What Is a Scientific Theory Really Made Of?
What Goes Into a Theory? A Three-Way Fight
Imagine you’re staring at your science fair project about pendulums. You’ve got a string, a weight, and a stopwatch. You’ve measured how long it takes to swing back and forth. You even wrote down a formula: the period depends on the length of the string, not the mass. But here’s a question you might not have asked: what is the theory behind that formula made of? Is it just the equation and the logical rules that connect it to other ideas? Is it the mathematical model that pictures all possible swings? Or is it something much messier—a blend of pictures, comparisons, values, and even sticky notes full of hunches?
For almost a century, philosophers of science have argued about three answers. Each one is like a different pair of glasses for looking at the same piece of science. Put them on, and a theory suddenly looks like a new kind of thing.
The Syntactic View: Science as a Perfect Language
In the early 1900s, a group of thinkers including Rudolf Carnap (1891–1970) and Carl Hempel (1905–1997) wanted to make science as clear as a grammar book. They believed a scientific theory is really a syntactic structure—a set of sentences written in a special logical language, like a language you’d find in a mathematics textbook but even more precise.
Under this view, every theory has two kinds of words. Theoretical terms name things you can’t see directly, like “electron,” “force,” or “gene.” Observational terms name things you can point to, like “the red liquid in the beaker” or “the pendulum swings left.” The theory itself is a pyramid of sentences. At the top, you have axioms—the basic, unprovable starting rules. Below them come theorems and laws, all built using logic. The whole thing is written in a metamathematical language, often first‑order predicate logic with symbols like ∀ (“for all”) and ∃ (“there exists”).
But how does all this logic touch the real world? The Syntactic View adds a special glue: correspondence rules. These sentences connect theoretical terms to things you can measure. For example, a correspondence rule might say, “The symbol T in Boyle’s law stands for the temperature shown on a properly calibrated thermometer.” Without that link, the theory would just be a beautiful string of symbols floating in the air.
An early triumph of this view was Hans Hermes’s attempt to define the mass ratio of two colliding particles using a long logical formula—a definition stripped down to bare syntax. To a syntactic thinker, that’s what the deep structure of science looks like.
The Semantic View: Mathematical Maps of Reality
By the 1960s, many philosophers began to grumble that the Syntactic View was too picky about language. Patrick Suppes (1922–2014) argued that trying to cram all of physics into a single logical language was “utterly impractical.” Why trap mathematics in a narrow grammar when you could just study the math directly?
The Semantic View says a scientific theory isn’t a set of sentences. Instead, it’s a class of mathematical models. A model here doesn’t mean a plastic volcano; it’s an abstract structure that might be a state space—an imagined N‑dimensional space where each point represents a possible state of a system.
Take a capsule drifting in deep space. To describe its motion, you could create a 9‑dimensional space: three dimensions for position, three for velocity, three for acceleration. Every possible journey the capsule could take—every swoop and curve—is a path through that space. The laws of physics are the rules that say which paths are allowed. Under the Semantic View, the theory is that collection of all allowed paths, not the symbols you wrote to describe them.
Another approach within the Semantic View uses model theory, a branch of mathematical logic. Here, a theory is a set of axioms, and a model is any structure that makes those axioms true. A tiny geometric figure of seven points and seven lines can be a model of the axioms for a projective plane. The theory lives in the relationship between axioms and models, not in the ink on the page.
Bas van Fraassen (born 1941) famously said that a theory is empirically adequate if the messy, “nasty, brutish, and short” phenomena we observe can be embedded—mapped perfectly—onto a neat corner of a mathematical model. On this view, science doesn’t have to tell a story in words; it builds a map of possibilities.
The Pragmatic View: Messy, Creative, and Full of Hidden Ingredients
Not everyone was satisfied with either the sentence‑builders or the map‑drawers. Around the 1980s, a third family of ideas began to take shape: the Pragmatic View. Its central claim is that a theory is much more than its logical or mathematical skeleton.
Nancy Cartwright (born 1944) pointed out that abstract laws like Newton’s F = ma rarely work all by themselves. To explain a real situation—say, the orbit of Mars—you need to combine many laws, add simplifying assumptions, and often use analogies. The theory of superconductivity, for example, was shaped by the London brothers’ comparison of a superconductor to a perfect diamagnet. That analogy wasn’t just a colorful illustration; it was part of the theory’s structure.
Cartwright and others view theories as nomological machines: arrangements of parts that, in a stable environment, produce regular behavior. But most of the world isn’t a tidy machine. When you apply a theory, you bring in non‑mathematical ingredients: metaphors, policy views, values, even the specific history of how the theory was built. In this picture, a theory is more like a recipe that calls for both exact measurements and a chef’s intuition.
The Pragmatic View also insists that theories come in many forms. A mechanistic theory in neuroscience might look like a flow chart. A historical theory in evolutionary biology might be a branching tree of species. Trying to force every theory into one mold—logical or mathematical—misses what scientists actually do.
One Idea, Three Lenses: The Hardy‑Weinberg Principle
To see how these views play out, let’s look at a real piece of science: the Hardy‑Weinberg Principle (HWP) in population genetics. In its simplest form, HWP says that in a large, randomly mating population with no evolution happening, the frequencies of genes stay the same generation after generation. It acts like a null model—a baseline for detecting when evolution is at work.
A Syntactic philosopher might ask: can we reduce all the concepts—allele, fitness, mating—to a crystal‑clear axiomatic system in first‑order logic? Woodger (1894–1981) attempted to rebuild Mendelian genetics with logical predicates, and HWP would be a theorem flowing from those axioms. The project is about finding the deepest grammatical structure of genetic theory.
A Semantic philosopher would treat HWP as a mathematical structure—a phase space of allele frequencies. The equation (p + q)² = p² + 2pq + q² describes a family of possible states. The interesting work is exploring how this model relates to other models (selection, drift) within the larger theory of population genetics.
A Pragmatic philosopher zooms in on how HWP is actually used. In studies of human genetic diversity, HWP appears in two very different projects. Diversity partitioning uses HWP to calculate how much genetic variation falls within versus between human groups—famously showing that only about 5% of variation is between continents, a result often taken to undermine the idea of biological race. Clustering analysis also relies on HWP, but it has been used to sort people into approximate continental clusters, sometimes offered as a limited defense of race. The same mathematical principle, the same null model, acquires different meanings depending on the purposes and values of the researchers. The Pragmatic View insists we can’t understand the theory without that context.
Why This Fight Matters for Your Next Science Project
You don’t need to be a historian or a logician to feel this debate. Every time you build a model for a school project, you’re facing the same puzzle. Is your explanation just the equation, and you’re done? Or do you also need to draw the state space, showing all the what‑ifs? Do you need to name the hidden analogies—like “this population grows like bacteria in a petri dish”—and think about what assumptions come with them?
The three views aren’t just dusty philosophy. If you see a theory as a pure set of sentences, you might miss a hidden flaw in your logic. If you see it only as a mathematical model, you might forget the values and purposes that shaped which model you built. And if you embrace the messiness, you might find that your theory is richer and more honest—but harder to defend in a single, clean sentence.
Scientists and philosophers are still arguing about which view gets closest to the truth. But knowing all three gives you a sharper eye. The next time you watch a pendulum swing or read about genes, you can ask: what kind of thing am I really holding in my head?
Think about it
- If you had to explain the movement of a playground swing to a friend, would you start with a mathematical model, a logical definition, or a story that includes comparisons like “it’s like a clock”? Which feels most complete, and what might it leave out?
- A biology textbook says a population is “in Hardy‑Weinberg equilibrium.” Does that describe a fact about the world, a useful fiction, or a tool that does different jobs for different scientists? How would you decide?
- Imagine someone claims that a theory is simply the collection of its mathematical equations. Can you think of a scientific idea—maybe from climate change or medicine—where the non‑mathematical parts (values, history, analogies) seem just as important as the math?
