Why I See Psychiatry as Naive and Simplistic

For me, the biggest general reason I am sceptical of what some of us term 'the medical model' of psychiatry is that, to me, it seems hopelessly, naively simplistic compared to the sheer complexity of the part of the body at the centre of psychiatric problems: the brain. The idea that a medical professional can talk to you for half an hour, once a week, and some nurses can make a few observations and write a few notes, and based on that alone, the medical professional has sufficient information to make an affirmative diagnosis from which an effective remedy can be prescribed? To me it seems absurd, and it is equally absurd that so many believe in it. The 'evidence base' of their 'evidence based medicine' is then very narrow and again hopelessly naively simplistic. The idea that two instances of what gets diagnosed as 'mania' are two instances of the same thing is an underlying assumption necessary for something like a randomised controlled trial for treatments for mania to even make sense. Yet what is the justification for such an assumption?

Consider if, to use one of my thought experiments, people categorised illnesses into three categories: legs, torso, and head, based on where the illness was primarily located. Angina? torso pain. Stomach ache? Torso pain. Liver cancer? Torso problem. Since they're all 'torso problems', we'll just diagnose them as 'torso problems' and seek some 'evidence based' treatment for 'torso problems'. Now if all we care about is pain, then most of the time sufficiently strong painkillers will work, and perhaps there will be a few cases of 'treatment resistant torso pain', and of course dying is an 'adverse' even that is sometimes observed in people with 'torso pain' even if treated by the known 'reliable treatment' of giving strong painkillers. If this sounds absurd to you, well I view the medical approach to psychiatry as equally absurd. I'm surprised and astonished that it works as well as it does, given its naively simplistic approach.

Or consider if we are in the cockpit of a plane, and we categorise problems based on the colour of the warning light. So there are 'green light' problems, 'amber light' problems and 'red light' problems. Then someone comes up with an 'evidence based' method of remedying 'red light' problems: simply deactivate the red lights. So now there are none of those annoying 'red light' warnings, so all is well, and our 'evidence based' method works. Every time the red warning lights are deactivated, we no longer see red warning lights. If this sounds absurd, again this is how absurd I see belief in the 'medical model' and the idea that 'evidence based medicine' as a paradigm is applicable to psychiatric problems.

Then there is the issue of the minuscule diagnostic manual: apparently a roughly 1000 page book listing roughly 300 distinct disorders is sufficient for the job. This strikes me as absurdly small, if knowing the diagnostic label suffices to pin down a problem sufficient that one can determine the solution. There are something like 10,000 different physical ailments. Given the complexity of the brain, and how its behaviour depends sensitively upon the configuration of neurons and what signalling is flowing around it, one should surely expect there to be more distinct problems. So a category like 'mania' is akin to the category of 'program crash' in software engineering. Imagine if we tried to lump all 'program crash' problems together in the same category and search for a solution to 'program crash' which doesn't require any more information on what is going on. Absurd? I find the minuscule scale of the DSM to be equally absurd.

The Alternative

So what is the alternative? The medical model where a trained professional takes a look, perhaps exchanges a few words, and then makes a definitive diagnosis and prescribes treatment misses something major. When a software engineer is trying to fix a problem, he (or she) will use things like a debugger which allows him (or her) to inspect the state of the running program. He (I'll stop adding 'or she', but this should be taken as implicit) will add code to the program which gives him information to better understand what is going on; he will add code to dump the state of the program to storage in the event of a crash. Debugging a program is a complex affair, and while there are common categories of errors such as 'buffer overflows' or 'off by one errors' or 'use after free', finding such errors is a complex affair. If a program crashes, just printing a modern-looking error message like 'Ooops, something went wrong', then finding out whether it was a use-after-free, or indeed whether or not the error was caused by the device the user is interacting with, or the server on the other end of a network connection, or some issue with the network in between, is a difficult if not impossible problem. Trying to do a 'clinical trial' of solutions for 'Ooops, something went wrong', is like what we see on many user forums with one person saying 'I hit it with a hammer and that sorted it', and 'just switch it off and on again' and so on. Unless we have enough information to pin down the problem, we don't have enough. And there are almost certainly more qualitatively distinct problems than there are lines in chess, by many orders of magnitude. And thus what we need is a more interactive approach, where the person with the problem is actively involved in figuring out what is going on, and actively involved in finding solutions. Communication is key, as only the patient sees their problem from the inside, and many things can only be seen from the perspective of the patient.

How we approach mental health needs a complete rethink. And the idea that medicine is the right discipline needs to be seriously questioned. And then the idea that randomised controlled trials are a sufficiently precise tool to investigate solutions for mental issues again needs to be seriously questioned: in particular, the statistical methods used, and how conclusions are drawn from them, needs scrutiny. One example of this is to ask, in the context of psychiatric disorders, what constitutes a 'representative sample'. A problem I see is that any sample big enough to be 'representative' needs way more people than there are on the planet (imagine trying to get a 'representative sample' of possible qualitatively distinct configurations of neurons: the number of such qualitatively distinct configurations is surely astronomical, and hence any representative sample must also be astronomical in size).