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

---

It is interesting to note that when certain types of paradoxes are fed to the Kalin-Burkhart machine it goes into an oscillating phase, switching rapidly back and forth from true to false. In a letter to Burkhart in 1947 Kalin described one such example and concluded, “This may be a version of Russell’s paradox. Anyway, it makes a hell of a racket.”

— Martin Gardner, Logic Machines and Diagrams (1958)

Life requires navigating the conflict between different parts of yourself.

Everything you do, all your behavior, is the result of different governors negotiating in an attempt to satisfy your drives and emotions. In most situations, governors come to a mutually-agreeable compromise and you can proceed from one thing to another in a way that satisfies all parties involved. 

But other times, things do not go so well. The inside of your head is like any palace intrigue: factions rise and fall, allies today are enemies tomorrow, and no one is ever fully in control.

Conflict

When your governors want two incompatible goals to be realized at once, the result is conflict.

Conflict can have different outcomes. When the opposing governors are closely matched in force and there’s a binary decision, it will lead to inaction. When they are closely matched in force and there’s a range of behavior, it will lead to half-measures. When one is much stronger, it can lead to countermeasures.

For example, your hunger governors might vote strongly in favor of eating a piece of cherry pie. But like many people, you have internalized the idea that eating cherry pie is a wicked, weak thing to do. So your shame governor votes strongly against it. The votes cancel each other out. You stand in the window of the bakery for a long time, staring at the pie and doing nothing. Here conflict has led to inaction.

A mouse’s hunger governor will vote to approach a feeding bowl (the mouse is hungry and the bowl is full of food), while its pain governor votes to avoid the feeding bowl (which has been rigged to give the mouse painful electric shocks). If these two governors are about equally strong, the mouse might go half way out towards the bowl of food, and no further. When it gets closer, the fear governor becomes more powerful and pushes it back. When it gets further, the fear governor becomes weaker and the hunger governor pushes it forward. Here conflict has led to an intermediate state, half-measures.

Even when one governor is strong enough to win, there can be ongoing conflict. You pull the cookies out of the cupboard because you’re hungry. Shame makes you throw them in the trash. But once they’re in the trash, hunger is in control again. So you fish the cookies out of the trash. This is conflict where the state is countermeasures.

For a real-world example, here’s Henrik Karlsson describing his own experience of a minor conflict:

Our emotions and intuitions are littered with contradictions.

To take a simple example: when I was at the gallery where I worked until last week, my low blood sugar cravings sometimes told me that it was ok to take a pastry from the café. But when I want to feel like a upright person, I don’t believe in taking stuff that isn’t mine. So which is it? If I follow my gut and eat the pastry, I will be true to myself in the moment, while betraying other versions of me.

The experience of conflict is stress. Staring at the pie and doing nothing is a fairly stressful experience. Hovering between fear and hunger is stressful for the mouse. Throwing out the cookies over and over again is no better.

Unlike the errors generated by your governors, stress is not an emotion. It’s a different kind of experience that happens when two or more actions are in direct competition.

It’s easiest to become stressed when two drives are in direct conflict. You want to ask someone out, but you’re afraid of rejection. You want to eat a whole pizza, but you know your family will laugh at you if you do. Here, two drives hold each other in check, there’s tension.

But you can also become stressed when drives are merely in competition. You might want to both go out and see friends (because you are lonely) and stay in and go to sleep (because you are tired). These are both positive desires, but you can’t do both at once, they are mutually exclusive. If they’re both about equally strong, you will do nothing, neither go out nor sleep, and it will be stressful. 

And you can become stressed when negative drives are in competition. A witness to an assault must choose between intervening (risking physical harm) or walking away (risking social condemnation as a coward). Both of these outcomes are things they would like to avoid, but they can’t avoid both. This is also stressful, and again, if they are equally matched the person will do nothing. 

Stress is a really negative experience, for two reasons. 

First of all, when they’re in conflict, governors are distracted from everything else. They commit all their resources to the fight, and clog up common parts of the system, like the voting channels, with constant bids for their concerns. 

Second, governors in conflict are absolutely gunning it. When a governor pushes and the signal doesn’t change, what does it do? That’s right, it pushes harder! If it’s pushing against another governor, then that governor pushes back. They both push ten, twenty, fifty times harder. Soon they are both pushing absolutely as hard as they can. 

This is an incredible waste. Whatever resources are involved in this contest will be burned through at an astounding rate, with the only result being a deadlock. But this is what you get when you’re in a double bind. For one governor to correct its error, the other governor must experience an increase in its error. There is no way for both systems to experience zero error at the same time.

This view of stress calls back to old theories like the approach-avoidance conflict, also sometimes called push/pull. Kurt Lewin, who was close with the early cyberneticists, was one of the people who argued for this approach.

“Conflict,” he wrote in his 1935 book, “is defined psychologically as the opposition of approximately equally strong field forces.” Kurt talks in slightly different terms, but the overall conclusion is the same. He offers this example: “The child faces something that has simultaneously both a positive and a negative valence. He wants, for example, to climb a tree, but is afraid.”

Kurt’s views were very influential back in the day, but psychologists don’t really focus on his models anymore. This might be because he insisted on explaining human motivation in terms of “psychical field forces” instead of drives, perhaps in an ill-fated attempt to try to make psychology sound more like physics. In fact, he explicitly rejected drives as “nothing more than the abstract selection of the features common to a group of acts that are of relatively frequent occurrence.” 

This model of stress has a surprising implication for self-control. You cannot alter your behavior by simply choosing to overcome the unwanted behavior.

There must be a drive already voting for that behavior, since the behavior exists. It must be controlling something. So to attempt to overcome a behavior can only lead to conflict. 

You can avoid situations that would put your governors at odds with each other. You can set one governor against another, suppress your unwanted impulses by the force of shame or fear. All of these defense mechanisms and more will “work”. They will keep you from accidentally doing the unwanted behaviors, at the cost of more conflict. Or they will let you avoid the agony of conflict, at the cost of making your life smaller and smaller. But the only fully healthy solution is to find a way to reorganize things so that both governors can fulfill their purpose without conflict.

William Powers said it best: 

The payment for a lifetime of “overcoming” one’s weaknesses, base desires, and forbidden habits is to spend one’s last years in a snarl of conflicts, one’s behavior restricted to that tiny part of the environment that leaves all conflicts quiescent, if any such place still remains. The rigidity of many elderly people is, I believe, the rigidity of almost total conflict, in which every move is made against massive inner resistance. 

…

Indeed, self-control is commonly taught as part of raising children … Through social custom and the use of reward and punishment, therefore, we have perpetuated the teaching of self-control and have thus all but guaranteed that essentially everyone will reach adulthood suffering severe inner conflict. Self-control is a mistake because it pits one control system against another, to the detriment of both. 

Anxiety

The word “anxiety” is only an abstraction. It groups together many things that seem similar, but may have different causes underneath. (See the prologue to learn more.)

But to take a stab at what all kinds of “anxiety” have in common, we could say that they all look like systems spending a huge amount of energy to make very little progress. 

We see two general ways that might happen.

The first is chronic conflict, two or more governors locked in a deadlock for a long time, spending a huge amount of energy fighting each other and getting nowhere.

The second is oscillation, a system wildly swinging back and forth, spending a huge amount of energy correcting and re-correcting, instead of efficiently settling towards a target or equilibrium.

Chronic Conflict

When you’re consistently stressed for a long time, that seems like one kind of anxiety.

Sometimes governors are briefly in deadlock, like gazing at the pie through the window. Eventually you will get cold enough to walk away, or hungry enough to buy the pie, or something will distract you. So this conflict can’t last for very long. 

But sometimes governors get locked, not just in conflict in the moment, but habitually in conflict all the time, so that you are constantly stressed. For many people, food is a source not only of stress but also of anxiety, because their feelings of shame will get into conflict with their desire to eat fat and sugar, not just one time, but over and over and over again.

In nature, stress tends to be limited to very brief experiences, where two drives happen to be perfectly balanced. These situations tend to be over pretty quickly. One of the drives will grow faster than the other, or the situation will change, and the conflict is resolved. 

However, in a manufactured environment, it’s easy to produce anxiety-inducing situations by accident. For example, a lab animal that only feels safe in a dark tunnel, but whose water bottle has been placed in the brightly-lit center of the cage, will go through repeated experiences of stress as its fear grapples with its thirst. Or a dog that wants to protect its family, but whenever the dog barks at passers-by, the family yells at it. These animals will be anxious, because their drives are habitually in conflict. 

Because of our notable collection of social emotions, humans seem to have the worst of this. Social emotions consistently come into conflict with the others. People want to yell at their boss out of anger but don’t want to suffer the social consequences of that outburst. They want to sleep with people they are socially forbidden to sleep with. They are terrified of something but are not able to act on their fear; like a student terrified of their teacher, or a professional driver terrified of getting in a crash. Our social norms around food seem practically designed to be anxiety-inducing; half of the things that a person might naturally be most excited to eat are considered “bad” or outright sinful. That’s a conflict right there.

It’s even possible this is the role social emotions serve in our psychology. Maybe social emotions are there to make us anxious. Humans are still by their nature angry, horny, violent, and so on. But our social emotions put some checks on these drives and may be the only thing that make it possible for us to work together over the long term. Social emotions are frequently called on to keep the other emotions in check, and stress is an unfortunate side-effect of this balance.

Hamlet was stressed because he has to both kill his uncle and not kill his uncle. He is presented over and over again with opportunities to kill or not kill his uncle, or at least to take steps in those directions. But he can’t do either, because the two drives are almost perfectly balanced. This is very stressful. 

Antigone was bound to bury her brother Polynices, but Creon had decreed that Polynices was not to be buried or mourned, on pain of death. Orestes avenges his father Agamemnon by killing his mother Clytemnestra, honoring his filial duty to his father but violating his filial duty to his mother. This is the source of tragedy. The ancients had it right.

Let’s see one very interesting example.

When two governors have very similar amounts of votes — let’s say within 10 votes of each other — neither one can win, and you are in a state of conflict. 

This causes a little bit of stress. But normally, one or the other of the options will soon get enough votes to beat that margin, or some new issue will come up that renders the decision moot. 

However, sometimes for one reason or another, all the vote totals get turned down. This is one of the malfunctions we call “depression”. 

This has a curious side-effect. Let’s say that normally you have a hard time deciding between staying at the party and going home. Your loneliness governor has 40 votes for “stay at the party” and your fatigue governor has 45 votes for “go home”. 

Since these are within 10 votes of each other, neither can really win. This is uncomfortable and you feel a little stressed. Instead of really engaging, you hover at the edge of the party. But eventually one or the other governor gets a big enough error that it gets enough votes to beat the margin. Probably you get a bit more fatigued, until fatigue hits 51 votes, wins the margin, and you go home.

But when all your errors are turned way down, something strange happens. At the party, you now have fewer votes overall, which makes it harder to break this tie. If errors are turned down to 50%, then you have 20 votes for “stay at the party” instead of 40, and your fatigue governor has 22.5 votes for “go home” instead of 45. 

Now to break the tie, your fatigue governor needs 7.5 more votes instead of 5, and each vote requires twice as much of an increase in fatigue. If errors are turned down to 10%, then you have only 4 votes for “stay at the party” and only 4.5 votes for “go home”! You will stay in deadlock for much longer, and it will be stressful the whole time. 

Worse than that, you will end up in deadlock more often, over more issues. Normally it is easy to choose to shower (70 votes) before eating breakfast (50 votes). We’re not talking about deciding between the two, just the decision to finish the one before the other. 

But if your vote totals are cut, you may find that this decision is suddenly 35 votes versus 25, just barely enough for the vote to resolve. If your votes are cut enough, you won’t be able to decide whether to shower first or eat breakfast first. You become indecisive about all kinds of things, even the smallest decisions.

This may explain why depression so often goes along with anxiety. When your vote totals are turned down, but the margin of votes by which an action has to win remains the same, you end up in a state of deadlock for much longer, and it will happen a lot more often. Since conflict makes you feel stress, you feel stressed all the time, over the kinds of decisions that would be simple or easily resolved before. That’s anxiety. 

There are probably many things that can cause anxiety. But any kind of depression that gives you fewer votes overall is going to almost always lead in this unfortunate direction. This also suggests that other forms of depression, that don’t give you fewer votes overall, shouldn’t lead to more conflict and shouldn’t go along with anxiety. 

Oscillation

The second way to waste a bunch of energy for no reason is when a system swings back and forth for a long time without settling.

One of the classic ways a control system can fail is that it goes into oscillation, wildly swinging back and forth, wasting a huge amount of energy instead of efficiently settling towards the set point and zero error.  

In psychology, this is most obvious in tremors. Damage to the control systems responsible for motor function leads to overshooting and very obvious physical oscillations. In Cybernetics, Norbert Wiener described a few cases:

A patient comes into a neurological clinic. … offer him a cigarette, and he will swing his hand past it in trying to pick it up. This will be followed by an equally futile swing in the other direction, and this by still a third swing back, until his motion becomes nothing but a futile and violent oscillation. Give him a glass of water, and he will empty it in these swings before he is able to bring it to his mouth. What is the matter with him?

… His injury is … in the cerebellum, and he is suffering from what is known as a cerebellar tremor or purpose tremor. It seems likely that the cerebellum has some function of proportioning the muscular response to the proprioceptive input, and if this proportioning is disturbed, a tremor may be one of the results.

This isn’t a problem just with the brain, this is characteristic of all control systems. Weiner notes it as, “… a badly designed thermostat may send the temperature of the house into violent oscillations not unlike the motions of the man suffering from cerebellar tremor.” And in fact, all control systems can oscillate if they become unstable.

Tremors are oscillations in low-level control systems responsible for muscle movements. That’s why you can see them — your arm or leg is actually waving back and forth. 

But oscillations might also happen at other levels of control. If systems oscillate at the level of behavior instead of at the level of arm/leg position, that might look like doing behaviors over and over again, or doing them and then undoing them. This would look kind of like compulsions, or like OCD.

If systems oscillate at the highest level, something like thought or intention, that might look like choosing one side of a decision, but then before acting, switching to the other side of the decision. The guy who decides to quit his job, then decides to stay at his job, 20 times per hour. This looks like a form of rumination. 

This would kind of explain why OCD and ruminations seem connected. They may be basically the same kind of problem just in slightly different parts of the system. Or maybe OCD is a more extreme form of rumination, an oscillation that makes it all the way into behavior, instead of just oscillating “within thought”.

The difference between oscillations and conflict is that conflict is always a struggle between two or more governors, while oscillation can happen in just one governor alone, especially if it is damaged or otherwise improperly tuned. 

Oscillation can happen for a few key reasons.

A governor with too much gain, that makes very aggressive corrections, can overshoot repeatedly instead of settling.

A governor with not enough damping can fail to slow down in time as it corrects its signal towards the target. Then it will overshoot, and have to bring the signal back. But then it may overshoot again. 

If there is any delay in feedback, where the governor is getting outdated information, it might keep making adjustments that are no longer needed, leading to overshooting and continuous corrections.

Oscillation is common because control systems often involve a tradeoff between speed and stability. If you want a fast response, you risk instability; if you dampen too much, you risk sluggish behavior.

Without getting too much into the weeds, just like depression can be caused by damage or malfunctions in different parts of your governors and selector, anxiety can be caused by damage or malfunctions in different parts of the ways that your governors are tuned, like their gain or damping, or by similar problems like delay in feedback.

Oscillation can also happen between two governors. It’s easiest to see this with an example. Let’s say that Danny’s hot and cold governors both have a problem where they have too much gain. So when he’s a little bit too cold, he does too much to correct it. He puts on socks and a sweater and gets a hot mug of tea and starts a fire in the fireplace. What happens now? Well, he soon becomes too warm. So he opens the windows and douses the fire and puts a fan on himself and strips down to his underwear. What next? Of course, he gets too cold. So it’s time to get warm again. He will keep oscillating until distracted.

You may even sometimes get oscillation inside a conflict. When two fine-tuned governors want mutually exclusive things, they will usually fight, putting out their more and more effort until they both reach their maximum output, and settle at a midpoint that is the balance between those two maximums. 

But if the governors are less well-tuned, they might oscillate. Danny’s fear and status governors are kind of deadlocked at work. He is afraid of his boss but he wants to crack jokes to impress his coworkers. A more “well-adjusted” man would be in a state of conflict. But Danny is poorly tuned. His status governor makes him crack a joke, and his fear governor is too slow to stop it from happening in time. His boss gives him a dirty look and Danny’s fear governor takes over. He shrinks down in his chair. But the fear subsides and soon he thinks of another joke. This is conflict, but it is also oscillation. 

Fear of the Future

A final thing we notice is that anxiety is often about the future. This might also be a kind of oscillation.

Consider Molly, a college student. Her parents and her community expect her to be a huge professional success (no pressure, Molly). Her status governor knows that it’s really important that she get a job when she graduates. If she doesn’t, her status governor faces a huge error, and since it can predict this, it wants to prevent it. But she has just started her senior year, so it’s not time to look for a job yet. The best thing she can do is focus on her studies.

This can lead to a weird cycle that looks kind of like a form of oscillation. She starts thinking about having to get a job. Her status governor leaps into action, panics, looks around for a way to start making a difference, but finds that there’s nothing it can do. Then it shuts off. But this can happen 100 times in an afternoon, and there’s not much she can do. There are no steps she can take to get a job now. She just has to wait. 

Because the governors are predictive, any promise of an extreme outcome can snipe you in this way. If your fear governor develops a fixation on car accidents, it might sometimes pipe up, “I predict we might get in a car accident. What can we do right now to make that less likely?” But you are in a work meeting, or at the grocery store. There’s nothing you can do at that moment to protect yourself from car crashes. But because the predicted error of a car crash is so huge (possibly death), your fear governor gets huge amounts of control over your attention and motivation when it makes this prediction. So for a while you are cowering in the cereal aisle, running through hypotheticals about dying in a 4-car pileup.

The ability to look at a hot stove, predict that it will burn you, and decide not to touch it, is a great adaptation. It’s why our governors are predictive — it’s great to be able to consider what will happen a few seconds in the future. But the human ability to look very far into the future is more of a mixed blessing. On the one hand, it means we can be motivated by things that may not happen for months. We can do long-term planning. But it also means we can be totally captured by far-off imaginary disasters (or imaginary blessings) that totally derail our ability to focus. 

You can lie awake in bed asking yourself, “will I be ready for my biology test on Friday?” The best thing to do, of course, would be to go to sleep. But the fatigue governor is being shouted down by the status governor, which is endlessly worried about failing the test. And there’s nothing you can do to make it quiet down. It is 2AM, there’s no way to prevent future status errors now.

Even worse is when you are taking concrete steps towards an outcome but none of your perceptions change. This is probably why founding a startup is so stressful. You work every day on your product, but there’s often no obvious change in your chances of success for weeks or even months. If you can get some metric like “number of users” that is constantly growing, that will help. But if not, you just have to keep plugging away and hope that you really are the next Google, or at least that you will be able to exit.

Or why dating can be so stressful. You can go on apps, go to events, meet people, go on first dates. But most of the actions you take don’t get you any closer to what you are trying to achieve. Each time you either meet the person or you don’t. 

This isn’t like most problems! When you are hungry, each apple or corn chip makes you slightly less hungry. When you are afraid, each step aways from the clown makes you slightly less afraid. But when you’re running a small business, most meetings cause no apparent change in your status or safety.

---

[Next: WHAT IS GOING ON?]

---