Why some conclusions are guaranteed and others are just really good guesses
Last time we learned the difference between an argument and a fight. An argument has premises leading to a conclusion. A fight has emotions leading to exhaustion.
Congratulations. You now know more about reasoning than most people who argue on the internet for a living.
Here’s the next piece. Not all arguments work the same way.
The logical bridge connecting premises to conclusion – philosophers call this the inference – comes in two very different types. Understanding which type you’re dealing with changes everything about how you evaluate what you’re hearing.
Any other recovering philosophy majors out there? Please beware of the following, as it might induce PTSD.
Sylvan Barnet and Hugo Bedau spent decades teaching this stuff at Tufts University. Barnet was a Shakespearean scholar who believed clear thinking and clear writing were inseparable. Bedau was a philosopher who spent his career fighting against the death penalty by showing, over and over, that the arguments for capital punishment simply didn’t hold up under logical scrutiny.
Together they wrote “From Critical Thinking to Argument,” a textbook that has introduced millions of students to the difference between deductive and inductive reasoning.
Here’s the short version. Deductive reasoning gives you certainty. If your premises are true, your conclusion must be true. No exceptions. Inductive reasoning gives you probability. If your premises are true, your conclusion is probably true. Maybe.
That difference matters more than you might think.
Deduction works from general principles to specific conclusions. The classic example: All humans are mortal. Socrates is a human. Therefore, Socrates is mortal.
If you accept those first two statements, you’re logically forced to accept the third. There’s no wiggle room. You can’t say “Well, maybe Socrates isn’t mortal” if you’ve already agreed that all humans are mortal and Socrates is a human. The structure of the argument locks you in.
Induction works in the opposite direction – from specific observations to general conclusions. Every swan I’ve ever seen is white. Therefore, all swans are white.
This seems reasonable. It was reasonable for centuries. Then Europeans showed up in Australia and discovered black swans. Oops.
This is the problem with induction. No matter how many observations you pile up, you can’t guarantee your conclusion. You might have seen ten thousand white swans. You might have seen a million. There’s still nothing logically preventing swan number one million and one from being purple.
So why use induction? Because we have to.
Most of what we know about the world comes from inductive reasoning. Science runs on it. Every time a researcher observes patterns and forms a hypothesis, that’s induction. Every time a study finds a correlation and suggests a cause, that’s induction.
We’re all just making educated guesses based on accumulated evidence.
This isn’t a flaw in science. It’s a feature. Induction is how humans figure out general principles from specific observations. But here’s where this connects to everything we’ve discussed in this series.
Remember System 1 from Essay 1? The fast, automatic part of your brain that jumps to conclusions based on pattern-matching? System 1 loves induction. It sees a few examples and immediately forms a general rule. Three Democrats do something annoying and System 1 concludes all Democrats are annoying. A handful of news stories feature crime by immigrants and System 1 concludes immigrants are criminals.
This is induction running on autopilot without any quality controls.
The whole point of learning about inference types is to slow down and ask the right questions. When someone presents a deductive argument, you check: Are the premises true? Is the logic valid? If both answers are yes, you’re stuck with the conclusion whether you like it or not.
When someone presents an inductive argument, you ask different questions: How much evidence supports this? How representative is that evidence? What would disprove this conclusion?
Political debate gets mangled because people treat inductive arguments like they’re deductive. “I saw three stories about voter fraud, therefore elections are rigged” uses the form of an argument but makes a massive logical leap. The evidence doesn’t come close to supporting that conclusion. An honest version would be: “I saw three stories about voter fraud, therefore some voter fraud exists somewhere.” Much less exciting. Much more accurate.
Here’s Bedau’s insight, earned through decades of arguing against the death penalty. Bad reasoning isn’t always obvious. It hides behind confident rhetoric and emotional appeal. The only defense is learning to identify what type of inference someone is making and whether the evidence actually supports it.
Deduction gives certainty but requires bulletproof premises.
Induction gives useful knowledge about the real world but requires epistemic humility – you might be wrong, and you should remain open to new evidence.
Neither type is better. They’re tools for different jobs. The mistake is using the wrong tool or pretending one type is the other.
As I have said before, skepticism should be your default about everything – that’s what turns System 2 on in the first place. People often want to point out things that I have posted that they feel make me untrustworthy – but that’s also a feature – not a bug. If you have to stop and think about something more thoroughly than you normally would, then this page is working as planned.
Here’s the uncomfortable truth I keep coming back to in this series. The issues we’re facing are more existential than most people are willing to admit. Getting rid of Donald Trump, or MAGA, or Republicans won’t really solve the problems you think they will. Because the reasoning failures we’ve been discussing for twelve essays now aren’t confined to one party. They’re human failures. System 1 doesn’t check your voter registration before hijacking your judgment.
Learning the difference between deduction and induction won’t save democracy by itself. But it’s one more tool for slowing down, asking better questions, and maybe – just maybe – having conversations that go somewhere other than exhaustion.