I can check on WolframAlpha for how to represent some logical operations using others, but I don't see anything for $\Leftrightarrow$. Is it because it is impossible?
2 Answers
HINT: By induction on the complexity of a formula $\varphi$ in two propositional variables $p, q$ built from $\neg$ and $\iff$, we can show that there are an even number (0, 2, or 4) of truth assignments making $\varphi$ true.
Actually, you use induction to completely classify the possible $\varphi$ which can be built in this way; this is a case of needing to assume a stronger claim in an induction argument. But this is the right idea.
This rules out the possibility of representing e.g. "$p\implies q$" (which is satisfied by three out of four truth assignments) in this way.
- 245,398
-
I am sorry, I don't understand this (English is my 2nd language). I see that you edited- I only understand that you CAN'T represent => :D – user3486470 Feb 10 '16 at 18:23
-
@user3486470 Which part don't you understand? – Noah Schweber Feb 10 '16 at 18:26
-
With <=> you can have 2 or 4 "truths", so it is impossible to represent => which has 3 "truths"? – user3486470 Feb 10 '16 at 18:34
Here is an informal approach:
(By 'is equivalent to', I mean 'has the same truth table'.)
Suppose you have some formula $\phi(x_1,...,x_n)$ created with the connectives mentioned.
Then either $\phi(\lnot x_1,...., \lnot x_n)$ is equivalent to $\phi(x_1,...,x_n)$ or $\phi(\lnot x_1,...., \lnot x_n)$ is equivalent to $\lnot \phi(x_1,...,x_n)$.
This is not hard to verify by induction.
However $\lnot x_1 \implies \lnot x_2$ is not equivalent to $\lnot (\lnot x_1 \implies \lnot x_2)$ nor is it equivalent to $x_1 \implies x_2$.
This can be seen by taking the truth assignment $x_1 = 0, x_2 =1$.
- 172,524