In a number of debates I've had in the past few years, I started to see a pattern in which I came to the same fundamental impasse with people again and again. It wasn't about disagreeing over facts, but about a semantic difference that I could not describe until around half a year ago (and have since found so daunting to write about that I've put it off for all those months.) The difference came up specifically in debates about things of enough complexity that we do not understand what drives its behavior on the inside, but have a good idea of how the more apparent and observable properties are related. The result was a constant battle of language games in which theories were seen as nonsensical because they were supposedly in contradiction to things that were much more apparent. How could carbs/genes/hormones/etc be responsible for obesity if the "real" cause was taking in too many calories? How could unemployment lead to less overall wealth if jobs are only a means to an end? If someone is depressed and behaving in self-destructive ways, why can't they simply choose to do something to help themselves? The problem is that none of these questions were dealing with things that were mutually exclusive; in all of these cases, they were dealing with two different types of theories.
The theories dealing with larger structures such as genetics, employment, and behavioral disorders are ones that I describe as cybernetic theories. Cybernetics is the study of how a system regulates its inputs and outputs in order to maintain stability. That can apply to something as simple as how a thermostat regulates its own parts, to how a human body regulates its metabolism, energy levels, and behavior in order to maintain homeostasis. Rather than looking at mere correlations between things that happen, it looks at the actual decision making of a system. But what makes a particular cybernetic theory? A cybernetic theory is a hypothesis that attempts to explain a mechanism by which a system's behavior can be predicted.
The more apparent causes that can be seen through observation are ones that I describe as phenomenological theories. In science, a phenomenological theory is "A theory that expresses mathematically the results of observed phenomena without paying detailed attention to their fundamental significance" (Thewlis, J. (Ed.) (1973). Concise Dictionary of Physics. Oxford: Pergamon Press, p. 248.) An example of this would be something like the fact that we can observe that an organism loses mass when it consumes less calories than it expends, and gains mass when it consumes more calories than it expends; we don't have to know why it's true to see that it is. One can also note that the prosperity of a nation is dependent not on abstract economic numbers, but on actual material wealth; fuel, food, infrastructure, etc. More abstract entities such as currency are things that help decide where resources are allocated and who gets what--so that is a cybernetic theory explaining how resources are acquired, distributed, and used; it doesn't make more resources pop out of the ground, but it does give people an incentive to look for resources that are in demand and helps prioritize who should get what resources. In the same way, everyone can agree that jobs are not an end in themselves (otherwise, it would just be useless work), but most of us see employment as an important number because if not enough people have jobs, it would require that we devise a completely new system for distributing wealth to people.
The theories dealing with larger structures such as genetics, employment, and behavioral disorders are ones that I describe as cybernetic theories. Cybernetics is the study of how a system regulates its inputs and outputs in order to maintain stability. That can apply to something as simple as how a thermostat regulates its own parts, to how a human body regulates its metabolism, energy levels, and behavior in order to maintain homeostasis. Rather than looking at mere correlations between things that happen, it looks at the actual decision making of a system. But what makes a particular cybernetic theory? A cybernetic theory is a hypothesis that attempts to explain a mechanism by which a system's behavior can be predicted.
The more apparent causes that can be seen through observation are ones that I describe as phenomenological theories. In science, a phenomenological theory is "A theory that expresses mathematically the results of observed phenomena without paying detailed attention to their fundamental significance" (Thewlis, J. (Ed.) (1973). Concise Dictionary of Physics. Oxford: Pergamon Press, p. 248.) An example of this would be something like the fact that we can observe that an organism loses mass when it consumes less calories than it expends, and gains mass when it consumes more calories than it expends; we don't have to know why it's true to see that it is. One can also note that the prosperity of a nation is dependent not on abstract economic numbers, but on actual material wealth; fuel, food, infrastructure, etc. More abstract entities such as currency are things that help decide where resources are allocated and who gets what--so that is a cybernetic theory explaining how resources are acquired, distributed, and used; it doesn't make more resources pop out of the ground, but it does give people an incentive to look for resources that are in demand and helps prioritize who should get what resources. In the same way, everyone can agree that jobs are not an end in themselves (otherwise, it would just be useless work), but most of us see employment as an important number because if not enough people have jobs, it would require that we devise a completely new system for distributing wealth to people.
| Domain | Cybernetic | Phenomenological |
| Nutrition | Hormones | Calories |
| Economics | Currency, Employment, Interest | Resources, Labor |
| Psychology | Pathology | Behavior |
Your Decisions vs. Your Body's Decisions
Now that I've gotten the general gist across, we can get into examples. In order to keep things clean, I'll only go into one: nutrition. This is where I've encountered endless language games in which many people make the ridiculous accusation that those who go beyond calories-in-calories-out are denying the rules of thermodynamics. Even among some of the smartest people I've read, I've seen this problem, such as a debate between Martin Berkhan of Leangains.com and Gary Taubes, author of Why We Get Fat, having an argument about the problems of overeating. Their views are mostly similar (though not entirely), but their biggest disagreements seem to largely come from arguments that are ultimately about semantics. Taubes says that overeating is not the true cause of obesity, but merely an inevitable side effect of the true cause, which is a bad diet. Berkhan responds by saying that you don't magically burn of all of the food if you eat more calories than you expend, but then says that the reason that dietary fat is less fattening is that fat is more satiating than carbohydrates. What Berkhan missed is that Taubes would agree--it's not that the calories magically disappear, it's that the amount of calories eaten is regulated by a mechanism that responds differently to carbohydrates than it does to fat. I would personally add that not only is that case, but that a good diet and a healthy body mean that the excess energy in your body is more easily accessible, and so you will not only have an easier time burning it, but will be naturally inclined to do it. Body fat is a battery, and obesity occurs when the body keeps charging the battery but not using any of it.*
In both cases, they're actually agreeing about the cybernetics of this: in both cases, eating more fat and protein and less carbohydrates leads to the decision to consume less calories; what we experience as hunger and satiety are expressions of more fundamental mechanisms that interact in order to regulate the system's decisions; the most central of these being hormones. Hormones in our body act as messengers and end up deciding how hungry we feel, where calories in our bodies go, how physically restless or restful we feel, and so on. If food is the natural resource base of our body, then metabolism is the web of economic links, with hormones perhaps acting as our financial and monetary system (interestingly, I believe that there is an analogue between the hormone insulin and the effect of interest rates on economies--a topic I'll briefly revisit later in this post.) At the same time, Taubes does not deny that there is an absolute correlation between calorie surplus and weight gain--the difference is that he rightfully points out that nobody is answering the question of why this calorie surplus is happening:
We don’t get fat because we overeat; we overeat because we’re getting fat.
Taubes, Gary (2010-12-28). Why We Get Fat: And What to Do About It (Kindle Locations 1431-1432). Knopf Doubleday Publishing Group. Kindle Edition.
This seems like gobbledygook, or at least weird wording, when one first reads it, but the logic is actually simple: obesity is the condition upon which the body decides to overeat and allocate the excess calories to fat. Sounds implausible? Then consider this: kids run a constant calorie surplus because their body is telling them that they need to grow--it's not like they consciously deciding to grow. In both cases, the word decision is key--obesity, just like growing, is a cybernetic phenomenon in which the body is accumulating calories because that's what it decided to do. A more detailed explanation of both what is believed to happen and the scientific evidence backing it up is a topic fit for entire books, so I can't go into it here, but Taubes is a great place to start. What's important to note here is that whether we end up running a surplus or deficit of calories is a decision made by the body.
But what about self-control? That's an important question, and it makes this post controversial because the first thing I have to say is, no, you don't have total autonomy over what your body does. Yes, you can use your conscious will to keep calories under a certain level, but will it work? If you are not taking in enough energy to get through a workout, then you'll become more sedentary in response. In fact, in severe cases of metabolic syndrome, the condition that causes obesity, starvation may cause the body to break down muscle, bone, and even organ mass before burning through all of its fat reserves. Why would the body do that? This requires understanding the essence of cybernetic systems: feedback.
This seems like gobbledygook, or at least weird wording, when one first reads it, but the logic is actually simple: obesity is the condition upon which the body decides to overeat and allocate the excess calories to fat. Sounds implausible? Then consider this: kids run a constant calorie surplus because their body is telling them that they need to grow--it's not like they consciously deciding to grow. In both cases, the word decision is key--obesity, just like growing, is a cybernetic phenomenon in which the body is accumulating calories because that's what it decided to do. A more detailed explanation of both what is believed to happen and the scientific evidence backing it up is a topic fit for entire books, so I can't go into it here, but Taubes is a great place to start. What's important to note here is that whether we end up running a surplus or deficit of calories is a decision made by the body.
But what about self-control? That's an important question, and it makes this post controversial because the first thing I have to say is, no, you don't have total autonomy over what your body does. Yes, you can use your conscious will to keep calories under a certain level, but will it work? If you are not taking in enough energy to get through a workout, then you'll become more sedentary in response. In fact, in severe cases of metabolic syndrome, the condition that causes obesity, starvation may cause the body to break down muscle, bone, and even organ mass before burning through all of its fat reserves. Why would the body do that? This requires understanding the essence of cybernetic systems: feedback.
Feedback or Die
The most basic example used to demonstrate cybernetics is a thermostat. It has a built in thermometer, which consists of mercury that either expands or contracts due to changes in temperature. This allows the thermostat to measure some discrepancy between its target temperature and the actual temperature outside--it will then turn on a heating or cooling system until the discrepancy goes away. This kind of feedback is known as negative feedback because the feedback causes the discrepancy to shrink. What's important to note here is that the information received by a cybernetic system is based on some difference between two absolutes--the thermostat behaves the way it does because it makes its decision based on whether the volume of the mercury inside its thermometer is less than, greater than, or roughly the same as some defined target.
The human body, while operating on these same principles, is much more complex in its rules. That said, there are still insights that can be gleaned from having a rough idea of how some of its key systems work. One such system is the hormone known as insulin. (NB: from here on, I am making a theoretical point with examples that may not exactly match up with the most up-to-date scientific theories. The point of the following is a thought experiment meant to give an intuitive sketch of how feedback works in a cybernetic system. I repeat, I am not making an empirical claim, I am using a simplification in order to illustrate a concept.) Insulin is a hormone that is charged with the task of absorbing any glucose that is found in the bloodstream and transporting it to various parts of the body (namely fat and muscle.) The fat cells and muscle cells that absorb the insulin do so by means of insulin receptors, which calibrate their sensitivity such that they absorb a certain amount of insulin before stopping.
These cells are very much like the thermostat, except that their target will be raised or lowered based on the relative amount of insulin running through the system. The reason for this is that the body's goal is to properly distribute nutrients and this distribution is determined by the insulin sensitivity of various parts of the body. If a receptor is receiving too much insulin, its sensitivity reduces so as not to take in more than it needs. Currency works like this as well: if an excess of money is flowing through the system but the amount of actual wealth (yes, loaded term, but bear with me) stays the same, then the purchasing power of the currency drops. Insulin works the same way: just as currency represents a non-fixed amount of wealth, insulin represents a non-fixed amount of nutrients.
The condition known as insulin resistance comes when these receptors become so insensitive that they are no longer absorbing any significant amount of insulin. The result of this is that the insulin, and any glucose that it might be transporting, remains circulating in the bloodstream. In order to get the glucose out of the bloodstream, more insulin is produced. In theory, this should be okay; eventually the same amount of glucose is being transported around, it just needs more total insulin to represent it. The same goes with money--100 years ago a dollar was worth a lot more, but nothing has broken down because it was gradual enough that at any given moment people had a stable sense of their purchasing power and there was enough time for wages to rise accordingly (it's not this simple, but my point stands that the system did not collapse.) In other words, everything is fine if enough of the system can recognize that everything is the same except that the yardstick has changed.
When the change happens too rapidly, however, the yardstick gets mismatched with reality and inefficient behaviors arise; and in extreme cases, the yardstick can become entirely useless. In economics, the former case matches up to the phenomenon of deflation, in which the purchasing power of money has increased due to lower prices, but unemployment results due to wages not lowering nearly as fast (Keynes called this "sticky wages".) In the case of hyperinflation, prices rise so fast that the currency is no longer a reliable yardstick, and any information the money represented about who owns what is vanished. While this may sound like an egalitarian's dream, the problem is that so many vital systems rely on this information that the result is terrible poverty.
But how do these breakdowns occur? What would make a discrepancy emerge so quickly and grow so fast that it can't be compensated for? The answer is positive feedback: where negative feedback closes a gap, positive feedback increases it. And since the positive feedback increases the gap, it's likely that the same behavior will repeat because the gap is still there. Although not all positive feedback is necessarily bad, systems break when they enter some cycle of positive feedback that they can't get out of. In the case of deflation, the unemployment caused by falling prices and lower wages means that people will spend even less. The result? Prices drop even further and more people are put out of work. Whether or not bailouts and stimulus packages are a good idea, their intent is to nip the cycle in the bud while it's still affordable to do so. In the case of metabolism, the issue is that once the cells are too insensitive to insulin, the body will produce even more massive amounts of insulin in order to compensate, but this will inevitably lower the insulin sensitivity of the already resistant receptors. This can go two ways: the insulin secretion eventually outpaces the receptors' reduction in insulin sensitivity, in which case some stable point is reached; sadly, this often happens through the body creating new fat cells and eventually becoming obese enough to stabilize the situation. For those who were wondering why the body would make a decision in which fat absorbs the lion's share of nutrients to the detriment of everything else, now you know: (relatively) insulin-sensitive fat cells have been recruited to help keep excess glucose out of the blood-stream; they are the nouveau riche of your metabolic system. The second way is much less pretty: insulin stops being secreted for good, the yardstick is gone; this is diabetes. From then on, insulin must be regulated through artificial means (injecting insulin manually whenever a meal is just eaten.)
The big jump to make is to realize that any sufficiently complex entity requires reliable feedback. All of the materials in the world are useless if they cannot work together to create the necessary complex behavior. If the feedback becomes too unreliable, the behavior becomes at best unpredictable, and at worst too incoherent for anything to work properly. The loss of the ability to produce insulin is the loss of an entire feedback mechanism, and the only reason that diabetes does not guarantee death is that humans have enough metacognition to use conscious regulation as a backup system to regulate glucose manually. But take that away, and a more fundamental point becomes clear: a system's health and survival depend on the integrity of its feedback.
Leveraged Phenomenology
This is not to say, however, that phenomenology is useless. On the contrary, it is actually essential to sound decision-making: the truth is that everything I've written above about insulin is an oversimplification of a very complex theory that even with all of its details and nuances cannot fully account for the complexity of the human body. But then why look at cybernetic theories at all? If we're interested in weight loss and we ultimately can only rely on phenomenological theories, wouldn't it just be best to look at calorie intake and expenditure?
Not so fast; there's is one caveat that has not been stated here: you don't have direct control over your calorie intake. It's not just that you don't directly control what goes to muscle and what goes to fat; your actual behavior with regards to diet and exercise is largely dependent on the messages of your metabolism. Too much sugar intake will most definitely affect your levels of hunger and your body's ability to process nutrients efficiently. Whereas the target of a thermostat is something we have complete control over (we just turn the dial), there is no equivalent part of our body that we have such direct control over. This doesn't, however, mean that we have no control; instead, we have differing degrees of control over different inputs.
Since different inputs allow different amounts of control, we need to go by the phenomenological theories that provide us the greatest degree of leverage. Calorie counting, when it works, works because we make the decision to eat foods that give us more bang for our buck. While it might be phenomenologically true that we'll lose weight should we take in less calories than we expend, this is something that on its own does not provide us very much in the way of leverage at all. On the other hand, there is much phenomenological evidence to show that cutting out certain types of food or engaging in intense exercise sessions a couple of times per week also does the same thing--and these are inputs over which we have a much greater degree of control. But how do we know what will provide us leverage and what won't? The answer is simple: for any input, get an idea of how dependent it is on feedback from the system's prior behavior. The more feedback-dependence, the less direct control.
Without understanding cybernetic theories, we would not be equipped to see this difference. Cybernetic theories offer us the ability to see how different phenomena are related through cascades of feedback and consequently allow us to see what phenomenological theories provide us the most control over future outcomes. But the example I've given here only scratches the surface--however powerful a framework cybernetics is for appreciating complex decision making, it is virtually impossible to decode the entirety of something as complex as the human body, let alone the world economy (which is likely even more complex due to the fact that financial numbers have no theoretical limit.) It goes without saying that this greatly complicates what began as something that felt simple--but the goal of this entry was to clear up a language game that hinders further inquiry into these ideas; and as such, I'll have to leave countless questions that I haven't even mentioned for another time.
*If I've misremembered or misphrased this argument in any way, please let me know. I have no interest in putting words in anyone's mouth.
The most basic example used to demonstrate cybernetics is a thermostat. It has a built in thermometer, which consists of mercury that either expands or contracts due to changes in temperature. This allows the thermostat to measure some discrepancy between its target temperature and the actual temperature outside--it will then turn on a heating or cooling system until the discrepancy goes away. This kind of feedback is known as negative feedback because the feedback causes the discrepancy to shrink. What's important to note here is that the information received by a cybernetic system is based on some difference between two absolutes--the thermostat behaves the way it does because it makes its decision based on whether the volume of the mercury inside its thermometer is less than, greater than, or roughly the same as some defined target.
The human body, while operating on these same principles, is much more complex in its rules. That said, there are still insights that can be gleaned from having a rough idea of how some of its key systems work. One such system is the hormone known as insulin. (NB: from here on, I am making a theoretical point with examples that may not exactly match up with the most up-to-date scientific theories. The point of the following is a thought experiment meant to give an intuitive sketch of how feedback works in a cybernetic system. I repeat, I am not making an empirical claim, I am using a simplification in order to illustrate a concept.) Insulin is a hormone that is charged with the task of absorbing any glucose that is found in the bloodstream and transporting it to various parts of the body (namely fat and muscle.) The fat cells and muscle cells that absorb the insulin do so by means of insulin receptors, which calibrate their sensitivity such that they absorb a certain amount of insulin before stopping.
These cells are very much like the thermostat, except that their target will be raised or lowered based on the relative amount of insulin running through the system. The reason for this is that the body's goal is to properly distribute nutrients and this distribution is determined by the insulin sensitivity of various parts of the body. If a receptor is receiving too much insulin, its sensitivity reduces so as not to take in more than it needs. Currency works like this as well: if an excess of money is flowing through the system but the amount of actual wealth (yes, loaded term, but bear with me) stays the same, then the purchasing power of the currency drops. Insulin works the same way: just as currency represents a non-fixed amount of wealth, insulin represents a non-fixed amount of nutrients.
The condition known as insulin resistance comes when these receptors become so insensitive that they are no longer absorbing any significant amount of insulin. The result of this is that the insulin, and any glucose that it might be transporting, remains circulating in the bloodstream. In order to get the glucose out of the bloodstream, more insulin is produced. In theory, this should be okay; eventually the same amount of glucose is being transported around, it just needs more total insulin to represent it. The same goes with money--100 years ago a dollar was worth a lot more, but nothing has broken down because it was gradual enough that at any given moment people had a stable sense of their purchasing power and there was enough time for wages to rise accordingly (it's not this simple, but my point stands that the system did not collapse.) In other words, everything is fine if enough of the system can recognize that everything is the same except that the yardstick has changed.
When the change happens too rapidly, however, the yardstick gets mismatched with reality and inefficient behaviors arise; and in extreme cases, the yardstick can become entirely useless. In economics, the former case matches up to the phenomenon of deflation, in which the purchasing power of money has increased due to lower prices, but unemployment results due to wages not lowering nearly as fast (Keynes called this "sticky wages".) In the case of hyperinflation, prices rise so fast that the currency is no longer a reliable yardstick, and any information the money represented about who owns what is vanished. While this may sound like an egalitarian's dream, the problem is that so many vital systems rely on this information that the result is terrible poverty.
But how do these breakdowns occur? What would make a discrepancy emerge so quickly and grow so fast that it can't be compensated for? The answer is positive feedback: where negative feedback closes a gap, positive feedback increases it. And since the positive feedback increases the gap, it's likely that the same behavior will repeat because the gap is still there. Although not all positive feedback is necessarily bad, systems break when they enter some cycle of positive feedback that they can't get out of. In the case of deflation, the unemployment caused by falling prices and lower wages means that people will spend even less. The result? Prices drop even further and more people are put out of work. Whether or not bailouts and stimulus packages are a good idea, their intent is to nip the cycle in the bud while it's still affordable to do so. In the case of metabolism, the issue is that once the cells are too insensitive to insulin, the body will produce even more massive amounts of insulin in order to compensate, but this will inevitably lower the insulin sensitivity of the already resistant receptors. This can go two ways: the insulin secretion eventually outpaces the receptors' reduction in insulin sensitivity, in which case some stable point is reached; sadly, this often happens through the body creating new fat cells and eventually becoming obese enough to stabilize the situation. For those who were wondering why the body would make a decision in which fat absorbs the lion's share of nutrients to the detriment of everything else, now you know: (relatively) insulin-sensitive fat cells have been recruited to help keep excess glucose out of the blood-stream; they are the nouveau riche of your metabolic system. The second way is much less pretty: insulin stops being secreted for good, the yardstick is gone; this is diabetes. From then on, insulin must be regulated through artificial means (injecting insulin manually whenever a meal is just eaten.)
The big jump to make is to realize that any sufficiently complex entity requires reliable feedback. All of the materials in the world are useless if they cannot work together to create the necessary complex behavior. If the feedback becomes too unreliable, the behavior becomes at best unpredictable, and at worst too incoherent for anything to work properly. The loss of the ability to produce insulin is the loss of an entire feedback mechanism, and the only reason that diabetes does not guarantee death is that humans have enough metacognition to use conscious regulation as a backup system to regulate glucose manually. But take that away, and a more fundamental point becomes clear: a system's health and survival depend on the integrity of its feedback.
Leveraged Phenomenology
This is not to say, however, that phenomenology is useless. On the contrary, it is actually essential to sound decision-making: the truth is that everything I've written above about insulin is an oversimplification of a very complex theory that even with all of its details and nuances cannot fully account for the complexity of the human body. But then why look at cybernetic theories at all? If we're interested in weight loss and we ultimately can only rely on phenomenological theories, wouldn't it just be best to look at calorie intake and expenditure?
Not so fast; there's is one caveat that has not been stated here: you don't have direct control over your calorie intake. It's not just that you don't directly control what goes to muscle and what goes to fat; your actual behavior with regards to diet and exercise is largely dependent on the messages of your metabolism. Too much sugar intake will most definitely affect your levels of hunger and your body's ability to process nutrients efficiently. Whereas the target of a thermostat is something we have complete control over (we just turn the dial), there is no equivalent part of our body that we have such direct control over. This doesn't, however, mean that we have no control; instead, we have differing degrees of control over different inputs.
Since different inputs allow different amounts of control, we need to go by the phenomenological theories that provide us the greatest degree of leverage. Calorie counting, when it works, works because we make the decision to eat foods that give us more bang for our buck. While it might be phenomenologically true that we'll lose weight should we take in less calories than we expend, this is something that on its own does not provide us very much in the way of leverage at all. On the other hand, there is much phenomenological evidence to show that cutting out certain types of food or engaging in intense exercise sessions a couple of times per week also does the same thing--and these are inputs over which we have a much greater degree of control. But how do we know what will provide us leverage and what won't? The answer is simple: for any input, get an idea of how dependent it is on feedback from the system's prior behavior. The more feedback-dependence, the less direct control.
Without understanding cybernetic theories, we would not be equipped to see this difference. Cybernetic theories offer us the ability to see how different phenomena are related through cascades of feedback and consequently allow us to see what phenomenological theories provide us the most control over future outcomes. But the example I've given here only scratches the surface--however powerful a framework cybernetics is for appreciating complex decision making, it is virtually impossible to decode the entirety of something as complex as the human body, let alone the world economy (which is likely even more complex due to the fact that financial numbers have no theoretical limit.) It goes without saying that this greatly complicates what began as something that felt simple--but the goal of this entry was to clear up a language game that hinders further inquiry into these ideas; and as such, I'll have to leave countless questions that I haven't even mentioned for another time.
*If I've misremembered or misphrased this argument in any way, please let me know. I have no interest in putting words in anyone's mouth.
Thanks! I really enjoyed this post (which says very well many things I have been struggling to say badly for several years now).
ReplyDeleteGlad to hear it--I had trouble articulating this idea for a long time myself. I have a vague idea of who you might be, but I'm guessing that you might want anonymity (sometimes I wonder if I should do the same thing myself.)
DeleteThis is a very interesting way of looking at theories, Alex. Phenomenological theories are the theory of choice for tinkerers.
ReplyDelete