Predetermined Variables

The way we understand the world often hinges on a simple question: what do we change, and what do we watch? This fundamental act of distinguishing between different types of variables is the bedrock of scientific inquiry. While the concepts of independent and dependent variables are familiar, modern science requires a more sophisticated toolkit to grapple with complex systems that evolve over time and respond to expectations about the future. This article addresses this need by delving into the crucial distinction between 'predetermined' variables, which are bound by the past, and 'jump' variables, which react instantaneously to new information. Across two main chapters, you will explore the theoretical foundations of this classification and witness its profound impact. The first chapter, "Principles and Mechanisms," will unpack the core concepts, from the basics of experimental design to the mathematical formalisms that govern stability in dynamic systems. Following this, "Applications and Interdisciplinary Connections" will demonstrate how these ideas provide a unifying lens to analyze causality and stability in fields as diverse as economics, ecology, and even the quantum nature of reality itself.

It’s a funny thing about science. We often imagine it as a passive observation of the world, like watching a film that’s already been made. But the truth is, the very act of doing science is more like directing a play. The scientist chooses which part of the scenery to light up, which actor to give a new line to, and then watches, with bated breath, how the rest of the cast reacts. This simple idea—the distinction between the things we change and the things we watch—is one of the most powerful in all of science. It’s the difference between an independent variable and a dependent one, a concept that starts simply but leads us to some of the most profound ideas in physics, chemistry, and even economics.

Let's imagine you're a scientist testing a new pill, A-734, that’s supposed to boost cognitive performance. You gather a group of students and divide them. One group gets the real pill; the other gets a sugar pill, a placebo. You are the director of this little drama. Your one, deliberate action is deciding who gets which pill. This choice—the presence or absence of compound A-734—is the ​​independent variable​​. It is independent because you, the experimenter, have set its value. It doesn't depend on anything else in the experiment.

Now, you sit back and watch. What happens? You measure the students' scores on a logic test after a month. This score is the ​​dependent variable​​. It's called dependent because its value, you hypothesize, depends on the independent variable. Did the students who took A-734 score higher? The entire experiment is a carefully constructed question: how does this dependent variable respond to the change we made in the independent variable? Everything else, from the duration of the study to the type of logic test used, is held constant. These are the ​​control variables​​, the stage setting that we try not to disturb. The beauty of a well-designed experiment lies in this isolation: one knob to turn (the independent variable), and one gauge to watch (the dependent variable).

This director-and-actor relationship isn’t just for lab coats; it's the fundamental grammar of the mathematical language we use to describe the universe. When a bioengineer models the population of yeast in a vat, they write down differential equations. They might have one equation for how the yeast population, $P$, changes over time, and another for how the nutrient concentration, $N$, changes over time:

$\frac{dP}{dt} = \dots \quad \text{and} \quad \frac{dN}{dt} = \dots$

Look closely at these equations. The "d-by-d-something" in the denominator tells you everything. Here, both derivatives are with respect to time, $t$. Time marches on relentlessly, not caring about the yeast or the nutrients. Time is the ultimate independent variable. The yeast population and the nutrient level, however, are constantly reacting to the passage of time and to each other. They are the dependent variables. The answer to "what is the independent variable?" in a dynamic system is often found by asking, "what variable are we taking the derivative with respect to?"

The number of independent variables defines the very nature of the problem. Consider the shape of a simple chain hanging between two posts under gravity. Its elegant curve, a catenary, is described by a function $y(x)$, where its vertical height $y$ depends on only one independent variable: the horizontal position $x$. Because there is only one, the equation describing its shape is an ​​Ordinary Differential Equation (ODE)​​.

But now imagine we are mapping the metabolic activity in a living brain using a PET scanner. The signal, let's call it $C$, isn't just a line; it's a value that changes at every single point in three-dimensional space ($x$, $y$, and $z$) and also changes from moment to moment in time ($t$). The signal is a function $C(x, y, z, t)$. It has four independent variables. An equation describing this field would be a ​​Partial Differential Equation (PDE)​​, a much more complex beast, because the signal can have different behaviors in different directions. The number of independent variables isn't just a detail; it sets the dimensionality of the world we are trying to describe.

Here’s where it gets really interesting. Is the designation of a variable as "independent" an absolute, God-given truth? Not at all. Often, it's a strategic choice made by the scientist for convenience or insight. It's a choice of perspective, like choosing the best camera angle to film a scene.

In thermodynamics, the internal energy $U$ of a gas is naturally expressed as a function of its entropy $S$ and its volume $V$. The fundamental equation is $dU = TdS - PdV$. In this view, $S$ and $V$ are the independent variables. But if you’re a chemist in a lab, controlling the volume of a gas can be a pain. It's much easier to control its pressure, $P$. So, can we switch our perspective? Can we treat $P$ as an independent variable instead of $V$?

Yes, we can! Through a beautiful mathematical technique called a ​​Legendre transform​​, we can define a new quantity. In this case, we define enthalpy as $H = U + PV$. A quick bit of calculus shows that the change in this new quantity is $dH = TdS + VdP$. Look at that! The natural variables for our new quantity, enthalpy, are now entropy $S$ and pressure $P$. We have successfully swapped a variable we didn't want to control ($V$) for one we do ($P$). We changed the independent variable to better suit our experimental reality.

This idea has revolutionary consequences. In medicine, for decades, doctors focused on blood pH (the concentration of hydrogen ions, [$\text{H}^+$]) as the primary variable to manage in acid-base disorders. The Stewart model of physiology turned this completely on its head. It argued that [$\text{H}^+$] is actually a ​​dependent variable​​. It's not a knob the body turns directly. Instead, the body independently controls three other parameters:

1. The ​​Strong Ion Difference (SID)​​: The electrical charge difference between all the fully dissociated ions, like $\text{Na}^+$ and $\text{Cl}^-$.
2. The total amount of weak acids ($A_{\text{tot}}$), mainly proteins like albumin.
3. The partial pressure of carbon dioxide ($P_{CO_2}$), controlled by breathing.

The body manipulates these three independent variables, and the laws of chemistry and electroneutrality force [$\text{H}^+$] to simply fall into line. A patient develops acidosis not because their body "produces acid" in a vacuum, but because, for instance, kidney failure causes a buildup of strong anions, which decreases the SID. This change in perspective, from treating [$\text{H}^+$] as the cause to seeing it as the effect, has fundamentally changed how doctors in critical care approach treatment. It all came from correctly identifying which variables are the true directors of the play.

As we move to systems that evolve over time and involve expectations about the future, like an economy, the distinction gets even more subtle and powerful. We now split our variables into two new camps: ​​predetermined​​ and ​​jump​​ variables.

The stage for this idea was actually set in quantum physics. In the famous EPR paradox, the question was whether the properties of an entangled particle are "real" before they are measured. A local hidden variable theory proposes that the result of any measurement is, in fact, ​​predetermined​​ from the moment the particle is created. If two particles fly apart with a total momentum of zero, and we measure the first to have momentum $\vec{p}_1$, then the momentum of the second must have been predetermined to be $-\vec{p}_1$ all along. Its fate was sealed at its creation by some "hidden variable". Quantum mechanics, of course, argues the opposite: the outcome is genuinely random, decided only at the instant of measurement. But this debate gives us a perfect intuitive picture of "predetermined": is the answer already written down, or is it decided on the spot?

This is exactly the distinction we need in economics. Think about the capital stock of a country—all its factories, machines, and infrastructure. The amount of capital we have today, $k_t$, is the result of how much we had yesterday, $k_{t-1}$, plus the investment we made yesterday, $i_{t-1}$, minus depreciation. Today's capital stock is a historical fact. It cannot change in an instant, no matter what surprising news comes out today. It is a ​​predetermined variable​​.

Now think about the price of a stock. If a company announces a revolutionary new invention, does its stock price wait until tomorrow to go up? Of course not. It adjusts instantaneously. It "jumps" to a new value the moment the information becomes public. Stock prices, exchange rates, and other such forward-looking prices are ​​jump variables​​. They are not tied to the past; they are determined by expectations of the future, and they can change in a flash.

"So what?" you might ask. "Why this obsession with categorizing variables?" The answer is profound: this classification is the key to understanding the stability of the world around us. Why do economies not spiral into hyperinflation or collapse into depression at the slightest disturbance?

The logic, formalized in what are known as the ​​Blanchard-Kahn conditions​​, is as elegant as it is powerful. Imagine a dynamic system has certain intrinsic tendencies. Some of these tendencies are explosive, pushing the system away from equilibrium (these correspond to "unstable eigenvalues"). Others are stabilizing, pulling the system back towards equilibrium ("stable eigenvalues").

The system's only hope for staying on a stable path is to use its jump variables. A jump variable is like an emergency lever. Because it can be chosen freely at any moment, it can be set to the exact value needed to perfectly counteract an explosive force. And so, we arrive at a beautiful balancing principle: for a system to have a single, unique, stable path forward, ​​the number of jump variables must be exactly equal to the number of unstable tendencies​​.

If there are more unstable tendencies than jump variables, there aren't enough levers to pull. The system is fundamentally unstable and will eventually explode. No stable equilibrium exists. If there are more jump variables than unstable tendencies, there are many different ways to keep the system stable. The future path is not uniquely determined; it's anyone's guess. Only when the two numbers match perfectly does the system have one, and only one, clear, stable path forward.

This principle, born from a simple distinction between variables, governs the behavior of everything from currency markets to climate models. It shows us that the world is a delicate dance between the unchangeable facts of the past and the instantaneous choices that shape the future. And it all begins with the simple act of asking: who is the director, and who is the actor?

In our previous discussion, we dissected the abstract machinery separating the world into two kinds of quantities: those that are "predetermined," carrying the inertia of the past, and those that can "jump," unbound, in anticipation of the future. This might have seemed like a formal, perhaps even sterile, classification. An accountant's exercise in sorting variables. But now, we are ready to see this idea in action. And you will find that this simple distinction is one of the most powerful lenses we have, allowing us to peer into the hidden workings of systems as diverse as computer circuits, lake ecosystems, financial markets, and the very fabric of reality itself. It is not just a classification; it is a key that unlocks a deeper understanding of cause, stability, and fate.

Let's begin with something concrete: a computer program or a digital circuit. The function of such a a system is, in a sense, to render a verdict—a final output based on a series of inputs. Now, suppose we want to make this circuit faster, more efficient. One of the first questions an engineer asks is, "Can we simplify this?" Imagine a complex Boolean function that depends on several variables. We might represent this function with a data structure called a Reduced Ordered Binary Decision Diagram (ROBDD). By simply inspecting the structure of this diagram, we can sometimes see that, along certain logical paths, an entire variable is missing. It never gets tested. What does this mean? It means the function's output is independent of that variable under those conditions. Its value was irrelevant to the final outcome. In this context, recognizing that a function is independent of a variable is recognizing a form of predetermination—the result is already set, regardless of what that particular switch does. This isn't just an academic curiosity; it's the heart of compiler optimization and efficient hardware design, saving energy and computational time by not bothering to ask questions to which the answer doesn't matter.

This principle of "independence" becomes even more crucial when we can't control the switches ourselves. Consider the plight of an ecologist studying a lake. They see a complex dance: the shimmering green biomass of phytoplankton, the nearly invisible zooplankton that graze upon it, and the fish that hunt the zooplankton. They want to answer a fundamental question: What controls the life in this lake? Is it "bottom-up" control, where the amount of nutrients determines everything? Or is it "top-down" control, where the number of fish dictates the entire food chain's structure through a cascade of effects?

The problem is that everything seems to affect everything else. More phytoplankton feeds more zooplankton, but more zooplankton eats more phytoplankton. It's a dizzying feedback loop. How can we untangle this knot of causality? Here, the idea of predetermined variables comes to the rescue, but under a different name: exogenous instruments. The ecologist recognizes that the total nutrient concentration in the water is driven by external factors, like runoff from the surrounding land, and is not meaningfully affected by the plankton in the short term. Likewise, the population of plankton-eating fish is largely predetermined by its own slow life cycle and is not immediately sensitive to daily fluctuations in their food source.

These two variables—nutrients ($N$) and fish ($F$)—act as our predetermined anchors in a sea of feedback. By observing how the system changes in lakes with different (predetermined) nutrient levels, and in lakes with different (predetermined) fish populations, the scientist can statistically isolate the effects. The nutrient level provides a clean "push" on the phytoplankton, while the fish population provides a clean "push" on the zooplankton. This allows the ecologist to measure the strength of the reciprocal relationship between phytoplankton and zooplankton without being confounded by the feedback loop. This powerful idea, formalized in Structural Equation Models, is what allows us to infer cause and effect from mere observation, and it all hinges on finding those variables that are, from the perspective of the interaction we care about, predetermined.

Nowhere does the distinction between predetermined and jump variables shine more brightly than in economics. This is because, unlike molecules in a gas, human beings have expectations. We live our lives looking forward, and our actions today are based on our beliefs about tomorrow. This forward-looking behavior gives rise to "jump" variables.

Think about your morning commute. The number of cars already on the highway when you leave your house is a fact. It's a legacy of the past few minutes. It is a classic ​​predetermined variable​​. But your decision to take that highway, and the decisions of thousands of others, is a ​​jump variable​​. It depends on your expectation of the traffic you will face. If everyone expects the highway to be clear, they all jump on it, and it instantly becomes congested. If everyone expects a jam, they jump to side streets, and the highway becomes clear. The stability of this entire system—whether it settles into a predictable pattern or lurches between gridlock and open road—depends critically on the interplay between the predetermined stock of cars and the forward-looking choices of commuters.

This same logic applies to far more consequential matters, like an international arms race. One nation's existing stockpile of weapons is a predetermined state variable, a hard fact inherited from past budgets. However, its current military spending is a jump variable, driven by its expectation of the rival's future actions. A model of such a system must treat these two variables, stock and spending, fundamentally differently. The Blanchard-Kahn conditions, a cornerstone of modern macroeconomics, formalize this. They state that for a dynamic system involving rational expectations to have a single, stable path into the future (a "unique equilibrium"), there must be a perfect balance: the number of unstable modes in the system's dynamics must exactly match the number of forward-looking jump variables. It's as if each jump variable is needed to "tame" an explosive root, steering the economy onto its lone stable trajectory.

The genius of this framework is its flexibility. What if a system has built-in delays? For instance, an investment in a new factory decided today might only result in a productive capital stock two years from now. The model at first seems to defy the first-order structure needed for the analysis. But we can be clever. We can define an auxiliary variable, say, "construction projects in the pipeline," which is predetermined today and becomes actual capital tomorrow. By adding these new, carefully defined state variables, we can transform a complex, higher-order problem into the standard form and once again analyze its stability by counting our variables and our eigenvalues.

But what happens when the conditions are not met? What if there are fewer unstable modes than jump variables? The system no longer has a single, unique path forward. It has an infinitude of them. This condition, called "indeterminacy," means that self-fulfilling prophecies are possible. And this brings us to the fascinating topic of rational bubbles. An asset price, like a stock, is partly based on its "fundamentals"—the expected future dividends. But could its price also contain a bubble component, a value that exists only because everyone believes it will exist? The model of predetermined and jump variables gives us a surprising answer: yes. If a system is indeterminate, it has "room" for a bubble. The bubble can be a perfectly rational phenomenon that exists by feeding on itself. Its existence is a direct consequence of a structural imbalance between the system's inherent dynamics and the number of forward-looking variables within it. Our framework doesn't just explain stability; it also explains the possibility of seemingly unstable phenomena like market manias.

We have traveled from circuits to lakes to markets. Now we take the final, most audacious step: to the nature of reality itself. We ask a question that vexed Einstein for his entire life: Is the universe deterministic? When we perform a measurement on a particle, say a quantum bit or "qubit," is the outcome we get the revelation of a pre-existing property? Or does the outcome spring into existence at the moment of measurement?

The first view, known as "local realism," supposes that every particle carries around a set of "hidden variables" that predetermine the outcome of any measurement we might choose to make. The particle has a definite spin, we just don't know it until we look. This is the ultimate "predetermined variable" theory. Quantum mechanics, on the other hand, suggests this is not so.

How can we decide? The physicist John Bell was the first to show that this philosophical debate could be settled by experiment. Later, brilliant "All-vs-Nothing" arguments made the contradiction even starker. Imagine we have four entangled particles, shared among four physicists, each ready to measure a property like spin along the $x$ or $y$ axis. Quantum mechanics makes exquisitely precise predictions for the correlations between their measurements. For example, it might predict that if they perform one specific set of measurements, the product of their outcomes will always be $-1$. For a second set, it's also $-1$. For a third, it's also $-1$ again.

Now, let's play the realist's game. Let's assume there are predetermined values, $v_{k, i} \in \{-1, +1\}$, for the outcome of each possible measurement. To agree with quantum mechanics, these hypothetical values must obey the relationships quantum theory predicts. But here the trap springs. If we take the first three quantum predictions and multiply them together, simple algebra on these assumed predetermined values shows that the outcome of a fourth set of measurements must also be $-1$. It is a logical necessity flowing from our assumption of predetermination.

Yet, when we turn to the rulebook of quantum mechanics, it predicts with utter certainty that the outcome of this fourth measurement will be $+1$. And every experiment that has ever been done confirms it. The prediction from the local, predetermined variable assumption (product = -1) and the prediction of quantum theory (product = +1) are in flat contradiction. It's not a matter of being slightly off; it's a clash between yes and no, black and white. All or Nothing.

The conclusion is staggering. The universe, at its most intimate level, does not seem to run on predetermined variables. The properties of a particle are not written down in some hidden instruction set before we measure them. The very act of measurement plays a role in creating the reality it purports to uncover. The simple idea of a predetermined variable, so useful in engineering and economics, seems to break down when we push it to the ultimate limit, revealing a universe more strange and more wonderful than we could have imagined.

From the practical logic of a computer chip to the profound mystery of the quantum world, the journey of this one idea—the distinction between what is fixed and what is free—reveals a deep and unifying thread running through the tapestry of science.