Can You Use Logic When You’re Not Sure?

The Picnic Problem: Why Certainty Is a Luxury

Imagine you’re planning a picnic. Your weather app says there’s a 70 percent chance of rain. You think: If it rains, I’ll get soaked. There might be rain. So… should I bring an umbrella? You just used two kinds of thinking at once. You used logic — the “If A then B” pattern — and you used a number that measures uncertainty, a probability. Most everyday decisions mix these two things. But for a long time, philosophers and mathematicians treated them as completely separate worlds.

Logic was supposed to be about airtight, 100-percent-certain reasoning. Probability was for gambling, weather, and guessing. This article is about what happens when you try to bring them together. It turns out they fit better than anyone expected — and the rules that emerge can help you think clearly even when you’re not totally sure.

Classical Logic’s Iron Rule: Truth Must Not Leak

To see why mixing logic and chance is tricky, start with classical logic. An argument is a list of starting claims (premises) and a final claim (conclusion). The argument is deductively valid if it’s impossible for all the premises to be true while the conclusion is false. Think of it as a promise: if you somehow know the premises are real, the conclusion is guaranteed.

For example:

• Premise 1: If the alarm clock rings, I will be late.
• Premise 2: The alarm clock rings.
• Conclusion: I will be late.

If both premises are true, the conclusion cannot be false. This is truth preservation — the truth of the premises cannot leak away; it must carry into the conclusion. The pattern works no matter what the sentences are about. Classical logic studies those patterns.

But here’s the problem. In real life, we are rarely certain about our premises. The alarm might not ring. We’re pretty sure it will, but not 100 percent. Classical logic has nothing to say about “pretty sure.” It’s all or nothing. So philosophers asked: can we build a logic that keeps the strong structure of classical arguments but replaces truth with degrees of certainty?

Turning Certainty into Numbers: The Probability Function

To build such a system, we first need a way to turn ideas like “I’m 90 percent sure” into precise numbers. This is the job of a probability function. It assigns a number from 0 (impossible) to 1 (certain) to every statement in our language. The rules are simple:

• No statement gets a negative number. The smallest is 0.
• If a statement is a tautology — something true no matter what, like “either it rains or it doesn’t” — it gets probability 1.
• If two statements can’t both be true at the same time (for example, “it rains” and “it doesn’t rain” at the same instant), then the probability that one or the other happens is just the sum of their separate probabilities.

These rules echo classical logic: tautologies still get top certainty, and the “or” of two incompatible claims behaves additively. Notice that classical truth values (0 and 1) are just the extreme ends of this system. So probability logic doesn’t throw logic away — it stretches it out along a number line.

Now we can ask: given an argument that is classically valid, what happens to the conclusion’s probability when the premises are not certain? The first answer is simple and satisfying. If every premise is completely certain (probability 1), then the conclusion must also be certain (probability 1). The classical promise still holds when you use probabilities. But things get more interesting when the premises are less than certain.

Adams’s Discovery: Uncertainty Adds Up

The philosopher E. W. Adams (1926–2004) wondered exactly how much uncertainty can travel from premises to conclusion. Instead of working with probabilities directly, Adams focused on uncertainty, defined as 1 minus the probability. If you’re 90 percent sure, your uncertainty is 0.1.

Adams proved a striking rule. Take any valid argument with a finite number of premises. The uncertainty of the conclusion is never greater than the sum of the uncertainties of all the premises. In other words, doubt piles up, but only additively.

Let’s try it. Suppose the alarm‑clock argument from earlier is valid. You think premise 1 (“if the alarm rings, I’ll be late”) is 90 percent sure — uncertainty 0.1. Premise 2 (“the alarm rings”) is also 90 percent sure — again uncertainty 0.1. Adams’s rule says the uncertainty of the conclusion cannot exceed 0.1 + 0.1 = 0.2. Therefore the conclusion (“I will be late”) must have probability at least 0.8, or 80 percent. Even though you’re not completely sure about either premise separately, you can still put a lower bound on how much you should trust the conclusion.

This is powerful. It means that logic still works as a kind of safety net: your conclusion can only be as shaky as your reasons, and no shakier than all their shakiness combined. The rule is even tight — in some cases you can design a situation where the conclusion’s uncertainty exactly matches that sum. So you can’t get a better guarantee without extra information.

Cutting Out the Noise: The Role of Irrelevant Premises

Adams’s sum rule has a weakness. Suppose you add a completely irrelevant premise to the argument — a random fact that doesn’t affect the conclusion at all. For example, you tack on “It is Tuesday” to the alarm‑clock argument. The argument stays valid, because validity only cares about the logical structure. But now your sum of uncertainties includes the uncertainty of “It is Tuesday,” which could be quite high (maybe you’re only 70 percent sure what day it is). The extra premise needlessly pushes up the worst‑case bound on the conclusion’s uncertainty, which seems unfair.

To fix this, Adams and others introduced the notion of essentialness. Not all premises are equally important. A premise is essential if removing it makes the argument invalid. Some premises work only in teams: together they are essential, but individually you could leave one out. The degree of essentialness of a premise measures how indispensable it really is.

Using degrees of essentialness as weights, you get a refined bound: the conclusion’s uncertainty cannot exceed the weighted sum of the uncertainties, where completely irrelevant premises get weight zero and utterly essential ones get weight one. In the alarm‑clock argument with an extra “It is Tuesday” premise, the irrelevant fact is assigned zero weight, so it doesn’t inflate your worst‑case doubt at all. This improved rule matches our intuition that you shouldn’t become less confident about a conclusion just because you muttered an extra, useless fact.

The math behind this is deeper — it uses techniques from linear programming — but the takeaway is simple. Logic can tell you not only that conclusions follow from premises, but how strongly they follow, and which premises actually do the pushing.

Why This Still Matters for You

Probability logic isn’t just a philosopher’s puzzle. It shapes the software that helps doctors interpret test results, helps artificial intelligence weigh conflicting clues, and helps courts reason about evidence when nobody is absolutely certain. Whenever a machine has to decide what to believe from messy data, it uses some version of these ideas.

But the most important application might be in your own head. Every day you make leaps from imperfect information. You hear a rumor, check the sky, read a review — and you judge what follows. You are, without thinking about it, tracking how much uncertainty each source adds and deciding when you’re confident enough to act.

Probability logic gives you a language for that hidden skill. It shows that even when the facts are blurry, the rules of good reasoning are not. Certainty is not the only friend of truth — disciplined maybe can be, too.

Think about it

1. If a scientist could perfectly predict every decision you’ll ever make, would it still make sense to blame people for bad choices? Why or why not?
2. Imagine you have two friends. One says “I’m 90% sure the movie starts at 7.” The other says “I’m 90% sure the movie starts at 7 because I checked the ticket.” Are these equally trustworthy? What if you combine both opinions?
3. You hear a rumor that 80% of all birds fly. But someone then says, “Yes, but penguins are birds.” Does the second sentence change how likely the first one seems? How does that connect to the idea of essential premises?