Leaving the Land of Woo

Tags: theory

When we are trying to explain something, typically we look for causes.  We are trying to relate what we know, to what seems to require explanation.  If we see a smashed flowerpot on the floor, we look for where it might have come from.  If it is outside, we'd look up to see if it might have come from the window ledge.  We do that because we are relating what we are trying to explain, to what we already know, and the window ledge is a possible source.  That's an important part of the explanation.

If we claimed that the flowerpot simply came into existence, already broken, on the floor, our explanation wouldn't be accepted because it didn't fit what we know of the world.  Things don't simply suddenly come into existence and disappear.  That's not something that happens in our world, so we do not accept it as a cause.

This point is very important because when we are looking for explanations, we have to be very careful to distinguish between when we are appealing to causes that we can demonstrate, and to those which we cannot.  An appeal to a cause that we cannot demonstrate, is not accepted as a cause at all - it would simply be another part of the account that needed further explanation.

So to be accepted as reasonable, we have to be able to demonstrate the effect of causes.  That's not an unusual statement because it's how we deal with the real world every day.  It's what provides us with a measure of control because identified causes give us predictable consequences.

Descriptions
A description tells us about what is being described.  Suppose I am describing what happened to a patient who was suffering from an illness and who subsequently became well again.  I can start by describing what the patient was suffering from, their appearance, the symptoms, how they were behaving.  Next I could describe what medication or other actions they took, and I could include descriptions of any changes that were noticed.  Their temperature came down, perhaps, or they said they felt better.

Such an account would provide useful details of the changes over a particular time, as perceived by the observer, and may also include first-person descriptions of how the person said they felt.

Here's an example of such an account, describing a patient, Jane, and what happened to her.

Jane went to the doctor mid-morning and complained of sharp pains in her back which she said had come on early that morning. She was clearly in some pain and the doctor placed his hand on her forehead for ten seconds, then told her to go home, take some painkillers and to rest.  She went home, stayed sat down after taking the painkillers, and reported later that day that she felt much better.  The pain had reduced and she felt more relaxed.

Such a description is factual but says nothing about causes.  Now let's broaden the description into an account which suggests causes.

Patient Jane presented with the symptoms of back pain, which appeared to have been brought on by an event early that morning. It was most likely a muscle spasm caused by over-exertion or moving awkwardly.  The patient was checked for fever by touching her forehead and since none was found, the doctor prescribed a painkiller and advised rest.  If the muscle spasm was not serious, it could be expected to ease in the course of the day.  The patient returned home and rested after taking painkillers and reported an improvement later that day.

The difference between the first and second descriptions is that the second explains both the diagnosis and suggests what actually happened.  The doctor diagnosed a muscle spasm causing acute pain and relying of her medical knowledge, knew that a minor spasm could clear itself up in a matter of hours.  The use of painkillers would alleviate the symptoms and rest would help the patient.  The doctor had reasons for acting as she did, and the reported improvement was as expected.

Now let's produce a proposed alternative causal account based on the same initial description.

Patient Jane presented with the symptoms of back pain caused by a blockage in the channels carrying her energy flows. The doctor placed a hand on her forehead to diagnose her chakras and immediately felt that they were misaligned.  A ten-second transfer of healing energy enabled the doctor to start in motion the healing process by removing the blockage.  To make sure the patient felt she was being treated, the doctor also prescribed painkillers, which were by then unnecessary but put the patient's mind at ease.  Later that day, the healing energy had produced a positive result and the patient reported that she felt better.

Both of these are suggested causal accounts and we could continue producing many more.  We could for example suggest that the doctor in fact hypnotised the patient and that the experienced ten seconds were in fact many minutes. We can be as creative as we like in generating these stories based on the initial description.  By introducing causes into the explanation, we are trying to account for the consequences in terms of the actions taken.

That's a crucial step in medical treatment of any kind because we require a diagnosis first.  We identify what we think is wrong, in order to identify possible causes. We then reason to eliminate those possible causes which are not active, and isolate the actual one.  This differential diagnosis enables a doctor to decide what treatment is most beneficial. The diagnosis identifies what is wrong, and the clinical reasoning suggests why. The process critically involves isolating possible causes.  So how do they decide what is possible and what isn't?

Explanations
We've now reached the point where we have a sketchy account of what happened to the patient, and some suggestions of accounts including causes.  How can we tell what really happened?  Did the spasm respond to analgesics and rest, or was it healing touch, aligning energy flows?  We can ask ourselves which parts of each theory are demonstrable, and which are not.

Both accounts agree that the patient experienced pain which later subsided. Both agree that the patient took painkillers. Both agree there was a ten-second touch on the patients forehead.  So let's examine each account and see how much explanation there is.

In the first example, the doctor diagnoses that there has been a muscle spasm, based on the presentation of symptoms.  That knowledge of the association of the symptoms with known causes enabled the doctor to identify the most likely source of the pain, namely a muscle spasm. The evidenced changes in the muscle during spasm have been clinically identified and are evidenced to cause acute pain. Everything the doctor did followed from there.  The doctor's reasoning is evidence-based so they are sure to be dealing with real causes.

In the second example though, the symptoms were not associated with the most likely known causes, but with a theory that the cause is a blockage in energy channels.  The problem with this approach is that the symptoms are being associated with causes which have not been evidenced. Use of the theory assumes that there is this form of energy, it assumes that there are channels, and assumes that they are blocked.  In the absence of evidence, literally any theory could have been used in this way.

The first example follows the path from observation of symptoms to a differential diagnosis, and from there to identifying the most likely clinically-evidenced cause. Only then is it possible to select the most effective treatment.  In the second case, the symptoms are observed and then directly related to the assumed explanation, invoking unevidenced energy sources, channels, and blockages. Without any form of differential diagnosis related to evidenced causes, a treatment is assumed to be appropriate.  Finally, the touch on the forehead is given special significance independent of any causal relationship to the symptoms.

Alternative Medicine Theories
These uniformly start from an assumed explanation and then seek an application of this theory to the symptoms observed.  They do not draw their theory from the analysis of data but try to find evidence from the data in support of their theory.  This is the wrong way round. In deliberately selecting anything that might lend support to their theory, they rule out that which doesn't correspond. Their selection of what they use as evidence is deliberately biased.  They are looking for supportive evidence, not developing a theory which explains the data.

Often they claim that it is not possible to investigate these therapies scientifically, thereby dismissing what is our most powerful method for investigating claims about the real world.  But they are incorrect.  Alternative therapies are perfectly susceptible to study and analysis, using controlled, double-blind, randomized clinical trials. Invariably, these trials consistently show up the alternative medicine claims to be no better than placebo.

A genuine theory draws on a mass of data and explains it consistently.  It produces a comprehensive explanation of the mechanism of treatment and where possible cure, and provides testable predictions of the form, when A is done, B happens.  By testing these predictions, the theory itself is tested. If the data contradicts the theory, the theory changes. In contrast, the theories behind alternative medicines consistently fail to produce successful testable predictions.  Where predictions of efficacy are made, they show themselves to be no better than placebo.  The alternative medicine practitioners, when faced with evidence that disproves their claims, simply move on to look for evidence somewhere else; instead of modifying their theory and retesting it, they deliberately ignore the results and maintain the belief that they are right. That's bias, plain and simple.

But the status of these alternative theories as theory is also false.  They are not theories, but stories which attempt to explain observable symptoms in terms of an unevidenced assumption of mystical causes.  These are not theories but stories.

A theory starts with the data, tries to explain them, produces a causal explanation which makes testable predictions, and crucially permits itself to be disproved. Alternative medicine theories do not do this, and that's why they cannot be accepted as theories at all.