You have probably worked with someone who fits the pattern. They walk into meetings first. They speak with confidence about topics they have not deeply studied. They accept compliments with practiced ease. From the outside, they read as one of the most self-assured people in the room. And then, in a specific and instructive moment, someone challenges them on something they got wrong, or contradicts them in front of another colleague, or fails to praise a piece of work they had expected to be praised for, and something shifts. The confidence, for a moment, does not hold. What replaces it, briefly, is something the person spends a considerable amount of energy trying to hide from the rest of the room.
Research over the past two decades has distinguished secure high self-esteem from forms that are more contingent, unstable, or defensive. Those constructs describe measured tendencies. They are not a reliable way to classify a confident-looking colleague from observation.
The two kinds of confidence
According to a foundational 2003 paper by Dr Michael H. Kernis at the University of Georgia’s Department of Psychology, published in Psychological Inquiry under the title “Toward a Conceptualization of Optimal Self-Esteem”, the research literature on self-esteem had, up to that point, largely treated the construct as a single thing measured along a single dimension. People had more self-esteem or less of it. Higher scores were treated as better than lower scores. Kernis argued that this framing had missed one of the most important distinctions in the psychology of the self, which was that high self-esteem itself came in two qualitatively different forms.
The first form Kernis called secure high self-esteem. This is confidence that is genuine, stable across situations, and not particularly reactive to threats or challenges. A person with secure high self-esteem can be criticised, disagreed with, or shown to be wrong about something, and their sense of worth stays approximately where it was. They do not need to prove anything. They can lose an argument without losing themselves.
The second form Kernis called fragile high self-esteem. This is confidence that is defensive, unstable, and highly reactive to external validation and threat. A person with fragile high self-esteem also, on the surface, presents as confident. They may present as more confident than the person with secure high self-esteem, because they invest more visible energy in maintaining the appearance. But their sense of worth is contingent on continuous validation from the outside, and small threats to that validation produce disproportionately large internal disturbances that the person is often working hard not to show.
The distinction is not, on Kernis’s argument, about how much self-esteem the person has. Both forms register as high on ordinary self-report measures. The distinction is about the underlying structure of the self-esteem, and about what happens to it under pressure.
There is another take on this – self-esteem vs ego. This video explains more:
What makes it fragile
The mechanism that produces fragile confidence has been documented separately, in a different corner of the literature. According to a 2001 paper by Dr Jennifer Crocker and Dr Connie T. Wolfe at the University of Michigan, published in Psychological Review under the title “Contingencies of Self-Worth”, most people’s sense of worth is not a single, freely-floating quantity. It is pegged to specific things. Academic performance. Physical appearance. Career success. Approval from other people. Being seen as morally good. Being seen as intelligent. Winning at whatever the person considers the important game to be winning at.
Crocker and Wolfe’s argument was that this contingency structure is what determines whether a person’s confidence is fragile or robust. If a person’s sense of worth is heavily contingent on external validation, then external threats to that validation are experienced as threats to the self. If a person’s sense of worth is contingent on more internal criteria, such as relationships or living according to their own values, then external threats do not carry the same weight. Both people may have high self-esteem on the standard measurements. Only one of them is quietly vulnerable to almost every ordinary social interaction that could go wrong.
The contingency framework explains why the confident-seeming person can be so reactive under specific conditions and so composed under others. Their confidence is not general. It is tied to the specific domains they have staked their worth on. Praise them for the thing they care about and the confidence swells. Threaten the same thing and the confidence collapses. The pattern is not a matter of general emotional volatility. It is a matter of what the person’s worth is contingent on being intact.
The subtle sign
The observable difference between the two forms of high self-esteem was demonstrated experimentally by a 2003 study by Christian H. Jordan, Steven J. Spencer, Mark P. Zanna, Etsuko Hoshino-Browne and Joshua Correll at the University of Waterloo, published in the Journal of Personality and Social Psychology under the title “Secure and Defensive High Self-Esteem”. The team measured participants’ explicit self-esteem, meaning what they reported about themselves on standard self-report questionnaires, and their implicit self-esteem, meaning their automatic, non-conscious self-associations as measured by the Implicit Association Test. Some people scored high on both. Others scored high on explicit self-esteem but low on the implicit measure.
Across three studies, the participants who scored high on the explicit measure but low on the implicit one showed the highest levels of defensiveness. They had the highest narcissism scores. They showed the strongest in-group bias when their group was threatened. They engaged in the most dissonance reduction after making a difficult choice. The pattern was clear across all three experiments. The specific configuration of high explicit self-esteem paired with low implicit self-esteem produced observable defensiveness in ways that high-and-secure self-esteem did not.
What this means, in practical terms, is that the difference between the two kinds of confident-seeming person is not visible in the confidence itself. Both look confident. What is visible is what happens when the confidence is tested. The person with secure high self-esteem can be shown to be wrong and remain the same person. The person with fragile high self-esteem, under the same challenge, produces the observable behavioural signature the Jordan team documented: defensiveness, self-justification, hostility toward the source of the challenge, and a small but detectable investment in restoring the threatened self-image.
Across these research programmes, reactions to challenge are relevant to the distinction between secure and fragile self-esteem. They remain group-level associations, not a diagnostic test. One disagreement or defensive reaction cannot reveal how secure another person’s self-esteem is.
Which is a quiet and slightly unsettling piece of information to carry into ordinary social life. It suggests that the confident person you have been reading as the most self-assured in the room may be the one whose sense of worth is most quietly under pressure, and that the person whose confidence has always looked more measured, less performed, and more willing to be challenged, may be the one who actually has the thing everyone else appears to have.
Kiran Athar is not a psychologist or a psychiatrist. She writes about psychology, mental health, and the ordinary corners of life where the two intersect, drawing on peer-reviewed research and primary-source scholarship.
Editor’s note: This article was updated on September 14, 2026, to align its headline and wording more closely with the evidence and sources.
