The idea that self-centered people give themselves away through specific, detectable behaviors has real research behind it, though not always the research most commonly cited for it. In a 2010 paper in the Journal of Personality and Social Psychology, Mitja Back, Stefan Schmukle, and Boris Egloff studied 73 psychology students meeting each other for the first time in a round-robin design, where every participant briefly interacted with every other participant, producing 2,628 total first-impression pairings. The researchers measured each participant’s narcissism separately, then asked what made narcissistic people come across as likeable within seconds of meeting them, and what gave them away.

The answer was specific and observable. People who scored higher on narcissism were independently rated by strangers as more physically attractive in grooming and dress, more expressive in facial expression, and more confident in body language, and these observable, external cues statistically explained why they were initially rated as more likeable. The charm was not an illusion exactly. It was mediated by a real, measurable set of self-presentation behaviors, more polished grooming, warmer initial facial expression, a more confident physical bearing, that a narcissistic person was consistently more likely to display in the first few seconds of contact.

This is one study, and the charm effect wore off

This is one dataset from psychology students at a single university, not proof that every narcissistic person is more attractive or that every stranger will read these cues identically. But the design, tracking the same set of first meetings across thousands of paired interactions, makes the finding considerably more than an anecdote. Notably, the researchers also tracked what happened over repeated interactions: the initial likeability boost narcissistic participants received did not hold up as the group got to know each other better over several weeks, a pattern that has been replicated in later work from the same research group tracking narcissism and popularity over longer stretches of time. The specific behaviors that create a strong first impression are not the same behaviors that sustain a good one.

The popular “tell” that turned out not to be real

Not every widely believed marker of selfishness holds up as well under scrutiny. One of the most commonly repeated claims, both among laypeople and in some earlier research, is that narcissistic people overuse first-person singular pronouns, “I,” “me,” “my”, a pattern sometimes called I-talk, treated as a kind of linguistic tell for self-focus. In a 2015 large-scale study in the Journal of Personality and Social Psychology, Angela Carey, Melanie Brucks, and their colleagues tested this claim directly and rigorously, pooling data from 4,811 participants across five separate labs, using five different established narcissism measures, in both English and German, across multiple types of writing and speech samples.

The result across this large, multi-lab dataset was essentially no relationship at all between narcissism and first-person pronoun use, an overall correlation the researchers reported as close to zero. This directly contradicts a claim that has circulated widely enough to be treated as common knowledge in some pop-psychology and self-help writing. It is a useful reminder that “obvious” behavioral tells are not automatically real ones, and that a plausible-sounding linguistic marker can fail a rigorous, adequately powered test even after years of being treated as settled.

What actually seems to hold up, and why the distinction matters

Read together, these two lines of research point toward a more precise account of what self-centered people reveal, and how. The Back, Schmukle, and Egloff findings suggest that narcissistic self-presentation shows up reliably in visible, physical channels: grooming, expressive display, confident bearing, the kind of cues that create a strong impression before anyone has spoken more than a few words, and that fade as a genuine, sustained relationship develops. The Carey and colleagues findings suggest that at least one widely assumed verbal tell, pronoun frequency, is not a reliable signal at all, despite feeling intuitively obvious.

That distinction matters for anyone trying to actually read people rather than rely on folk psychology. The visible, physical self-presentation cues identified in controlled first-impression research have real, replicated statistical backing. A specific word-choice habit that sounds like it should be diagnostic, and gets repeated as though it were established fact, does not.

Part of why the pronoun myth spread so easily is that it has surface plausibility working in its favor. It seems intuitive that a person preoccupied with themselves would talk about themselves more literally, using “I” and “me” at a higher rate than everyone else. But Carey and colleagues note that first-person singular pronoun use is nearly universal and heavily shaped by grammatical necessity, topic, and context, a person describing their own weekend will naturally use more “I” statements than one describing a news event, regardless of how self-focused they actually are as a person. That context-dependence appears to have swamped any signal from personality in their data, across five labs and two languages, which is part of why the finding held up as a genuine null result rather than a fluke of one small sample.

What this does not prove

Neither study offers a checklist that reliably identifies a selfish or narcissistic person from a single interaction, and neither is intended to function as a diagnostic tool for use on any specific individual. The Back, Schmukle, and Egloff sample was drawn from psychology students at one German university, and it is not established how fully the pattern generalizes to other age groups, cultures, or higher-stakes settings like a job interview or a first date outside a university context. The Carey and colleagues null result on pronoun use, while unusually well powered, cannot rule out that some other, still untested linguistic marker might show a real association with narcissism where I-talk did not. What both studies do support, taken together, is a genuinely useful caution: some intuitive behavioral tells for selfishness have real evidence behind them, and some do not, and the difference is not something a person can reliably sort out just by trusting what sounds plausible.