In the 1970s, David Rosenhan showed that healthy people could be diagnosed as mentally ill

Originally published at:


To be fair to the psychiatrists, having someone behave perfectly normally but report hearing voices is a little bit like sending a perfectly healthy person to the hospital after inserting a mass of strange tissue into their abdomen. The hospital is likely to focus the abnormal thing and try to figure out what the underlying cause is.

Of course there is no reason to be fair to psychiatrists in the 1970s.


So, is that photo of Saint Elizabeth’s in DC?


Pretty much, “I convinced folks to admit me to a psych ward with very serious symptoms and then told them that I lied about the symptoms much the same way someone with a very serious psychiatric problem would, why won’t they believe that I’m not a danger to myself or others after concocting a crazy story!!!”.

As someone trained in this field, fuck yeah I’d keep them and do an involuntary hold if they attempted to leave. The problem with psychology is that we don’t have blood tests for this, we don’t have imaging tests, we don’t have the same sorts of tests that someone may have that had broken their arm. It just doesn’t exist.

So we do the best we can, knowing it is an imperfect system designed to help the most amount of people.

Now I got into psychology through AI and testing…kinda sad, I’ve been a part of a few projects that CAN identify illnesses but since AI depends on a blackbox and we can’t really prove it, it is hard to sell it. We did a tap test that could identify folks with ADHD and those just faking it to get the drugs…or their parents forcing them to do it because they were shitty shitty parents that wanted either an excuse for their shitty parenting or a drug that would calm them down (sadly, ADHD drugs for those without ADHD work just like the amphetamines that they are). Couldn’t sell it even though it outperformed any other test. Most would simply use a self-reported checklist and be done with it. And this was one of our more benign assessment tools.

The point is, we don’t have the proper tools to do the assessments, so we err on the side of treating the patient to the best of our ability. We are sometimes wrong, but our aim is never to harm someone. And if someone comes in hearing voices? And they are trained to do it in a way that we could tell they weren’t faking-- i.e., someone ‘seeing’ invisible people…kinda a give away that they are faking…audio hallucinations are common, visual? Almost never unless someone is on a specific type of drug that will get out of their system. If they see someone, we still try to wait out the drugs (72 hours is sufficient). Most people faking will actually say they see someone because they don’t know this…which if they just say voices? They’ve been coached to be an at-risk patient.

If these were actually folks coming in with breaks in their reality and they were released or turned away in the '70s…we’d be reading from the same folks posting these stories complaining my field just ain’t care and it’s only for the elite. These types of posts always get under my skin.

That said, I haven’t worked in the field in years…I use my skills these days for other researchy type things.


Yes. Though it’s also worth noting that governments actually used to fund mental health care. Odious as the provision of it was.


Psychiatry is not purely medicine. If it were, symptoms (subjective criteria such as aural hallucinations) would be considered merely an indication for a more thorough assessment of carefully defined objective criteria (i.e. signs), the diagnosis for which would be defined physiologically (e.g. ‘pre-synaptic hypodopaminergia of the anterior cingulate cortex likely secondary to enhanced expression of DAT due to the rs27072 SNP of the SLC6A3 gene’).

Instead, psychiatric diagnoses are, in most cases, just meticulously well-researched clusters of symptoms.

This phenomenological aspect of psychiatry, both in practice and theory may be considered a flaw but only to the extent that it obscures an upstream physiological cause. It remains essential, however, because it involves the one organ that relies on consciousness for self-modification. Your thoughts and feelings will always, always matter, no matter how biologically hardwired your particular brain cooties may seem.

Which is exactly why psychiatric clinicians take it pretty f—king seriously when you tell them you’re hearing voices.


What? You don’t hear voices? Shit.


Actually I do! [not even a joke, though I am perfectly jovial about it]

I have experiences of having voices in my head that I recognize are my own thoughts but that come to me a bit like auditory experiences. Also while on one particular brain-drug I had full blown aliens-beaming-thoughts-into-my-head experiences (which I fortunately intellectually realized were drug-induced).

I recently took a family member to the ER who was reporting hearing voices. I thought they made latched way too much onto the “hearing voices” part and didn’t pay enough attention to the other parts, but when the psychiatrist arrived they correctly teased out that it was hearing-voice-as-a-interpretation-of-negative-self-talk from hearing-voices-because-of-a-psychotic-episode and advised us accordingly.

Most of us have “heard things” or “seen things” at some point in our lives. I think in the case of the people doing this experiment, they knew just what to say to convince the psychiatrists they spoke to that they were hearing voices in the worst possible sense of the term. Having psychotic episodes isn’t like having the sniffles. If you have an episode one day and you are fine the next, “I guess you got over it” is not a good explanation. So the doctors tried to come up with diagnoses that fit a person who had the symptoms one day and not the next, and never thought of the correct diagnosis: person who lied to us to prove a point.


Your average diagnosis by your primary care physician isn’t that different than those made by psychiatrists. Internal medicine isn’t that definitive for a lot of things, it’s an art your doctor uses to make an evidence-based decision, but not necessarily a definitive one. Except in cases where there’s an actual test for a condition, most of the time they’re just making educated guesses on the limited information that can be gathered.

There’s still an important difference, though. Psychiatry using the DSM actually defines disorders as clusters of symptoms. If a doctor looks at my throat and my ears and listens to my chest and says, “You probably have a virus, get some rest” they are taking an educated guess, but they are taking an educated guess about an underlying physiological cause - we actually think the virus exists.

If a psychiatrist tells me I have dysthymia or bipolar or BPD they aren’t guessing at an underlying cause of my symptoms at all. They are just categorizing me by a system that they believe is helpful in determining treatment.


I guess if it were me personally faced with a diagnosis from a machine learning algorithm, I would be hesitant to want to fully trust it too. Not being able to reason about the why a decision was made, means I can’t independently reason through the test’s process, and verify it’s conclusion.

You guys should definitely give the podcast a listen, and give Futility Closet some love and a few dollars, too!

My impression is that this study is not an indictment of the psychiatric establishment. It’s simply a reminder to continuously challenge your own assumptions. Each patient was given an initial diagnosis, and everything that followed from that point was based on the assumption that the initial diagnosis was correct. Here are some of the salient points from the podcast, which I am blurring in the hopes that you’re all going to listen in.

[spoiler]In this case, all the “patients” reported hearing the same three words: empty, thud, and one other one (hollow?). That was enough to get them admitted. Subsequently they acted perfectly normally and reported no further symptoms. They were all prescribed tranquilizers for the duration of their stay. All took notes on their experiences during their stay, leading the clinical staff to document graphomania-like behavior. In one case, they met up outside the cafeteria before it opened for lunch. Clinical staff documented that as “oral acquisitive behavior.” And one of the researchers was a painter and did some painting while committed, which was documented as “painting behavior.” One patient was released with a diagnosis of schizophrenia, one with a diagnosis of manic depressive or bipolar behavior, and the rest with diagnoses of schizophrenia in remission.

After reading about Rosenhan’s experience, a different hospital challenged him to send his researcher-patients to them, confident that they would quickly identify them. He told the hospital he would send them a patient. That hospital identified more than 40 different actual patients as his researchers, but of course he didn’t send any at all.[/spoiler]


Not necessarily. Giving an diagnosis (like schizophrenia) initially and then viewing everything through that lens would be a mistake. But that’s not what happened here, what happened is that they believed the facts they got on the first day. On day 1 this person reported hearing voices. That doesn’t change if the behave normally on day 2. The doctors tried to come up with some explanation that fit both the day 1 facts and the day 2 facts and the day 3 facts.

And when I read:

That strikes me as something that would easily be taken to be a weird thing to say outside of psychiatry, but that sounds just like how psychiatrists talk (I’ve read many psychiatrists’ assessments of me). The person was literally, factually exhibiting “painting behaviour.” That might or might not be useful to understanding what is going on inside them. It is almost certainly useful in trying to relate to them to get better insight into what is going on inside them.

There’s tons to criticize about psychiatry and its history. The doctors who were diagnosing these people probably thought that being trans was a mental illness on par with schizophrenia. Some of them may have performed lobotomies in the 50s. But “lying to people results in them being wrong” isn’t actually a super interesting result to me.


So true. The first thing that came to my mind were healthy people diagnosed as mentally ill after being committed in order to rob them of their autonomy and/or assets.


Yeah, usually patients don’t get the backgrounds on the instruments before we give them to them. And if a patient KNOWS about an instrument before I were to give it to them, it is a sign that I might need to switch instruments – because while most more involved tests look for fakery, why would one even bother with it? It is why a lot of instruments use to be considered protected and others weren’t allowed to even see them unless you had specific credentials. Things like the Rorschach has undergone factor analysis that tells us a LOT about what certain individuals might say when confronted with a blot, and done properly is a tool that can identify certain things – however it still requires a human on the other side to test this with more specific testing. This is a waste of time if I know there might be some trickery.

That said, we don’t just throw random samples at machines via whole cloth and let them figure things out. We have an idea of why something works the way it does FIRST and then let the machine see if they can identify the same patterns – or even discriminate from patterns we don’t see – or look at things at a lower granularity than humans can efficiently.

So there is a reason they are identifying this stuff…and with any instrument, there is human interpretation and corroboration included.

Anyhoo…still one of my pet peeves, journalists that don’t understand a subject getting upset when they make shit up and then we believe them when they do and then they get upset that we believed it…

1 Like

I’ve listened to a few here and there over the years, but just a week ago filled my Casts app with shows and binged. Without fail it has been a useful distraction from the interminable frustrations of air travel.


I talk to myself a lot, especially when under stress, and preferably in English which is not my native language. Do I qualify? FORGET I SAID THAT. It’s hot today. Why am I still in the office. Verdammt.


Same. We just flew on Frontier like a bunch of chumps over the weekend. These podcasts helped get me across the finish line.

1 Like

I’m with you on all of the above, but I’m a little surprised that a single report of auditory hallucinations is enough to diagnose someone as schizophrenic. But my background on this topic is limited to ninth grade psychology. Based on that, I’m pretty sure that 50% of psychiatry is electrocuting pigeons.