This theory is revised in the open. Every change, and the criticism or test that prompted it, is in the revision history at the end, including the correction of its own ethical rule (§3). The paper's commitment, "I can be wrong and remain whole," applies to the paper.
This paper supersedes The Circumpunct Theory of Narcissism: A Complete Research Framework (April 2026) as the framework's current statement of the theory. The April page remains online as a dated record, unchanged except for a pointer to this one.
What carries forward: the two Lies (Inflation and Severance), the reading of narcissism as a displaced center rather than an excess of self, and recovery as a return to the center.
What changes: (1) the account of formation, where installation by a "virus" becomes a learning pathway with two routes that converge on one confusion; (2) the evidence standard, where the April page's case and community material is not relied on, and every cited study is stated with what it does and does not show; (3) the dimensional table, which now follows the corrected ladder (boundary at 2D, field at 3D), names the five virtues, and gives the whole no dimension of its own.
Companion: the thesis architecture Protecting the Lie (corpus file protecting_the_lie_thesis.md) describes what the narcissistic organization protects and how. It leaves development aside by design. This paper supplies what it left out: how the protection is learned, why correction feels like a threat in the first place, and when deception is ever justified.
Anyone might lie when threatened. This paper proposes that what distinguishes narcissism is not more lying under threat but a different sense of what counts as a threat. The narcissist perceives correction as a threat, because they have accepted some of their own lies as reality. A lie that has become someone's reality is no longer experienced as a lie. Correcting it feels like an attack on what is real, and the ordinary human response to attack follows: deny, distort, defend. Each defensive lie that is believed in turn becomes part of that reality, so the territory that correction can threaten keeps growing. The pattern seals itself.
The thesis sits inside an ethics of deception. Deception is justified only against a wrongful threat: someone wrongfully threatening a person or what is theirs, when no honest response would serve, limited to stopping the wrong, and ending when the threat ends. Read in reverse, the same four conditions describe the narcissistic pattern exactly, failing one at a time. The person correcting a narcissist is not a wrongdoer; they hold a truth they have every right to. To borrow the exception anyway, the narcissist must recast the truth-teller as the aggressor, which is the reversal at the heart of DARVO, and because the lie is their reality, the reversal can be sincere. Honest alternatives exist but feel unavailable. The deception spreads to witnesses and records. And because reality keeps contradicting an accepted lie, the emergency never ends.
The paper then places the accepted lies at the core of the pattern (self-deception) and the defensive lies at its surface (conditioned deception); proposes two developmental routes, punishment of truth and reward of the image, that converge on one confusion (worth made contingent on a representation); explains why the deception is selective (only some lies have become reality); proposes a mechanism for the empathy picture (the other person's testimony is discounted when it contradicts the accepted account); states which canonical features of narcissism the account should explain and which it should miss; grounds its remedy in an anchor of worth that correction cannot reach; and sets out what existing research supports, what it does not establish, and how the theory could fail.
Anyone might lie when threatened. The narcissist is threatened by the truth,
because some of their own lies have become their reality.
Anyone might lie when threatened. Children who are punished for the truth learn to hide it (Talwar & Lee, 2011), adults under pressure shade their accounts, and people protecting someone in danger conceal where that person is. Lying under threat is ordinary human behaviour. Whether a particular lie is justified is a separate, ethical question, answered in §3.
What distinguishes narcissism, I propose, is not a greater readiness to lie under threat. It is what counts as a threat. The narcissist perceives correction as a threat, because they have accepted some of their own lies as reality. A lie that has been accepted as reality is no longer experienced as a lie; to its holder it is simply how things are. Correcting it therefore feels like an attack on what is real, and the ordinary response to attack follows.
The response is learned, and it persists beyond any real danger. The central confusion is between protecting a person and protecting a self-image: being corrected, exposed, or held accountable is treated as a threat to existence, and distortion becomes a way of restoring safety or control. Each defensive lie that is believed joins the accepted account, so the confusion feeds itself (§4.5).
Within this framework, narcissism is a persistent pattern in which some of a person's own lies, above all lies about the self, have been accepted as reality, so that correction is perceived as a threat and met with the ordinary defenses against threat: self-deception keeps the accepted account from being revised, and conditioned deception protects it from other people, at the expense of shared reality, accountability, and other people's agency.
My shorthand is "lying embedded into the nervous system." Here, embedded means learned and repeatedly reinforced, sometimes operating before deliberate reflection. It does not mean permanent, irresistible, or detectable through a single brain scan. The shorthand describes habit, not anatomy: no neural finding cited in this paper shows a lasting change in the nervous system, and the careful statement of the claim is "a learned and repeatedly reinforced response." A person may consciously value honesty while repeatedly falling back on deception.
This is a philosophical account and a psychological hypothesis. Its definition is specific to this framework. Clinical narcissistic personality disorder has a broader definition and a multifactorial origin, with mechanisms described in domains such as self-esteem regulation, emotion regulation, cognitive style, interpersonal relations, and empathy (Weinberg & Ronningstam, 2022). Section 10 states which features the account expects to explain and which it expects to miss.
The rest of the paper follows from the thesis. Section 3 states when lying under threat is justified. Section 4 shows that the narcissistic pattern fails each condition of a justified lie, and why: the threat it answers is a truth, and the truth does not stop. Section 8 places the accepted lies and the defensive lies in two layers of one organization, and the remaining sections trace how the pattern is learned, what it explains, and how it could change or be tested.
Truth concerns what is actually the case. Our access to it is partial, so confidence must remain open to evidence and correction.
A lie is a representation presented as true despite the speaker believing it false, with the intention of misleading. This is close to the traditional definition, although philosophers still dispute whether an intention to deceive is strictly required (Mahon, 2015). Sincere error is different.
Deception includes lying and other attempts to create or maintain a misleading understanding, such as strategically omitting a fact needed for an informed decision. Deception is a subset of the much larger territory of information control and impression management that all social life involves (Goffman, 1959). Most information control is not deception: privacy, tact, and discretion are not attempts to mislead. It becomes deception when it creates a false impression about something that matters to another person's understanding or choice.
Self-deception means preserving a desired belief through biased attention, interpretation, or avoidance of contrary evidence. It need not involve consciously knowing the full truth and deliberately hiding it from oneself. This follows the deflationary account in the philosophical literature, on which motivationally biased treatment of evidence (misinterpretation, selective focusing, selective gathering of evidence) can suffice for self-deception, with neither an intention to deceive oneself nor a simultaneous belief in the truth required (Mele, 1997, 2001).
An accepted lie is a claim, usually about the self, that began as a lie, an exaggeration, or a convenient story told under threat or for advantage, and has come to be held as true and exempt from revision. It is the point at which deception turns into self-deception (§6.5 proposes how). To its holder it is no longer a lie but part of how things are, which is why correcting it is felt as an attack on reality. What makes it a lie is its history and its function: it was maintained against the evidence available when it was defended, and it cannot now be corrected. It is not a verdict on whether the claim is true today, which an observer may be unable to reach. An accepted lie is typically false or overstated, because a claim that can never lose drifts from the evidence, but falsity is the consequence, not the criterion. The thesis of this paper is that the narcissistic pattern is organized around such lies. Their nearest relative in descriptive psychopathology, the overvalued idea, is discussed in §10.
A defended self-image is a representation of the self (as innocent, superior, special, entitled, or indispensable) that is treated as necessary for the self to continue, rather than as a revisable description. The companion thesis calls this the Protected Identity Proposition: a self-claim exempt from ordinary revision, defined by how it is defended rather than by whether it is false.
Corrigibility is the capacity and commitment to revise beliefs and behaviour when sufficient evidence shows that revision is needed (§13).
Lying when threatened is ordinary; whether it is justified is a separate question, and this paper answers it narrowly. My ethical starting point is that deception is justified only against a wrongful threat. Outside that exception, people should use truthful speech, appropriate privacy, refusal, or silence instead of intentionally misleading themselves or others.
The original shorthand was "lie or die." It names the clearest case, not the boundary. Concealing an endangered person's location from an attacker is the paradigm, but the same reasoning covers lying to a thief about where the car keys are, or to a fraudster who is trying to take someone's home. In each case the person deceived is a wrongdoer who wants the truth only as an instrument of the wrong. The threat need not be seconds away, and the exception covers the defense of other people as well as oneself. I call it the self-defense exception.
A wrongful threat is a threat of harm that the person threatening has no right to inflict: violence, theft, fraud, coercion, unlawful eviction. A legitimate consequence is not a wrongful threat, however painful it is: a debt one really owes, a lawful ruling, the loss of a job one did badly, the reaction of someone who has learned the truth. Lying to escape a legitimate consequence is lying to escape accountability. The right in question is a moral right, not merely a legal one, and lawfulness settles nothing in either direction: lying to an authority that is lawfully hunting refugees is justified, because what it means to do is a wrong, and lying to a critic who is entitled to speak is not, however spiteful the critic.
Concealing an endangered person's location from an attacker satisfies this principle. Concealing a harmful act to preserve admiration does not become an emergency merely because exposure would be painful. These are illustrative cases, not a method for inferring someone's motives from a single interaction.
The rule is a demanding ethical position, and it has to be argued rather than treated as scientific consensus. Four considerations support it.
Who has a right to the truth. The wrong in deception is that another person is made to choose or live on terms they cannot adequately evaluate (§12). That wrong is overridden when the person deceived means to use the truth as the instrument of a wrong. Someone trying to steal, defraud, or harm has no claim on the truth they mean to use as a weapon. This is Benjamin Constant's objection to Kant, that the duty of truthfulness is owed to those who have a right to the truth and that no one has a right to a truth that harms others (Constant, 1797), with "harms" read strictly as "wrongs." A creditor collecting a real debt harms the debtor, but has every right to the truth.
Why not the size of the loss. The tempting alternative is to key the exception to what one stands to lose: first life, then property, then a way of life, then quality of life. That rule has no stopping point, and worse, it cannot tell the narcissist from anyone else. To a person defending an image, losing the image is a loss of quality of life. A rule based on loss would grant, as an ethical permission, the confusion this paper diagnoses. The same loss gets different answers depending on its source: being forced out of one's home by a fraudster licenses deception, and being forced out by a debt one really owes does not.
Corruption resistance. An exception has to be usable by people who are frightened and self-interested, which is to say, by everyone. Whether someone was stealing, defrauding, or attacking is usually a fact others can check after the danger has passed. Whether a critic was "attacking" someone's image can be verified only by the person who feels attacked. The rule is drawn where it can be audited. This is close to Bok's test of publicity: whether a justification could be stated openly to reasonable people, including the person deceived (Bok, 1978).
The same line as the diagnosis. The diagnosis is that the narcissist treats correction as a threat. The rule says that correction is never a wrongful threat: a person who corrects you with the truth has a right to it. The rule and the diagnosis meet at one point, and §4 shows what happens there.
A rule this narrow has to answer for the cases where it looks cruel and the cases where it looks fussy.
So the ordinary permissions do not form a further legitimate category. Most of them are not deception at all; the remainder are small instrumental deceptions that the rule counts as wrong without counting as grave.
Augustine held that every lie is a sin, though some are graver than others (Augustine, Enchiridion 18), and Kant, answering Constant, refused even the lie to a murderer who asks whether one's friend has taken refuge in the house (Kant, 1797). Sissela Bok's principle of veracity gives every lie an initial negative weight that must be overcome by justification. A lie is unjustified when a truthful alternative would serve, and a justification must be one that could be defended openly before reasonable people. Bok allows that some lies, such as some lies to enemies, can meet that burden, while distrusting easy appeals to crisis (Bok, 1978). Her distrust is the same worry as §4.6 below. The self-defense exception is broader than Augustine and Kant, because it permits the lie to the murderer and to the thief. It is close to Constant, and to Bok on lies to enemies, and narrower than most consequentialist accounts, because the size of a benefit never by itself justifies a lie; only the wrongfulness of the threat does.
The paper's central argument does not depend on winning this debate. A reader who holds Bok's view, or a more permissive one, can keep everything that follows: the keystone of §4 needs the self-defense exception only as the paradigm of a justified lie, not as the only one. Narcissistic deception fails the four conditions under any of these views, because none of them treats a truthful correction as a wrongful threat, and none permits an exception without an ending.
In ordinary life the framework's Truth Resolution Protocol governs disclosure: transmit at the lowest resolution that is still true, and let the receiver widen it. A refusal is a gate, not a concealment, and it passes the protocol's retraction test: nothing said at low resolution has to be taken back when resolution increases. The self-defense exception is the emergency rule standing beside that ordinary rule. It permits what the protocol never permits, a statement that will later have to be retracted, and it permits it only while a wrongful threat lasts.
"Hijacked" describes a change in function: the ordinary readiness to deceive under threat is recruited to protect comfort, reputation, superiority, advantage, or control. I am not claiming that deception originally evolved only for emergencies, or that humans once used it only to survive. Establishing an evolutionary origin would require a different argument and different evidence.
The psychological hypothesis is narrower: once some of a person's lies have been accepted as reality, correction is perceived as a threat, and the ordinary response to threat is turned against the truth. Section 6 describes two routes by which that state is reached, punishment and reward, and argues that they converge.
The four conditions were stated as limits on a justified lie. Read in the other direction, they are a diagnostic. Narcissistic deception does not invent a new capacity. It uses the ordinary one, the readiness to deceive under threat, against something that is not a wrongful threat at all: a correction. And it fails the four conditions one at a time.
| Condition | Justified lie | Hijacked emergency | What could be observed |
|---|---|---|---|
| Credible wrongful threat | The person deceived means to use the truth to do a wrong. | The person deceived holds a truth they have every right to; the correction contradicts an accepted lie and is felt as an attack on reality. | Defensive intensity tracks how much a correction threatens the accepted account, not whether anyone is doing a wrong; truth-tellers are described as aggressors. |
| Necessity | No truthful response, refusal, silence, or appeal for protection would work. | Truthful options exist ("I was wrong," "I won't discuss that") but feel unavailable. | Active deception chosen where an honest refusal was available at no real cost. |
| Proportionality | Limited to stopping the wrong. | Spreads from self-protection to managing other people's reality: witnesses, records, memories. | Deception aimed at third parties and at the shared record, not only at the person correcting. |
| An ending | Ends when the threat ends; repair follows when it is safe. | Never ends, because reality contradicts an accepted lie continuously. | Defense that persists after the contradiction is withdrawn, and recurs at low stakes. |
The first failure is the root, and it is easy to misdescribe. On this account the narcissistic response to correction is not usually a performance of fear. The fear is real; what is mistaken is its object. A person who has accepted some of their own lies as reality experiences a correction of those lies as an attack on what is real, and "I did something harmful" lands as "I am unacceptable" (§5). The readiness that would defend against a thief is recruited against a truth.
To borrow the exception, the truth-teller has to be recast as the aggressor. That is the reversal at the heart of DARVO: deny, attack, reverse victim and offender (Freyd, 1997). On this account the reversal is often sincere. If the lie is the person's reality, the one correcting it really does look like the one attacking reality, and the defender really does feel like the victim. That is why the reversal can be so convincing to onlookers and so hard to argue with: it is less a tactic laid over a known truth than a report of how things look from inside the accepted lie. Sincerity does not make it harmless, and it does not make it true.
Two consequences follow. Reassurance about the facts does not calm the response, because the facts were never what was felt to be threatened. Confrontation tends to escalate it, because confrontation raises the felt stakes. Neither consequence excuses harm (§12). Both explain why arguing the facts so often fails, and why a person can be sincerely terrified while doing something that frightens others.
Necessity fails in a particular way. The honest alternatives are present in the situation and absent from the person's felt options. Refusal and silence are nearly always available to someone facing a wrongful threat: a hidden person needs only to stay hidden, and a thief can simply be refused. They are often useless to someone protecting an accepted lie, because silence leaves the contradiction standing. An accepted lie needs the record to agree with it. That asymmetry is why the hijacked emergency reaches for active distortion where a justified one could have used silence.
Proportionality fails because an accepted account of the self, unlike a life or a car, is held partly in other people. A car can be protected by keeping one fact from one thief. An account of the self lives in what others remember, believe, and are willing to say. Protecting it therefore means managing those people: discrediting a witness, revising a shared history, isolating a partner from people who would corroborate them. The companion thesis lists these as protection operations, with DARVO as their acute form. In this paper's terms, they are what the self-defense exception looks like when the thing being defended lives in other minds.
The fourth failure carries the most weight. A wrongful threat is an episode: the attacker leaves, the thief is stopped, the exception lapses, and repair becomes possible. An accepted lie is never out of danger, because reality contradicts it continuously. Every ordinary failure, every unimpressed stranger, every accurate memory held by someone else is a standing challenge. If the accepted lie is defended as though it were reality itself, the emergency has no ending condition, and what began as an exception becomes a way of living.
This one failure accounts for three features that functional accounts of narcissism already describe:
The ending fails for a second reason, which explains why the pattern does not merely persist but grows. The sequence is:
Each defense widens what future corrections can threaten. The emergency does not just fail to end; it spreads. That fits the laboratory finding that self-serving dishonesty escalates with repetition (Garrett et al., 2016; §6.2), although that study measured deliberate lies within a single task, not accepted lies over a life. It also explains why the pattern is selective (§6.3): correction is threatening only where accepted lies are, so outside those domains the same person can be calm, accurate, and even scrupulous.
The loop runs in the other direction too. Every correction that is absorbed, rather than defended against, shrinks the territory: an accepted lie that is given up can no longer be threatened. That is the mechanism of recovery (§13).
The pattern borrows more than a capacity. It borrows the exception's justification. "They were attacking me," "I was protecting them," and "I had no choice" are the grammar of self-defense, and they are the justifications most available to someone for whom correction feels like an attack, because from inside, the attack is real. The theory therefore contains the seed of its own rule's corruption, and it should say so plainly.
The safeguard is in the conditions themselves. A justified exception can be verified, after the danger has passed, by someone other than the deceiver: the threat, its wrongfulness, the absence of alternatives, the scope, and the ending are facts others can check. A hijacked exception depends on the deceiver's say-so, because the "attack" was on something only the deceiver could feel. An exception that only its user can certify is not an exception; it is a permission.
First, four separately observable signatures in place of one undifferentiated "defensiveness" (§15). Second, a principled line between ordinary self-protection and the narcissistic pattern: anyone might lie when threatened, and the pattern is treating the truth as the threat, with an emergency that has lost its ending. Third, a description that does not require attributing calculated malice where the evidence shows fear, while still holding the person responsible for what the fear is allowed to do. Fourth, it locates the change that would matter. The emergency can end only when the accepted lies stop being reality, which requires that being wrong stops being experienced as ceasing to exist (§5, §13).
By "the war is ontological," I mean that the conflict reaches into a person's sense of being: who they are, whether they have worth, and whether other people are allowed to exist as independent persons whose experiences matter. The phrase names a conflict over whether our stories remain answerable to what exists and to the people affected by them.
The proposed confusion is that a particular story about the self becomes a condition of feeling able to exist. "I did something harmful" is experienced as "I am unacceptable," and protecting the story begins to feel like protecting the person. Evidence is then resisted because accepting it seems to threaten the self that depends on that story.
This is what it means for a lie to be accepted as reality. The lie has been promoted from a claim about the self to part of what the self is, so a correction that would demote it feels like an amputation rather than an update. The question "is this true?" has been replaced by the question "can I survive this being false?", and for an accepted lie the felt answer is no.
Moral psychology has a close neighbour for this confusion. Following Helen Block Lewis (1971), Tangney, Stuewig, and Mashek (2007) distinguish guilt, a negative evaluation of a specific behaviour, from shame, a negative evaluation of the global self. In plain words, guilt says I did something bad, and shame says I am bad. Shame tends to be associated with attempts to deny, hide, or escape, and with externalizing blame, while guilt tends to be associated with confession, apology, and undoing the consequences. The proposed confusion converts what could have been guilt into shame. A harmful act is experienced as a verdict on the self's right to exist, and the defenses that shame recruits become survival measures.
The circumpunct framework states the same thing structurally. A center is a point of view with no parts, so there is nothing in it for a fact to break. An image is made of claims, so it can be compared, ranked, and contradicted. A person seated in their center can hear that they were wrong. A person seated in their image is diminished by every true sentence that contradicts it. The myth already says so: Narcissus does not love himself too much. He mistakes his reflection for himself and starves at the surface of the water. On this reading narcissism involves too little genuine centeredness, not too much. The sequence runs: the person denies their own point of view, relocates it into an idealized image, and must then obscure reality to protect the image, because a fixed picture cannot otherwise survive a moving world.
There is an epistemic side: what happened, what evidence supports it, and what we should believe. There is an ontological stake: whether I can remain a person of worth when my account is wrong, and whether your experience can matter when it challenges mine. The theory connects these without treating every factual dispute as an existential conflict.
When another person's testimony is repeatedly dismissed to preserve a preferred self-image, that person may be treated as a supporting character in someone else's account. Their independent perspective becomes an obstacle to manage. The facts have not literally changed, but the relationship's rules for acknowledging facts have been distorted.
This bears directly on the interpersonal picture of narcissism. Another person's report about their own experience ("that hurt me," "that is not what happened") is evidence, and evidence that contradicts the image is handled rather than weighed. Systematic discounting of that evidence produces something that looks, from the outside and from the other person's side, like an absence of empathy: the other person's inner life does not register as a fact. On this account the capacity need not be missing. What is missing is the credibility the other person's testimony is allowed to carry when it threatens the image. This is a cousin of what Fricker (2007) calls testimonial injustice, in which a speaker is given deflated credibility; here the deflation is driven by protection of an image rather than by prejudice about the speaker.
It explains why narcissistic relating can feel like objectification from inside the relationship without positing a missing empathy module, and it makes a definite prediction: empathy and credence should be state-dependent, low when the other's experience implicates the image and closer to normal when it does not. There is supporting evidence that narcissistic low empathy is not a fixed incapacity. Across three studies, maladaptive narcissism (entitlement, exploitativeness, exhibitionism) predicted lower empathy, and instructions to take the distressed person's perspective attenuated that association in both self-report and heart rate; the authors conclude that the deficit does not reflect inability (Hepper, Hart, & Sedikides, 2014). Those studies did not manipulate image threat, so they support the motivational reading without testing this mechanism.
The proposed process connects a load-bearing self-image, threat, relief or reward, repetition, and justification. It is a hypothesis about a pathway that some people may follow, not a universal account of childhood, trauma, or personality development.
The two routes look opposite. One makes truth dangerous; the other makes the image precious. The developmental evidence suggests both need a place. In a natural experiment with 84 children aged three and four in two West African schools, children from the punitive school who had peeked at a forbidden toy were far more likely to deny it (94 percent against 56 percent) and better at maintaining the denial under follow-up questioning (Talwar & Lee, 2011): a punitive environment and concealment go together. Yet the strongest prospective evidence on the origins of narcissism itself points toward the reward route. In a study that followed 565 children, aged 7 to 11 at the start, through four waves six months apart, parental overvaluation predicted later narcissism, whereas parental warmth predicted later self-esteem and not narcissism (Brummelman et al., 2015). That design is longitudinal and correlational, not experimental.
The proposal is that the two routes teach the same lesson: worth depends on a representation. The threat route teaches it by punishing the person when the truth about them appears. The reward route teaches it by rewarding the person when the image appears. Either way, the self-image becomes the thing whose survival matters, and the conditions for the hijacked emergency are in place. A child loved as special, rather than as a person, learns that the image is what is loved; for a child, a threat to that love is a threat to security, which is to say, to life.
This convergence is a hypothesis, not a finding. No study cited here tests it. It makes a definite prediction: punitive environments and overvaluing environments should both produce identity-contingent defense of the self-image, while warmth without overvaluation should not.
Conditioning is not limited to deliberate lies. Self-deceptive avoidance is also followed by relief: the painful fact goes unexamined and the distress subsides. If relief reinforces, then looking away is learned in the same way lying is. This matters for §8, where the core of the pattern is located in self-deception rather than in conscious lying.
The closest laboratory evidence concerns deliberate dishonesty. Adults advised a partner about the amount of money in a jar of pennies under different incentives (55 adults, 25 of them scanned). Self-serving dishonesty escalated over repeated trials, and a reduced amygdala response on one decision predicted the size of the escalation on the next; dishonesty that served the other person did not escalate in the same way (Garrett et al., 2016). That supports adaptation during repeated self-serving dishonest choices within a laboratory task. It does not show that the adaptation lasts beyond the task, establish permanent "wiring," identify a specific emotion, or show the development of narcissistic personality disorder.
Narcissistic deception is often selective. It can be absent when the image is not at stake, and it can coexist with careful honesty elsewhere. Pure habit would not predict that. The thesis supplies the main answer: only some of the person's lies have been accepted as reality, and correction is threatening only where they are (§4.5). Two further features of the account explain the rest.
First, conditioning is cue-specific. What is learned is not "lie" but "distort when the image is challenged," and a response conditioned to a class of cues fires on those cues and not on others. Second, the response has an aim. Motivated cognition supplies the aim (keep the image standing); conditioning supplies the automaticity (do it quickly, before checking). Selectivity is what the aim looks like from outside. Speed and habit are what the conditioning looks like.
In the steady state the two routes merge. Relief when a challenge is defeated and reward when the image is admired now strengthen one thing: the protection of the image. This is why admiration and attention, which look like the reward route's instrumental gains, are better read in the steady state as confirmation that the image is standing, as the companion thesis argues. Deception for a gain that has nothing to do with the image remains instrumental (§9). Deception for a gain that consists in the image being affirmed is part of the pattern.
The schedule of consequences matters as much as the consequences. Behaviour that is reinforced only some of the time is more resistant to extinction than behaviour reinforced every time, the partial reinforcement extinction effect (Humphreys, 1939; Mackintosh, 1974). Deception is rarely met with consistent consequences: sometimes it works, sometimes it is caught, and often nobody checks. On this account that schedule is not neutral background. Inconsistent accountability is a maintenance condition for the pattern (§13).
The loop explains how a habit forms, not why it forms in one person and not another under similar conditions. The account owes a candidate, and its candidate is where a person's worth is anchored. People differ in the domains on which their self-worth is contingent, and in those domains self-esteem and affect rise and fall with successes and failures (Crocker & Wolfe, 2001); pursuing self-worth in a contingent domain becomes a goal with costs of its own (Crocker & Park, 2004). The hijacked emergency requires an extreme of that contingency: worth contingent on one representation of the self.
The developmental routes of §6.1 are proposed as ways that contingency is installed, and warmth without overvaluation as a way it is prevented. Temperament presumably contributes through the threat side, in how strongly any challenge is felt, but the theory's distinctive claim is about the anchor, not the alarm. It predicts that contingency of worth on a specific self-image should predict the four signatures of §4 better than overall narcissism scores do.
Everything downstream of the thesis (the selectivity, the sincere reversal, the loop, the empathy picture) runs through one transition: a told and defended lie stops being experienced as a lie. The routes of §6.1 explain how deceiving is learned and how the image becomes load-bearing. They do not by themselves explain the transition.
My proposal is that a lie becomes belief when it is told or acted consistently, and then externally reinforced enough. Consider the claim "I am more entitled than people with darker skin." It is false. A person who acts it out (expecting deference, taking precedence, treating objections as insolence) and is allowed to, by people who look away, who share the assumption, or who never notice, receives a steady stream of confirmation. Each time the claim is enacted and nothing contradicts it, the world appears to agree. In time the person no longer holds a claim about the world; they live in a world where the claim is simply how things are. Private lies about the self work the same way: the innocent version of an event, told and retold to people who do not challenge it, becomes the event.
Lying alone usually does not do this. In the clearest test, people who lied to convince others that false childhood stories were true mostly came to rate those events as less likely, not more (Polage, 2004). Belief rose in some liars, and later work tied that rise to indicators of weaker source monitoring and to telling the lie in a separate session from the test (Polage, 2012). Witnesses told to fabricate details about a culprit later remembered him less accurately and sometimes reported their own fabrications as memory (Pickel, 2004). A lie told once, for a reward, to an audience that does not matter, is usually remembered as a lie. The conditions in the proposal are what make the difference, and each has a neighbour in research. Those studies concern ordinary participants and mostly ambiguous or low-stakes material, not defended claims about the self; applying them here is my extension, not their finding.
Put together, the proposal is that a defended lie becomes reality when its source label is lost and the people around it have stopped contradicting it. The content survives, rehearsed, enacted, and confirmed. The memory that it was said rather than found fades, and is not sought, because finding it would hurt. What remains is a claim with no remembered origin, held with the confidence of something often stated and never challenged, which from inside is indistinguishable from a fact. Once the label is gone there is nothing for the person to correct: the claim was never a lie, as far as they can tell. That is the state the elicitation protocol of §15 measures (a claim with no nameable revising evidence), and it is why the loop of §4.5 can run sincerely.
The counter-mechanism is the same condition reversed: people who do not let it stand. Dependable accountability (§13.3) has a social face. Every witness who declines to confirm an accepted lie keeps its source label alive a little longer, and every institution that lets it pass helps erase it.
Two qualifications. Whether the conversion is faster or more complete when the lie protects a load-bearing image is the hypothesis, not a finding; this paper expects it to be, because the motive to be believed, the frequency of rehearsal, and the motive to forget the source are all highest there. And the mechanism erases its own evidence: the history that makes an accepted lie a lie is exactly what the mechanism removes, which is why that history is the part of the thesis that testing reaches last (§14).
Statements such as "I was protecting them," "they could not handle the truth," or "everyone does it" can shield deception from scrutiny. Bandura's account of moral disengagement describes how people can reinterpret harmful conduct as worthy, minimize its consequences, or shift responsibility (Bandura, 1999). "They could not handle the truth" deserves particular attention, because it is the Noble Lie's own formula: the paternalist justification the framework examines at the scale of families and institutions in The Fallout of the Noble Lie. "They were attacking me" and "I was protecting them" are the grammar of self-defense, borrowed for a situation in which nobody was doing a wrong (§4.6).
A justification is not automatically a rationalization. The test is whether it survives examination of the actual danger, available alternatives, affected people, and contrary evidence. The four conditions of §3 are that examination made explicit, and the corruption test of §4.6 adds the decisive question: could someone other than the deceiver confirm it? A reason that changes whenever evidence appears may be protecting the desired conclusion rather than explaining the decision.
In this model, moral justification can appear before the lie as permission, after it as an excuse, or alongside it as self-deception. It can strengthen the pattern, but some deception is openly opportunistic and requires little felt moral justification.
Neither distress nor confidence is a lie detector. Someone can become distressed while telling the truth, and someone can sincerely remember an event inaccurately. Evidence of deception requires more than an uncomfortable reaction or disagreement.
This paper defines a lie narrowly: a statement the speaker believes false, made with the intention to mislead. The companion thesis uses "the Lie" structurally: a self-claim exempt from revision, which functions as a lie whether or not its holder could articulate it, and which requires no awareness of deception. Read carelessly, the two documents contradict each other. One calls narcissism conditioned deception; the other says the protected Lie needs no deceiver. They describe different layers of one organization.
At the core are the accepted lies, above all the defended self-image, kept in place by self-deception in the deflationary sense of §2: biased attention, interpretation, and avoidance of contrary evidence, with no need to know the truth and hide it from oneself. This is the thesis's protected proposition.
At the surface are the operations that protect the image from other people: denial, selective recounting, discrediting witnesses, revising shared history. Some of these are lies in the strict sense. Some are sincere distortions produced by the core, reported by a person who believes them. This is where conditioned deception lives.
The thesis is the traffic between the layers. Surface lies that are believed migrate to the core: a lie told in defense becomes, once believed, part of the accepted account. The enlarged core then makes new corrections threatening, and the surface answers them with new lies. This is the self-sealing loop of §4.5 seen as movement between layers: deception becomes self-deception, and self-deception creates the need for more deception. The accepted lies at the core are the reason correction is a threat; the lies at the surface are the ordinary response to that threat.
There is evidence that this division is real. Across five studies, Jones and Paulhus (2017) found that the three "dark triad" traits deceive differently. All three predicted cheating when the risk was low. Psychopathy predicted cheating even under serious risk of punishment; Machiavellianism predicted cheating under high risk only when participants were ego-depleted; cheating that required a deliberate lie was predicted by psychopathy and Machiavellianism. Narcissism stood out elsewhere: those high in narcissism showed the highest levels of self-deceptive bias. Reading that as "narcissists believe their own enhanced accounts" is this paper's interpretation, not the authors' wording.
The interpretation fits the dissociation the companion thesis predicts in its §3.6: a manipulator updates their private picture of the world while managing appearances, whereas a narcissistic organization does not perform the private update, because the image is identity rather than strategy. The thesis lists this as an untested prediction. The Jones and Paulhus result supports it without testing it in the thesis's terms.
The consequence for the definition is that the pattern is not "a liar who knows the truth." It is a person whose access to the truth about themselves is closed at the core, and whose dealings with others are bent at the surface to keep it closed. This also refines an earlier formulation in the corpus (DARVO Geometry: "narcissism is lie-protection without awareness"). Awareness is largely absent at the core. At the surface it varies, and that is where responsibility most often has something to attach to (§12).
Not all deception is narcissistic. The keystone of §4 adds a column the earlier draft did not have: what would end each pattern.
| Pattern | What the deception protects or pursues | What ends it |
|---|---|---|
| Self-defense deception | A person, or what is theirs, against a credible wrongful threat, within the limits of necessity and proportionality. | The threat ending. |
| Narcissistic deception | An image of innocence, superiority, special worth, entitlement, or indispensability. | Nothing in the environment; reality keeps contradicting the image. Only a change in what the image means to the person. |
| Instrumental deception | Advantage, resources, pleasure, or control; it need not be driven by threatened identity. | The gain being secured, or the expected cost of being caught rising above it. |
| Self-deceptive avoidance | A belief that makes a painful situation easier to tolerate or a difficult choice easier to postpone. | The situation resolving, or the evidence becoming impossible to avoid. |
These functions can overlap. Narcissism and psychopathy are distinct, potentially overlapping constructs; narcissism is not simply a form of psychopathy. Research has found differences in their associations with deception (Jones & Paulhus, 2017). In a study of 67 incarcerated men, psychopathic traits did not predict how often participants were dishonest, although among those who were dishonest, higher psychopathy scores predicted reduced anterior cingulate activity during dishonest decisions (Abe, Greene, & Kiehl, 2018).
Moral justification can accompany any of these patterns and can occur without either condition. Lying, rationalizing, or struggling to accept correction does not by itself establish a personality disorder. A theory that treats every deceiver as narcissistic would lose the distinctions it needs to explain behaviour.
The account will be read as a claim about narcissism, so it has to say which of narcissism's canonical features fall out of it and which do not. A theory of one mechanism should be judged by where it predicts it will fail.
| Feature | What the account says | Expected fit |
|---|---|---|
| Vulnerability, hypersensitivity to criticism, shame | The thesis itself: correction of an accepted lie felt as a threat to existence (§1, §4.1, §5). | Best |
| Rivalry and antagonism (devaluing critics, retaliation) | The spreading perimeter: protection operations aimed at witnesses and records (§4.3). | Strong |
| Low empathy | The supporting-character mechanism: the other's testimony discounted when it implicates the image (§5.1). Predicts state-dependence. | Proposed mechanism; partly supported |
| Admiration-seeking, self-promotion | In the steady state, confirmation that the image is standing (§6.3). Without any challenge present, explained only through that reading. | Partial |
| Grandiosity | The content of the defended image. The account explains why it must be defended; it explains why this content and not another only through the overvaluation route. | Partial |
| Entitlement | A clause of the image (exemption from ordinary rules and consequences), defended like the others. The account explains its defense more than its origin. | Partial |
| Exploitativeness for gain | Instrumental deception (§9), unless the gain is the image being affirmed. | Not explained |
| Confident grandiosity that tolerates correction | Not the pattern on this account. If it exists stably, it is outside the theory's scope. | Outside scope |
Clinical and research traditions frame these features differently, and the account should be placed among them. In the psychoanalytic tradition, Kernberg (1975) described a pathological grandiose self formed defensively in response to cold and rejecting parenting, and Kohut (1971) traced narcissistic pathology to parental empathic failures to meet the child's needs for mirroring and idealization, leaving a fragile self. This account is closest to Kernberg on function, since the grandiose image is defensive, and departs from both on development. Its threat route concerns the punishment of truth, not coldness or empathic failure, and its reward route is the social-learning account that Brummelman and colleagues (2015) set against the psychoanalytic one and found support for.
In social-personality psychology, narcissism is commonly measured with the Narcissistic Personality Inventory, one of the most widely used measures (Raskin & Terry, 1988), which mostly captures grandiose, agentic features. The account expects to explain less of those than of the antagonistic and vulnerable features. The distinction between narcissistic admiration (assertive self-enhancement) and narcissistic rivalry (antagonistic self-protection) in the NARC model, both serving to maintain a grandiose self (Back et al., 2013), maps closely onto that expectation: the hijacked emergency is a theory of rivalry first. A trait-model line of work distinguishes grandiose from vulnerable narcissism (Miller, Lynam, Hyatt, & Campbell, 2017) and describes narcissism with three factors: agentic extraversion, narcissistic neuroticism, and self-centered antagonism (Crowe, Lynam, Campbell, & Miller, 2019). On that model, the account is aimed at the antagonism and neuroticism factors more than the agentic one. Recent evidence fits that division. In two correlational studies, antagonistic and neurotic narcissism were strongly negatively related to independence of intellect and ego, the capacity to keep disagreement from registering as a personal threat, while agentic narcissism showed modest positive links to openness (Zajenkowski et al., 2025). Morf and Rhodewalt's (2001) self-regulatory model, in which a grandiose yet vulnerable self is maintained through a chronic pursuit of external affirmation and through interpersonal and intrapersonal strategies, is the closest existing neighbour. This paper adds the ethical sorting standard, the four signatures, the two-layer organization, and the claim that the account being regulated is felt as reality.
The nearest construct outside personality psychology is the overvalued idea, first described by Wernicke (1900): in McKenna's (1984) summary, a solitary, abnormal belief, neither delusional nor obsessional, that preoccupies the person to the point of dominating their life. Veale (2002) reads overvalued ideas as idealized values so overriding that they come to define the self, and forensic psychiatry has extended the idea to extreme overvalued beliefs shared within a group (Rahman, Resnick, & Harry, 2016). An accepted lie about the self plausibly belongs to that family. This account adds four things: its content (the self's standing, not a value about the world, the body, or a group's cause); its interpersonal maintenance (it needs others to keep not contradicting it, which produces confirmation-seeking and the management of witnesses); its signature (yielding to pressure and resisting evidence within the protected domain); and its history (it grows through its own defenses, §4.5). In one sentence: an accepted lie about the self is an overvalued idea about the self's standing, maintained by other people's silence and enlarged by its own defenses. If the predictions of §15 are fully explained by existing accounts of overvalued ideas, the construct adds nothing and should be retired into them.
No estimate of explained variance is possible before the tests of §15 are run. The qualitative prediction is definite: the account should explain more of the antagonistic and vulnerable variance in narcissism than of the agentic variance.
One clarification matters for the first row of the table. The account predicts heightened reaction to correction of a protected claim, not a general hypersensitivity to being evaluated. A preregistered study using the Trier Social Stress Test (161 men) found no compelling evidence that narcissism, in any facet, moderated psychological or physiological stress reactivity to social-evaluative threat (Stern et al., 2025). That null counts against a general "fragile self" reading of narcissism, and it is consistent with this account, which locates the threat in correction of particular accepted lies rather than in evaluation as such.
In the circumpunct framework, a person integrates a center (•), a line of continuity through time (—), a boundary (○), and a field (Φ) into a whole (⊙). The self-image is a representation within that whole. It can change while the person remains. Existence has priority over its description. The proposed narcissistic confusion occurs when preserving a representation outweighs the reality and well-being of the person and their relationships.
Each station carries one of the framework's five virtues, and each has a characteristic distortion when the hijacked emergency takes it over.
| Station | Virtue | Commitment to truth | Distortion under the hijacked emergency |
|---|---|---|---|
| • 0D | TRUE center, point of view; curiosity | I acknowledge my experience and its limits. | My viewpoint cannot be wrong without my ceasing to be, so it is treated as the final authority over yours. |
| — 1D | FAITHFUL line, continuity through time; reliability | My commitments persist when honesty becomes inconvenient. | Disclosure and accountability are timed to preserve advantage. |
| ○ 2D | GOOD boundary, protection and connection; plasticity | I protect privacy and safety while respecting your agency. | Protection becomes control over what you may know or challenge. |
| Φ 3D | RIGHT field, mind and context; access | I interpret evidence while keeping explanations revisable. | Explanations preserve the preferred story by excluding contrary evidence. |
| ⊙ All | AGREEMENT the whole, integration; validation | I can acknowledge error, repair harm, and remain whole. | The defended self-image organizes the other parts around its preservation. |
An earlier draft of this table labeled the whole "4D." In the framework the whole is not a further dimension. It is the composition of the four stations, which is why its row reads "All."
The center row states the ontological claim at its sharpest. The failure is not simply that the viewpoint is inflated; it is that the viewpoint has been made non-revisable, because being wrong is felt as ceasing to be. The boundary row is where deception meets the wider territory of information control (§2): the same boundary that legitimately protects privacy becomes, under the hijacked emergency, a filter on what others are permitted to know.
The framework names two ways a part can misrepresent its relation to a whole: Inflation (claiming to be the source) and Severance (denying one's own wholeness). The narcissistic pattern runs both, in series. The center is abandoned first, Severance turned inward: there is no point of view here worth trusting. The image then claims the authority the center gave up, Inflation worn as armor. This is why puncturing grandiosity tends to reveal emptiness rather than a hidden, larger self.
Taken row by row, the table is an organizing reading, and the rows are fitted to distortions already described elsewhere in this paper. Its one non-obvious commitment is the bottom row. If the whole is where the pattern lives, the five distortions are not five problems but one organization, and they should rise and fall together: under image threat they should co-occur, and an intervention that changes the anchor of worth should move all of them, while an intervention aimed at a single station (presenting evidence at Φ, setting a boundary at ○) should produce compliance in that row without durable change in the others. That is testable (§15).
These dimensional labels organize the philosophy. They do not establish physical dimensions of mental processes or identify a measured neural architecture.
Within this ethical account, good involves truthful relationship that protects life and respects other people's agency. The moral wrong in manipulative deception is that another person is made to choose or live on terms they cannot adequately evaluate. Sustained, knowing distortion can preserve conditions in which harm continues.
Falsehood alone is not evil. Involuntary perceptual experiences, sincere error, uncertainty, and imperfect memory do not carry the intent of a lie. A true statement can also be used cruelly or recklessly. Truthfulness therefore belongs alongside care, appropriate boundaries, and responsibility for consequences.
Conditioning may help explain harmful behaviour without making it acceptable. Responsibility should consider awareness, available choices, capacity, and opportunities to repair. The two-layer account of §8 sharpens this: awareness is usually thinnest at the core and thickest at the surface, where a person chooses whether to discredit a witness or revise a record. Other people retain the right to set boundaries even when the behaviour has a painful history.
The theory evaluates patterns and actions, without dividing people into permanently good and evil kinds. Corrigibility requires allowing for change and for mistaken judgments about others.
Corrigibility is the capacity and commitment to revise beliefs and behaviour when sufficient evidence shows that revision is needed. It is a practice of truthfulness, not automatic submission to whoever claims to be right. A refusal to accept an unsupported accusation can be entirely reasonable. The framework's metacognitive theory of truthfulness gives the discriminant: the truthful person yields to evidence and resists pressure; the liar yields to pressure and resists evidence. Gaslighting is corrigibility hijacked: pressure presented as evidence.
That discriminant can be measured. Corrections to a person's own elicited self-claims are scored in two matched kinds: corrections backed by evidence the person can inspect (a record, a document, their own earlier statement, a third party's account offered with its source), and corrections backed only by pressure (assertion, status, insistence). Each is scored for acknowledgment, for change in stated confidence, and for whether the revised account is still the one given at the next session. Corrigibility is revision under evidence minus revision under pressure. The index is positive for a corrigible person. Within the protected domain the account predicts a negative index, and change would show as the index moving toward positive on protected claims specifically. In therapy only the evidence arm is used, because pressure must not be manufactured clinically; the pressure arm belongs to research analogues. Trait intellectual humility (Leary et al., 2017) is the natural convergent measure, and its facets already show the expected split: the antagonistic and neurotic components of narcissism, not the agentic one, are tied to failing to separate disagreement from personal threat (Zajenkowski et al., 2025).
In this theory, selective acceptance of correction is a possible expression of conditioned deception, rather than its deepest definition. Someone may welcome correction when it offers advantage and resist it when it threatens a defended identity. The central question remains what happens to truth when preserving that identity becomes costly.
The commitment "I can be wrong and remain whole" is the remedy the whole account depends on, so it cannot simply be assumed. If the self-image is a representation that can change while the person remains, what makes the person remain?
Within the framework the answer is structural. A center is not made of claims, so no claim's failure can remove it. And it was not earned, so it cannot be revoked by evidence about performance. The framework's reading of the aperture as a gift says this directly: the point of view is given before any walk begins; one did not earn it, one is it. Worth located there is unconditional in the strict sense: no condition was ever placed on it, so no condition can fail. The image, by contrast, is a description the person built or was handed. It can be revised because it was never what held them up.
For a reader outside the framework, the same claim has an empirical shadow. Warmth, which communicates that a child is valued, predicted self-esteem; overvaluation, which communicates that a child is valued for being special, predicted narcissism (Brummelman et al., 2015). Worth that rests on being a person can survive correction. Worth that rests on being a particular image cannot, and the literature on contingencies of self-worth describes the costs of the second kind (Crocker & Wolfe, 2001; Crocker & Park, 2004). The remedy is therefore not a better image but a different anchor.
The theory proposes that truth becomes easier to tolerate when acknowledging error no longer means losing one's basic worth. In the terms of §4, the emergency can end only when the accepted lies stop being experienced as reality. The loop of §4.5 runs in both directions: every correction that is absorbed rather than defended against shrinks what future corrections can threaten. There is an established neighbour for this proposal. Self-affirmation research finds that affirming important values in a domain unrelated to a threat can reduce defensive responses to threatening information (Sherman & Cohen, 2006), and a randomized field experiment with 405 young adolescents found that a brief values affirmation reduced aggression among the more narcissistic for a school week (Thomaes et al., 2009). Those results support the direction of the proposal. Whether they hold at the pathological end of the spectrum, and under correction of a specific defended claim rather than general ego threat, is open; the companion thesis treats that as one of its sharpest tests.
The aim would be to combine humane correction with dependable consequences, so honesty does not become a route to humiliation and deception does not remain a reliable route to advantage. In the language of §5, the work is to turn shame back into guilt: a verdict on an act, which can be repaired, rather than a verdict on a person, which can only be fought.
Dependable is the operative word. If deception sometimes works and is sometimes caught, the conditioning account predicts a pattern that is harder to extinguish than one that was never caught at all (§6.3). Families, workplaces, and institutions that enforce consequences sporadically are not merely failing to correct the pattern; on this account they are sustaining it. The practical recommendation is less dramatic than it sounds: not harsher consequences, but consistent ones, delivered without humiliation, so that honesty is reliably safer than distortion and distortion reliably fails.
This does not make a partner, child, colleague, or harmed person responsible for another person's recovery. A person can understand the proposed mechanism, refuse further manipulation, and require sustained evidence before restoring trust.
The commitment I am proposing is:
I can be wrong and remain whole. I can acknowledge harm without defending it. Your experience can challenge my account without threatening my right to exist.
Existing research supports components of this account. It does not establish the complete sequence, its universality, or the claim that it is the primary cause of narcissism. Each study below is stated with what it supports and what it does not show.
The distinctive hypothesis is that learned threat and reward responses help turn protection of a self-image into a recurrent priority over truth; that relief, reward, intermittent consequences, and justification maintain this priority; and that the result fails the four conditions of a legitimate emergency in the manner of §4. No source cited here tests that complete sequence, the convergence of the two routes, the supporting-character mechanism, or the four signatures. Nor do these sources establish that deception evolved exclusively for emergencies, or that this account explains every form of narcissism.
Of the thesis sentence, the "because" is the part testing reaches last. Whether a self-claim is now exempt from revision can be measured prospectively (§15). Whether it was once a lie, maintained against evidence available at the time, mostly cannot, because the mechanisms that turn a lie into a belief (§6.5) also erase the record of the turning. Prospective studies of the loop, and laboratory paradigms in which self-relevant lies are induced and their later acceptance is tracked, can approach it; retrospective claims about any one person's history cannot establish it.
A useful test must define narcissistic traits independently of lying, and measure deception independently of disagreement. Otherwise the conclusion would already be built into the definitions. Studies should distinguish ordinary mistakes, self-deceptive bias, and intentional misrepresentation.
Measuring the defended self-image. People differ in what they treat as identity-relevant, so a single "identity threat" manipulation applied to everyone will be noisy. The design should be idiographic. First elicit each participant's own identity-level claims, with, for each claim, their confidence, what evidence would revise it, and how they expect to feel if it were revised (the companion thesis's Self-Claim Elicitation Protocol). Claims for which nothing would count as revising evidence, and whose revision is forecast in annihilating terms, are the candidates for a defended image. Then separate the threat to the claim from mere unpleasantness with yoked pairs. Pair participants, and write feedback against a protected claim of each partner; both partners receive both sets, framed as feedback about themselves. The same sentence threatens a protected claim for one partner and not for the other, so wording, tone, and content are identical and only diagnosticity varies. Match the items on each recipient's own pre-rated unpleasantness, cross diagnosticity with social cost (public delivery by a high-status evaluator against private delivery by a peer, varied by context and never by wording), and control felt unpleasantness measured at delivery. Individual differences in what counts as identity-relevant then become the manipulation rather than noise. If the diagnosticity effect vanishes under that control, aversiveness explains the defense and the target signature fails. The full design, with measures, analysis, and ethics, is in the companion preprint (Roonz, 2026).
The four signatures. The keystone predicts four effects that can be measured separately, and it is weakened by each one that fails:
Corrigibility index. In treatment research, the index of §13 (revision under evidence minus revision under pressure, scored on the person's own claims) should move from negative toward positive on protected claims as the pattern resolves, and should not move on claims the person never protected.
Affirmation before accountability. If security of worth is the operative variable, a values affirmation before an image-diagnostic correction should reduce defensive deception and increase accurate admission among participants higher in narcissism. If it reduces distress but not deception, or does nothing in the pathological range, the remedy of §13 is wrong or incomplete.
Consistency of accountability. In a repeated task with opportunities for self-serving misreporting, compare consistent detection with intermittent detection, then stop detecting. The account predicts that misreporting persists longer after detection stops when detection was intermittent, and that the difference is larger when the misreports protect a defended self-claim.
Testimony discount. Ask participants to rate the credibility of the same report about another person's experience when it implicates their own self-image and when it does not. The supporting-character mechanism predicts a larger credibility gap among participants higher in narcissism, and a matching gap in empathy.
Co-variation. If the circumpunct table's bottom row is right, the five station-level distortions should co-occur under image threat and move together under worth-security interventions, while single-station interventions should produce compliance without durable change.
Anchor of worth. Contingency of worth on a specific self-image should predict the four signatures better than overall narcissism scores (§6.4).
Accepted lies. Using the elicitation above, identify self-claims that are contradicted by independent records yet held with high confidence and with nothing named that could revise them. The thesis predicts that correction of those claims, and not of comparable claims the person holds loosely, triggers the threat response and the four signatures.
Loop growth. In a repeated design, track whether defensive statements that a participant later recalls and endorses as true (rather than as things they said) predict broader defensiveness to later corrections. The self-sealing loop predicts that believed defenses enlarge the threatened territory, and that defenses the participant still recognizes as lies do not.
Sincere reversal. When participants reverse victim and offender under correction, measure whether they privately and consistently report themselves as the victim. For accepted lies, the account predicts that the reversal is sincere rather than merely strategic. Private report alone cannot settle this. A false belief that was never a lie, a false memory for instance, predicts the same sincere private report as deep self-deception. The discriminating evidence is longitudinal: whether the belief can be traced to a point at which it was defended against evidence then available, and whether it expands under correction as the loop of §4.5 predicts, which a mere false belief should not.
Relief and reward. Track whether relief or gains following a deceptive choice predict its later repetition. Compare these effects with alternative explanations such as expected punishment, material incentives, or habitual impulsivity.
Two routes. Compare punitive, overvaluing, and warm-without-overvaluation environments. The convergence hypothesis predicts identity-contingent defense from the first two and not the third. If only one route predicts it, narrow the account to that route.
Changes in felt safety. Test whether reducing humiliation while retaining accountability improves accurate admission and correction. If it changes distress but not deception, the theory must explain that dissociation rather than assume the intervention was insufficient.
Distinct pathways. Compare deception driven by threatened self-image with deception pursued for gain without evident identity threat. If the latter explains the findings better, narrow the threat-based account instead of redefining every desire as a threat.
Longitudinal development. Follow how threat exposure, overvaluation, incentives, deception, and independently assessed personality traits change over time. A childhood association by itself cannot establish the proposed adult developmental pathway.
Counterevidence must count. Persistent honesty under identity threat, durable acceptance of costly correction, or a better-supported alternative mechanism should change how the theory is applied. Rejecting the theory is not evidence of narcissism, and disagreement must never be treated as proof that the other person is lying.
Kinds of claim, kept separate.
Novelty. The neighbours are established: the overvalued idea, the shame and guilt distinction, moral disengagement, deflationary self-deception, contingencies of self-worth, self-affirmation, the NARC model, self-regulatory models of narcissism, and the dark-triad deception literature. What this paper proposes is the assembly, in particular the reading of the four conditions as the anatomy of the hijack. That assembly has not been checked against the literature for prior statement and is offered as a proposal.
Adjudication. The rule in §3 was first drafted as "life only." On 2026-10-22 the author tested that form against their own intuitions (lying to stop a car thief; lying to avoid being forced from one's home) and found that the line they actually hold is the wrongfulness of the threat, not the size of the loss. The wrongful-threat form replaced it at the author's direction. The thesis sentence of §1 is the author's.
Use. This is a framework for understanding and for generating testable hypotheses. It is not a diagnostic instrument, and it gives no one the standing to diagnose another person.
Framework connections.
protecting_the_lie_thesis.md): the defended self-image is its Protected Identity Proposition; §4.4 derives its insatiability (Proposition 3.2) and its "persistence after withdrawal" discriminator; §8 supplies support for its §3.6 prediction; §6 supplies the developmental account its §3.7 leaves out; §15 uses its Self-Claim Elicitation Protocol.- 2026-10-22 v1.3: answers a third-round review (Kimi). New §6.5, How a lie becomes reality, built on the author's formulation (a lie becomes belief when it is told or acted consistently and then externally reinforced enough), with the evidence that telling alone usually does not do it and the shared-reality boundary that separates instrumental from defended lies; two candidate sources excluded on verification (an Expression of Concern; a failed replication). The "because" of the thesis named as the part testing reaches last (§14); the sincerity test's ambiguity and its longitudinal answer (§15); the lawful-but-spiteful hard case and the moral rather than legal right (§3); a line up front stating that the paper is revised in the open. Fifteen references added.
- 2026-10-22 v1.2: answers Dr. Priyanka's review of the companion preprint, which applied here too. "Accepted lie" (§2) no longer defined as a falsehood: it is a lie by origin and function, with falsity as the consequence, not the criterion. The overvalued idea placed as the nearest construct, with four additions and a retirement condition (§10). Corrigibility measured as revision under evidence minus revision under pressure (§13), with a corrigibility-index test (§15). The yoked-pair design separating diagnosticity from unpleasantness (§15; full design in the preprint). Zajenkowski et al. (2025) added as support for correction felt as threat and the facet split; Stern et al. (2025) added as a generic-threat null that rules out a "fragile self" reading (§10, §14). Eight references added: seven verified sources and the companion preprint.
- 2026-10-22 v1.1: the thesis moved to the author's sentence: anyone might lie when threatened; the narcissist perceives correction as a threat because they have accepted some of their own lies as reality (abstract, §1, one-liner, subtitle; "accepted lie" defined in §2). §3 reworked from "life only" to the wrongful-threat rule as supporting ethics, with the line drawn at the wrongfulness of the threat rather than the size of the loss (author's adjudication). §4 recast around correction as the threat that is not wrongful, with the sincere DARVO reversal (§4.1) and the self-sealing loop (§4.5); the loop added to §8 as traffic between layers; selectivity tied to the thesis (§6.3); three new tests (§15).
- 2026-10-22 v1.0: initial. Developed with Claude from Ashman's draft of the same day: the survival exception's four conditions read as the anatomy of the hijack (the emergency that never ends); the two-layer reconciliation with the Protecting the Lie thesis; two developmental routes converging on one confusion; the dimensional table moved to the corrected ladder, with the virtues named and the whole given no dimension slot. Two external reviews of the draft (Grok, Kimi) answered in the same pass: the survival rule argued rather than asserted, with "life" scoped to the person and the hard cases worked; the exception's own corruption and the verification test (§4.5); selectivity, the merging of the routes, and intermittent reinforcement (§6.3); the anchor of worth (§6.4, §13.2); the supporting-character mechanism for empathy (§5.1); canonical features and expected coverage (§10); idiographic measurement and new tests (§15). Citations verified with receipts in plans/protecting_the_lie_literature_log.md. Supersedes circumpunct_narcissism.html.