6

I will propose a counterexample to $$((p \land q) \Rightarrow z) \Rightarrow ((p \Rightarrow z) \lor (q \Rightarrow z)).$$ Let's assume that $p$ is "$n$ is divisible by $2$", $q$ is "$n$ is divisible by $5$" and $z$ is "$n$ divisible by $10$". Then, while it's true that ($n$ is divisible by $2$ and $n$ is divisible by $5$) is sufficient for ($n$ is divisible by $10$), the disjunction ($n$ is divisible by $2$ is sufficient for $n$ is divisible by $10$) OR ($n$ is divisible by $5$ is sufficient for $n$ is divisible by $10$) is actually false.

What is wrong with my above counterexample and understanding? Why is $$((p \land q) \Rightarrow z) \Rightarrow ((p \Rightarrow z) \lor (q \Rightarrow z))$$ true?

ryang
  • 38,879
  • 14
  • 81
  • 179
  • Have you checked it with a truth table? – Mauro ALLEGRANZA Feb 11 '24 at 15:51
  • 3
    The are proposition; thus the "n" must be the same for all three. – Mauro ALLEGRANZA Feb 11 '24 at 15:52
  • 3
    Your counterexample shows what I think is exactly what trips people up about propositional implication. Your counterexample has implicitly inserted some universal quantifiers - really you've shown that $(\forall n)((p(n) \land q(n)) \implies z(n)) \implies ((\forall n)(p(n) \implies z(n)) \lor (\forall n)(q(n) \implies z(n)))$ is false. I agree that this is actually what most mathematicians mean in day-to-day-life when they say things like "if $p$ and $q$ implies $z$, then $p$ implies $z$ or $q$ implies $z$"... – Izaak van Dongen Feb 11 '24 at 15:54
  • 1
    You may find it helps to focus on a simpler "weird tautology" like $(p \implies q) \lor (q \implies p)$. These things have been discussed before on the site, eg here and here. I would also ask.. what happens when you try and fully fill in the details to show your counterexample is a counterexample? Your counterexample does sound like a counterexample, but can you show it actually is one using the rigorous definitions of propositional logic? (hopefully, you won't be able to!) – Izaak van Dongen Feb 11 '24 at 16:21
  • @IzaakvanDongen thanks. Clearly the statement with 3 quantifiers is false.

    So I understand that ((p∧q)⇒z)⇒((p⇒z)∨(q⇒z)), is regarded as (∀n)(( ( p(n) ∧ q(n) ) ⟹ z(n) )⟹( (p(n)⟹z(n)) ∨ (q(n)⟹z(n)) )) in math logic, am I right? But even if yes I'm not sure how to interpret this: that the implication occurs only for this particular number n?

    – Maciej Jałocha Feb 11 '24 at 18:55
  • 1
    Remember that when we define $\implies$ in logic, we really are making a definition, as with any other definition. It has some aesthetic resemblance to the concept of implication that you're familiar with, but there are some technical details of the definition that means sometimes it doesn't mean exactly what you think it does. When we say that some expression is a tautology, that (usually) means that for all possible assignments truth values of the propositions involved, the expression evaluates to true. In this case indeed once you fix any particular $n$, the formula will indeed be true. – Izaak van Dongen Feb 11 '24 at 21:54

3 Answers3

5

The sentences \begin{align}\Big((P \land Q) → Z\Big) → \Big((P → Z) \lor (Q→ Z)\Big)\tag{✔️1}\end{align} and \begin{align}∀\color{green}n\bigg(\quad\Big(\big(P\color{green}n ∧ Q\color{green}n\big) → Z\color{green}n\Big) \;→\; \Big(\;\big(P\color{green}n → Z\color{green}n\big) \;∨\; \big(Q\color{green}n → Z\color{green}n\big)\;\Big)\quad\bigg)\tag{✔️2}\end{align} are both indeed true regardless of interpretation (use a truth table to check this for Sentence $1$), because for any given value of $\color{green}n,$ if $P$ and $Q$ jointly implies $Z$ then at least one of $P$ and $Q$ implies $Z.$

On the other hand, your counterexample

while it's true that ($\color\red n$ is divisible by $2$ and $\color\red n$ is divisible by $5$) is sufficient for ($\color\red n$ is divisible by $10$), the disjunction ($\color{blue} n$ is divisible by $2$ is sufficient for $\color{blue} n$ is divisible by $10$) OR ($\color\violet n$ is divisible by $5$ is sufficient for $\color\violet n$ is divisible by $10$) is actually false

disproves the logical validity of \begin{align}∀\color{red}n\Big(\big(P\color{red}n ∧ Q\color{red}n\big) → Z\color{red}n\Big) \;→\; \bigg(∀\color{blue}n\big(P\color{blue}n → Z\color{blue}n\big) \;∨\; ∀\color{violet}n\big(Q\color{violet}n → Z\color{violet}n\big)\bigg),\tag{❌3}\end{align} which is more clearly written as \begin{align}∀\color\red a\Big(\big(P\color\red a ∧ Q\color\red a\big) → Z\color\red a\Big) \;→\; \bigg(∀\color{blue}b\big(P\color{blue}b → Z\color{blue}b\big) \;∨\; ∀\color{violet}c\big(Q\color{violet}c → Z\color{violet}c\big)\bigg).\tag{❌3}\end{align} Your counterexample isn't disproving Sentence $2,$ because it isn't premised on its seven occurrences of $n$ having a common value.

ryang
  • 38,879
  • 14
  • 81
  • 179
  • Thanks. I do agree that the intuition coming from the direct logic's reasoning is fine. Regarding the 3rd sentence: Why incidentally? If, ∀n((P(n)∧Q(n))→Z(n))→(∀n(P(n)→Z(n))∨∀n(Q(n)→Z(n))) is not what is understood as ((p∧q)⇒z)⇒((p⇒z)∨(q⇒z)), then isn't this ∀n(((P(n)∧Q(n))→Z(n))→((P(n)→Z(n))∨(Q(n)→Z(n))))? What's the regarded version of ((P∧Q)→Z)→((P→Z)∨(Q→Z)) with quantifiers? – Maciej Jałocha Feb 11 '24 at 19:01
  • 1
    @MaciejJałocha Are you asking about the difference between Sentences 2 and 3? I've just added colours to clarify (I also renumbered the sentences, so the "3rd sentence" in your comment now actually refers to Sentence 2; apologies). – ryang Feb 13 '24 at 05:37
  • Sure, it's good you added the last sentence to make it verbose. – Maciej Jałocha Feb 16 '24 at 06:57
1

I will first present a speedy proof of this, and then an explanation as to why this works that I hope is more clear than the others that have been given.

Assume that $((p \land q) \implies z) \implies ((p \implies z) \lor (q \implies z))$ is false. Then it must be the case that If $p \land q \implies z$ is true, it can be that $p \implies z \lor q \implies z $ is false, meaning $$(p \land q \implies z) \land \neg ((p \implies z) \lor (q \implies z))\\ (\neg(p \land q) \lor z) \land ((p \land \neg z) \land (q \land \neg z))$$ This is forced to be a contradiction. By the second parenthesis term, p and q are true and z is false, but by the right hand side if $z$ is false the $p$ or $q$ is false which is a contradiction, and if $z$ is true then it contradicts that $z$ is true. Therefore, the statement cannot be false and is a tautology. $\square$

Now, let's talk about what the sentence means and why your example doesn't make sense. I think the issue here is your prepositions don't exactly match up with what the sentence is written as. You would like to say that for a certain number $n$, if $2|n \wedge 5|n \implies 10|n$, which it does, then either $2|n \implies 10|n$ or $5|n \implies 10|n$. These are not causally true statements, but they are logically true. If we let $n=7$ the statement is clearly vacuously true. If $n=10$, you showed that it holds already. But what if $n = 15$? Then, $2 |15$ is false, $5 | 15$ is true, and $10 | 15$ is false. But since $2|15 \implies 10|15$ is vacuously true, the statement $$(2|n \implies 10|n) \lor (5|n \implies 10|n) \equiv (T) \lor (5|n\implies 10|n) \equiv T$$ Still evaluates to true. So in my own very loose words, this statement is true by vacuous truth mainly, and when its not equivalent it is just equivalent to $p \land q \implies z$.

This is a really good question however, as it does raise an excellent point to something most people don't realize, which is when and where vacuous truth applies, and why, even though many people say it just makes sense, it is very often confusing.

0

First things first: Draw out a truth table, always helpful for understanding this type of question!

I believe the confusion comes from the difference between how we use the implication sign in common language and what it means as a connective in propositional calculus: I can't add an image of the truth table , but I would be grateful if someone would.

If either p or q were False, then the right hand side is True. So the expression is True.

So we only need to consider both p and q True, then if z false, the LHS is True (As False--> False is True), then the implication is True( As False-->True and False--> False is True so we don't need to look at the RHS).

If both p and q True, and z True, then the LHS is True, and the RHS is clearly true, so the implication is True.. So under any possible truth assignment the statement is true. That's known as a tautology!

J.Dmaths
  • 715