Previously in this series:
N=1: Introduction
N=1: Single-Subject Research

I.

Our psychology is focused on behavior. We focus on behavior because we want to figure out what actions we can take to influence the world around us. But a focus on our actions can also make us superstitious. 

The classic example is from a BF Skinner study, where he put a bunch of pigeons in a box and dropped in food at random intervals. Instead of realizing that the food drops were random, the pigeons assumed that they were somehow responsible and tried to figure out what they had done to make the food appear. 

Whatever they were doing at the time the food dropped, they tried again. A pigeon who had just turned counterclockwise when the food arrived would turn counterclockwise again and again. When more food eventually did arrive, the counterclockwise-turning was validated. “The experiment might be said to demonstrate a sort of superstition,” wrote Skinner. “The bird behaves as if there were a causal relation between its behavior and the presentation of food, although such a relation is lacking.” [1]

Compare this to a rat confronted with a set of buttons, trying to figure out which of the buttons give food and which give painful electric shocks. Unlike the pigeons, the rat is faced with a deterministic system where her actions lead directly to reward and punishment, so her focus on behavior is justified and leads to a correct understanding of the system. The pigeon is faced with a random system where his actions have nothing to do with the arrival of food, so his focus on behavior is pointless and leads only to superstition and confusion.

II.

We worry this is a common problem in chronic illness. Let’s say that Mary develops chronic fatigue syndrome (CFS). She is proactive and wants to solve the problem, so she comes up with a plan of 26 different treatments, which we’ll call A, B, C, D, and so on. Maybe A is “cut out dairy”, B is “walk 20 minutes every day”, etc. but the specific plans don’t really matter. She starts implementing each plan for two weeks, first plan A, then plan B, etc. 

But Mary is working from the wrong assumption. She thinks her chronic fatigue comes from something she’s doing or not doing. In short, she thinks it comes from her behavior. This is a common assumption because our psychology is focused on behavior — we look for things we are doing right or doing wrong. But what really happened is that last month she bought a bag of rice that was grown in a field that was contaminated with cadmium, and developed low-level cadmium poisoning, which is entirely responsible for her chronic fatigue. Cutting out dairy or walking to the corner store won’t do a thing, because the cadmium is the only cause of her illness. None of the interventions she has planned will help. 

But the cadmium is slowly being cleared from her system by natural means at the same time as she works her way through the 26 treatments. What happens is this: Mary reaches treatment L (“take omega-3 supplements”) just as the cadmium in her system drops below critical levels, and Mary is immediately “cured”. 

Since her symptoms stop almost immediately after starting treatment L, Mary assumes that the omega-3 supplements are what cured her, and continues taking them indefinitely. In reality, the omega-3 supplements do nothing for her — as long as her cadmium levels are low, she doesn’t have CFS, and if she ever gets exposed to high enough levels of cadmium again, her chronic fatigue will come right back.

III. 

What Mary should do is she should run a self-experiment with the omega-3 supplements. She should randomly assign some weeks to be on omega-3 supplementation, and some weeks to be off. If she did this, she would quickly find that the omega-3 makes no difference to her chronic fatigue.

It’s understandable why she doesn’t try this — she is worried that if she stops taking the omega-3, her chronic fatigue will come back, and she doesn’t want to risk it. Also, we suspect she wants a world that makes mechanical sense (“I just needed to take more omega-3”) rather than a world where she randomly gets sick and there’s nothing she can do to stop it. It’s hard to blame her for that.

This is how the focus falls on behavior and misses hidden variables. By “behavior”, we mean actions that are directly under people’s control. Eating more or less of something, getting up earlier or later, trying more or different kinds of exercise, and so on. By “hidden variables” we mean essentially any variable you wouldn’t normally think of, especially one not connected to your actions. For example, heavy metals in your drinking water, additives in your food, viruses you contracted from your friends, air pollution from forest fires hundreds of miles away, mold in your ceiling, or things you’re exposed to at work. 

Most of these hidden variables can be influenced by your actions, but they’re not the kinds of behaviors that come to mind. You can always quit your job, but for most people, that doesn’t come to mind as a possible treatment for their illness. You can cut out spinach or dairy, because “eat less dairy” is psychologically simple — but “consume less sulfites” isn’t a clear action for most people because “foods with sulfites” isn’t a category to most people. They may not always know which foods contain sulfites, and they may not know what sulfites are. 

IV.

This is what chronic illness looks like for Mary as an individual. At the group level, things look somewhat different. 

If a chronic disease is caused by a hidden variable (like cadmium randomly being in some foods but not others), you should see something like this: People get sick for apparently no reason. They all try many different treatments, and most treatments don’t seem to work for anyone. Sometimes a treatment will seem to work for a bit, but then it will unexpectedly stop working. Whenever you feel like you start to get a firm grip on things, all the rules you learn go out the window. However, there are many individual stories of trying some new treatment and suddenly being cured. Unfortunately, the cures in all of these stories are entirely different treatments, and the cures that work for one person never seem to work for anyone else.

And this does sound like what we see in many chronic illnesses, which makes us suspect that some chronic illnesses are being caused by a hidden variable. It could be contamination in food, water, or air, like our hypothetical Mary’s experience with cadmium. But it could also be any other unexpected variable that doesn’t have to do with personal behavior. For example, it could be the result of a virus, or an allergy to something in your household, or a curse put on you by the local witch. When taken as a group, chronic illness communities look exactly how we would expect them to look if the illnesses were caused by some hidden variable, and that makes us suspect that they are caused by some hidden variable.

Naturally this makes us wonder if there is any way to figure out what these hidden variables might be, assuming you believe they exist. The fact that they are hidden does make it inherently tricky, but we have a couple of ideas, here they are.

TRY BIG ELIMINATIONS

Your chronic illness may be triggered by something in your environment (your home, work, local food, local water, etc.). To test this, you can change as much of your environment as possible all at the same time, for example by taking an extended trip to Nepal. 

If you start feeling better or your symptoms disappear, this strongly suggests that something in your home environment is causing your illness. If you don’t feel any better, it suggests that your symptoms 1) aren’t caused by your environment, 2) are caused by elements of your environment that you brought with you (e.g. your clothes, your shampoo), or 3) are caused by elements of your environment that are common to both your home and Nepal (car exhaust?).

Your chronic illness might also be triggered by something you eat. To test this, you can change as much of your diet as possible all at the same time, for example by trying the potato diet, where you eat essentially nothing but potatoes. The potato diet is good because potatoes are simple, contain no additives, and are more or less nutritionally complete. Many people can survive happily on nothing but potatoes, salt, water, and hot sauce for up to four weeks (we have good data on this!). 

If your symptoms disappear or get better, this strongly suggests that either you had some deficiency that the potato diet fixed, or something in your normal diet is causing your illness. If your illness is just as bad as ever, it suggests that either your symptoms aren’t caused by your diet, or are caused by elements of your diet that are also in the potato diet.

Neither of these approaches will tell you what is causing your illness, but both have the potential to narrow things down enormously. If you go to Finland for a month and your migraines stop three days in and don’t come back until you get home, that’s pretty clear evidence that something at home is causing your migraines. You don’t know if it’s your laundry detergent, your well water, or something at your job, but you can take steps to narrow it down further, and you can stop worrying about your diet so much. 

Similarly, if you try the potato diet for a month and your executive function issues disappear, you can stop worrying about fumes from your boiler and can try to figure out what part of your diet is giving you brainfog.

Even a null result is informative. If you go on the potato diet for a month and your migraines carry on as normal, that’s a pretty clear sign that it’s not something in your diet, and you should look elsewhere. 

Some people find this approach surprising, because scientific investigation usually involves isolating a small number of variables and putting them under tight control. This works fine when you have a small number of variables to start with, or you know which variables you’re interested in. But in the search for hidden variables, there are a nearly infinite number of things that could be the cause of your illness. We need a technique that lets us rule out lots of theories at once, so doing these big splits can be extremely productive.

There’s a classic genre of logic puzzles often called balance puzzles. In these puzzles you have several coins, one of which is lighter than all the others, and you have to use a balance scale to find the light coin in the smallest number of weighings. The way you solve these problems is by splitting the coins into groups and comparing the groups directly. If you split the coins into two groups and the group on the right weighs less than the group on the left, the light coin must be in that group.

Consider a version of this puzzle where there is an illuminated lightbulb and a row of 1,000 switches. You want to find the switch that controls the lightbulb, but you don’t know which it is. You could go down the row of switches and try them one by one, but this would probably take you several hundred steps. On the other hand, if you have some kind of opportunity to flip a bunch of switches at once, that can narrow things down really quickly. 

Let’s say that half the switches are red and half are blue, and you can flip all the switches of a single color at once. If you flip all the red switches and the light goes off, then the master switch must be red. If you flip all the red switches and the light stays on, then the master switch must be blue. Either way, you now have only 500 switches to try. 

This is the same situation we’re in with chronic illness, except that there are something like 1,000,000 switches on the wall, and in some cases the lightbulb might be controlled by complicated interactions between multiple switches. It still makes sense to toggle big groups of switches all at once when you can, because that can narrow things down drastically.

One limitation of this approach is that it’s only really good at finding triggers. If you’re suffering from an iron deficiency, big eliminations probably won’t help find that.

The other limitation of this approach is that it’s not always clear how long you have to eliminate things for. Do you need to eliminate the mystery trigger for a week? A month? Longer? Ideally we could send you to Nepal for a year, or put you on the potato diet for a year, but in reality this won’t be practical for most people.

If Mary is getting poisoned by cadmium, and it takes two months to clear all the cadmium from her system, then going on a restrictive diet for only one month won’t help. But the problem is, she can’t know this in advance. How is she supposed to know about the clearance rate when she doesn’t even know what’s poisoning her?

So we’re stuck with an asymmetry. If one of these eliminations helps you, that narrows things down quite a bit. But if they don’t help you, then it’s more ambiguous. Maybe the half-life of whatever is making you sick is just too long. Still, it seems like it would be worth trying.

TRY SOME LIKELY VARIABLES AT RANDOM

Another possibility is to just try various things and see what works. This is grasping in the dark, but we can still do a lot to cover our bases.

For example, there are a finite number of vitamins. “Vitamin deficiency” is a plausible type of hidden variable, so you could just cycle through all the vitamins and see if any of them happen to be an immediate cure. It seems unlikely that you will get this kind of miracle result, but vitamins are pretty safe so the risk is very low, and you could at least check “vitamin deficiency” off your list.

Similarly, there are a finite number of elements. Some of them, like iron and potassium, are necessary for human health. You could try supplementing these and see if that treats your illness. Other elements, like lead and mercury, are known to be bad for your health. You could try getting blood and/or urine tests, or testing your local water supply, and see if you have higher exposure to any of these known toxins. 

Again these are all shots in the dark, but they’re all plausible variables that could be affecting your health. If it turns out your blood mercury levels are way higher than normal, that would be good to know.

You could also try to hit your illness with some generic treatments, basically anything where the name starts with the prefix “anti-”. If you can convince your doctor, you could maybe get them to put you on a broad-spectrum antibiotic (in case your chronic illness is bacterial), antiviral (in case your chronic illness is… etc.), antifungal, anti-inflammatory, or antihistamine. This is a little more risky, but there’s some chance your chronic illness might be fungal and this is one of the only ways you would ever find out. 

This broad-spectrum approach will generally be better for finding deficiencies, but in some cases it might also help identify triggers.

GET RID OF YOUR BLOOD

Haha, but no, seriously. Donating blood is easy, safe, and it’s a nice thing to do for your community. You might save a life. And if there’s something nasty building up in your blood, you might be able to get rid of some of it. There’s already some evidence that donating blood can reduce your serum PFAS levels. Maybe it can clear some other things from your system.

Again, this is a pretty blind approach. It probably won’t work for most people. It may not work for anyone at all. But if you donate blood and your symptoms immediately get better, that would be pretty interesting, right? 

V.

In the beginning, we’ll be taking shots in the dark. People will try dozens of things with little or no success. This will be quite frustrating.

But the hope is that eventually, we will start to get our bearings. If a couple people with chronic fatigue find that they have high levels of cadmium in their blood, then other people with chronic fatigue will want to check their cadmium levels before trying other interventions. Conversely, if a couple dozen people with chronic fatigue check for cadmium and find nothing, checking for cadmium should be moved lower down on the list for chronic fatigue. 

Over time, some people will get very skilled at organizing and interpreting this kind of research. And with enough people trying things and going over the data together, tricky bugs quickly become shallow. 

Depending on the success of this approach, you can even imagine this being somewhat formalized. Someone could make a centralized list of things to try or to have tested, and people could report what they had tried and whether it worked out. Tests that seem to be helpful could be moved up in the rankings so people could know to try them first; tests that don’t seem to help people could be moved down and left for the last ditch attempt.

Will this work at all? Who knows, but it seems like it’s worth trying. And at the very least, we may be able to rule out some hypotheses. After all, if we can establish that it’s none of the things on this list… then what the hell is it? 

Endnotes

[1]: 

“A pigeon is brought to a stable state of hunger by reducing it to 75 percent of its weight when well fed. It is put into an experimental cage for a few minutes each day. A food hopper attached to the cage may be swung into place so that the pigeon can eat from it. A solenoid and a timing relay hold the hopper in place for five sec. at each reinforcement.

If a clock is now arranged to present the food hopper at regular intervals with no reference whatsoever to the bird’s behavior, operant conditioning usually takes place. In six out of eight cases the resulting responses were so clearly defined that two observers could agree perfectly in counting instances. One bird was conditioned to turn counter-clockwise about the cage, making two or three turns between reinforcements. Another repeatedly thrust its head into one of the upper corners of the cage. A third developed a ‘tossing’ response, as if placing its head beneath an invisible bar and lifting it repeatedly. Two birds developed a pendulum motion of the head and body, in which the head was extended forward and swung from right to left with a sharp movement followed by a somewhat slower return. The body generally followed the movement and a few steps might be taken when it was extensive. Another bird was conditioned to make incomplete pecking or brushing movements directed toward but not touching the floor. None of these responses appeared in any noticeable strength during adaptation to the cage or until the food hopper was periodically presented. In the remaining two cases, conditioned responses were not clearly marked.

The conditioning process is usually obvious. The bird happens to be executing some response as the hopper appears; as a result it tends to repeat this response. If the interval before the next presentation is not so great that extinction takes place, a second ‘contingency’ is probable. This strengthens the response still further and subsequent reinforcement becomes more probable. It is true that some responses go unreinforced and some reinforcements appear when the response has not just been made, but the net result is the development of a considerable state of strength.”