Source: p 502, A Concise Introduction to Logic (12 Ed, 2014) by Patrick Hurley

enter image description here

Because line 1 cannot be instantiated, the only strategy is to derive the antecedent of the conditional with the aim of deriving the consequent via modus ponens. [...]

  1. Why is the bolded true? Why can you not instantly replace 1A with 9, and 1C with 13?

  2. This proof appears overcomplicated.
    Instead of an Indirect Proof, why not use Conditional Proof?
    Then using 1A as the Assumption of the Conditional Proof, you can start the proof at 11;
    3-10 above can be skipped and would never be needed.


2 Answers 2


Because UI and EI rules "works" only with outermost quantifiers, i.e. when we have e.g. (∃x)Px.

If we have instead (∃x)Px ⊃ (∃y)Qy, we cannot apply the quantifier rules to remove them.

According to your question 2, you think that the proof should be 'simpler' as follows :

[1A] i.e. 11) (∃x)(∃y)(Jxy ∨ Kxy)   --- premise

[1C] i.e. 12) (∃x)Lx         --- premise

2) (x)(y)(Lx ⊃ ¬Ly)         --- premise

13) Lo               --- from 12) by EI

14)-15) Lo ⊃ ¬Lo          --- from 2) by UI twice

16) ¬Lo              --- from 13) and 15) by MP

17) Lo ∧ ¬Lo            --- from 13) and 16) by Conj

Now we have a contradiction and thus - according to the IP rule - we have to "discharge" one of the premises to derive this one premise's negation. We may discharge 12), deriving :

18) ¬(∃x)Lx.

The derivation is formally correct, but what we have proved is:

from the premises : (∃x)(∃y)(Jxy ∨ Kxy), (∃x)Lx, (x)(y)(Lx ⊃ ¬Ly),

it follows ¬(∃x)Lx.

Please, note that the derivation does not use [1A] at all; thus (as we can imagine) the derivation boils down to :

(x)(y)(Lx ⊃ ¬Ly) implies ¬(∃x)Lx.

The textbook's original proof was instead :

from the premises : (∃x)(∃y)(Jxy ∨ Kxy) ⊃ (∃x)Lx, (x)(y)(Lx ⊃ ¬Ly),

it follows (x)(y)¬Jxy,

which is quite different.

  • Thanks for the reply. 14 is from UI applied to 2, which is presumed as a premise whether you start from 3 or 11. So I see nothing wrong?
    – user8572
    Jan 20, 2016 at 20:48
  1. "If there are licorns then there are horses with horns" does not entail that there are licorns. Equivalently in logic, "(∃x)(Lx) ⊃(∃x)(Hx)" does not entail "(∃x)(Lx)". The existentials cannot be directly instantiated because they are nested in a more complex logical form.

  2. The assumption (3) is used to derive a contradiction. Then we can say that the assumption was false: (18) is the negation of (3). If we assume (11) instead we will not prove (18) but the negation of (11) which is not what we're looking for.

  • Thanks. In your answer to my 2, you wrote If we assume (9), but I asked about assuming 11. Can you please clarify?
    – user8572
    Jan 20, 2016 at 17:25
  • @LePressentiment it works with 11 (or whatever) as well. Fixed it! Jan 21, 2016 at 11:34
  • +1. Thanks. For your 1, can you please clarify and match your Substitution Instances (licorns, horses, horns) with the letters in my OP? What is x, y, J, K, L?
    – user8572
    Jan 21, 2016 at 20:30
  • 1
    No because it's not exactly the same logical form. This was intended as an example. The best I can do is write my example in a logical syntax. Jan 22, 2016 at 10:51

You must log in to answer this question.