context: the DSM’s approach to narcissistic personality disorder.
NPD was added to the DSM-3 in 1980. Contextualized as one of 11 personality disorders, narcissism’s diagnostic definition was heavily influenced by psychoanalytic theory.
Today, the DSM-5-TR defines NPD as “a pervasive pattern of grandiosity (in fantasy or behavior), need for admiration, and lack of empathy, beginning by early adulthood and present in a variety of contexts,” and sets forth nine sub-criteria, five of which must be present for diagnosis.
Although the DSM presents NPD as a single disorder, by 2009 researchers were proposing two distinct phenotypes based on clear variability within the NPD population: grandiose narcissism and vulnerable narcissism.
What’s the difference between these two types?
Grandiose narcissism is marked by arrogant, aggressive, attention-seeking, and exploitative behavior, a lack of empathy, and self-serving beliefs about one’s own importance and entitlement. “These individuals can be socially charming, despite being oblivious to the needs of others, and are interpersonally exploitative,” a 2015 study notes.
Vulnerable narcissism is marked by hypersensitivity to criticism or failure, avoidant or inhibited social behavior, perfectionism, and acute, intrusive shame. “Interpersonally these individuals are often shy, outwardly self-effacing, and hypersensitive to slights, while harboring secret grandiosity.”
Growing appreciation of narcissism’s variability prompted a slight change in 2013’s DSM-5 edition, which allowed for more nuanced presentations.
(Illustrating the DMS’s definition-by-committee approach, NPD was initially proposed to be dropped from the DSM-5 for lack of empirical support, but a group of clinicians and researchers successfully lobbied for its continued inclusion. This supports my point above, that the DSM is biased toward conventional thinking.)
The DSM today describes a “single, relatively homogenous syndrome” that doesn’t account for the wide range of real-world presentations. “Individuals with narcissistic personality disorder may be grandiose or self-loathing, extraverted or socially isolated, captains of industry or unable to maintain steady employment, model citizens or prone to antisocial activities,” the 2015 study noted.
Nor does the DSM acknowledge what researchers now align on: there are (at least) two different kinds of narcissism. (Some researchers present evidence for a third, “high-functioning” category alongside “grandiose” and “vulnerable.”)
Researchers aren’t sure if these types even belong under the same label. Grandiose and vulnerable narcissism could be extreme ends of the same continuum, but it’s also possible they’re different conditions entirely. This is a focus of current research.
The broader problem with the DSM and personality disorders
We just saw problems with the DSM’s treatment of NPD. But there’s a growing belief that the DSM’s approach to all personality disorders is flawed.
Researchers increasingly believe that personality disorders aren’t distinct, diagnosable pathologies. Instead, they are loosely defined, overlapping clusters of traits that lack strong empirical support.
A core goal of diagnostic manuals like the DSM is to guide treatment through diagnosis. But in addition to lacking empirical support, the DSM’s personality disorders don’t reliably map onto effective treatment plans. There’s simply too much variability within each category.
If these diagnostic labels don’t reflect well-defined conditions, nor do they inform effective treatment, is the current system really serving us?
Many researchers think not. There’s a growing trend toward using the Five Factor Model (FFM)—a trait-based system that categorizes personality along five broad dimensions and their facets—to study psychological conditions. Against that backdrop, some experts believe that personality “disorders” are extreme or maladaptive variants of normal personality traits.
This aligns with a trend in autism research. Guided by genetic evidence, researchers increasingly see most forms of autism as a concentration of “normal” traits. (Normal because they circulate in the general population). Under this framework, autism’s individual traits are not atypical. What is atypical is their concentration in a single individual.
(I covered this polygenic understanding of autism in a prior post.)
New research: Autism and vulnerable narcissism
All of this sets the stage for a recent study that found a link between autism and vulnerable narcissism.
Researchers had already observed through the lens of the Five Factor Model that autistic individuals and people high in vulnerable (but not grandiose) narcissism tend to score high on neuroticism. Neuroticism is one of the eponymous Five Factors and is marked by emotional reactivity, self-doubt, and sensitivity to stress.
These shared links to neuroticism raised an interesting possibility: could autism and vulnerable narcissism be directly connected?
A team of researchers in Milan set out to test that idea. In 2024, they published the first study to directly measure traits of vulnerable and grandiose narcissism in autistic adults (without intellectual disability).
They administered a widely used narcissism questionnaire (the PNI-52) to 87 autistic adults and compared their responses to a large normative sample.
The results were striking: autistic participants scored significantly higher on vulnerable narcissism traits, but not on grandiose traits.
Roughly one-third of participants scored high enough to suggest a pathological narcissism diagnosis. Pathological narcissism is when narcissistic traits are severe enough to interfere with emotional functioning, relationships, or self-regulation. This risk was driven almost entirely by the group’s scores on the vulnerable subscale, not the grandiose one.
The study also found that vulnerable narcissism was strongly correlated with autistic social communication difficulties. The more social difficulties someone reported on an autism measure, the more likely they were to score high in vulnerable narcissism.
A shared sensitivity?
So why the link between autism and vulnerable narcissism?
The researchers propose a couple explanations.
One is based in neuroticism, the personality trait associated with emotional sensitivity, self-doubt, and difficulty regulating negative emotions. Neuroticism tends to be elevated both in autistic individuals and in people high in vulnerable narcissism. This was the common link that initially prompted the study. In other words, this presence of this trait is a shared causal factor.
Another explanation involves the more specific trait of sensory sensitivity. This trait is commonly found in autism but was only recently linked to vulnerable narcissism. It involves heightened sensitivity to sensory and emotional input, which can make social environments feel overwhelming.
The researchers propose that both autism and vulnerable narcissism result from this underlying sensitivity. In people who are especially reactive to criticism, uncertainty, or social overload, these sensitivities may lead to coping strategies like withdrawal, perfectionism, or internalized shame, which are patterns that appear in both profiles.
This raises a deeper question: to what extent are vulnerable narcissism and autism truly distinct conditions, versus diverse presentations of a shared genetic foundation?
That question is likely to shape future research.
Why this link was missed before
And here’s the aspect of the study that underscores shortcomings in the DSM’s approach to NPD.
There were earlier studies that found no correlation between autism and narcissism. At first glance, the 2024 study seems to contradict those studies.
But there’s a clear explanation: the prior research treated narcissism as a single construct, following the DSM’s lead. Those studies did not disentangle grandiose and vulnerable narcissism scores and test them separately.
Once the two kinds of narcissism are independently tested, a different picture emerges. The association between autism and vulnerable narcissism, previously hidden, becomes visible.
This is what I meant earlier when I said that conflicting research findings can sometimes expose methodological flaws.
In this case, the flaw was relying on the DSM’s definition of narcissism as a single condition — an approach that has been criticized for more than a decade — in studies of narcissism and autism. The Milan team broke from that pattern by treating grandiose and vulnerable narcissism as distinct constructs. That innovation allowed them to uncover links to autism that had previously been obscured.
What else are we getting wrong?
This study illustrates the value of getting granular with psychological profiles and staying open to new frameworks. What we currently label as a single disorder may turn out to be multiple, distinct conditions that never should have been grouped together in the first place. Or, distinct conditions may end up having a common basis and should be linked together.
More broadly, our system of “disorder” labeling may be flawed. Many so-called pathologies might just be maladaptive concentrations of common traits, deemed maladaptive only because they clash with a society built around the more prevalent templates.
Rethinking “disorders” this way could open the door to a better understanding of difference, and to more effective support for traits that cause distress across a range of conditions, rather than siloed treatments based on arbitrary line-drawing.
Did you enjoy this post? Ways to support my work—for free!
1. Subscribe for regular updates and 2. Tap below to heart this post so others discover it.
Looking for more to read? Check out these past posts:
My autistic special interests: the fire that burns itself out
Evolutionary mismatch: just one part of the neurodivergence story
Research cited in this post:
Eve Caligor, Kenneth N. Levy, Frank E. Yeomans. 2015. Narcissistic Personality Disorder: Diagnostic and Clinical Challenges. American Journal of Psychiatry 172(5):415-422. Read it online.
Giovanni Broglia, Veronica Nisticò, Bianca Di Paolo, Raffaella Faggioli, Angelo Bertani, Orsola Gambini, Benedetta Demartini. 2024. Traits of narcissistic vulnerability in adults with autism spectrum disorders without intellectual disabilities. Autism Research 2024 17:138-147. Read it online.
Stay curious,
Laura
This is fascinating. I’d love to know more about the etiology of both forms of narcissism and a revisit to the psychoanalytic frames - identity formation, trauma/abuse, etc.
This is really interesting. Thanks. I have often wondered whether psychopathy is actually antisocial personality disorder, as dsm calls it, with comorbid autism. I would love to see this tested. And there is an additional neglected aspect of vulnerable and grandiose narcissism: Some people oscillate between both. I havent read up on your references. But my first query would be whether a questionnaire measures of narcissism are validated for detecting narcissism in those with autism. Most arent. This sounds like a potential problem with interpreting the results. But i think as a research question its important.