5

The following two exercises come from Logic for Mathematicians by J.B. Rosser, chapter 2 section one page 17. I am not so sure how to interpret the words "unless" and "except".

Notation: $\sim P$ represents negation the negation of $P$, and $PQ$ denotes $P\& Q$ which the author refers to as the logical product of $P$ and $Q$. Also, $P\equiv Q$ denotes "$P$ if and only if $Q$".

Exercise 3. If “$P$ ”, “$Q$ ”, and “$R$ ” are translations for “$x=y$ ” , “$x/z=y/z$ ”, and "$z\ne0$" , write a translation for “if $x=y$ , then $x/z=y/z$ except when $z=0$ ”

Solution: It is not the case that $(P\rightarrow Q)\sim R$ , i.e. we have $\sim[(P\rightarrow Q)\sim R]$ which is equivalent to $\sim[\sim(P\sim Q)\sim R]$ which is equivalent to $\sim\{[\sim P\sim(\sim Q)]\sim R\}$ (Note: the last step was an invalid manipulation, hence my conclusion is wrong) . Since $\sim(\sim Q)$ is equivalent to $Q$ and because the logical product is both associative and commutative, we have $\sim[Q\sim(PR)]$ hence $Q\rightarrow RP$.

My confusion is that I am pretty sure that $Q\equiv RP\space$ holds but i don't think that it would be the right translation. Am I right?

Exercise 4. If “$P$ ” and “$Q$ ” are the translations of “$(n-1)!+1$ is divisible by $n$ ” and “$n$ is prime” then write a translation for “$(n-1)!+1$ is not divisible by $n$ , unless $n$ is prime”

Solution: In other words we have that “the only time $(n-1)!+1$ is divisible by $n$ when $n$ is prime” or “$(n-1)!+1$ is divisible by $n$ when and only when $n$ is prime”, hence $P\equiv Q$.

I am not sure if I interpreted `unless' correctly here.


More Reasoning for Exercise 3:

I arrive at the conclusion $\sim[(P\rightarrow Q)\sim R]$ because it says "$P\rightarrow Q$ except when $\sim R$", which I take to mean that "it is false that both $\sim R$ and $P\rightarrow Q$ are true at the same time"; i.e. it is not the case that $(P\rightarrow Q)\&\sim R$, hence $\sim[(P\rightarrow Q)\sim R]$. From there mechanical manipulation of symbols gets me to $Q\rightarrow RP$ (again this is wrong).

Additionally, though I am certain that $RP \rightarrow Q$ I am not quite sure how to reason this from the sentence “if $x=y$, then $x/z=y/z$ except when $z=0$”. Furthermore, with my interpretation of it being $\sim[(P\rightarrow Q)\sim R]$ I am not sure what symbolic manipulations would give $RP \rightarrow Q$.

Bram28
  • 100,612
  • 6
  • 70
  • 118
JimmyJackson
  • 1,517
  • 1
    For ex. 3, I think it should be $(P(\sim R)\to Q)$. Ex. 4 seems fine to me – Ragnar Dec 21 '13 at 01:21
  • But $P(\sim R)$ means that $z=0$ and $x=y$ in which case mathematically we can't imply that $x/z=y/z$, however the author makes it clear that symbolic logic is void of meaning. – JimmyJackson Dec 21 '13 at 01:29
  • Moreover, mathematically $PR$ means the $x=y$ and $z\neq 0$ which certainly implies that $x/z=y/z$. That is why am pretty sure that $Q\equiv RP$ holds. – JimmyJackson Dec 21 '13 at 01:38
  • Oops, I missed the negation in $R$. The exercise is only in one direction, so the $Q\to PR$ part isn't needed here. (although it is true.) – Ragnar Dec 21 '13 at 01:44
  • @Ragnar I added some more thought on exercise 3, if you could take a look at it and let me knwo what you think. – JimmyJackson Dec 21 '13 at 02:21
  • @MaliceVidrine It is a very good book, i like the dot notation and he goes into great detail about restricted predicate calculus. – JimmyJackson Dec 21 '13 at 03:38
  • Dot delimiters are pretty great. It's also a historically important reference on NF set theory in its later chapters. One of my favorite texts all 'round. – Malice Vidrine Dec 21 '13 at 04:04
  • @MaliceVidrine Do you know if there is a solutions manual in circulation? I am trying to do all the exercises and I would like to check my answers. – JimmyJackson Dec 21 '13 at 04:17
  • Not that I'm aware of. The book was out of print for a long while, so it wouldn't surprise me if any such manual quietly disappeared in the interim. – Malice Vidrine Dec 21 '13 at 04:25
  • Did you mean $x|z\equiv y|z$? – Yiorgos S. Smyrlis Dec 23 '13 at 16:38

5 Answers5

5

Considering "unless" means "without", "if not", we have several ways to translate "$P$ unless/except $R$". Note that

  • Without $R$, $P$ is true. We can write "$\sim R\to P$".
  • Either $P$ or $R$ is true. We can write "$P$ or $R$"
  • If $P$ fails, $R$ is the reason. We can write "$\sim P\to R$"

Moreover, "$P\to Q$" is logically equivalent to "$\sim P$ or $Q$", also "$\sim(P$ and $\sim Q)$"

So don't confuse by what $P,R$ is. They are only symbols representing propositions.


For Ex3, denote if part "$P\to Q$" by $S$ and unless part "$\sim R$" by $T$, then it's routine to write $$\sim T\to S\equiv\sim(\sim R)\to(P\to Q)\equiv\sim(R\sim(\sim(P\sim Q)))\equiv\sim(R(P\sim Q))$$

Ex4 is similar and more straightforward.


Yes, sometimes we explain "$P$ unless $Q$" as $P\equiv\sim Q$. That means $P$ is true if $Q$ is not and vice versa. But that depends.

If "$P$ unless $Q$" means "$P\equiv\sim Q$", then whenever $Q$ is true, $P$ cannot be true, or "$Q\to\sim P$". Let's check it for Ex3. Assume $z=0$ and $x=y$, we require $x/z=y/z$ is false. However, $x/z=y/z$ is neither true nor false because it's undefined. And Ex4, similarly we assume $n$ is prime, it's expected that $(n−1)!+1$ is not divisible by $n$ is false, or $n\vert (n-1)!+1$ and so is the fact, in which case you can say "$P\equiv\sim Q$".

In above examples, we see "unless" can have different meanings in different context. However for propositional logic, we don't have any context. No one knows what a proposition symbol represents. That's why in propositional logic, "unless" means "if not" instead of "if not and only if not".

Shuchang
  • 9,800
  • I have added to the question above and posted a bounty. If you are interested, perhaps you could expound on your current answer. Thanks. – JimmyJackson Dec 23 '13 at 01:32
  • @JimmyJackson Do you mean explain two ex's? And is there any notation for "or"? – Shuchang Dec 23 '13 at 01:35
  • Yes I would like the final answer to both exercises with separate lines of reasoning for each one. And yes we are permitted to use the "or" symbol, i.e. $P\vee Q$, but we are encouraged to keep in mind that this is really equivalent to $\sim(\sim P \sim Q )$; "or" does not actually belong to our symbolic logic which we build using only $\sim$ and &. – JimmyJackson Dec 23 '13 at 01:42
  • If I take your explanation in the first dot I will, arrive $RP \rightarrow Q$. The conundrum is this, I can give a rigorous proof that "except" leads to $Q \rightarrow RP$. Now we must decide something, are you wrong or am I wrong? Or are we both wrong? See in the given case it is obvious that we really have $Q \equiv RP$, BUT our symbolic logic is void of meaning so this is unhelpful information. So now I am left asking does "except" really mean a bi-conditional exists, i.e. without $R$, $P$ is true hence $P$ if and only if $\sim R$? This is why I can't accept your answer yet, please expound. – JimmyJackson Dec 23 '13 at 03:03
  • I was wrong to say that $Q\rightarrow RP$ could be achieved from $\sim[\sim(P\sim Q)\sim R]$, I made a major mistake in my manipulations. However, I feel confident in my conclusion which I have posted below, and "except" or "without" is a bi-conditional connective. That is without $R$, $P$ is true, is to say $\sim R \equiv P$. My reasoning is below, thanks for your time. – JimmyJackson Dec 23 '13 at 06:55
  • Inadvertently down-voted. How to undo? – Dan Christensen Dec 23 '13 at 19:02
  • @DanChristensen: Unintuitively, down-vote again to undo. – user21820 Dec 28 '13 at 03:55
4

I will follow the suggestion of S.C.Kleene, Mathematical Logic (1967), where he summarize in a table a list of expressions and their possible translation in symbols (pag.64) :

$A \lor B$ is $A$ unless $B$ [usually] and is $A$ except when $B$ [usually].

Try now with the exercises :

Exercise 3 : If $P$, $Q$ and $R$ are translations for “$x=y$” , “$x/z=y/z$” and "$z = 0$", write a translation for “if $x=y$, then $x/z=y/z$ except when $z=0$”

I will start from the mathematical condition, rewritten as : $\lnot z = 0 \rightarrow (x=y \rightarrow x/z=y/z)$ i.e. $ z = 0 \lor (x=y \rightarrow x/z=y/z)$.

We have that : $R \lor (P \rightarrow Q)$, i.e. $(P \rightarrow Q) \lor R$.

We can read it as : “if $x=y$, then $x/z=y/z$, except when $z=0$”

Exercise 4 : If $P$ and $Q$ are the translations of “$(n-1)!+1$ is divisible by $n$” and “$n$ is prime” then write a translation for “$(n-1)!+1$ is not divisible by $n$, unless $n$ is prime”

The mathematical condition is : [$(n-1)!+1$ is divisible by $n$] $\rightarrow$ ($n$ is prime) i.e. $\lnot [(n-1)!+1$ is divisible by $n$] $\lor$ ($n$ is prime).

We have that : $\lnot P \lor Q$.

  • Sorry it took a while to accept this answer, I haven't been to the library in a while. I found S. C. Kleene's book to be great source for this question. In fact, both J.B. Rosser and S.C. Kleene studied under Alonzo Church, so I suppose both authors have a similar interpretation of the words. – JimmyJackson Dec 28 '13 at 21:24
1

Definitions:

except: with the exclusion or exception of

unless: 1) except on the condition that : under any other circumstance than 2) without the accompanying circumstance or condition that

Retrieved from here

Exercise 3:

Solution: The first step to forming our translation is to strip the statement of its meaning, we notice that our statement is of the form “$P\rightarrow Q$ except when $\sim R$”. Now we are left to interpret what “except when” means. With our definition above we can write “$P\rightarrow Q$ excluding the case $\sim R$” which can be interpret two ways “When $\sim(\sim R)$ is true then $(P\rightarrow Q)$ is true” in symbols this says “$\sim(R\sim(P\rightarrow Q))$”, i.e. $R\rightarrow(P\rightarrow Q)$. We could have also interpret the statement as “It is not the case that $\sim R$ and $P\rightarrow Q$ are both true” which is simply to say that “$\sim(\sim R(P\rightarrow Q))$ ” . i.e.$(P\rightarrow Q)\rightarrow R$. If we were to consider the conjunction of both of the interpretations above we would then have to conclude “$P\rightarrow Q$ except when $\sim R$ ” is to be translated as “$(P\rightarrow Q)\equiv R$”.

Yet another way that we could translated “$P\rightarrow Q$ except when $\sim R$ ” is to say “$\sim(P\rightarrow Q)$ when and only when $\sim R$ ” or “$P\rightarrow Q$ when and only when $\sim(\sim R)$ ”; in either case we have to conclude again “$(P\rightarrow Q)\equiv R$”. Our conclusion is also in line with the following source on slide 114; moreover, I have found another source here on page 157 that also encourages us to be careful with the word “unless”. As explained in the first source we should discern between a strong “unless” and a weak “unless”, the strong “$P$ unless $Q$ ” is to be translated as “$P\equiv\sim Q$ ” while the weak version is to be interpreted as “$\sim Q\rightarrow P$ ” (as suggested by many in the other answers). The author goes on to explain that when in doubt we should assume the weak “unless” is being used. However, when trying to demonstrate the strong form of “unless” the author chose to substitute “unless” for “except” which might lead one to believe that “except” implicitly mean strong “unless” as I have concluded above.

Nonetheless, this is only speculation on my part from the little information given by those whom I assume to know more about the subject than myself. I can only conclude from this exercise that “except” and “unless” are interjected into our logic because of common English, that is,at times we have no choice but to deal with the wording that we are presented. However, there are many alternatives for these word which eliminate ambiguity, and thus we should strive to avoid the use of the "unless" and "except" for they are sloppy due to their nature.

Exercise 4:

To begin our statement says “$\sim P$ unless $Q$ ”, in which case we usually consider a “unless” which is to be interpreted as “If not $Q$ then $\sim P$ ” or “$\sim Q\rightarrow\sim P$ ”, which is equivlent to “$P\rightarrow Q$ ”. (The preceding was explained in detail in Exercise 3). However from context it is also clear that “$Q\rightarrow P$ ”, in which case we have “$P\equiv Q$ ”. Nonetheless as a rule of thumb we will always interpret “$P$ unless $Q$ ” as “$\sim Q\rightarrow P$ ” and not as “$P\equiv\sim Q$ ” which is equivelent to “$\sim P\equiv Q$”.

Now on a side note, in my original question I claimed that $R\rightarrow(P\rightarrow Q)$ is equivalent to $RP\rightarrow Q$. I also claimed that $(P\rightarrow Q)\rightarrow R$ was logically equivalent to $Q\rightarrow RP$ which I disprove in the truth table below: enter image description here

Additionally I added the following truth table to confirm $RP\rightarrow Q$:

enter image description here

JimmyJackson
  • 1,517
1

Write a translation for "if $x=y$, then $x/z=y/z$ except when $z=0$".

This can also be written, "if $z\neq 0$ then (if $x=y$, then $x/z=y/z)$".

Using the implication operator, we have

$z\neq 0 \implies [x=y \implies x/z = y/z]$.

But you cannot use implication, only conjuction and negation. By definition,

$A\implies B \equiv \neg[A \land \neg B]$

Applying this definition twice, we would have

$\neg[z\neq 0 \land [x=y \land \neg [x/z= y/z]]]$

which can be easily translated into your special notation as

~(R(P(~Q)))

A similar analysis applies to the other exercise. Note that, as used here, "unless" and "except when" are interchangeable.

  • I think you meant to put a $\neg$ to the right of the first (or strongest) conjunction, so that we end with ~(R~(P(~Q))) right? – JimmyJackson Dec 23 '13 at 19:34
  • No, that was a double negation. – Dan Christensen Dec 23 '13 at 19:47
  • I didn't catch that at first. Interesting, I actually came the same conclusion in my answer, that is that $R\rightarrow (P\rightarrow Q)$. In fact, this is the reason that I claim that there exists a bi-condition not only a condition; we should first interpret the statement in terms of & and ~, and then from there we can add in implications as we find them (since $\implies$ does not actually belong to our symbolic logic, it is only shorthand). – JimmyJackson Dec 23 '13 at 20:20
1

Please, note that the Dover edition of J.B.Rosser, Logic for Mathematicians (2008) has (pag.17) in Ex.II.1.3 : $R$ for "$z = 0$".