depression

The Mind in the Wheel – Part V: Depression and other Diagnoses

slimemoldtimemold, , , , , , 12 Comments

[PROLOGUE – EVERYBODY WANTS A ROCK]
[PART I – THERMOSTAT]
[PART II – MOTIVATION]
[PART III – PERSONALITY AND INDIVIDUAL DIFFERENCES]
[PART IV – LEARNING]

There’s a thermostat that regulates the temperature
That might not be reliable
That should be disconnected

— Thermostat, They Might Be Giants

Here are some mysteries about depression: 

In most illnesses, the list of symptoms is hit or miss. Not every patient gets every symptom, or even every common symptom. If a common symptom of an illness is breaking out in hives, many people will break out in hives, but some people won’t. 

Depression is much stranger. Like other diseases, you sometimes get symptoms and sometimes do not. But on top of that, you also sometimes get symptoms, and other times get their opposites. 

One common symptom of depression is eating too little. Another common symptom of depression is eating too much. 

One common symptom of depression is gaining weight. Another common symptom of depression is losing weight. 

One common symptom of depression is insomnia, not being able to sleep. Another common symptom of depression is sleeping too much. 

The most typical symptom of depression is feeling really bad. Except in other cases, when the most typical symptom of depression is feeling nothing at all. 

There are other weird mysteries when we look at how depression is treated. Even though insomnia itself is a symptom, sleep deprivation often seems to treat depression. (In this context it’s sometimes called Wake Therapy.) This is effective in as many as 50% of cases, though the relief is usually only short-term. 

We think these mysteries make sense when you start looking at depression as a disorder of negative-feedback-loop emotions. 

You see symptoms and their opposites because many pairs of emotions cover two ends of a single variable. There’s one set of emotions that make sure you eat enough and another set that make sure you don’t eat too much. Since depression can interfere with either side of the scale, you sometimes get opposite symptoms. We have one set of emotions that make you go to sleep and another set that make you wake up. Since depression can interfere with either, some depressoids have insomnia, and others sleep through their alarm. 

Though it’s not usually listed as an official symptom, we would also expect there to be some cases of depression where people end up overheating, and other cases where people end up getting much too cold. Just like for eating and sleeping, there are governors on both ends of the scale, and the governors that would normally take care of these errors are being interfered with. 

Symptoms of depression like “loss of interest in sex” don’t have an opposite symptom because while there is a governor making sure that you’re interested in getting a certain amount of sex, there isn’t a corresponding governor making sure you’re not getting too much sex. 

Sleep deprivation may be a treatment in some cases because as we’ve previously mentioned, happiness is created by producing and then correcting errors. That means that a very large error created by getting very sleepy might be big enough to register, even when something is wrong with the happiness machinery. If nothing else, it might shake things up enough that something will register. 

We also think that close examination shows that depression is not just one disorder, it’s several different disorders. They share a surface-level similarity, but can be clearly divided into types. 

The surface-level similarity that all different kinds of “depression” have in common is that they are all disorders where the person very rarely experiences happiness. This is why our culture formed the category “depression”, because we noticed that sometimes people had a persistent lack of happiness, even when they found themselves in situations that should normally make them happy. 

But the systems that produce happiness are complex. There are many things that can go wrong, so there are many different kinds of “depression”. And besides the fact that they all present similarly — the person has a hard time experiencing normal happiness — different kinds of depression don’t always have very much else in common. 

If we take a look at different ways this model can malfunction, we’ll see different outcomes that all look kind of like depression. But we’ll also notice that despite their basic similarity, most of these different problems have at least slightly different symptoms, so it may sometimes be possible to distinguish them by symptoms alone. 

In fact, almost anything that goes wrong in the motivational system will cause something that looks like depression, which is probably why “depression” is so common. Our job is to look past these superficial similarities and try to figure out exactly what is malfunctioning, so we can have some hope of treating it. 

Components

If you have two patients with very similar symptoms, you might be tempted to assume they have the same disease. But they might also be experiencing the same symptoms for totally different reasons. For example, a cough could be caused by a viral infection, bacterial infection, or a non-infectious condition like asthma or acid reflux. Or from accidentally inhaling your Dr. Pepper.

You want to give the antibiotics to the person with a cough from a bacterial infection, and the antivirals to the person with a cough from a viral infection, and you don’t want to mix them up. The underlying cause determines the appropriate medical approach, not the symptoms. If you don’t know what’s causing the problem, you can’t treat it. 

Imagine you are working on an internal combustion engine. To work correctly, an engine requires both gas and a spark. The battery is fully charged, so you know you have the power needed to create a spark. When you turn the key, the engine turns over, but doesn’t start. 

This could be caused by at least two problems. First of all, maybe no gas is getting into the cylinder. Second, maybe the spark plug doesn’t work.

Those sound like two different causes. But they are not, at least not quite. It’s true that we can narrow things down to these two different branches — it almost certainly has something to do with the gas or with the spark plug. But the real causes are much more complicated.

Maybe there’s no gas getting into the cylinder, but “no gas getting into the cylinder” could happen in a number of different ways. First of all, the vent to the gas tank is clogged, creating a vacuum that stops gas from being drawn into the line. Second, there might just not be any gas in the tank. Third, the gas filter could be clogged. Fourth, the gas pump could be broken; for example, there might be a hole in the diaphragm.

The spark plug is also made up of many smaller components. If any component fails, then there’s no spark.

Imagine there are 100 old cars. One misty, rainy morning, you discover that none of them start. This was a real issue back in the day — a big rainstorm one night, and in the morning, half the cars in town stop working. 

You figure it must be the spark plug wires, moisture kills ’em. But replacing the spark plug wires only fixes some of the cars. You eventually find out that some of them actually got water in the gas instead, and need a different fix.

Same symptoms (old car doesn’t start), same distal cause (rain storm), but a different proximal cause. So we can have cases with the exact same symptoms (engine turns over but doesn’t start), with two possible diagnoses. And even within those two diagnoses, there are potentially dozens of causes, each requiring a different fix. Even if you figure out for sure that the problem is a lack of gas, you still don’t know if you need to replace the gas filter, or part of the gas pump.

Any system is made up of many smaller parts, all of which can break in several different ways, so you can usually get the exact same disorder as result of issues with different parts. A specific part of the chain is broken, but you can’t tell which one.

Ultimately, examining the symptoms that arise when different systems malfunction will be helpful for treatment. But it’s not law — systems can break for more than one reason. System malfunctions from two different causes may look just the same — from the outside, all you notice is that this system isn’t carrying out its function, but that doesn’t tell you how to fix it.

Malfunction: Too Much Success 

Some people have such perfect control over their life that they never meaningfully get hungry, thirsty, tired, lonely, cold, etc. etc. 

This sounds good, great even. But in fact the person ends up very depressed, for a simple reason. Nothing is actually wrong with this person, there’s no damage, it’s not even really a malfunction. But the fact that they almost never correct major errors means they very rarely produce any happiness.

This is the depression of the idle rich, which we mentioned before. It can be treated by intentionally creating errors and then correcting them. For example, you might intentionally get very tired and thirsty from running an ultramarathon, and then rest and rehydrate, which will grant nothing short of ecstasy. Or you might expose yourself to pain from some other extreme sport, then recover, and again reap the happiness. People often discover this treatment on their own, which is why the idle rich are often into certain kinds of (no judgment) self-destructive hobbies. 

Sleep deprivation therapy seems like it would work pretty well for this kind of depression. It’s the same logic as extreme sports. Staying up all night and then going to sleep would be an almost euphoric experience that would provide you with some happiness, at least for a while. 

As a bit of a tangent, something else that may explain some behavior of the rich is that there may be an emotion that drives us not only to maintain our current status, but to increase our status at some constant rate. In other words, there may be a governor whose target is the rate of change, or first derivative, in status.

For someone of normal status, this expresses itself as normal ambition. The average person can always become more important. But as you become higher and higher status, this becomes a problem, because the higher status you are, the harder it is to increase your status further. 

Someone who reaches maximum status for their social group finds themselves in a bind. They still feel the drive to increase their status, but they are already top dog. A person might become CEO, or in an earlier age might become King, due to their drive to increase their status. And these people will usually be the ones with the strongest status governors, making their inability to increase their status any further especially painful for them. Those who reach the top are left with a hunger they can no longer feed. “When Alexander of Macedonia was 33, he cried salt tears because there were no more worlds to conquer.”

Kings like Alexander often tried to overcome this by establishing their divinity. This approach doesn’t work so well any more, but modern people sometimes handle it by realizing that while they cannot advance their status any further in their own field, they can still advance their status in new areas. This is how Bill Gates ends up advancing his status by becoming a philanthropist, or how Mark Zuckerberg advances his status by training in jiu jitsu. Both of them had already maxed out their status as tech guys, so to keep increasing in status, they had to find new kinds of status in new arenas. 

It’s reasonable to ask if this kind of “depression” is actually a problem. Is it so bad to not be very happy, as long as all your needs are met? If you don’t understand what’s going on, you might be concerned that there’s something wrong with you. But if it’s this simple, and it’s just a side effect of all your needs being met, then is there anything to be concerned about? 

We think there might be. First of all, while happiness isn’t everything, it’s nice to be happy, and it’s reasonable to think about ways you could be happier. 

Second, we suspect that happiness regulates the balance between explore versus exploit. If that’s true, then we would expect that over time, people who are depressed pursue stranger and stranger strategies as they increase their tendency to explore new ideas, in an effort to find something that “works”. But this never brings them happiness, because the problem is internal. 

If this is the case, then people who suffer from long-term depression, of any kind, should appear to act crazier and crazier over time, as they explore more and more unusual strategies.

Malfunction: Happiness not Generated

Under normal circumstances, correcting an error creates some amount of happiness. Somewhere in the system is a mechanism that registers when a correction has taken place, and creates a corresponding amount of happiness. The bigger the correction, the more happiness is created.

If something happens to this mechanism — the signal is turned down really low, part of it gets broken, the shipments of neurotransmitters it depends on never arrive — then you get a very characteristic form of depression.

This person experiences emotions as normal, generally behaves as normal, and has successful behavior. After all, their governors are all functioning exactly as normal, errors are getting corrected just like before. They are surviving, even thriving. But despite their successful behavior, they never feel happiness. They appear normal to casual observers, but describe themselves as “dead inside”. 

This kind of experience comes out pretty clearly in patient descriptions, like this one reported by William James:

I have not a moment of comfort, and no human sensations. Surrounded by all that can render life happy and agreeable, still to me the faculty of enjoyment and of feeling is wanting — both have become physical impossibilities. In everything, even in the most tender caresses of my children, I find only bitterness. I cover them with kisses, but there is something between their lips and mine; and this horrid something is between me and all the enjoyments of life. My existence is incomplete. The functions and acts of ordinary life, it is true, still remain to me; but in every one of them there is something wanting — to wit, the feeling which is proper to them, and the pleasure which follows them…All this would be a small matter enough, but for its frightful result, which is that of the impossibility of any other kind of feeling and of any sort of enjoyment, although I experience a need and desire of them that render my life an incomprehensible torture.

If happiness is still generated, just at much lower rates than usual, you would get a less extreme version of this experience. Someone who generated 50% as much happiness as usual would feel a little down in the dumps, but not terrible. Someone who generated 10% as much happiness as usual would feel pretty depressed, but not quite entirely dead inside. 

This is a good chance to give an example of what we meant when we were talking about how every system is made of many components, how the spark plugs can break in many ways. Even in this very simple model, many different problems will create the same kind of malfunction. 

For example, maybe the mechanisms that actually generate happiness are working as intended, but the connections that transmit the correction signal to those mechanisms are malfunctioning. In this case, happiness isn’t generated, because the signal that should trigger happiness never reaches its destination:

Or, maybe the connection is working just fine, but the mechanisms that should generate the happiness are malfunctioning. So the signal arrives just as it should, but nothing is done in response: 

While these are malfunctions in different parts of the system, a person with a malfunctioning correction-connection would behave almost exactly the same as a person with a malfunctioning happiness-generator. They would probably benefit from different treatments, since the cause of their depression is different, but they would be very hard to tell apart based on their symptoms. 

And of course, this is one of the simplest possible models. In real life, there are more than just two components; there must be dozens. 

Malfunction: Errors not Generated

Despite being responsible for different signals, your governors all run on basically the same architecture. If you want to think in terms of mechanical engineering, maybe they all share the same fuel, or the same lubricant, or they’re supplied by the same pump. If you think more in terms of programming, consider them as using many of the same functions, inheriting from the same class, or relying on the same set of libraries. 

Since they’re all supplied by the same metaphorical pump, if something goes wrong with that pump, something can go wrong with all of the governors at once. If there’s a function that you use all over your program, and you accidentally comment out an important line in the function, everything that uses that function will be affected. Maybe every part of the program will be impacted the same way, but depending on how the function is used, maybe in different ways. 

The most basic job of a governor is to compare the incoming signal to the set point and generate an error. If it does this correctly, its second job is to try to correct that error. But first it has to successfully generate the error.

So if this ability to correctly generate an error ever breaks, that’s a big deal. In a minor malfunction, error signals will still be generated as normal, but all error signals would be turned down, let’s say by 50%. In this case you’ll mostly behave as normal, but you will have to be twice as far out of alignment — get twice as cold, go twice as long without sleeping, etc. — before you take the same amount of action you normally would. And for a given level of actual distress, you will feel only about 50% as tired, hungry, thirsty, lonely, etc. 

If this happens to you, you’ll also experience less happiness than normal. Your errors don’t grow as fast as normal, so when you correct your errors, they’ll tend to be unusually small. Eating a meal that would normally correct 10 points of hunger error and create 10 points of happiness will instead correct 5 points of hunger error and create 5 points of happiness. So you’re not joyless at 50% error, but your actions won’t bring the sense of satisfaction that they used to. 

Of course, you could wait until you had a subjective experience of 10 hunger before eating a meal. Then you would get the same amount of happiness from correcting it. But if you do that, you will notice that you eat only half as often as usual, and you’ll still ultimately be getting less happiness over the long term, since you are correcting the same magnitude of error, but only half as often. You’ll also notice that you feel weak and brainfoggy, since you are only getting about half of your actual nutritional needs. 

However, lots of people eat more by routine than by hunger. So most people would probably stick to their three-meals-a-day approach, through the normal drumbeat of social routine, or just out of habit. These people will eat as much as normal, but get half as much satisfaction. 

If something more serious goes wrong, and all error signals are turned down to 10% instead, your motivation would become extremely sluggish, and you will generate much less happiness than usual. It will take you a lot longer to take action than it would otherwise, because it takes much longer for your error to reach a given level. This might be called procrastination.

Let’s simplify and say that hunger is driven entirely by blood sugar (not true, but this is for the sake of example). For a person whose error signals are turned down to 10%, their blood sugar will slowly drop lower and lower without causing an appreciable error. Without any error, it’s hard for them to have any motivation to eat, let alone cook. 

Eventually blood sugar gets very low and the governor is finally generating the minimum amount of error needed to get over the gate’s threshold. This person still won’t be very motivated, and the hunger still won’t be very pressing. And if they do eat, it won’t make them very happy, because the correction is quite small. This person is trapped in a world of low motivation, very dulled emotions, difficulty telling whether they are hungry / thirsty / tired / etc., and little happiness.

In the extreme case, if your error-generating systems are so busted that almost all your errors are close to zero no matter how far out of alignment you are, things get pretty bad. This person barely experiences emotions, so almost no behavior happens. This is classic “can’t get out of bed” or “bedrot” depression — the person has hygiene problems and only eats, sleeps, or moves when extremely hungry/tired/etc. Since their error signals never get very big, there’s no opportunity to correct them, and this person experiences almost no happiness.

If errors actually become zero, the person is effectively immobile, in a sense almost comatose or paralyzed. This lines up pretty well with the symptoms of having a serious basal ganglia injury. Consider this case report from Treating Organic Abulia with Bromocriptine and Lisuride: Four Case Studies:

During the preceding three years he had become increasingly withdrawn and unspontaneous. In the month before admission he had deteriorated to the point where he was doubly incontinent, answered only yes or no questions, and would sit or stand unmoving if not prompted. He only ate with prompting, and would sometimes continue putting spoon to mouth, sometimes for as long as two minutes after his plate was empty. Similarly, he would flush the toilet repeatedly until asked to stop.

Antipsychotic drugs like haloperidol seem like they might be messing with the same system. If you take too much haloperidol, you’ll sit there and do nothing, possibly until you die.

Malfunction: Errors Generated Too Much 

Earlier we used the analogy of a pump being damaged and working at only 50% capacity. The pump can also be damaged in a way that sends it into overdrive, where it careens along at 200% capacity.

If this happens, all your errors are twice as large as normal. You are more driven, and driven to do more things. You become wildly active. Since your errors are larger, correcting them makes you even happier. The most normal successes, like drinking a glass of water, create almost ecstatic happiness, and your mood soon goes off the charts. This sounds a lot like the high phases of manic depression / bipolar disorder.

The real question here is why manic depression is so common, but pure mania — mania just by itself — is so rare. People are pure depressed all the time, many people are bipolar, so why aren’t many people suffering from pure mania? 

We certainly don’t know, but here are some rough hypotheses. 

One very mechanical answer is that overdrive is simply unsustainable. A person can’t be manic all the time, because eventually they will run out of juice. 

Consider that example with the pump. In this case, the pump that supplies all the different engines has malfunctioned and is running at 200%. The pump circulates oil drawn from a reservoir. Normally this reservoir re-fills with oil (perhaps as it’s filtered, or drawn from a larger reservoir) faster than the pump circulates it. But when the pump is rushing along at 200%, it drains the reservoir faster than it can be filled. So it empties the reservoir and triggers some kind of emergency stop, during which it can draw no oil at all. Eventually the reservoir refills and the stop is lifted, and the pump goes back into overdrive mode again. 

In this explanation, bipolar disorder *is* pure mania. During your manic phases you overuse some kind of limited resource. When it runs out, you’re cast into the depressive phase caused by the lack of that resource. This lasts until the resource has built up back to some minimum level, at which point you start running at 200% again. But the real nature of the problem is disguised, because the human nervous system has checks and limits. 

(A quick research check suggests that in bipolar disorder, manic phases last only days or weeks, while depressive phases last months. That’s a pretty interesting pattern, why aren’t they more symmetrical? Perhaps it is because the limited resource takes longer to regenerate than it does for the manic phase to burn through it.)

Another possibility is that bipolar disorder isn’t a malfunction in our motivational system, it’s a disorder in some other system that’s connected to motivation. Think about the circadian rhythm. This is a system that, roughly speaking, drives us to be active during the day and sleep during the night. If something were to happen to our circadian rhythm — if the daytime highs were dizzyingly high and the nighttime lows were crushingly low, if the cycle were an awkward 108 hours long instead of a nice 24 hours long — that might also look a lot like bipolar disorder. 

Malfunction: Errors Reduced in Other Ways 

One thing we’d like to explain is why you get the weird pattern of opposite symptoms in depression, where (for example) one person eats too much and another person eats too little. Most diseases don’t cause their own opposite symptoms.

This problem seems like it has something to do with the balance between governors that come in pairs and watch two ends of the same variable. In the undereating/overeating example, that would be the balance between the hunger governor (“make sure to eat enough”) and the satiety governor (“but don’t eat too much!”).

But if you turn down the errors on both of these governors equally, they should remain in perfect balance. So turning down all governors by a flat amount shouldn’t cause this kind of symptom, we shouldn’t see this weird pattern. 

No actual problem is ever so clean. Let’s go back to our pump metaphor. We can say things like “everything is turned down by 40%”, but in practice if 10 different motors are all supplied with lubricant from the same pump, and that pump gets jammed and starts working at only 40% capacity, some of the motors will be worse off than others. Motors that are further away from the pump, and have longer tubes, will probably be worse off. Motors close to the pump will be better off. Some of the closer motors might even keep functioning as normal. At 40% capacity, clogs may form in the lines, but they will form in some lines and not in others.  

The point is, in any kind of real nuts-and-bolts system, a 40% loss of capacity won’t lead to a performance drop of exactly 40% in all parts. Some parts will be more affected than others. So even if you take a general hit to something that supplies all your governors, it might still affect your hunger governor more than your satiety governor, leading you to undereat. It might sometimes turn governors down and other times turn them up, leading to either insomnia or sleeping too much.  

But there are other ways to get this pattern too. 

First of all, this could have something to do with the weights on the governors, what we think of as personality. We’ve been assuming that any change will be a percent of the original signal (if the original signal was 10, and there’s a flat reduction to 30%, the new signal will be 3), but if every governor is cut down by a flat amount instead, then the balance between the governors will end up somewhat different than it was before. 

Let’s say two people have a malfunction with their error-generation systems, but instead of reducing all errors by 50%, this malfunction reduces the weights on all their governors by 0.6 across the board.

One guy has a starting hunger weight of 1.3 and a starting satiety weight of 0.8. After being reduced by 0.6, his new hunger weight is 0.7 and his new satiety weight is only 0.2. His drive to eat was always a bit more powerful than his drive to stop eating. But the relative strength of this relationship has changed enormously. Now, his satiety governor is barely active at all. He is definitely at risk of eating too much, the satiety signals just don’t come through like they are supposed to. So this guy gets one pattern of symptoms: overeating.

The other guy is the exact opposite, a starting hunger weight of 0.8 and a starting satiety weight of 1.3. After depression, his new hunger weight is 0.2 and his new satiety weight is 0.7. His drive to eat was never very powerful, but now it’s almost nonexistent. He will definitely end up with a different symptom: not eating enough. The hunger signals just don’t come through like they are supposed to. 

If this is one way to become depressed, then the symptoms of this kind of depression, especially the asymmetric symptoms, will tend to be more extreme versions of someone’s normal personality traits. When they become depressed, someone who has always had some trouble falling asleep will get the symptom of insomnia — while someone who has always had some trouble waking up will get the symptom of oversleeping.

Finally, we’d like to note that everything in the world is at least a little bit random. If you have a general problem with the cybernetic governors in your brain, odds are that some of them will be more affected than others, for no particular reason. 

For people with this kind of malfunction, since some of their emotions are functioning correctly, they can still sometimes correct their errors. When only some governors are affected, you will experience some happiness, though usually less than before, so this is often hard to diagnose as depression. If one governor is particularly knocked out, then it may be diagnosed as something else — like if your sleep governor is particularly suppressed, it might be diagnosed as insomnia.

Malfunctions in Voting and Gating

There’s some set of mechanisms that generate, transmit, and assign votes. Like anything else, these can break or get jammed. 

Any malfunction that makes a person get fewer votes than normal will lead them to take less action. Any malfunction that makes someone get almost no votes will lead them to take almost no actions. This is basically the same as the malfunctions in error generation described above: the person will take fewer actions and generate less happiness than usual when they do. 

A particularly interesting part of the selector is the gate. Remember that the gate has a threshold for a minimum amount of votes, and it suppresses votes for actions below this total to keep us from dithering or wasting resources. To use some arbitrary numbers for the sake of illustration, the gate might have a threshold of 5, meaning that when an action gets 5 votes or less, the gate clips that to zero, and the action gets effectively no support at all.

Any malfunction that raises someone’s gate threshold a bit will lead them to take fewer actions, because actions that would normally pass the threshold will no longer be able to clear it. Any malfunction that raises someone’s gate threshold a lot will lead them to take almost no actions, once the threshold is so high that almost no action can afford it. 

Since action is needed to correct most of your errors, and correcting your errors is the source of happiness, if one of these malfunctions happens in your head, you will get less happy. Again, this looks like depression.

On the other hand, malfunctions that lower your gate threshold, so that actions can be performed even when they don’t get very many votes, will lead you to take more actions. In particular, you will have more of a bias towards action, because any small discomfort will more easily translate into doing something about it. 

Malfunctions in these systems are a little hard to talk about, because they cause very similar behavior as the malfunctions in generating errors, described just above. In both cases, people will take less and less action, becoming sad and unmotivated. So they will be very hard to tell apart.

The most likely difference is subjective. When the mechanisms that generate errors are malfunctioning, you get weaker errors, or no errors at all. Since errors are emotions, these people feel no emotions — no hunger, no thirst, no pain, no loneliness, etc. 

But with malfunctions in voting or in the gate, emotions / error signals are being generated as normal. It’s just that the governor never gets the votes it needs to correct them, or the votes can’t get through. These people experience all their emotions just as strong as ever, but cannot take action to correct them.

If the malfunction is severe enough, this would present a lot like bedrot — the person can’t get out of bed, they barely eat or sleep, etc. But subjectively it is very different. With bedrot caused by a malfunction where your governors can’t generate their normal error signals, you don’t do anything, but you also don’t feel anything. With bedrot caused by a malfunction in voting or gating, you still feel hungry, tired, thirsty, gross, etc. as much as ever, but you’re trapped and cannot bring yourself to take even the smallest action to help yourself. 

Malfunctions in Pricing

A governor needs to be able to act on the world to succeed, but it also needs to be able to recognize success when success arrives. If you succeed and you don’t remember it, then you can’t benefit from the experience. So any kind of malfunction in the learning process can make behavior get very strange.

Consider what would happen if each experience were only recorded as a fraction of its true value. You go out with friends and find that this reduces loneliness by 10 points. But through some strange error, it is experienced or recorded as reducing loneliness by only 1 point. This gives you a very skewed view of whether or not you should go out with friends, when you are trying to decide what to do in the future. 

If this happens across the board, then every action will gradually but consistently get underestimated. Over time, your governors learn to estimate all behaviors as being only 10% as effective as they really are. 

This wouldn’t be such a big deal, except that all actions have costs. Forget about going out in the cold to see friends. It used to be 5 points of effort to drive to the bar, plus one point because the cold governor is always voting against it, but seeing your friends reduces loneliness by 10 points, so it was worth net 4 points on average. But now that “seeing friends” is estimated at only 1 point, it’s no longer calculated to be worth it! This guy will sit at home and feel bad, wanting to do something, but feeling like nothing is worth the effort.

In this kind of depression, you mistakenly believe that no action is worth the effort it would take, so you end up sad, because you choose not to do anything. This choice makes a certain internal sense — according to your best recordkeeping, choosing to do things isn’t worth it. From an outside perspective we know what is happening — your recordkeeping is all wrong! But it’s not obvious from the inside.

If you did take those actions, you would find that they create happiness as normal. Nothing is wrong with your ability to experience things. But you still wouldn’t learn from that experience, because there is something wrong with your memory.

Other malfunctions in the machinery of learning and memory would have similar effects. For example, if values were stored properly (“let’s write this down, eating a burger makes me 50 points less hungry”) but retrieved improperly (“hmmm, according to my notes a burger only corrects 5 points”), the effect would be almost the same. 

Recap

  • Happiness is generated when one of your governors corrects an error, and the bigger the correction, the more happiness. 
  • If you are too successful at keeping your errors in check, then the corrections you make will always be very small, so you will be generating very little happiness. This looks kind of like depression.
  • If something goes wrong with the mechanisms that generate happiness, then your governors will keep generating their errors and will keep correcting them, but these corrections will not create any happiness. This looks like depression.
  • If something goes wrong with your governors’ ability to generate their errors, so that all their errors are smaller than normal, then you will be more lethargic, have blunted emotions, and when these errors are corrected, the correction will create less happiness. This looks like depression.
  • If something goes wrong so that sometimes governors generate smaller errors and sometimes generate larger errors, then you will go through periods of intense activity with huge amounts of happiness, and periods of serious lethargy with almost no happiness to speak of. This looks like manic depression.
  • If something goes wrong with the systems that handle turning the errors into votes, counting the votes, weighing the vote totals, etc., then you will still feel those errors as normal emotions, but you will not be able to successfully take the actions needed to correct those errors. This looks like depression.
  • If something goes wrong with the systems that keep track of the effects of your actions, that learn or remember these values, so that all actions are remembered as less valuable than they really are, then bad things stop seeming as bad, good things stop seeming as good, and nothing seems worth doing relative to the cost of the effort required to do it, so you don’t do much at all. This looks like depression.

And this may not be a complete list. 

[Next: ANXIETY]