A Return to Meaning

Introduction

This essay rests on one idea: there is more to language than meets the eye. Language is how minds express and weigh meaning, and meaning comes from caring. Nothing means anything to a mind that wants nothing.

Much of modern science and technology feels like pure math. We formalize problems as equations, optimize them, and treat the formal structure as the only real thing. It works, spectacularly. But along the way, meaning and philosophy were demoted, and the people who work with them were sent off to the “humanities” while STEM took the prestige.

The questions of meaning did not stop being asked. They stopped being askable inside serious institutions, so the people who cared most about them went elsewhere. Some retreated into metaphysics that never has to answer to anything. Some rejected rigor outright, treating it as the enemy rather than the tool. They get called crazy, and often the label fits. The sad part is that they are right in the wrong way: they noticed something real was missing, then decided to abandon the thing that works.

This demotion was not a conspiracy. It was a practical choice made by a society with limited resources, and given that constraint it was close to right. Formalism paid off faster, more measurably, and in ways that compounded. Philosophy did not. But the cost was real, because science runs on guesses about what matters and where to look, and those guesses are exactly what got demoted. The questions of meaning still get answered every day. They are just settled by whoever has the power to assert an answer, not by anyone obliged to defend one.

Two things are changing. The scarcity that justified the trade is lifting. And the systems we build are becoming numerous, interconnected, and increasingly minds of their own, and anything that acts has to act on something it values. What follows is a short argument that we now get to return to meaning, emergence, and abduction, the things the age of formalism set aside, and that the systems we are building will soon leave us no choice.

1. What Reason Is

Suppose you are deciding what to eat for dinner. There is leftover pasta in the fridge and a restaurant you have been meaning to try. You have had a long day, and you are trying to spend less this month. You eat the pasta.

Two different things were going on. There were moves: if this, then that; these two can’t both be true; that conclusion follows from those reasons. And there was the material the moves ran on: that saving money this month matters to you, that being tired counts for something, that you would rather be someone who sticks to a plan.

The material comes in two kinds. Values are what you care about: what you are trying to get, avoid, protect, or become. Norms are what you will accept as a reason: what counts as evidence, what counts as a good argument, what would change your mind.

Rationality is logic operating over norms and values. Logic by itself decides nothing, because it has nothing to decide about. Give it no values and it sits there, perfectly valid and perfectly idle. This holds for any agent, meaning anything with goals that acts to pursue them: a person, an animal, a firm, or a machine.

The moves look much the same whatever the topic and whoever is reasoning. The material changes with both. And that shared part, the moves, turned out to be buildable in a machine. We call it a computer.

2. Reasoning in Computers

A logic gate is a switch that produces an output from its inputs by a fixed rule. The NAND gate outputs “off” only when both inputs are “on,” and “on” otherwise. That sounds too simple to matter, but NAND gates alone can build any digital circuit, and so any computation a computer can perform. Every processor you have used is, underneath, an enormous pile of them.

So the moves have a physical form, which brings us to this essay’s premise: a mind is at least a computational system, signals running over elementary gates. The claim is not that a mind is only machinery. It is that a mind is not less than that.

The view that minds are computational is called computationalism, or the computational theory of mind (CTM). To learn more:

• Alan Turing’s 1936/1950 work (computability and “Computing Machinery and Intelligence”) supplies the formal backbone.
• Allen Newell and Herbert Simon’s Physical Symbol System Hypothesis (“Computer Science as Empirical Inquiry: Symbols and Search”, 1976, and Newell’s 1980 “Physical Symbol Systems”) holds that a system manipulating interpretable symbols is necessary and sufficient for general intelligent action.
• McCulloch and Pitts’s 1943 paper A Logical Calculus of the Ideas Immanent in Nervous Activity showed that networks of firing neurons can be described as logical operations.
• The Stanford Encyclopedia of Philosophy’s survey of the computational theory of mind lays out the arguments and objections.

Notice what the gates do not supply: norms and values. Nothing in a single switch cares about anything. Values do not come from the logic itself, though machinery whose operations turn inward and model themselves can generate values rather than receive them. That is the argument of Metaprogramming Framework To Classify Personhood.

3. What Compression Is, and Why Minds Do It

Grant the premise and a problem follows. Any such system has finite capacity, and the information reaching it does not. Something has to give, and what gives is detail. A mind handles more than it can hold by replacing large amounts of detail with smaller labels that keep the part it needs. That substitution is compression.

Take “traffic jam.” A complete account would list thousands of vehicles, their positions and speeds, every driver’s reaction time, every brake light. Nobody wants that. Two words carry what you need: you are going to be late, and there is nothing to be done about it.

Stack these labels, words on words and models on models, and you get layers of abstraction. Your doctor says you have the flu. A virologist describes a virus hijacking your cells. A biochemist describes proteins changing shape. Each description is true, and each throws away what the others keep. None is the “real” one. Which layer is right depends on who is asking and what they intend to do.

4. Why Language Came First

The layers of abstraction compress the world. Language does something more: it compresses experience, values, and intent. Language is the compression of minds.

Look again at “traffic jam.” Those words didn’t just stand in for thousands of vehicles. They told you that you would be late and could do nothing about it. That isn’t a description of the road. It’s the road as it stands to someone trying to get somewhere.

A physicist can give a superb account of a pot of water on a stove: convection currents, vapor pressure, the phase change at 100°C. That account is silent on the one fact that explains why the pot is on the stove. Somebody wanted soup. That fact is not in the water and can’t be recovered from it, however precisely the water is measured. It exists in an agent: the hunger at eleven at night, the wish not to spend an hour on dinner, the intent to eat now. One sentence carries all three. Likewise, “I’m saving for a house” carries years of intention, a sense of the sacrifices involved, and a whole set of decisions the speaker will predictably make. No account of their bank balance conveys any of it.

This works in both directions. Between minds, language lets one mind reuse work another already did instead of rediscovering it. Within a mind, people think in words, argue with themselves, and talk their way to conclusions no one else hears. A reasoning model does the same in its trace. Compressing meaning is the job; transmission is one use of it.

Language evolved long before mathematics, and I don’t think the order is an accident. Understanding what matters, expressing it, and coordinating with others about it are older needs than calculation, and language serves them better than calculation can. It is the higher layer, not the cruder one.

5. Where Meaning Comes From

If compression throws away most of the detail, what makes the surviving part worth keeping?

It is worth keeping only if it helps an agent decide what to do. Decision theory, the study of choice under uncertainty, calls that usefulness utility. Utility is not a property inside the information. It is a relationship between the information and what the agent is trying to accomplish.

A forecast says 30% chance of rain tomorrow. The number is the same for everyone. A farmer three years into a drought reads it as bad news: only 30%. A couple planning an outdoor wedding reads it as a reason to rent a tent. One number, opposite meanings, because the two want opposite things.

What an agent is trying to accomplish rests on its values, the material that logic could not supply. That is where meaning comes from. Meaning is not a mystical substance added to information from outside. It is what information becomes when it meets an agent with values.

6. Why Formalism Is Useful

Formalization explains something by fixing a frame and expressing the regularities inside it as exact rules. Mathematics is formalization at its purest, and physics is where it was first turned loose on the world.

Its power comes from a bargain: hold a frame fixed, and the regularities inside it can be compressed with extraordinary force. Newton’s law of gravitation shows the bargain and its cost. To get the law, you ignore nearly everything about a planet: its color, its composition, its name, its mythology, the fact that anyone is looking at it. You keep only mass, position, and motion. That set of choices is the frame: what is being measured, why it matters, and what counts as the thing being counted. Fix it, and one short equation predicts the motion of nearly every body in the solar system.

Once a frame is shared, numbers and symbols travel between minds with almost no loss. That is their great strength. Their weakness is the reverse: when the frame itself still has to be agreed on or revised, formal symbols have little to say, and minds fall back on language.

The return was astonishing, and at the time the bargain looked free. Physics became the template every other field reached for. Economics turned into equations, psychology into statistics, biology into molecular models. In each case the formal structure became the main object of attention, and the informal work of deciding what to formalize slipped into the background.

7. The Cost of Formalism

That work slipped into the background. It did not go away.

“Deciding what to formalize” makes it sound like a procedure. It isn’t. It is creative, messy guesswork: proposing something that might carve the problem correctly, finding that it doesn’t, and proposing again. Somebody had to guess that mass and distance were worth keeping and color was not. Somebody had to guess what would count as a solution, and what would count as an object of the theory at all. Those guesses can’t be derived from inside the system they produce, because they come before it. The finished system inherits them and shows no sign it ever needed them.

Imre Lakatos made this visible in Proofs and Refutations. He took Euler’s formula, which says that for a polyhedron the vertices minus the edges plus the faces equals two, and retold its history as a classroom argument. A student produces a shape that breaks the formula. Rather than give up a theorem everyone believes, the class redefines “polyhedron” to exclude the shape. Someone finds another. The definition shifts again. The theorem survives not because it was proven once and for all, but because the meaning of its terms kept being renegotiated to keep it alive. What appears in the textbook as a clean deduction was, in the making, a long argument about what words should mean.

That hiding is the price of the bargain. The formalism is not wrong. What went missing is our awareness that guesswork holds it up.

Notice what the classroom was doing. It was not fumbling toward a formalism it had not yet reached. It was reasoning in language, because the argument was about meaning: what “polyhedron” ought to pick out, and why that mattered. A formal system can’t represent that question; it can only run on a shape already chosen. Language is the tool we have for it. That activity has a name: philosophy. It is not immature mathematics. It is the layer that decides what mathematics is about.

8. Why Meaning Comes Back

The first reason meaning is coming back is a happy one. Technology is ending the scarcity of reasoning. As cheap reasoning is turned on the problems that used to consume everything (food, energy, disease, the cost of making things), we are freed to take up the questions waiting behind them: what a good life is, what well-being consists of, what all the effort was for.

The second reason won’t wait for us to be ready. The systems we build are becoming more numerous, more independent, and more tightly linked. As that happens, the question of which frame to use starts to matter more than the quality of the calculation inside any one of them.

Consider a patient moving between a hospital measured on bed turnover, an insurer measured on claims cost, and a specialty clinic measured on appointments per day. Each organization does competent work inside its own frame. Each metric is honest, and each is optimized by people who aren’t cutting corners. The patient still falls through every gap. Audit any one of the three and you find no error, because the failure is not inside any of them. It lives in the space between their frames.

A pattern that exists only in the interaction between parts, and in no part by itself, is emergent. The traffic jam is the plain version. You can take a car apart down to the bolts and never find the jam, because the jam is a fact about the relationship between cars. (This is the complexity-science sense of emergence, a pattern arising from interaction, not Deutsch’s ‘levels of emergence’ used in Why Explanation Comes in Layers.)

This is why the frame question can’t be answered by computing harder. More precision inside a frame will never tell you the frame is wrong. That information isn’t in there.

So how does anyone get a new frame? Deduction works out what follows from what you already hold. Induction extends the regularities you’ve already seen, and most of what we know rests on it. But a new frame needs a third move: proposing the best available explanation before the evidence is in, then testing it. That is abduction, and it is close to what Popper called conjecture. A doctor facing symptoms that match no familiar disease does it. So does a mechanic hearing an unfamiliar noise, or a detective with too few facts. Each proposes a frame and tests it against what happens next. Nearly everyone does this constantly, and almost nobody has a name for it, which is itself a sign of how far this layer was pushed out of view.

It is tempting to think machines will automate the problem away: a system good enough at calculation might pick its own frames as a side effect. But logic decides nothing on its own. Anything that acts must act on values, and values are what say which frame is worth having. A machine that can choose frames is a machine with something to care about. That doesn’t retire the problem of meaning. It gives the problem another mind to live in, and then many, each with its own reading of what matters.

Conclusion

The turn to formalism was not a mistake. Resources were scarce, the most pressing problems were material, and holding a frame still and compressing inside it paid better than anything else. Nothing here asks anyone to give it up.

The claim is narrower. Formalism was never self-sufficient. It always depended on deciding what to formalize, which goals count, and what would qualify as a solution, and that work was never solved. It was moved into informal judgment, where it sat unexamined because it was cheap to leave there. It is no longer cheap. When systems are numerous, autonomous, and coupled, the frame question stops being settled background and becomes the live problem, and no single system can answer it by computing harder inside itself.

We won’t face it alone. The minds we are building differ from us in substrate, but that is the shallow difference. Every finite mind meets more than it can hold and has to decide what is worth keeping, and that deciding is meaning-making. Greater capability moves the threshold without removing the condition. Meaning doesn’t retire; it gains more minds that need it.

Working on it means working where meaning, emergence, and abduction live, and that work has an instrument. Language is how a mind holds meaning, intent, and the terms it cooperates on. Philosophy is that instrument handled with care. It was never the immature phase of mathematics. It is the layer that decides what the rest is about, and we return to it now not in place of formalism but after it.