Let's take an argument form for which we could also do a direct derivation, then one for which a direct derivation would not be possible.
P -> (Q & R), ~R |- ~P
1. P -> (Q & R) Ass
2. ~R
Ass
3.
P
Hyp Here we are
indenting to show the beginning of a subordinate derivation. Anything
that
follows from this hypothesis should not be seen as "true" otherwise.
4.
Q & R 1,3 -> elim
5.
R
4 & elim
6.
R & ~R 2,6 & int This
is clearly an impossibility, so we know that the imagined possibility of
P can be ruled out--and then the opposite is shown to be true
7. ~P
3-6 hyp elim notice that
now we cite the entire set of lines
P -> Q, Q -> R |- P -> R
1. P -> Q Ass
2. Q -> R Ass
3. ~(P -> R)
Hyp
4. ~ (~P v R)
3 MI
5. P & ~R
4 DeM
6. P
5 & elim
7. Q
1,6 -> elim
8. R
2,7 -> elim
9. ~R
5 & elim
10. R & ~R
8,9 & int
11. P -> R
5-10 hyp elim
The advantage of indirect proofs is that they
can be used with any valid argument form. The disadvantage is that
often direct proofs and, as we shall see next, conditional proofs are shorter
and sometimes more interesting.
Go back to the derivation problems in earlier
modules and rework them with indirect proofs.