Intuitively, people often answer “not that long”. We assume that if we behave normally, reality will inevitably assert itself.

Rosenhan starts from a simple, almost naive question — and explosive for that very reason: how, concretely, do we distinguish a sane person from an ill one? And at what point does what we believe we are observing say more about the system of observation… than about the person observed?

The experiment

Rosenhan recruits eight people with no psychiatric disorder. They present themselves at various psychiatric hospitals and report one symptom: they say they hear voices, or noises, in their head. Nothing else. The result: they are admitted. From then on the instruction is simple: behave perfectly normally, and try to get out as fast as possible.

If the system worked the way we imagine, the reasoning would be: the symptom has gone, observation finds nothing pathological, the hypothesis can be dropped. In an ideal world, the label would fall away as quickly as it was applied.

Except that is not what happens. Very quickly, diagnoses are assigned — the most frequent one in the account of this experiment: schizophrenia. And there everything tips over, because once the label is applied, it becomes a pair of glasses. And those glasses colour everything.

From the diagnosis onwards, every behaviour of the pseudopatients is interpreted through the prism of pathology. An ordinary behaviour turns into a clinical sign. A neutral intention becomes a clue. A contradiction becomes further proof. A few examples have remained famous — and chilling, because they are plausible.

• Arriving early at the canteen? That can be read as unusual anxiety.
• Saying “I am fine, I do not hear anything any more”? That can be recorded as denial of illness.
• Taking notes to remember things? That can become a “compulsion”.

This mechanism has a name in social and cognitive psychology: once a hypothesis is in place, the mind — and institutions — spontaneously start looking for what confirms it, and reclassifying what contradicts it. The diagnosis is no longer merely a working hypothesis. It becomes an all-encompassing frame of reading.

That is the disturbing point: error is not merely possible… it becomes coherent. The system holds, it self-validates, because it turns counter-examples into new arguments. And the more it does so, the sturdier it looks. This is not a simplistic criticism of individuals. It is a criticism of contexts, and of the power of frames.

The second round

After the results are published, some heads of psychiatric institutions react with anger and defiance: “Fine. Send us pseudopatients now, and you will see that we spot them.” Rosenhan accepts the principle… then announces that he has sent people in.

The teams go on alert. They scrutinise, they assess, they sort. Over the months, some forty people are identified as suspicious, sometimes turned away, sometimes labelled as frauds.

When you expect fakers, you mechanically increase false positives. You see what you are looking for. This second part is almost more vertiginous than the first, because it shows that the problem is not only overestimating illness, but manufacturing certainties in both directions, depending on the mood, the pressure and the dominant narrative of the moment.

What labels do to ordinary bonds

This is where the experiment becomes genuinely interesting for our daily lives, our couples, our families, our institutions, and even our consultations. As soon as we stick on a label, we change the way we perceive the person. And, more discreetly still, we change the way the person perceives themselves.

In a family, you see it very early: “He is the restless one.” “She is the sensitive one.” “He is the difficult one.” “She is the mature one.” “He is the clown.” Some labels look positive, but they confine too: the strong one, the nice one, the responsible one. Others are outright disqualifying, and their violence is immediately visible.

In every case the mechanism is the same: a person is frozen inside a description, and that description becomes a script. Then everything that happens is reread through that script. In a couple it is the same: “You are selfish.” “You are hysterical.” “You are cold.” “You are useless.” From then on, every gesture becomes “proof” and every attempt at nuance a “manoeuvre”. We no longer discuss a behaviour within a context: we argue about an identity.

Separating the person from what happens to them

The point is not to say that disorders do not exist, nor that diagnoses are useless. They can guide, structure, protect, open up entitlements, organise care. The problem begins when the diagnosis stops being a tool and becomes a definition of the person.

This is an idea narrative therapists have formalised with remarkable power: externalising the problem. In plain terms, we separate the person from what is happening to them. We stop confusing “who you are” with “what you are going through”. Instead of “I am anxious”, we explore how anxiety invites itself in, when it gains ground, what it is trying to protect, when it retreats, and what the person is already doing to resist it.

The Rosenhan experiment — whether we take it as a historical electric shock, a critique of institutions, or a lesson in social psychology — leaves a lasting trace: it is not only the person who “has a problem”. It is the relation between a person, a context, and a frame of reading that manufactures what we believe we see.

References

• Rosenhan, D. L. (1973). On Being Sane in Insane Places. Science.
• Goffman, E. (1961). Asylums.
• Becker, H. (1963). Outsiders.
• White, M., & Epston, D. (1990). Narrative Means to Therapeutic Ends.
• Cahalan, S. (2019). The Great Pretender.

The full account of the experiment, with both of its instalments, is told on screen just below — in French.