On my recent post on autism, several people chimed in to say that “autism” wasn’t a unitary/homogenous category. It probably lumps together many different conditions with many different causes. It’s useless to speculate on the characteristics of “autism” until it can be separated out further.
I get this every time I talk about a psychiatric condition. The proponents of this view seem to think they’re speaking a shocking heresy that overturns the psychiatric establishment. But guys, we know this kind of stuff. Psychiatric diagnoses don’t have to perfectly match underlying root causes to be useful.
Suppose a patient comes to you with difficulty breathing, excessive sweating, anxiety, and extreme discomfort when lying down flat. You recognize these as potential signs of pulmonary edema, ie fluid in the lungs. You do an x-ray, confirm the diagnosis, and prescribe symptomatic treatment – in this case, supplemental oxygen. All of this is good work.
But you can have fluid in your lungs for lots of different reasons. Most of the time it’s heart failure, but sometimes it’s kidney failure, pneumonia, drug overdose, smoke inhalation, or altitude sickness. Some of these causes will have slightly different symptoms, which an alert doctor can notice.
Suppose the real cause of your pulmonary edema is heroin overdose. In that case, it wouldn’t be fair to call pulmonary edema a “root cause”. The root cause of your problem is the heroin. But you also can’t call pulmonary edema merely a “symptom”. No patient comes in saying “Doc, I’m feeling a bit pulmonary edemic today”. The symptoms of pulmonary edema are difficulty breathing, excessive sweating, anxiety, etc. So what is pulmonary edema?
I don’t know the technical philosophy-of-medicine term for this, but let’s call it a “condition”. A condition which nobody has yet matched with a biological process gets dubbed a syndrome – a set of symptoms that go together even if we remain agnostic about why. A condition which has been matched a biological process ends up like pulmonary edema – such a well-known part of the medical canon that nobody feels a need to do philosophy around it.
Lots of things are conditions like this. Even some universally-known diseases like stroke are better thought of as conditions than root causes. Strokes can be caused either by ischaemia (usually a blood vessel blocked by a clot) or haemorrhage (a blood vessel bursting and bleeding out). These two causes have differing risk factors (anticoagulants cause haemorrhagic stroke but protect against ischaemic) and differing treatments (tPA relieves ischaemic stroke but catastrophically worsens haemorrhagic).
But nobody ever bursts into neurology conferences shouting “STROKE ISN’T A REAL DISEASE, IT’S A COBBLED-TOGETHER BASKET OF MULTIPLE DIFFERENT ROOT CAUSES!” Everyone realizes that conditions are a useful intermediate level to work at.
This is how I feel about things like depression too. No psychiatrist would be even a tiny bit surprised to hear that depression is many different conditions with many different causes. For example, everyone knows some depressions are caused by hypothyroidism, and others aren’t.
The biggest difference between the philosophical status of depression vs. stroke is that we know what biological process stroke corresponds to. Stroke is brain cells dying from lack of oxygen. It can be caused by arterial blockage or by bleeding, sometimes it can even have more distal causes like cocaine use or Moyamoya disease, but it all ends with brain cells dying from lack of oxygen. That in turn produces classic symptoms like sudden-onset slurred speech, hemiparalysis, and facial asymmetry.
We don’t have as good an idea what biological process depression corresponds to. There are some theories – maybe a failure of synaptogenesis – but they’re all pretty speculative right now. Still, I think it’s reasonable to propose that they correspond to some process.
First, because depression includes a lot of surprising symptoms mysteriously clustered together. Just as without the concept of “stroke” you can’t explain why slurred speech and hemiparalysis happen together so often, so without the concept of “depression” it’s hard to explain why SIGECAPS tend to go together. The only good alternative I’ve heard here is the idea of symptom networks. But I no longer find this very convincing, and it never seems to be what the people talking about how “depression isn’t a single disorder” mean.
Second, because at this point we don’t even know what biological process normal low mood corresponds to, but it seems like it has to be something, and it would be strange for a single biological process to cause low mood and not be related to depression.
My (very wild) guess is that in the end psychiatric disorders will mostly turn out to be computational conditions. That is, something like “the learning rate of this system is set too high” or “the threshold for errors in this error-detector is too low”. There will be lots of different things that will cause that, from biological (because these computations are implemented on biological systems including the usual range of things like serotonin and dopamine and synapses) to psychological (because the brain is plastic enough that its computational parameters can change with experience) to environmental (because if you pour a bucket of battery acid onto a computer, probably its computational parameters will change in some way). This is just my personal bias towards computational explanations speaking, and it could be that these disorders will be better explained by regional stories (ie “the amygdala is broken” or “the hippocampus is broken”), by biochemical stories (“there’s too much serotonin”), by structural stories (“there are too few synapses”), by some combination of these, by something totally different, or by something that’s on a totally different level than any of this.
If something like this story is true, it means that research that treats depression as a single condition might or might not work. Returning to the analogy of stroke, I think (though I’m not an expert) that the prognoses for ischaemic and haemorrhagic stroke are mostly similar, since both depend on how long it takes the brain to adapt after some cells have died. But the risk factors for these two kinds of stroke are different (again, anticoagulants protect against one and cause the other). Scientists who were researching “stroke”, without understanding the different causes, would get some things right and end up confused about others.
Some people, upon hearing this, say that we should be trying to figure out the different kinds of depression so we can do real research on those. People have been trying this for a century, and every one of their leads have been false. Traditional psychiatry flirts with admitting two subtypes of depression, but you can also find papers claiming to have found three subtypes, four subtypes, five subtypes, etc. Even papers that agree on how many subtypes there are often identify the subtypes totally differently. This has not been a productive research program and I think better understanding of what depression is will be more valuable than bashing our heads against the subtype identification problem further. At least this is how it has always worked in regular medicine, where once we realized what eg pulmonary edema was, everything fell into place (including potential root causes) and nobody felt like figuring out exactly how many subtypes there were was a very interesting problem.
The saying goes: all models are wrong, some models are useful. I don’t think existing psychiatric diagnosis is particularly accurate, but I think it’s the most useful thing we have right now. And I don’t think talking about how each condition is probably made up of many root causes is a particularly damaging objection to it. We should keep the likely heterogeneity in mind and pull it out when we need it, but we shouldn’t use that as an excuse to abandon the whole nosology.