A housekeeper who has served the same house through two owners is dismissed on a fabricated pretext. She returns that night in the rain, soaked, and asks to be let in for two minutes. She is begging: hands together, voice breaking, no leverage of any kind. Then she goes to the basement, and on the way back up she learns something — that the four staff who replaced her are one family, hired by fraud. The next time we see her face she is holding a phone, and the negotiation that follows is conducted from a position of total strength. The interesting thing is that nothing about her has changed. What changed is a fact about the other side.
// The theoryThe calculation is not the first step
The classical tradition from Beccaria onwards models the offender as a chooser: costs, benefits, probabilities. Its modern descendants added realism — Cornish and Clarke's bounded rationality, the crime-specific decision — but kept the structure, and that structure has an empirical problem. When researchers measure the variables, perceived risk of sanction turns out to be a weak predictor of offending while moral inhibition is consistently a strong one; and the two interact in a way the simple model does not anticipate. Per-Olof Wikström's Situational Action Theory formalized the finding: people do not deliberate about acts they regard as unthinkable. A moral filter, formed by the interaction of a person's moral rules with the moral norms of the setting they are in, determines the option set; only what passes the filter reaches the stage where costs and benefits are weighed. The practical consequence is sharp and well replicated: deterrence operates almost exclusively on people whose filter has already admitted the act. For everyone else, the threat is irrelevant because the option was never live.
Shared guilt is the cheapest way to lower a moral cost
Moon-gwang's reversal is so fast that it looks like a character reveal, and it is not. Read it through the filter and the sequence is mechanical. Before: blackmailing an employer is not in her option set, because she is a person with a job, a reputation and a self-image as the competent keeper of this house — the moral cost is prohibitive and the calculation never starts. After: she learns that the people opposite her have committed fraud, forged documents and conspired to get her fired. That single fact does three things at once. It supplies a justification (they did it first), a neutralization in Sykes and Matza's sense (denial of the victim: they are not victims, they are perpetrators), and — decisively — it removes the enforcement risk, because a party engaged in fraud cannot call the police. Moral cost and sanction cost collapse in the same instant, which is why the shift needs no interval. The theory's prediction is exactly this: the filter is situational, not just dispositional, and information about the setting can change it faster than anything can change a person.
// The subjectTwo frauds, and only one of them gets a plan
It is worth noticing what she actually asks for, because it is modest and it dates the moral filter precisely. She does not demand money, a share or a job back. She asks that her husband be allowed to continue existing in the basement — that the arrangement she has maintained for four years be permitted to continue. That is not the request of someone who has become a criminal; it is the request of someone whose entire moral world is still organized around keeping a secret she considers shameful, and who has discovered a lever that lets her keep it. The escalation into cruelty comes afterwards and comes from both sides, and the film is careful to distribute it.
That places her precisely in this hub. Ki-woo's filter admits a forged document and stops there. Ki-taek's never admits anything until it is bypassed entirely by affect. Geun-sae's admitted nothing for four years and produced a living burial. Moon-gwang's is the only one that moves because of information rather than emotion or opportunity, which makes her the cleanest case in the film of the mechanism Wikström describes — and the one whose downfall is caused by exactly the miscalculation the theory warns about, because a filter that drops fast drops for both parties.
Moon-gwang
Get the map of theories for free
Subscribe to the criminal archive —one case a week, no spam— and I'll send you The map of theories: the great currents of criminology on a single sheet.
// The crackA theory that is very hard to prove wrong
Situational Action Theory is elegant, well supported in parts, and carries three serious problems. The first is falsifiability. If offending occurs, we infer the filter admitted the act; if it does not, we infer the filter blocked it. Escaping the circle requires measuring the filter independently and in advance, which SAT's own studies — principally the Peterborough Adolescent and Young Adult Development Study, a genuinely impressive longitudinal design — attempt through scenario instruments and moral-rule scales. Those are self-reports about hypothetical situations, and the gap between what people say they would find unthinkable and what they do under pressure is the oldest problem in this kind of measurement. The second is scope: the theory is strongest for acquisitive and situational offending among adolescents, which is what its data covers, and much weaker for offenses involving intoxication, intimate relationships or organisational settings, where the notion of a discrete option set fits badly. The third is that the “moral cost” term in rational choice models absorbs too much: once shame, guilt, self-image, reputation and commitment are all folded into a single cost, the model can accommodate any result, which is the same criticism that eventually landed on deterrence. What survives, and it is genuinely useful for policy, is the asymmetry: measures that raise the perceived cost of sanction work only on the minority whose filter already admits the act, while measures that shape the filter — moral education, setting norms, changing what is unthinkable in a given place — act on the much larger group, and act earlier.
An untested filter
Moon-gwang blackmails people whose fraud harmed her, and then participates enthusiastically in a good deal of cruelty before the film is done with her. The theory explains the speed of her turn and does not license any of it. The part worth keeping is the uncomfortable inversion it produces: the reason she had never done anything like this in years of service was not that she is a better person than the family she is confronting, but that nothing in her situation had previously made it thinkable. Most of us are held where we are by a filter that has never been tested, and the finding this literature keeps producing is that the filter is a property of the situation at least as much as of the person.
Three key ideas
- The calculation is not the first step. A moral filter, formed by a person's rules interacting with the setting's norms, determines the option set; only what passes it reaches cost-benefit. Deterrence therefore works almost only on those whose filter already admits the act.
- Shared guilt collapses moral cost and sanction cost at once: it supplies a justification, a denial of the victim, and immunity from enforcement, because a party committing fraud cannot call the police. That is why her reversal needs no interval.
- SAT is hard to falsify, measures the filter by self-reported scenarios, fits acquisitive adolescent offending better than intoxicated or organisational offending, and the “moral cost” term absorbs too much. The surviving asymmetry: shaping the filter reaches far more people than raising the penalty.
Sources · to keep reading
- Wikström, P.-O. H. et al. (2012). Breaking Rules: The Social and Situational Dynamics of Young People's Urban Crime. Oxford University Press.
- Wikström, P.-O. H. (2006). “Individuals, Settings, and Acts of Crime”, en The Explanation of Crime. Cambridge University Press.
- Paternoster, R. y Simpson, S. (1996). “Sanction Threats and Appeals to Morality”, Law & Society Review, 30(3).
- Sykes, G. M. y Matza, D. (1957). “Techniques of Neutralization”, American Sociological Review, 22(6).
- Pauwels, L. J. R. et al. (2018). “A Meta-Analysis of Situational Action Theory”, Criminal Justice and Behavior, 45(9).
- Cornish, D. B. y Clarke, R. V. (eds.) (1986). The Reasoning Criminal. Springer-Verlag.
Case-file questionsFrequently asked questions
Who is Moon-gwang?
Moon-gwang (“El ama de llaves”) is a villain from Parasite, analysed on KRIMINIS through the lens of criminology. She begs on her knees and, sixty seconds later, records a video.
Why does Moon-gwang (The housekeeper) become a villain?
Moon-gwang and the moral cost-benefit: the housekeeper who discovers the other party is guilty too and stops having anything to lose. The file on the moral filter that decides which acts even reach the calculation.
Which criminological theory explains Moon-gwang?
Moon-gwang's case is analysed through Moral cost-benefit. Classical criminology assumes the offender weighs costs and benefits. What modern research added is that the calculation is almost never the first step: before it there is a moral filter that decides which actions even enter the option set. Per-Olof Wikström put it plainly — acts of crime are moral actions and must be explained as such — and rational choice studies confirm it the uncomfortable way: moral cost predicts offending far better than perceived sanction risk. Moon-gwang is the demonstration in real time. For years she does nothing because the act is not on her list. The instant she discovers her rivals are equally guilty, the filter drops and the calculation begins.
