How can I evaluate $(\forall x)(\phi \rightarrow (\exists x) \psi )$?

149 Views Asked by At

In the book Logic and Structure written by Dirk van Dalen, something like $(\forall x)(\phi \rightarrow (\exists x) \psi ) $ (where x is the only free variable in $\psi$ and $\phi$) is a sentence.

But I just don't know how to evaluate this, the book doesn't explain this or show any example like this, but it does appear in the some of the exercices.

I tried to treat it as if it was equivalent to $((\forall x)\phi \rightarrow (\exists x)\psi)$ but sadly, it's not.

If anyone is familiar with these concepts, I'd appreciate any kind of help :)

2

There are 2 best solutions below

0
On BEST ANSWER

What do you mean by "evaluate"? Determine its truth value in a structure? It goes just like any other formula, according to the definitions of interpretation:

$\begin{align} & \mathfrak{A} \vDash \forall x (\phi \to \exists x \psi)\\ \Leftrightarrow\ & \text{for all } a \in A, \mathfrak{A} \vDash (\phi \to \exists x \psi)[\overline{a}/x]\\ \Leftrightarrow\ & \text{for all } a \in A, \mathfrak{A} \vDash \phi[\overline{a}/x] \to \exists x \psi\\ \Leftrightarrow\ & \text{for all } a \in A, [\mathfrak{A} \nvDash \phi[\overline{a}/x] \text{ or } \mathfrak{A} \vDash \exists x \psi]\\ \Leftrightarrow\ & \text{for all } a \in A, [\mathfrak{A} \nvDash \phi[\overline{a}/x] \text{ or for some } b \in A, \mathfrak{A} \vDash \psi[\overline{b}/x] ] \end{align}$

So for all objects, either $\phi$ has to be false of that object, or for that object there has to be another object of which $\psi$ is true.

Note that the $\exists x$ "overwrites" the binding of $x$ by $\forall$; the $x$'es that occur in $\psi$ are existentially quantified and not effectively affected by the universal quantifier although it techically has scope over the entire formula.


In general we have the following logical equivalences (in van Dalen they appear as exercise 2 in chapter 2.5):

  • provided that $x$ does not occur free in $\psi$:
    $\begin{align} & \forall x (\phi \to \psi)\\ ⟚ \ & \exists x \phi \to \psi \end{align}$
    and
    $\begin{align} & \exists x (\phi \to \psi)\\ ⟚ \ & \forall x \phi \to \psi \end{align}$
  • provided that $x$ does not occur free in $\phi$:
    $\begin{align} & \forall x (\phi \to \psi)\\ ⟚ \ & \phi \to \forall x \psi \end{align}$
    and
    $\begin{align} & \exists x (\phi \to \psi)\\ ⟚ \ & \phi \to \exists x \psi \end{align}$

That is, quantifiers switch to their dual when moving them between outside the formula and the left-hand side of an implication, and remain the same when moving them between outside the formula the right-hand side of an implication.
Thus, since $x$ only occurs bound on the RHS of the implication, $\forall x (\phi \to \exists x \psi)$ is logically equivalent to

$$(\exists x \phi \to \exists x \psi)$$

So the above can more easily be rephrased as

Either there is no object of which $\phi$ is true, or there is at least one object of which $\psi$ is true.

0
On

$(\forall x)(\phi_x\rightarrow (\exists x)\psi_x)\tag{0}$

The definition of the implication symbol via truth table is $(1)$.

$$A\rightarrow B\tag{1}$$

\begin{array}{c|cc} &\lnot A&A\\\hline \lnot B&\color{orange}{1}&0\\ B&\color{orange}{1}&\color{blue}{1} \end{array}

$$\color{orange}{\lnot A}\lor \color{blue}{B}\tag{2}$$ Looking at this truth table, it is clear that an equivalent definition is $(2)$. As such, your attempt $(3)$ is equal to $(4)$, which is clearly different to the original formula $(0)$.

$(\forall x)\phi_x\rightarrow (\exists x)\psi_x\tag{3}$ $(\exists x)\lnot\phi_x\lor (\exists x)\psi_x\tag{4}$


The best way to interpret $(0)$ is to relabel the variables $(5)$, so $x$ does not appear twice in the sentence meaning different things each time.

$(\forall x)(\phi_x\rightarrow (\exists \color{red}{y})\psi_\color{red}{y})\tag{5}$

After relabelling the variables, you can expand the implication in the centre using $(2)$, and bring the universal quantifier from the front inside the brackets, noting that it only applies to the first formula $\phi$ $(6)$.

$(\forall x)\lnot\phi_x\lor (\exists y)\psi_y\tag{6}$

Finally, you can apply the definition of equality $(2)$ again to arrive at $(7)$, which is of a similar form to your attempt $(3)$.

$(\exists x)\phi_x\rightarrow (\exists y)\psi_y\tag{7}$