I have to give a deduction of the existential generalization axiom without using the axiom itself. I'm given universal instantiation and the quantifier rules $(\psi\rightarrow \varphi, \psi\rightarrow (\forall x\varphi))$ and $(\varphi\rightarrow\psi,(\exists x\varphi)\rightarrow \psi)$ but that's about it. My first step I think should be to assume $\forall x\varphi$ which will allow me to use instantiation. After that I feel I should use the quantifier rules but can't find a valid application of the rules. A hint would be appreciated. Thanks.
Asked
Active
Viewed 137 times
1 Answers
1
Recall that (page 32 of the 2nd edition) the existential quantifier is not primitive; it is defined as an abbreviation for $\lnot ((\forall x) (\lnot \alpha))$.
Thus, from the first quantifier axiom:
$(\forall x) \lnot \varphi \to \lnot \varphi^x_t$
by propositional consequence :
$\lnot \lnot \varphi^x_t \to \lnot ((\forall x) \lnot \varphi)$
and by propositional consequence again and abbreviation:
$\varphi^x_t \to (\exists x) \varphi$.

Mauro ALLEGRANZA
- 94,169
-
Thank you! So basically I just have to assume $\forall x\neg\varphi$ and $\neg\neg\varphi_t^x$. – Jeff M. Mar 16 '17 at 16:09
-
@JeffM. - No; this is not a correct instance of axiom : you must have only one $\lnot$. – Mauro ALLEGRANZA Mar 16 '17 at 16:22
-
I'm getting there, but I'm still confused as to what assumptions I need to make. If I assume $\forall x\neg\varphi$, don't I need to assume $\neg\neg\varphi_t^x$ in order to get the propositional consequence on the second line? But at the same time if I assume both of these, then I also have $\neg((\forall)\neg\varphi)$ which means both the assumption and its negative are true, a contradiction... – Jeff M. Mar 16 '17 at 16:27
-
1@JeffM.- not necessarily; see Denition 2.4.5 Rule of inference of type (PC) page 53. I'm using $(P \to Q) \vDash (\lnot Q \to \lnot P)$ for the 1st to 2nd line and $P \to \lnot \lnot P, \lnot \lnot P \to Q \vdash P \to Q$ for the 2nd to 3rd line. – Mauro ALLEGRANZA Mar 16 '17 at 16:52