Chapter Two — The Violence of Being Finished

The Child on the Screen

In 1983 John Darley and Paget Gross showed different groups of participants the same child taking the same academic test. The child was Hannah, a fourth grader. The videotaped performance was deliberately mixed: some easy items missed, some difficult ones answered, enough competence and enough error to permit more than one plausible reading. What differed was not Hannah’s behavior. What differed was the social information supplied before participants watched it. One group had been led to understand Hannah as coming from a relatively affluent background; another as coming from a poor one. Without the performance tape, the class cue did little to separate estimates of her ability. After participants watched the identical ambiguous performance, however, the cue reorganized what they believed they had seen. The high-status Hannah was judged to be performing above grade level; the low-status Hannah below it. Participants could point to behavior on the tape in support of both conclusions.¹

The experiment is often summarized as an illustration of stereotyping. That is true, but too weak for the problem before us. The more important fact is that the label did not simply substitute for observation. Observation occurred. The problem was that observation arrived already inside a hypothesis. The same performance could be recruited as evidence for incompatible accounts of the same child.

This is the elementary structure of finalization.

Not every first impression is finalizing. Not every stereotype is finalizing. Not every trait judgment is finalizing. Not every mistaken belief is finalizing. Human beings cannot encounter one another without selection, categorization, inference, and compression; Chapter One has already made that concession. Finalization begins at a more demanding threshold: a representation of a person becomes sufficiently organizing that later information is interpreted through it, sufficiently durable that contradiction carries a special burden, and sufficiently authoritative that the representation can begin to govern the conditions under which the person will next be seen.

The last condition matters most. A thought in one stranger’s head may be unfair without becoming socially consequential. The same thought in a teacher, physician, supervisor, judge, parent, partner, therapist, or institution can change opportunities, questions, permissions, scrutiny, credibility, and risk. A model can become part of the target’s environment. Once that happens, the perceiver is no longer only predicting the person. The perceiver may be participating in the production of the evidence by which the prediction will later be judged.

The violence named in this chapter title is therefore not a synonym for misperception. Most misperception is not violence. The term should be reserved for cases in which representational closure acquires enough relational or institutional force to constrain a person’s intelligibility, credibility, opportunity, treatment, or capacity to supply counterevidence. To call every inaccurate judgment violent would make the word morally useless. To deny that some representations can become coercive because they begin as cognition would make social cognition politically naïve.

The chapter’s claim is narrower: a person is “finished” when a partial representation stops behaving like a revisable hypothesis and begins behaving like an interpretive rule.

Impression Before Evidence

Psychology did not discover with Hannah that impressions are organized rather than additive. Solomon Asch’s classic 1946 experiments had already shown how a single descriptor could alter the meaning of an otherwise identical set of traits. Participants heard a person described as intelligent, skillful, industrious, determined, practical, and cautious. One group also heard warm; another heard cold. The substitution changed the total impression and the traits participants considered fitting. Asch’s point was not that one word magically determines a person. It was that an impression is a configuration: information is interpreted in relation to information already present.²

Later models of impression formation made the mechanism more explicit. Susan Fiske and Steven Neuberg’s continuum model treats person perception as ranging from relatively category-based processing toward more individuated processing, depending in part on information, attention, motivation, and whether the perceiver has reason to move beyond an initial categorization.³ James Uleman, Leonard Newman, and Gordon Moskowitz reviewed a large experimental literature showing that trait inferences can arise spontaneously when people merely comprehend descriptions of behavior, without having deliberately set out to make trait judgments.⁴ The social mind is not a neutral archive into which behavior is deposited and later interpreted. Interpretation begins during perception.

That fact is not itself pathological. If someone returns a lost wallet, interrupts every person at dinner, lies about a trivial matter, notices a frightened child, or humiliates a subordinate, observers do not store each act as an isolated video file. They infer generosity, dominance, dishonesty, attentiveness, cruelty, insecurity, courage, or some other dispositional possibility. Social knowledge requires abstraction. A person who could remember behavior but never infer anything from it would be nearly as socially disabled as a person who inferred too quickly.

The problem begins when abstraction hides its own scale change. Behavior occurs at one level; character is inferred at another. “She interrupted me three times” and “she is domineering” are not the same proposition. The first can be established by observation. The second is a theory about a distribution of behavior across situations. It may be correct. It may also be a reasonable working hypothesis. But the evidentiary burden has changed.

The correspondence-bias literature is useful precisely because it shows how easily that change can be undernoticed. Daniel Gilbert and Patrick Malone define correspondence bias as the tendency to infer enduring dispositions from behavior that can be explained by situational constraints. Their review identifies multiple mechanisms rather than one universal error: observers may fail to notice constraints, hold unrealistic expectations about how much situations matter, categorize behavior too strongly, or fail to correct an initial dispositional inference sufficiently.⁵ The danger for this book is obvious. If a person acts curtly under humiliation, avoidantly under threat, passively inside rigid hierarchy, or chaotically during overload, the observer may be looking at real behavior while still drawing an overlarge conclusion about the person.

Even here, however, the literature requires restraint. The familiar claim that actors explain themselves situationally while observers explain them dispositionally became a textbook certainty, yet Bertram Malle’s meta-analysis of 173 studies found no general actor-observer asymmetry across valence and designs; the classic pattern appeared only under particular conditions.⁶ This matters methodologically. A book about finalization would reproduce its own object if it turned every observer into a caricature of dispositional error. Social psychology contains famous biases that weaken when their boundary conditions are examined. The argument must be built from mechanisms that survive qualification, not from a mythology in which perceivers are always wrong and targets always possess superior self-knowledge.

The distinction can now be stated more sharply. Impression formation is inevitable. Trait inference is common. Dispositional inference can be accurate. Finalization is not any of those things. It is what happens when an inference acquires premature immunity from the conditions that should revise it.

From Description to Essence

One way a representation becomes harder to revise is by changing ontological register. A behavioral tendency becomes a nature. A repeated pattern becomes what the person really is. A social category becomes evidence of an inner essence presumed to explain its members.

The language of psychological essentialism is useful here if it is kept precise. Douglas Medin and Andrew Ortony introduced psychological essentialism as an account of how people may treat categories as if observable features are generated by deeper, nonobvious realities. Their concern was not limited to social identity, and essentialism in their sense does not require a person to articulate a metaphysical doctrine.⁷ The cognitive move is enough: surface regularities appear to point toward something underlying that makes the category what it is.

Applied to persons, the move has enormous intuitive force. Someone who lies is not merely a person who lied; perhaps the lie revealed dishonesty. Someone who repeatedly withdraws under conflict is not merely withdrawing; perhaps the behavior reveals avoidance. Someone who dominates subordinates may be revealing cruelty. Sometimes these inferences are exactly what moral life requires. We need language for character because actions are not independent draws from nowhere.

Yet essence is where a useful pattern can become a trap. Once behavior is understood as an expression of a hidden core, contradictory acts become easier to discount. Generosity by a person judged selfish can be recoded as impression management. Competence by a person judged incompetent can become luck, coaching, or an unusually easy task. Warmth by a person judged manipulative can become strategic charm. The model grows stronger not because it predicts more accurately but because it learns to explain away every possible disconfirmation.

Nick Haslam et al.’s empirical work on essentialist beliefs about social categories is especially important because it refuses a simple good-versus-bad account. They found that essentialist beliefs were multidimensional rather than a single syndrome; ratings of forty social categories produced separable dimensions corresponding roughly to natural-kind beliefs and to entitativity or reification. Essentialism, they cautioned, is not monolithically related to prejudice or devaluation.⁸ That caution should govern our use of the concept. The claim is not that seeing stability is oppressive. The claim is that reification supplies one cognitive route by which a probabilistic pattern can come to feel like an inhering core.

Finalization is therefore not essentialism. A person can essentialize a category without finalizing a particular individual, and a person can finalize an individual without believing in social essences. Essentialism is one mechanism that can raise the cost of revision because it converts observed regularity into presumed nature.

The question then becomes motivational: why would a mind prefer a finished account when uncertainty remains?

Seizing, Freezing, and the Relief of Knowing

Donna Webster and Arie Kruglanski’s work on need for cognitive closure provides one answer. Their measure treated need for closure as a dispositional tendency expressed through desires for predictability and order, discomfort with ambiguity, decisiveness, and closed-mindedness, while later theory emphasized that closure can also be situationally evoked.⁹ Kruglanski and Webster described two broad tendencies: urgency, which favors obtaining a definite answer quickly, and permanence, which favors preserving that answer once obtained. They called the corresponding processes seizing and freezing.¹⁰

The language is almost too perfect for this book, which makes it dangerous. Finalization must not become a literary renaming of need for closure. Need for closure is a studied motivational construct with a specific theoretical and measurement history. Finalization is a proposed relational configuration. A perceiver can have high need for closure without finalizing a particular person, and a low-closure perceiver can still become finalizing when reputation, fear, hierarchy, ideology, or self-protection make revision costly. The value of closure theory here is mechanistic: it shows that the desire for definiteness can change how much information a person processes and how strongly early cues govern later judgment.

Motivation can enter by another route as well. Ziva Kunda’s review of motivated reasoning argued that people do not simply choose conclusions at will. Desired conclusions bias the cognitive strategies through which information is accessed, generated, and evaluated, but the conclusion still needs a plausible justificatory route. People are constrained by what they can make seem reasonable to themselves.¹¹ That constraint is precisely why finalization can feel evidentiary from the inside. The finalizing person need not think, I have decided what you are and will ignore all facts. They may instead experience a continuous stream of reasons why each new fact fits the established account.

This is the first phenomenological asymmetry worth preserving. From the outside, closure can look like refusal. From the inside, it often feels like accumulating coherence.

Suppose a supervisor believes that an employee is politically difficult. A forceful disagreement confirms the theory. Silence confirms it too: now the employee is withholding. A successful collaboration with one colleague can be interpreted as selective alliance-building. A failed collaboration becomes direct evidence. A careful written record can look defensive. The absence of a record can look irresponsible. None of these interpretations is logically impossible. That is what makes the system durable. Finalization rarely requires obviously false statements. It can survive through locally plausible reinterpretations whose cumulative architecture has become unfalsifiable.

The same process occurs in intimate life. Once a partner is understood as selfish, every gift can be examined for motive. Once a child is understood as fragile, every risk can be interpreted through anticipated collapse. Once a parent is understood as controlling, every request can be heard as domination. Again, the representations may contain truth. The danger is not that a stable pattern has been named. The danger is that the pattern has become a universal solvent: everything dissolves into it.

This is why finalization should be defined partly by its treatment of counterevidence. But counterevidence behaves more strangely than the simple metaphor of “confirmation bias” suggests.

What Contradiction Actually Does

The easiest version of this chapter would claim that people remember what confirms their expectations and forget what does not. Some studies do show that pattern. Myron Rothbart et al., for example, found better recall and higher frequency estimates for stereotype-confirming behaviors when an expectancy was induced before participants encountered the behavioral information.¹² If we stopped there, finalization could be described as selective memory: the model reproduces itself because contradiction disappears.

The broader literature does not permit that simplicity.

Charles Stangor and David McMillan’s meta-analysis of fifty-four experiments found that, overall, recall and recognition sensitivity were better for expectancy-incongruent than expectancy-congruent information. Incongruity often attracts processing because it requires explanation. Uncorrected recognition measures did show a tendency to report congruent information as having been seen, but the central result is an important rebuke to any theory in which human beings are cognitively blind to contradiction.¹³

The more interesting question is what happens after contradiction is noticed.

Sometimes it changes the model. Sometimes it does not. Research on stereotype change has long distinguished several possibilities. In a bookkeeping model, disconfirming cases gradually accumulate until the representation shifts. In a conversion model, striking disconfirmation can produce a more abrupt change. In a subtyping model, the discrepant person is placed into an exception category, preserving the larger representation. Lucy Johnston and Miles Hewstone found that stereotype change depended in part on whether inconsistent information was concentrated in a few members or dispersed across many; disconfirming members could be isolated as atypical rather than allowed to revise the category.¹⁴

Kunda and Kathryn Oleson demonstrated an especially revealing version of the same process. Participants encountered targets who strongly violated stereotypes. When there was no obvious additional information by which to explain the deviant case, participants generalized more from the counterstereotypic person and changed their stereotype. But when even a neutral attribute was supplied, participants could use it as grounds for subtyping. The neutral feature came to function as an explanation for why this person did not count as evidence about the category.¹⁵

The mechanism is devastatingly ordinary. Counterevidence need not be denied. It can be quarantined.

The excellent woman is exceptional. The gentle man is not representative. The generous rich person is unusual. The competent employee is good only under unusually favorable conditions. The addicted person who sustains recovery was never “really” as severe a case. The once-anxious child who now travels alone has simply learned a script. The colleague who succeeds after being judged incapable was carried by the team. The partner who changes after years of conflict is behaving temporarily because the relationship is at risk.

Some of these explanations will sometimes be true. That is the permanent difficulty. Finalization is not detectable by asking whether a single explanation is possible. It appears in the asymmetry of explanation across time: confirming evidence is allowed to update the model directly, while disconfirming evidence is repeatedly assigned to a special causal category that protects the model from revision.

A genuinely revisable representation therefore requires more than noticing anomalies. It requires rules for when anomalies are allowed to matter.

Recent work on impression updating makes the point from the other direction. Peter Mende-Siedlecki, Sean Baron, and Alexander Todorov showed that people update moral and ability impressions asymmetrically according to the diagnostic value they assign to behavior: rare, therefore more informative, behaviors carried disproportionate weight. Highly competent behavior was especially diagnostic of ability, while immoral behavior was especially diagnostic of morality.¹⁶ In another line of work, Jenifer Siegel and colleagues found that beliefs about agents learned to be morally bad could remain comparatively uncertain and volatile, permitting revision when later evidence improved.¹⁷ Human person perception, in other words, includes machinery for change. The mind is not condemned to preserve first impressions.

This finding changes the theory. Finalization cannot be defined as cognitive stability. Stable beliefs can be rational when evidence is redundant and diagnostic. Nor can it be defined as reluctance to update. Some evidence should barely move a well-supported judgment. If a person has betrayed ten confidences, one afternoon of discretion should not reset the posterior to innocence. Revisability does not mean equal weighting of every new observation.

The relevant defect is miscalibrated revision: the evidence required to worsen the judgment and the evidence required to improve it become radically unequal for reasons the model itself cannot justify.

That inequality may be epistemic, emotional, relational, or political. It may arise because negative behavior is genuinely more diagnostic in a domain. It may arise because the perceiver fears being fooled twice. It may arise because trust has been injured and appropriately recovers slowly. Or it may arise because the category has become self-protective. A serious account cannot decide among those possibilities by moral intuition alone.

The question is not whether the model changes slowly. It is whether its revision rules are themselves answerable to evidence.

When Expectation Becomes Environment

The strongest form of finalization begins when a representation stops remaining private.

Carl Word, Mark Zanna, and Joel Cooper’s 1974 experiments on interracial interaction remain among the clearest demonstrations of the transition from expectation to environment. In the first experiment, White interviewers interacting with Black job applicants behaved differently from those interacting with White applicants: they sat farther away, conducted shorter interviews, and displayed patterns of speech and nonverbal behavior consistent with lower interpersonal immediacy. In a second experiment, the researchers trained interviewers to reproduce those different interaction styles while interviewing White applicants. Applicants exposed to the style previously directed toward Black candidates were subsequently judged as performing less adequately and appeared more nervous.¹⁸

The design matters because it separates two propositions that are often confused. The first is that prejudice can distort judgment. The second is more radical: the perceiver’s behavior can alter the target’s behavior, creating evidence that appears to justify the original representation. The target is no longer being observed under neutral conditions. The hypothesis has entered the situation.

Mark Snyder, Elizabeth Tanke, and Ellen Berscheid demonstrated a parallel process in heterosexual interaction. Male participants who believed, on the basis of a manipulated photograph, that a female conversation partner was physically attractive behaved toward her differently from men who believed their partner was unattractive. Independent raters, hearing only the women’s side of the conversations, then judged women paired with the “attractive” expectancy as behaving in ways that were more sociable, poised, and likeable.¹⁹ The initial belief was false with respect to the actual partner, yet it helped create the interpersonal pattern that seemed to confirm it.

Snyder and William Swann subsequently showed comparable processes with hostility expectations. Perceivers led to expect a hostile partner elicited more hostile behavior from that partner than perceivers led to expect a nonhostile one. Their work also suggested an important limit: confirmation effects did not simply become permanent personality changes. The continuation of behavior into later interactions depended on whether the target had come to internalize the behavior elicited during the first encounter.²⁰ In a related series of experiments, Snyder and Swann showed that perceivers could select questions designed to test an introvert or extravert hypothesis in ways that made confirming answers more likely.²¹

Joseph Darley and Russell Fazio later organized this family of findings into an interactional sequence. A perceiver’s expectation can alter the perceiver’s conduct; that conduct becomes part of the target’s situation; the target responds; and the response can then be interpreted as evidence that the expectation was correct.²²

This is the point at which the language of finalization earns its relational meaning. A representation is not finalizing simply because one observer privately believes it with high confidence. It becomes more consequential when the belief affects access to situations in which contrary evidence could appear, changes the questions through which the person is examined, alters the tone and warmth of interaction, modifies the risk the perceiver is willing to take, or determines which behaviors will be considered diagnostic.

A teacher who assumes a child lacks aptitude may offer less demanding material and then observe slower development. A manager who believes an employee lacks judgment may reserve consequential assignments for others and later cite the absence of demonstrated judgment. A partner who expects evasion may interrogate rather than inquire, encounter defensiveness, and treat the defensiveness as proof. A clinician who expects exaggeration may ask fewer exploratory questions and then record the absence of corroborating detail. In each case, there may still be genuine information about the target. The epistemic defect lies in forgetting that the conditions under which the information was produced were partly structured by the perceiver’s prior model.

The danger grows when the perceiver controls opportunity. The person who assigns work also decides which capacities can become visible. The person who controls a diagnosis influences which future clinicians will read the case through that diagnosis. The parent who decides a child is fragile controls which risks the child is permitted to attempt. The institution that marks a person high-risk can reduce the very opportunities through which lower risk would become observable.

This does not mean social expectations manufacture persons at will. The strongest versions of the self-fulfilling-prophecy story are false. People resist, reinterpret, compensate, seek other environments, and carry dispositions that survive hostile expectations. The effect of a perceiver depends on power, duration, the target’s alternatives, the availability of other relationships, and the extent to which the target internalizes the interaction. But the minimal conclusion is secure: some representations do not only describe social reality. They help configure it.

Once a model can influence the evidence by which it will later be judged, ordinary standards of confirmation become insufficient. The fact that a prediction came true no longer tells us how much of the truth belonged to the target before the prediction entered the scene.

The Accuracy Objection

At this point, a serious critic should object that the argument is becoming too easy. Psychology has spent decades cataloguing biases, self-fulfilling prophecies, stereotyping, motivated reasoning, and attributional errors. A reader could emerge with the flattering impression that social judgment is mostly projection and that the person being judged possesses the more accurate view by default.

That conclusion would be indefensible.

David Funder’s Realistic Accuracy Model begins from a premise that bias-centered accounts can obscure: personality judgments can be accurate, and understanding accuracy requires asking whether relevant behavioral information exists, becomes available, is detected by the judge, and is utilized correctly.²³ Accuracy is not the absence of interpretation. It is successful interpretation under conditions that permit valid information to travel from behavior to judgment.

Longitudinal acquaintance research makes the same point empirically. Jeremy Biesanz, Stephen West, and Allison Millevoi found that as people became better acquainted, consensus and accuracy in personality judgments increased in meaningful ways; additional contact can improve rather than merely entrench representation.²⁴ Other people sometimes know us better with time because they have seen more behavior under more conditions.

Lee Jussim’s work on interpersonal expectations makes the objection sharper still. Reviewing teacher-expectancy research with Jacquelynne Eccles and Stephanie Madon, Jussim argued that teacher expectations predict student achievement largely because they are often accurate to begin with; bias and self-fulfilling effects exist but account for a smaller share of the association.²⁵ Jussim and Kent Harber’s later review of thirty-five years of research concluded that teacher self-fulfilling prophecies are real but generally modest, tend to dissipate rather than accumulate without limit, and coexist with substantial perceptual accuracy.²⁶ Jussim’s broader synthesis presses the same claim: social perception is not dominated by error, and any theory that treats bias or self-fulfilling prophecy as the ordinary condition of interpersonal judgment badly misstates the evidence.²⁷

This is not an inconvenience for the theory of finalization. It is its necessary boundary condition.

If a supervisor has watched an employee mishandle judgment repeatedly across different assignments, independent observers identify the same pattern, the pattern survives changes in incentives and context, and later evidence continues to support it, a stable negative assessment may be warranted. If a partner has experienced repeated deception over years, distrust may be rational even after a period of apparent improvement. If a court considers a documented history of violence when making a risk decision, it is not automatically “flattening” the person by treating history as predictive evidence. A model does not become unjust because it is durable.

Accuracy, however, does not answer every question this book cares about. A representation can be substantially accurate at one level and still exceed its evidence at another. “This employee has repeatedly failed to escalate risk in this role” can be well supported. “This person lacks judgment in every consequential domain” is a larger claim. “Because this person lacks judgment, no future evidence of good judgment can count” is larger still. And “because this person lacks judgment, I am entitled to govern unrelated domains of their life” adds jurisdiction that accuracy about the original behavior cannot supply.

The accuracy objection therefore improves the construct by separating four problems that moral language often bundles together: falsity, overgeneralization, nonrevisability, and authority inflation. Finalization does not require all four. It does not even require falsity. A person can be finished by a representation containing substantial truth if the truth is granted scope, permanence, or jurisdiction beyond what the evidence warrants.

Conversely, a false judgment is not automatically finalizing. A stranger can form an absurd opinion, revise it an hour later, and leave no consequence behind. Error matters, but error is not the whole phenomenon.

The stronger criterion is proportionality. Does the representation preserve a defensible relation among evidence, scope, confidence, revisability, and consequence?

Where that relation is preserved, we have judgment. Where it fails systematically, we approach finalization.

When Stability Becomes Sentence

The construct can now be stated more formally.

Finalization is a relational condition in which a partial representation of a person acquires more scope, permanence, or authority than its evidence warrants, such that new information is interpreted under revision rules that systematically protect the representation from appropriate correction and, in consequential settings, may shape the opportunities through which further evidence can be produced.

This definition contains four separable dimensions.

The first is scale inflation. Evidence gathered at one level is promoted to a broader level without sufficient warrant: event becomes disposition; disposition becomes character; character becomes destiny. The mistake is not the act of generalization but the unmarked increase in scope.

The second is revision asymmetry. Evidence consistent with the representation is allowed to update it directly, while inconsistent evidence repeatedly requires special explanation, subtyping, discounting, or an impossible evidentiary threshold. The problem is not that all evidence receives equal weight—it should not—but that the rules for weighting have become structurally self-protective.

The third is evidentiary production. The representation influences the conditions under which the target acts, which capacities become visible, what questions are asked, which opportunities are offered, what risks are permitted, or how behavior is elicited. Later “confirmation” therefore contains an interactional component that the perceiver may fail to recognize.

The fourth is authority inflation. The representation is attached to consequences outside the domain that generated it or is treated as conferring a right to decide more than the evidence licenses. Expertise in one domain becomes presumed competence over the whole person; lawful classification for one purpose becomes ontological judgment; relational injury becomes jurisdiction over identity.

No single dimension by itself is sufficient in every case. A one-time overgeneralization can be corrected. A stable judgment can be accurate. A self-fulfilling interaction can occur without durable closure. An institution can exercise broad authority legitimately when law and evidence warrant it. Finalization concerns the configuration: the representation becomes difficult to revise for reasons that exceed its evidentiary strength, and its consequences become large enough that the person must increasingly live inside the model.

This also clarifies the chapter title. Not every finalization is violence. The word becomes appropriate when finalization is joined to power: when the model materially constrains a person’s credibility, opportunity, treatment, relationship, liberty, or ability to generate the counterevidence required for revision. The violence is not that another mind has described you. Human life is impossible without description. The violence lies in converting a description into a sentence whose own operation helps keep the sentence true.

The Violence of Credibility

Power does not enter only through opportunity. It enters through credibility.

Miranda Fricker’s account of testimonial injustice names one form of this problem with greater precision than the general language of misrecognition can supply. A speaker suffers testimonial injustice when prejudice causes a hearer to assign the speaker less credibility than the speaker deserves. Fricker’s central cases concern identity power: socially available prejudices can alter, often without explicit deliberation, whether a person is treated as a competent knower of what they have experienced.²⁸

The concept should not be dissolved into the claim that it hurts when people do not believe us. Disbelief can be reasonable. Testimony can be false. A person can sincerely report an experience and still misunderstand it. Fricker’s point is narrower and more political: credibility itself is distributed through social perception, and identity prejudice can corrupt that distribution.

Kristie Dotson extends the pressure in a different direction. Her account of epistemic violence examines practices of silencing in which a hearer’s pernicious ignorance produces failures of uptake that render a speaker’s communicative agency vulnerable. The violence is not simply that two people disagree about what happened. It concerns recurrent failures in the conditions under which certain speakers can successfully participate in testimonial exchange at all.²⁹

These concepts are neighboring theories, not synonyms for finalization. Testimonial injustice can occur without a durable, person-specific finalizing model. Finalization can occur without identity prejudice. A parent can finalize a child, a spouse can finalize a partner, and an organization can finalize an employee for reasons having little to do with socially subordinated identity. Dotson’s epistemic violence also carries a more specific political and communicative burden than this book’s general construct should appropriate.

The comparison is useful because it reveals what happens when a representation affects not only what observers think about a person but what they permit the person to know.

Once someone has been coded unreliable, unstable, manipulative, naïve, ideological, hysterical, defensive, difficult, or incapable of insight, their own testimony can become recursively disqualified. “That is not what happened” becomes further evidence of denial. “I have changed” becomes evidence of impression management. “You misunderstood my motive” becomes proof of lack of self-awareness. “This system is treating me differently” becomes proof of grievance orientation. The person’s attempt to supply counterevidence is converted into confirming evidence for the reason counterevidence cannot be trusted.

This is one of finalization’s strongest forms because it attacks the channel through which revision would otherwise occur.

The point is not that self-testimony deserves sovereignty. People lie, confabulate, rationalize, forget, and misunderstand themselves. The Introduction rejected complete self-sovereignty for exactly this reason. Other people can see patterns we cannot. Institutions sometimes possess records more reliable than memory. A victim is not obligated to accept an offender’s preferred interpretation of the harm. A clinician is not required to treat every patient theory as medically correct. Epistemic respect does not mean evidentiary equality among all claims.

What it requires is that first-person testimony remain evidence rather than becoming inadmissible by definition. A person may not control the verdict, but they must not be excluded from the evidentiary system simply because the existing verdict predicts that anything they say will be self-serving.

The justice problem therefore appears at a second level. Some people confront finalizing representations whose authority is amplified by race, gender, disability, class, criminalization, psychiatric diagnosis, immigration status, or other historically organized structures of credibility. In those cases the representational problem cannot be repaired by telling the individual to become more articulate. The audience’s conditions of uptake are themselves part of the system.

This is where a universal psychology of being misjudged reaches its political limit. Everyone can be misunderstood. Not everyone is equally exposed to institutions whose misrecognition is backed by inherited credibility hierarchies and material power.

The Counterfeit of Openness

There is, however, an equal and opposite way to corrupt this theory: turn revisability into compulsory doubt.

Imagine a friend who has borrowed money repeatedly and lied about repayment. A spouse who has become violent. A manager whose judgment has failed in the same high-consequence domain across years. A clinician facing repeated evidence that a treatment is harming a patient. A court evaluating a person with a documented pattern of serious violence. In each case, the demand “remain open to who I might become” can become an attempt to shift the cost of uncertainty onto the person who bears the risk.

No theory of unfinalized personhood can require that.

Revisability concerns the epistemic status of a representation. It does not entail unlimited access, renewed trust, restored authority, repeated exposure, or suspension of consequence. A person can remain unfinalized in your understanding while being permanently excluded from your life. You can believe that someone may become kinder and still refuse another conversation. You can believe that a violent offender is capable of transformation and still support confinement proportionate to present risk and lawful punishment. You can accept that an employee’s future elsewhere remains open while concluding that the person should no longer hold this role. You can refuse to reduce a person to an addiction while refusing to lend them money again.

This distinction is essential because consequences themselves produce information. Trust is not a philosophical compliment paid to human potential. It is a decision under uncertainty about exposure to another person. Different decisions properly require different thresholds. The evidence needed to believe that someone has changed is not identical to the evidence needed to risk a child’s safety, restore fiduciary authority, reopen an intimate relationship, or return a person to a role in which prior failures were catastrophic.

A mature representation can therefore be epistemically open and behaviorally closed.

That sentence should be difficult. Contemporary moral language often treats openness as virtue and closure as failure. But some forms of closure are precisely how human beings recognize consequence. The injured person need not become the laboratory in which the offender demonstrates transformation. The organization need not recreate the same high-risk conditions solely to discover whether its previous judgment can be falsified. The victim is not obligated to provide new opportunities for the accused to generate exculpatory evidence.

Finalization begins only when the protective decision is silently promoted into total knowledge. “I will never trust you with my finances again” may be a bounded consequence. “You are incapable of honesty in any relationship and nothing you ever do can count against that conclusion” is a different proposition. “You cannot return to this role” can be legitimate. “Your failure here proves that no institution should ever treat you as capable” changes scale and jurisdiction. “I am ending this marriage because I cannot safely remain” does not require the further claim that the other person is exhausted by the worst thing they did within it.

The counterfeit of revisability says: because persons can change, every judgment must remain practically reversible.

The stronger position says: because persons can change, no consequence should pretend to prove more about the person than the consequence requires.

This is how accountability and unfinalizability coexist. We can close doors without declaring that no other door should ever open. We can retain memory without treating memory as destiny. We can make decisions whose consequences are lasting while admitting that the person upon whom they fall remains temporally unfinished.

The Person Inside the Verdict

We can now return to Hannah.

Nothing in Darley and Gross’s experiment tells us what Hannah would have said about the observers’ judgments. The child exists on the screen as an object of interpretation. Participants receive a social category, watch an ambiguous performance, and infer ability. Psychology records the judgment. Hannah does not answer back.

Real persons do.

They say, You are treating one failure as the whole pattern. They say, You have already decided what every new action means. They say, You are confusing what happened in that relationship with who I am everywhere. They say, You never gave me the assignment by which I could have shown you otherwise. They say, You trust everyone else’s account of me more than mine. They say, I am not that person anymore. They say, That part is true, but not the conclusion you have built around it.

Sometimes they are right.

Sometimes the protest is itself a defense.

A person confronted with an accurate pattern can experience specificity as reduction because the truth threatens a preferred self-concept. The employee who has repeatedly failed to manage conflict may hear any account of the pattern as stereotyping. The parent whose behavior has been controlling may insist that the family has “flattened” a lifetime of love into a few incidents. The partner who has betrayed trust may experience sustained distrust as refusal to recognize change before enough evidence of change exists. The charismatic outsider may call ordinary accountability an institution’s inability to tolerate complexity.

The theory therefore reaches a point where the perceiver’s errors can no longer carry the whole argument. We have shown how impressions form, how traits are inferred, how essence can be presumed, how closure can become motivationally attractive, how contradiction can be quarantined, how expectations can help produce confirming behavior, how accuracy complicates every accusation of bias, and how power can turn representation into consequence. None of that tells us, by itself, what to do when the represented person objects.

The protest “Do not flatten me” has at least three possible meanings.

It can mean: your representation is false.

It can mean: your representation contains truth, but you have expanded its scope or authority beyond what the evidence supports.

And it can mean: your representation is accurate enough to require a change in me that I do not yet want to make.

Those possibilities can feel nearly identical from inside the threatened self.

That is why resisting finalization cannot mean demanding belief. The right at stake is more exacting: the right to remain representable by evidence that has not yet been collected, without claiming immunity from the evidence already here.

A person is not unflattened because every account remains undecided. A person is unflattened when no account is allowed to become more final than its evidence, scope, revision rules, and legitimate authority permit.

The unresolved burden now sits inside the first person. If another mind can finish us too quickly, the self can also protect itself too completely. The difficult question is no longer whether other people have the right to judge. They do. It is how to recognize the difference between being reduced and being corrected when both can hurt in exactly the same place.

The protest is simple.

Do not flatten me.

Its truth is not.

Notes

1. John M. Darley and Paget H. Gross, “A Hypothesis-Confirming Bias in Labeling Effects,” Journal of Personality and Social Psychology 44, no. 1 (1983): 20–33, https://doi.org/10.1037/0022-3514.44.1.20.

2. Solomon E. Asch, “Forming Impressions of Personality,” Journal of Abnormal and Social Psychology 41, no. 3 (1946): 258–290, https://doi.org/10.1037/h0055756.

3. Fiske and Neuberg, “A Continuum of Impression Formation,” 1–74.

4. James S. Uleman et al., “People as Flexible Interpreters: Evidence and Issues from Spontaneous Trait Inference,” Advances in Experimental Social Psychology 28 (1996): 211–279, https://doi.org/10.1016/S0065-2601(08)60239-7.

5. Daniel T. Gilbert and Patrick S. Malone, “The Correspondence Bias,” Psychological Bulletin 117, no. 1 (1995): 21–38, https://doi.org/10.1037/0033-2909.117.1.21.

6. Bertram F. Malle, “The Actor-Observer Asymmetry in Attribution: A (Surprising) Meta-Analysis,” Psychological Bulletin 132, no. 6 (2006): 895–919, https://doi.org/10.1037/0033-2909.132.6.895. Across 173 studies, Malle found no general actor-observer asymmetry of the classic textbook form; the pattern depended on particular conditions and measurement choices.

7. Douglas L. Medin and Andrew Ortony, “Psychological Essentialism,” in Similarity and Analogical Reasoning, ed. Stella Vosniadou and Andrew Ortony (Cambridge University Press, 1989), 179–195.

8. Nick Haslam et al., “Essentialist Beliefs about Social Categories,” British Journal of Social Psychology 39, no. 1 (2000): 113–127, https://doi.org/10.1348/014466600164363. Their analyses distinguished separable dimensions of essentialist belief and caution against treating essentialism as a unitary or uniformly prejudicial tendency.

9. Donna M. Webster and Arie W. Kruglanski, “Individual Differences in Need for Cognitive Closure,” Journal of Personality and Social Psychology 67, no. 6 (1994): 1049–1062, https://doi.org/10.1037/0022-3514.67.6.1049.

10. Arie W. Kruglanski and Donna M. Webster, “Motivated Closing of the Mind: ‘Seizing’ and ‘Freezing,’” Psychological Review 103, no. 2 (1996): 263–283, https://doi.org/10.1037/0033-295X.103.2.263.

11. Ziva Kunda, “The Case for Motivated Reasoning,” Psychological Bulletin 108, no. 3 (1990): 480–498, https://doi.org/10.1037/0033-2909.108.3.480.

12. Myron Rothbart et al., “Recall for Confirming Events: Memory Processes and the Maintenance of Social Stereotypes,” Journal of Experimental Social Psychology 15, no. 4 (1979): 343–355, https://doi.org/10.1016/0022-1031(79)90043-X.

13. Charles Stangor and David McMillan, “Memory for Expectancy-Congruent and Expectancy-Incongruent Information: A Review of the Social and Social-Developmental Literatures,” Psychological Bulletin 111, no. 1 (1992): 42–61, https://doi.org/10.1037/0033-2909.111.1.42. Their meta-analysis synthesized fifty-four experiments and found an overall advantage for expectancy-incongruent information on recall and recognition-sensitivity measures, complicating simple confirmation-only accounts of social memory.

14. Lucy Johnston and Miles Hewstone, “Cognitive Models of Stereotype Change: 3. Subtyping and the Perceived Typicality of Disconfirming Group Members,” Journal of Experimental Social Psychology 28, no. 4 (1992): 360–386, https://doi.org/10.1016/0022-1031(92)90051-K.

15. Ziva Kunda and Kathryn C. Oleson, “Maintaining Stereotypes in the Face of Disconfirmation: Constructing Grounds for Subtyping Deviants,” Journal of Personality and Social Psychology 68, no. 4 (1995): 565–579, https://doi.org/10.1037/0022-3514.68.4.565.

16. Peter Mende-Siedlecki et al., “Diagnostic Value Underlies Asymmetric Updating of Impressions in the Morality and Ability Domains,” Journal of Neuroscience 33, no. 50 (2013): 19406–19415, https://doi.org/10.1523/JNEUROSCI.2334-13.2013.

17. Jenifer Z. Siegel et al., “Beliefs about Bad People Are Volatile,” Nature Human Behaviour 2 (2018): 750–756, https://doi.org/10.1038/s41562-018-0425-1. The experiments found greater uncertainty and volatility in beliefs about agents learned to be morally bad, leaving greater room for upward revision when later evidence improved.

18. Carl O. Word et al., “The Nonverbal Mediation of Self-Fulfilling Prophecies in Interracial Interaction,” Journal of Experimental Social Psychology 10, no. 2 (1974): 109–120, https://doi.org/10.1016/0022-1031(74)90059-6.

19. Mark Snyder et al., “Social Perception and Interpersonal Behavior: On the Self-Fulfilling Nature of Social Stereotypes,” Journal of Personality and Social Psychology 35, no. 9 (1977): 656–666, https://doi.org/10.1037/0022-3514.35.9.656.

20. Snyder and Swann, “Behavioral Confirmation in Social Interaction,” 148–162.

21. Mark Snyder and William B. Swann Jr., “Hypothesis-Testing Processes in Social Interaction,” Journal of Personality and Social Psychology 36, no. 11 (1978): 1202–1212, https://doi.org/10.1037/0022-3514.36.11.1202.

22. John M. Darley and Russell H. Fazio, “Expectancy Confirmation Processes Arising in the Social Interaction Sequence,” American Psychologist 35, no. 10 (1980): 867–881, https://doi.org/10.1037/0003-066X.35.10.867.

23. David C. Funder, “On the Accuracy of Personality Judgment: A Realistic Approach,” Psychological Review 102, no. 4 (1995): 652–670, https://doi.org/10.1037/0033-295X.102.4.652.

24. Jeremy C. Biesanz et al., “What Do You Learn about Someone over Time? The Relationship between Length of Acquaintance and Consensus and Self-Other Agreement in Judgments of Personality,” Journal of Personality and Social Psychology 92, no. 1 (2007): 119–135, https://doi.org/10.1037/0022-3514.92.1.119.

25. Lee Jussim et al., “Social Perception, Social Stereotypes, and Teacher Expectations: Accuracy and the Quest for the Powerful Self-Fulfilling Prophecy,” Advances in Experimental Social Psychology 28 (1996): 281–388, https://doi.org/10.1016/S0065-2601(08)60240-3.

26. Lee Jussim and Kent D. Harber, “Teacher Expectations and Self-Fulfilling Prophecies: Knowns and Unknowns, Resolved and Unresolved Controversies,” Personality and Social Psychology Review 9, no. 2 (2005): 131–155, https://doi.org/10.1207/s15327957pspr0902_3.

27. Lee Jussim, Social Perception and Social Reality: Why Accuracy Dominates Bias and Self-Fulfilling Prophecy (Oxford University Press, 2012), https://doi.org/10.1093/acprof:oso/9780195366600.001.0001.

28. Miranda Fricker, Epistemic Injustice: Power and the Ethics of Knowing (Oxford University Press, 2007), esp. chap. 1, “Testimonial Injustice,” 9–29.

29. Kristie Dotson, “Tracking Epistemic Violence, Tracking Practices of Silencing,” Hypatia 26, no. 2 (2011): 236–257, https://doi.org/10.1111/j.1527-2001.2011.01177.x.

← Chapter One · Book contents · Chapter Three →