The 7 Levels of Logical Thinking

The 7 Levels of Logical Thinking

I watched a video essay by the philosopher Joe Folly called The 7 Levels of Logical Thinking expecting a listicle. It's not one. It's a map of how "what makes an argument good" turns, rung by rung, into research most trained philosophers can't read either. I've spent enough time arguing with strangers online to recognize every early rung personally, and enough time around actual mathematicians to know I will never touch the last two.

Here's the ladder, and my honest opinion of which rungs are worth your time.

One thing first: don't confuse this with intelligence. Someone stuck at level one can still be sharp, well-read, hard to fool in practice. What they're missing isn't IQ — it's a vocabulary for why an argument works or fails. That's a narrower gap than it sounds, and it closes fast once you start climbing.

1. Pre-logic

Everyone starts here: a gut sense that "you're a bad person, therefore you're wrong" doesn't add up, with no ability to say why it doesn't add up. That's it. That's the whole level. Most people never leave it, and most days that's completely fine.

2. The fallacy monger

You learn the names — ad hominem, straw man, appeal to authority — and suddenly the internet looks full of people committing crimes against reason. Aristotle built this list for a reason: bad arguments are often the most persuasive ones, because they exploit shortcuts nobody's named yet.

But here's where I part ways with the fallacy-hunters a little. Someone citing a physicist over a random commenter on quantum mechanics gets flagged as "appeal to authority" constantly, and it's usually a bad call by the person doing the flagging. Deferring to earned expertise, when you genuinely can't verify the claim yourself, isn't a logical crime — it's just reasoning under uncertainty. Most things called "logical fallacies" online aren't structural errors at all. They're informal fallacies, which means you can't spot them from shape alone — you have to know the specific claim and the specific context. Someone's character is irrelevant to whether their math proof checks out. It's entirely relevant to whether they'd make an honest mayor. Same move, opposite verdict, because the context changed underneath it.

This is also, not coincidentally, the level where people get insufferable. I say that having been insufferable at this level myself.

3. Basic formal logic

This is the one worth actually stopping for.

Propositional logic swaps "that sounds off" for a testable structure. Propositions get letters — P, Q, R. Connectives (and, or, if-then, not) get exact truth conditions instead of vibes. "If P then Q; P; therefore Q" — modus ponens — either holds by structure or it doesn't, and your opinion of the conclusion is irrelevant to whether it holds.

The conditional is where it gets genuinely weird. "If P then Q" is false only when P is true and Q is false — which means "if cats can fly, then 1+1=2" counts as technically true, since cats can't fly and the antecedent never triggers. Philosophers call this "vacuously true," and it's bothered people since the ancient Stoics, because it feels completely disconnected from how the words work in a normal sentence. Formal logic keeps the weird definition anyway, because the alternative — requiring some "relevant connection" between the two halves — is nearly impossible to define mathematically. That trade-off, precision over intuition, is basically the theme of everything above this level.

What you get for the trouble: validity (if the premises are true, the conclusion can't be false) and soundness (valid, and the premises are actually true). Two tools, maybe five hours of focused study, and you can diagnose exactly what's wrong with almost any argument you'll hear this year. I'd push back slightly on the video's claim that this covers "99% of everyday reasoning" — it's closer to 90%, because a decent chunk of real disputes turn on disputed premises, not structure, and propositional logic has nothing to say about which premises are true. Still: rung three is the last one that pays for itself in daily life.

4. First-order logic and friends

Past whole propositions, you start quantifying: all, some, every. "All cakes are delicious" becomes "for all X, if X is a cake, X is delicious" — clunky to say out loud, precise on paper, and finally capable of handling categories and exceptions instead of single flat facts. Basic set theory rides along here too (union, intersection, complement map cleanly onto or/and/not), plus basic probability theory, which is really just level 2's "defer to the physicist" instinct wearing a number instead of a gut feeling.

Useful if you're headed into philosophy, math, computer science, or linguistics. Genuinely optional otherwise. This is the fork in the road: recreational logic mostly ends here, and professional logic begins.

5. Specialized systems

Past level four, "logic" stops being one subject. Modal logic formalizes necessity and possibility through Kripke models — think of it as a cluster of possible worlds connected by lines, and something counts as necessary if it's true in every world you can reach from where you're standing. Tweak the same machinery slightly and you get doxastic and epistemic logic, formalizing belief and knowledge instead — which is how this stuff ends up in computer science and game theory, of all places.

Model theory does something different: it strips a mathematical structure down to its bare minimum symbols (integer arithmetic reduces to (Z, +, ×, 0, 1)) and forces a distinction between syntax — the symbols — and semantics — what they actually refer to. That distinction sounds pedantic until you try to ask whether a formal system is being honest about the thing it claims to describe.

And advanced set theory proves something that still doesn't sit right in my head: the set of all fractions is the exact same size as the set of whole numbers, despite there being infinitely many fractions crammed between any two whole numbers. It's provable. It still feels wrong every time I revisit it.

A linguist working in formal semantics and a mathematician doing model theory both say "I study logic," and they'd struggle to follow each other's papers. That's the honest state of level five: not one subject anymore, several dialects wearing the same name.

6. Metamathematics and metalogic

Here logic turns around and studies itself. Does a formal system contradict itself? Does it prove only true things (soundness)? Does it prove all true things (completeness)?

Gödel's incompleteness theorems live here, and they're probably the most misquoted result in mathematics. The actual claim: no consistent, sufficiently powerful formal system can prove every true statement about arithmetic, and no such system can prove its own consistency. The claim you'll hear secondhand: "Gödel proved logic is broken." It didn't. Arithmetic is fine. That secondhand version is a level-2 fallacy-monger move wearing a level-6 costume, and it deserves the same skepticism.

Set theory has its own version: the continuum hypothesis — is there an infinity strictly between the size of the naturals and the size of the reals? — was proven independent of standard set theory (ZFC). Not unsolved. Independent. The axioms simply don't have an opinion, and never will.

Second-order logic sits in the middle of an unresolved fight here too: it can express things first-order logic can't (the induction principle itself, for one), but the extra expressive power comes with formal machinery heavy enough that plenty of logicians think the trade isn't worth it. I don't have a strong view on who's right. I mention it mostly because "more expressive" sounding like a pure upgrade is exactly the kind of thing that stops being true once you're the one doing the proofs.

7. Current research

A recent published abstract in a formal logic journal characterizes "ultrahomogeneity in the class of colored D-sets" and classifies certain structures as "monadically NIP." I read it twice. I still don't know what it means, and neither, honestly, does most of the field outside that sub-specialty — which tells you something about how far a discipline can drift from its founding question. Researchers here often hold joint appointments across philosophy, math, and computer science departments, because at this altitude the subject stops being any one of those things.

Aristotle wanted logic to be topic-neutral: a tool that applies to anything. The frontier doesn't fully live up to that, but the ambition survives. There aren't many corners where linguistics, computer science, philosophy, and pure math keep genuinely talking to each other. This is one.

Seven levels, and my actual advice after sitting with all of them: learn the fallacies at level two, but don't let them make you insufferable. Learn propositional logic at level three properly, because it's the only rung that pays rent on the five hours it costs. Everything past level four is a real, fascinating, legitimately hard field — just don't mistake how deep the hole goes for a reason you personally need to climb into it.

A caveat on the ladder itself: this is a pattern in how logic tends to get taught and practiced, not a law everyone follows in order. Some people skip rungs. Some loop back to level two forever. Treat it as a map, not a schedule.