Sei sulla pagina 1di 8

PHIL 201:

Introduction to Formal Logic

Spring 2009

Session 21: Overview

Well look at two additional rules that are unique to PL derivation:

Existential-In
Universal Derivation

Existential-In
Existential-In (I)
If one has an available line of the
form F[n], where F[n] is a formula
that results from substituting all free
instances of the variable in F[]
with the constant n, then then one
is entitled to infer F[]

I:

F[n]
*
F[]

Existential-In

Lets look at some examples of applying I


Example 1:

(F[n] = Fb, n = b)

Fb
*

Example 2:

xFx

(F[] = Fx, = x)

Rjk

(F[n] = Rjk, n = j)
*
(F[] = Ryk, = y)

yRyk

Existential-In

Example 3:

(F[n] = Fb & Hb, n = b)

Fb & Hb
*

Example 4:

x(Fx & Hx)

(F[] = Fx & Hx, = x)

Fb & Hb

(F[n] = Fb & Hb, n = b)


*

x(Fx & Hb)

(F[] = Fx & Hb, = x)

Existential-In

The previous example serves to emphasize that when we apply I, we


neednt replace all occurrences of the n in the formula were taking to
be F[n] with the variable

Although the pattern in the previous example is not used very often,
the following pattern is:

Example 5:

(F[n] = Rkk, n = k)

Rkk
*
xRxk

(F[] = Rxk, = x)

Existential-In

Suppose that Rkk means Kay respects herself


From this it follows that:

Someone respects him/herself xRxx


Kay respects someone (i.e. Kay) xRkx
Someone respects Kay xRxk
Someone respects someone xyRxy

To obtain this last formula, we apply I twice

Using Existential-In: Examples

Lets look at some examples of applying I


Example 1:

(1)

x(Fx Hx)

Pr

(2)

Fa

Pr

(3)

SHOW: xHx

(4)

Fa Ha

(5)

Ha

(6)

xHx

DD
1, O
2,4, O
5, I

Using Existential-In: Examples

Example 2:

(1)

x(Gx Hx)

Pr

(2)

Gb

Pr

(3)

SHOW: x(Gx & Hx)

(4)

Gb Hb

(5)

Hb

(6)

Gb & Hb

(7)

x(Gx & Hx)

DD
1, O
2,4, O
2,5, &I
6, I

Using Existential-In: Examples

Example 3:

(1)

x~Rxa ~xRax

Pr

(2)

~Raa

Pr

(3)

SHOW: ~Rab

ID

(4)

Rab

As

(5)

SHOW: X

DD

(6)

x~Rxa

2, I

(7)

~xRax

1,6, O

(8)

xRax

(9)

4, I
XI

Using Existential-In: Examples

Example 4:

(1)

x(yRxy yRxy)

Pr

(2)

Raa

Pr

(3)

SHOW: Rab

(4)

yRay yRay

(5)

yRay

2, I

(6)

yRay

4,5, O

(7)

Rab

DD
1, O

6, O

Universal Derivation

Suppose we have the following derivation problem:


(1)

x(Fx Gx)

Pr

All frogs are green

(2)

xFx

Pr

Everything is a frog

(3)

SHOW: xGx

...

Everything is green

This is clearly a valid argument, but how can we prove it?

But once weve gotten rid of the universals, we have no way of getting
them back!

We could obtain Gn (for any constant n) by using O on (1) and (2),


and then applying O

Universal Derivation

To deal with situations like this, where were trying to SHOW a


universally quantified formula, we introduce a new show rule

Heres the reasoning behind the rule: When we say that xGx, were
saying that for everything in our universe of discourse U = {a, b, c, ...},
the predicate is G applies to it:

For every x in U = {a, b, c, ...}, x is G


a is G, and b is G, and c is G, ...
Ga & Gb & Gc ...

Any universally quantified statement is equivalent to a (possibly


infinite) conjunction

Universal Derivation

To show that xGx follows from the premises of the earlier


argument, we could show that Ga, Gb, Gc, etc. follow from the
premises:
(1)

x(Fx Gx)

Pr

(2)

xFx

Pr

(3)

SHOW: Ga

(4)

Fa Ga

1, O

(5)

Fa

2, O

(6)

Ga

4,5, O

DD

We could do this for Ga, Gb, Gc,


and so on

But its not feasible to prove Gn


for all n in the universe of
discourse!

How many do we have to prove?


ONE!

Universal Derivation

We just showed that we can derive Ga from the premises

If we can show that Ga, then since theres nothing special about a in
this context, we can also show that Gb, Gc, Gd, etc.

If we can show that Gx for any x in the universe U, then weve shown
that xGx

So it looks like we can adopt the following rule: to show F[], it is


sufficient to show a substitution instance F[n] of F[]

But there was nothing special about the constant a. We could have
used the same method to show that Gb, Gc, Gd, etc.

This is assuming that the n in F[n] is not special...

Universal Derivation

Lets try a couple of problems using this new rule, universal derivation
(UD):
(1)

x(Fx Gx)

Pr

(2)

xFx

Pr

(3)

SHOW: xGx

UD

(4)

SHOW: Ga

DD

(5)

Fa Ga

1, O

(6)

Fa

2, O

(7)

Ga

5,6, O

Universal Derivation

Lets try a couple of problems using this new rule, universal derivation
(UD):
(1)

x(Fx Gx)

Pr

(2)

SHOW: xFx xGx

(3)

xFx

(4)

SHOW: xGx

UD

(5)

SHOW: Ga

DD

(6)

Fa Ga

1, O

(7)

Fa

3, O

(8)

Ga

6,7 O

CD
As

Universal Derivation

What about the following:


(1)

Fa & Ga

(2)

SHOW: xGx

(3)

SHOW: Ga

(4)

Ga

Pr
UD
DD
1, &O

This is clearly an invalid argument

From the fact that Alex is a green frog, it doesnt follow that
everything is green!

Universal Derivation

What went wrong?

This is because the method of showing Ga here is special to Ga: it


depends on the fact that we already know Ga from (1)

Because of this, we cant show xGx by showing Ga; wed need to


show it by showing some other substitution instance of Gx, such as Gb

In general, when we want to use UD to show F[], it is sufficient to


show that F[n], where n is a new constant that doesnt already appear in
the derivation

Although we can show Ga here, we cant show Gb, Gc, Gd, etc. using
the same method: The method of showing Ga doesnt generalize

Universal Derivation
Universal Derivation (UD)

SHOW: F[]

If one has a show-line of the form


SHOW: F[], then if one has SHOW:
F[n] as a later available line, where F[n]
is a substitution instance of F[] and n is
a new constant that hasnt appeared in
previous lines, then one may box and
cancel SHOW: F[]

UD

SHOW: F[n]

...

...
...
Note: the n must be new

Universal Derivation

Lets try a couple examples:

(1)
(2)

Fa & Ga
x(Fx yGy)

Pr

(3)

x(Gx Fx)

Pr

(4)

SHOW: xFx

UD

(5)

SHOW: Fb

DD

(6)
(7)

Fa
Fa yGy

(8)

yGy

(9)

Gb

8, O

(10)

Gb Fb

3, O

(11)

Fb

Pr

1, &O
2, O
6,7, O

9,10, O

Universal Derivation

Lets try a couple examples:

(1)

Raa

Pr

(2)

xy(Rxy xyRxy)

Pr

(3)

SHOW: xyRyx

UD

(4)

SHOW: yRyb

UD

(5)

SHOW: Rcb

DD

(6)

y(Ray xyRxy)

2, O

(7)

Raa xyRxy

6, O

(8)

xyRxy

(9)

yRcy

8, O

(10)

Rcb

9, O

1,7, O

Potrebbero piacerti anche