Vacuous truth
|
|
Informally, a logical statement is vacuously true if it is true but doesn't say anything; examples are statements of the form "everything with property A also has property B", where there is nothing with property A.
It is tempting to dismiss this concept as vacuous or silly. It does, however, have useful applications. One example is the empty product -- the fact that the result of multiplying no numbers at all is 1 -- which is useful in a variety of mathematical fields including probability theory, combinatorics, and power series. The eminent mathematician Gian-Carlo Rota speaking before an audience of perhaps 1000 mathematicians in Baltimore in January 1998, stated that physicists in particular like to dismiss this idea as ivory-towerish (Rota's rhetorical style often used exaggeration and hyperbole), and mentioned as an example that the elementary symmetric polynomial in no variables at all is 1. Then he went on to say that this led to the remarkable discovery that the Euler characteristic is actually one of the finitely additive "measures" treated in Hadwiger's theorem, so that "pure" mathematicians who attach importance to this kind of "vacuity" have the last laugh.
| Contents |
Examples
The statement
- All elephants inside a loaf of bread are pink.
is vacuously true since there are no elephants inside a loaf of bread; here property A is "being an elephant inside a loaf of bread", and property B is "being pink". Another example is
- If a prime number is even and bigger than two, then it must be divisible by three.
There are no such prime numbers, so in a sense the truth of this statement "doesn't matter".
The statement "0 mathematicians can change a lightbulb" is not vacuously true (or, indeed, true at all); the lightbulb joke "in a group of 0 mathematicians, any one of them can change a lightbulb" however is vacuously true.
Vacuous truth should be compared to tautology, with which it is sometimes conflated.
The remainder of this article uses mathematical symbols.
Scope of the concept
The term "vacuously true" is generally applied to a statement S if S has a form similar to:
- P ⇒ Q, where P is false.
- ∀ x, P(x) ⇒ Q(x), where it is the case that ∀ x, ¬ P(x).
- ∀ x ∈ A, Q(x), where the set A is empty.
- ∀ ξ, Q(ξ), where the symbol ξ is restricted to a type that has no representatives.
The first instance is the most basic one; the other three can be reduced to the first with suitable transformations.
Vacuous truth is usually applied in classical logic, which in particular is two-valued, and most of the arguments in the next section will be based on this assumption. However, vacuous truth also appears in, for example, intuitionistic logic in the same situations given above. Indeed, the first 2 forms above will yield vacuous truth in any logic that uses material implication, but there are other logics which do not.
Arguments of the semantic "truth" of vacuously true logical statements
This is a complex question and, for simplicity of exposition, we will here consider only vacuous truth as concerns logical implication, i.e., the case when S has the form P ⇒ Q, and P is false. This case strikes many people as odd, and it's not immediately obvious whether all such statements are true, all such statements are false, some are true while others are false, or what.
Arguments that at least some vacuously true statements are true
Consider the implication "if I am in Boston, then I am in Massachusetts", which we might alternatively express as, "if I were in Boston, then I would be in Massachusetts". There is something inherently reasonable about this claim, even if one is not currently in Boston. It seems that someone in Seattle, for example, would still have good reason to assert this proposition. Thus at least one vacuously true statement seems to actually be true.
Arguments against taking all vacuously true statements to be false
Making implies and logical AND logically equivalent
Second, the most obvious alternative to taking all vacuously true statements to be true -- i.e., taking all vacuously true statements to be false --, has some unsavory consequences. Suppose we're willing to accept that P ⇒ Q should be true when both P and Q are true, and false when P is true but Q is false. That is, suppose we accept this as a partial truth table for implies:
| P | Q | P ⇒ Q |
| T | T | T |
| T | F | F |
| F | T | ? |
| F | F | ? |
Suppose we decide that the unknown values should be F. In this case, then implies turns out to be logically equivalent to logical AND, as we can see in the following table:
| P | Q | P⇒Q | P AND Q |
| T | T | T | T |
| T | F | F | F |
| F | T | F | F |
| F | F | F | F |
Intuitively this is odd, because it certainly seems like "if" and "and" ought to have different meanings; if they didn't, then it's confusing why we should have a separate logical symbol for each one.
Perhaps more disturbing, we must also accept that the following arguments are logically valid:
- P ⇒ Q
- P AND Q
- P
and
- P ⇒ Q
- P AND Q
- Q
That is, we can conclude that P is true (or that Q is true) based solely on the logical connection of the two.
Intuition from mathematical arguments
Picking "true" as the truth value makes many mathematical propositions that people tend to think are true come out as true. For example, most people would say that the statement
- For all integers x, if x is even, then x + 2 is even.
is true. Now suppose that we decide to say that all vacuously true statements are false. In that case, the vacuously true statement
- If 3 is even, then 3 + 2 is even
is false. But in this case, there is an integer value for x (namely, x=3), for which it does not hold that
- if x is even, then x + 2 is even
Therefore our first statement isn't true, as we said before, but false. This doesn't seem to be how people use language, however.
A linguistic argument
First, calling vacuously true sentences false may extend the term "lying" to too many different situations. Note that lying could be defined as knowingly making a false statement. Now suppose two male friends, Peter and Ned, read this very article on some June 4, and both (perhaps unwisely) concluded that "vacuously true" sentences, despite their name, are actually false. Suppose the same day, Peter tells Ned the following statement S:
- If I am female today, i.e., June 4, then I will buy you a new house tomorrow, i.e., June 5.
Suppose June 5 goes by without Ned getting his new house. Now according to Peter and Ned's common understanding that vacuously true sentences are false, S is a false statement. Moreover, since Peter knew that he was was not female when he uttered S, we can assume he knew, at that time, that S was vacuously true, and hence false. But if this is true, then Ned has every right to accuse Peter of having lied to him. This doesn't seem right, however.
Arguments for taking all vacuously true statements to be true
The main argument that all vacuously true statements are true is as follows: As explained in the article on logical conditionals, the axioms of the propositional calculus entail that if P is false, then P ⇒ Q is true. That is, if we accept those axioms, we must accept that vacuously true statements are indeed true. For many people, the axioms of the propositional calculus are obviously truth-preserving. These people, then, really ought to accept that vacuously true statements are indeed true. On the other hand, if one is willing to question whether all vacuously true statements are indeed true, one may also be quite willing to question the validity of the propositional calculus, in which case this argument begs the question.
Arguments that only some vacuously true statements are true
One objection to saying that all vacuously true statements are true is that this makes the following deduction valid:
- ¬ P
- P ⇒ Q
Many people have trouble with or are bothered by this because, unless we know about some a priori connection between P and Q, what should the truth of P have to do with the implication of P and Q? Shouldn't the truth value of P in this situation be irrelevant? Logicians bothered by this have developed alternative logics (e.g. relevant logic) where this sort of deduction is valid only when P is known a priori to be relevant to the truth of Q.
Note that this "relevance" objection really applies to logical implication as a whole, and not merely to the case of vacuous truth. For example, it's commonly accepted that the sun is made of gas, on one hand, and that 3 is a prime number, on the other. By the standard definition of implication, we can conclude that: the sun's being made of gas implies that 3 is a prime number. Note that since the premise is indeed true, this is not a case of vacuous truth. Nonetheless, there seems to be something fishy about this assertion.
Summary
So there are a number of justifications for saying that vacuously true statements are indeed true. Nonetheless, there is still something odd about the choice. There seems to be no direct reason to pick true; it's just that things blow up in our face if we don't. Thus we say S is vacuously true; it is true, but in a way that doesn't seem entirely free from arbitrariness. Furthermore, the fact that S is true doesn't really provide us with any information, nor can we make useful deductions from it; it is only a choice we made about how our logical system works, and can't represent any fact of the real world.
Difficulties with the use of vacuous truth
All pink rhinoceros are carnivores. All pink rhinoceros are vegetarians.
Both of these seemingly contradictory statements are true using classical or two-valued logic - so long as the set of pink rhinoceros remains empty.
Certainly, one would think it should be easy to avoid falling into the trap of employing vacuously true statements in rigorous proofs, but the history of mathematics contains many 'proofs' based on the negation of some accepted truth and subsequently demonstrating how this leads to a contradiction.
One fundamental problem with such 'demonstrations' is the uncertainty of the truth-value of any of the statements which follow (or even whether they do follow) when our initial supposition is false. Stated another way, we should ask ourselves which rules of mathematics or inference should still be applicable after we first suppose that pi is an integer less than two?
The problem occurs when it is not immediately obvious that we are dealing with a vacuous truth. For example, if we have two propositions, neither of which imply the other, then we can reasonably conclude that they are different; counter-intuitively, we can also conclude that the two propositions are the same since this is a vacuous truth because (P⇒Q)∨(Q⇒P) is a tautology in classical logic.
Avoidance of such paradox is the impetus behind the development of non-classical systems of logic relevant logic and paraconsistent logic which refuse to admit the validity of one or two of the axioms of classical logic. Unfortunately the resulting systems are often too weak to prove anything but the most trivial of truths.
Vacuous truths in mathematics
Vacuous truths occur commonly in mathematics. For instance, when making a general statement about arbitrary sets, said statement ought hold for all sets including the empty set. But for the empty set the statement may very well reduce to a vacuous truth. So by taking this vacuous truth to be true, our general statement stands and we are not forced to make an exception for the empty set.
There are however vacuous truths that even most mathematicians will outright dismiss as "nonsense" and would never publish in a mathematical journal (even if grudgingly admitting that they are true). An example would be the true statement
- Every infinite subset of the set {1,2,3} has seven elements.
Further reading
- When is truth vacuous? Is infinity a bunch of nothing? (http://www.geocities.com/n_fold/vactruth.html): a transcript of a discussion in which some professional and amateur mathematicians try to find a definition for vacuous truth and debate its properties