[epistemic status: speculative]


Millgram et al (2015) find that depressed people prefer to listen to sad rather than happy music. This matches personal experience; when I’m feeling down, I also prefer sad music. But why? Try setting aside all your internal human knowledge: wouldn’t it make more sense for sad people to listen to happy music, to cheer themselves up?

A later study asks depressed people why they do this. They say that sad music makes them feel better, because it’s more “relaxing” than happy music. They’re wrong. Other studies have shown that listening to sad music makes depressed people feel worse, just like you’d expect. And listening to happy music makes them feel better; they just won’t do it.

I prefer Millgram’s explanation: there’s something strange about depressed people’s mood regulation. They deliberately choose activities that push them into sadder rather than happier moods. This explains not just why they prefer sad music, but sad environments (eg staying in a dark room), sad activities (avoiding their friends and hobbies), and sad trains of thought (ruminating on their worst features and on everything wrong with their lives).

Why should this be?


Let’s review control theory, ie the theory of homeostasis and bodily set points.

Many of your body systems have set points. For example, your temperature set point is usually around 98.6 degrees F. If you’re out in the snow and get colder than 98.6, your body will kick in various heating mechanisms (like shivering) until it’s back at the set point. If you’re out in the desert and get hotter than 98.6, it will kick in cooling mechanisms (like sweating) until it’s back.

Your inner thermostat acts through both conscious and unconscious processes. The examples above - shivering and sweating - are mostly unconscious. The conscious process is that when your body goes too far below 98.6, it makes “your conscious mind” “feel” “cold”. That “incentivizes” “you”, the “conscious” “actor”, to do things like go indoors, or put on a jacket, or turn on your space heater. You can think of the feeling of coldness as the conscious projection of the wider homeostatic drive to become warmer.

Although specific set points (eg 98.6) are set by evolution, they’re not hard-coded. Master regulatory systems can change set points in response to changing demands. For example, when you get infected by a heat-sensitive pathogen, your immune system might choose to boil it away, and increase your inner thermostat to (let’s say) 102 (instead of 98.6). Now you have a fever.

The funny thing about fevers is that you feel cold. Someone with a fever shivers. They demand to cover themselves in blankets. All of this makes sense, right? Your inner thermostat notices you’re at 98.6, and that’s colder than the desired temperature of 102. So it activates unconscious regulatory processes (like shivering) and conscious regulatory processes (like making you feel cold). Since you consciously feel cold, you engage in heat-seeking behaviors. You cover yourself in blankets, or turn up the space heater. This seems paradoxical (why does someone with a fever, ie someone who is too hot, feel cold?!) but it’s perfectly logical from the control theory perspective.


I’ve previously argued that anorexia is another condition that only makes sense in the context of set points.

Just as your body has an inner thermostat regulating temperature, it has an inner “lipostat” regulating body weight. The lipostat is why you feel hungry when you haven’t eaten enough and full when you have. It’s why, after you overeat, you might fidget a lot (fidgeting burns calories) - or why, if you’re starving, you’ll get weak and tired and not move very much (staying in one place conserves calories).

In the modern era, people do get fat pretty often. I think of this more as a disorder of the lipostat, caused by damage from years of unhealthy eating, rather than a failure of it. In the short term, the lipostat does a great job preventing obesity, returning people back to the same weight even after 10,000+ calorie diets through extreme fidgeting and subsequent fasting. It’s only long-run exposure to whatever modern food is doing that messes with the factory settings.

I argue that anorexia is a lipostat disorder, where it’s set permanently low. I realize this conflicts with the many stories of people becoming anorexic for psychosocial reasons (eg they wanted to be a ballerina and their coach made fun of their weight), but I think the psychosocial reasons (and the subsequent extreme dieting) cause the lipostat to permanently re-set at a lower weight.

Why do I think this? Partly because those ballerinas report that even after they stop caring about ballet, and realize their anorexia is killing them, and really really want to gain weight, they can’t. They no longer feel hungry. The thought of eating a normal-sized meal feels as disgusting to them as the thought of eating twenty pounds of steak in one sitting feels to me. But also, because studies find anorexic people fidget more - their unconscious lipo-regulatory processes are activating to maintain their dangerously-thin weight! And also because the conscious processes are equally messed up: other studies find that anorexics seem to genuinely believe on some gut level they’re fatter than they are (eg they will unconsciously flinch when walking through a narrow doorway that a thin person could easily pass but a fat person couldn’t). Even when you convince them that this is irrational, they still “feel” fat on a gut level.

(also, you can just directly cause anorexia - including subjective feelings of being too fat - by lesions in parts of the brain, and I assume some of these are in liporegulatory circuits)

Just to hammer in the analogy to fevers:

  • In a fever, you are dangerously warm, but instead of trying to cool down, you feel “driven” to perform behaviors that make you even warmer.

  • In anorexia, you are dangerously thin, but instead of trying to gain weight, you feel “driven” to perform behaviors that make you even thinner.

Now let’s take it all the way:

  • In depression, you are dangerously sad, but instead of trying to cheer up, you feel “driven” to perform behaviors that make you even sadder.


In anorexia, some psychosocial event (like criticism from a ballet coach and subsequent voluntary self-starvation) causes a shock to the lipostat. Instead of correctly activating regulatory processes to get body weight back to normal, it accepts the new level as its new set point, and tries to defend it.

Depression is often precipitated by some psychosocial event (like loss of a job, or the death of a loved one). It’s natural to feel sad for a little while after this. But instead of correctly activating regulatory processes to get mood back to normal, the body accepts the new level as its new set point, and tries to defend it.

By “defend it”, I mean that healthy people have a variety of mechanisms to stop being sad and get their mood back to a normal level. In depression, the patient appears to fight very hard to prevent mood getting back to a normal level. They stay in a dark room and avoid their friends. They even deliberately listen to sad music!

The feverish person feels too cold, and the anorexic person feels too fat, so we might expect the depressed person to feel too happy. I think something like this is true, if we put strong emphasis on the “too”. One of the official DSM symptoms of depression is “feelings of guilt/worthlessness”. A depressed person will frequently think things like “I don’t deserve my friends / job / money / talents.” In other words, they believe they’re too happy! They think they deserve to be sadder!

Depressed people seem to purposefully seek out the most depressing thoughts they can. They find that, unbidden, they are forced to think about the most humiliating thing they ever did, dwell on their worst failures, consider all the things that could go wrong in the future. They’ll be trying to cook dinner, and their brain will tell them “Consider the possibility that you could die alone and unloved.” Why is their brain so insistent that they spend time considering this possibility? Maybe it’s for the same reason that a feverish person’s brain makes them shiver: it’s trying to maintain an extreme state, and it needs to pull out all the stops.

We know that if we make depressed people stop doing these things, they feel happier. This is the principle behind behavioral activation, opposite action, and cognitive behavioral therapy, three of the most powerful therapies for depression. If you depression tells you to do something, do the opposite. Go on a nice walk in the park! Listen to happy music! Spend time with your friends! If you do these things, your depression is pretty likely to go away. The problem isn’t that they don’t work, the problem is that it’s like a feverish person trying to take an ice bath, or an anorexic trying to eat a big meal - all their instincts are telling them not to do it. And if your depression tries to get you to think in a specific way, think in a different way. When it tells you that you should still feel bad for that embarrassing thing you did in third grade, tell it that makes no sense, and that you’ve done plenty of things you’re proud of since then. Again, this often works if you do it. It’s just really hard.

Psychologists already suspect the existence of a happiness set point (thymostat?); this is the principle behind ideas like the “hedonic treadmill”. So my theory here is that at least some cases of depression involve recalibrated happiness set points. A set point can either recalibrate randomly (ie for poorly understood biological reasons) or after a specific shock (ie interpreting a prolonged period of sadness as “the new normal”). Once a patient has a new, lower, happiness set point, their control system works to defend it. It enlists both biological systems (possibly changing the levels of various neurotransmitters?) and behavioral systems to defend the new set point. If it “succeeds”, the person maintains an abnormally low mood.

Taking this theory seriously would suggest a research program focusing on some of the following points:

  1. Which other conditions seem like cases of miscalibrated set points? Some of these are obvious, eg primary polydipsia. Others are more questionable; can hypertension be considered a recalibration of blood pressure set point? Opiate addiction a recalibration of endorphin set point? I’m not sure. What would it mean, philosophically, to answer yes vs. no to these questions?

  2. Do any other conditions display a pattern of voluntary/natural/logical derangement of a parameter, followed by involuntary/surprising/fixed derangement of the same parameter (eg first voluntary weight loss to become attractive, then involuntary maintenance of low weight)? I’ve never heard of anyone giving themselves a primary polydipsia by voluntarily drinking more water (maybe because they heard that bogus eight glasses per day statistic) and then being unable to stop; if there were cases like this, it would lend this theory significant support. Is there some common factor that makes set points “looser” vs. “stickier” (hint: anxiety and trauma history)

  3. Miscalibrated set points seem to sometimes recalibrate themselves to correct values; depressed people usually recover, anorexic people sometimes regain normal-ish weight. Should this be thought of as the system “naturally righting itself”? What determines whether or not this happens? Is it the opposite of the original derangement process? IE if an obese person diets for long enough, does their lipostat eventually recalibrate to the diet as a new set point?

  4. Am I eliding some important differences in which these conditions are vs. aren’t ego-dystonic? Some anorexic people, when told that a treatment will help them gain weight, often refuse out of fear of becoming fat (though others are happy to accept treatment). Depressed people, when told a treatment will make them happy, very occasionally refuse on grounds that “I don’t deserve happiness”, but this is pretty rare; most of them are glad to accept the treatment. Is this just a quirk of how each of these different drives is implemented, or is it a strike against the theory?

I’ve previously endorsed predictive coding theories of depression and other illnesses. How does that interact with this perspective? This is even more speculative than the rest, and I don’t feel like I entirely get it, but here’s the completion my internal pattern-generator has spit out:

Many of the claims of predictive coding can be rephrased as claims about control theory, and vice versa. You have to slightly fudge things to make this work on homeostatic bodily processes, but this is the kind of fudging that Karl Friston has already worked into his free energy concept.

In predictive coding, the equivalent of control theory’s “set point” is the “prior”. This suggests an elegant equivalence: an incorrectly fixed set point, like those in anorexia and depression, are the same thing as a trapped prior.

Depression is a trapped prior on low mood, which can also be thought of as a thymostat set to low mood. From a cognitive point of view, you can think of this as a deranged prior leading to confirmation bias across thoughts and activities; from an enactive point of view, you can think of it as a control system maintaining a set point with effectively-regulatory thoughts and activities. The cognitive point of view is helpful when you’re thinking about cognitive half of CBT, and the enactive point of view is helpful when you’re thinking about the behavioral activation half of CBT.

Or when you’re trying to figure out why depressed people listen to blues music.