5

I am trying to clear my doubts about various terms: tautology, contradiction, contingent, satisifiable, unsatisfiable, valid and invalid. I have read on them from various sources, and am putting all my understanding below in point form. I don't know whether I am overthinking. I just want exhaustive understanding. Could someone please check whether my understanding below is correct, and where my mistakes are?


Some of these definitions are straight up from other sources (so they must be correct), and the rest are written from my own understanding:

  1. Tautology: a formula or assertion that is true in every possible interpretation (that is, for all assignment of values to its propositional variables). Ref
  2. Contradiction: a formula or assertion that is false in every possible interpretation (that is, for all assignment of values to its propositional variables).
  3. A formula that is neither a tautology nor a contradiction is said to be logically contingent.
    Such a formula can be either true or false based on the values assigned to its propositional variables.
  4. A formula is satisfiable if it is true under at least one interpretation.
    So, it's either contingent or a tautology. Ref
  5. If a logic is a contradiction then it is said to be unsatisfiable.
  6. A formula is valid only if it is a tautology. Ref
  7. A formula is invalid if it is contingent or a contradiction.

Based on these definitions, I prepared a diagram showing how these concepts overlap:

enter image description here

Based on this diagram, I tried to answer questions like, "what is the negation of a tautology?" I felt that it could be either contingent or a contradiction: the above diagram means, "Given an assertion, if it is not a tautology, it can be either contingent or a contradiction." But it seems that I was wrong: the above diagram does not mean, "The negation of a tautology can be contingent or a contradiction." J.G.'s comment pointed out that I was simply negating definitions above, where I should have actually tried investigating how models (sets of values assigned to variables of formulae) satisfying a given formula behave for the negation of that formula. It took a while for me to understand that, and I have now come up with the following relations between any given assertion and its negation (I have given examples in brackets):

enter image description here

enter image description here

ryang
  • 44,428
RajS
  • 1,343
  • The bottom part of this question is new and not included in the previous question. – Doug Spoonwood Sep 08 '19 at 19:08
  • I will like people to hold on. I am doing an edit to extend bottom part. Also that old question is not fully answered. I touched this topic long back and left without conclusion as I did not had time to dig deep due to exams. I completely forgot that question. I wanted exhaustive relations between all those terms. The answer on older question touches only few terms (answerer also said so in the comments on the answer, may be because some terms are old usage). – RajS Sep 08 '19 at 19:09
  • @DougSpoonwood done with the editing. Added all facts which I can think exhaustively. Can you please confirm if they are correct? – RajS Sep 09 '19 at 06:02
  • Your previous question already received answers. – Mauro ALLEGRANZA Sep 09 '19 at 06:23
  • The table above with 15 rows is quite complicated... As per previous answers, there are two basic cases : (i) the formula $\varphi$ is a tautology, in which case its negation $\lnot \varphi$ is unsatisfiable. Reason why : if $\varphi$ is TAUT, then every truth assignment satisfies it. Thus, no truth assignment will satisfy its negation. – Mauro ALLEGRANZA Sep 09 '19 at 06:28
  • (ii) The formula $\varphi$ is satisfiable : this means that some (one, many, possible all, but at least one) truth assignment will satisfy it. From it, the only thing we can conclude is that its negation $\varphi$ is not a tautology (same argument above). – Mauro ALLEGRANZA Sep 09 '19 at 06:42
  • Yes I get both cases. I feel case (i) corresponds to point 10 in my question and case (ii) corresponds to point 9. But exam asks True/False questions like: (a) if assertion is invalid, then its negation can be invalid (b) if assertion is invalid, then its negation can be unsatisfiable (c) if assertion is not a tautology, then it can be valid (d) if assertion is not invalid, then it can be unsatisfiable. As per facts in tables: (a) True (b) False (c) False (d) False. But first need someone to confirm those tables. (I just realized I missed, "negation of contingent" and "not a contingent") – RajS Sep 09 '19 at 06:58

2 Answers2

2

enter image description here

  1. Following the truth-functional formulation of your Definitions 1 to 3, invalid satisfiable formulae are actually a proper subset of contingent formulae. I have revised your diagram to reflect this.

    As such, in first-order logic, your Definitions 5 to 7 are all incorrect. To wit:

    • $(\exists x\,x\ne x)$ is contingent and unsatisfiable.
    • $(\forall x\:x=x)$ is contingent and valid.
    • Valid Formulae and Invalid Formulae are complementary sets.
    • Satisfiable Formulae and Unsatisfiable Formulae are complementary sets.
    • Every formula that isn’t valid (non-valid formula) is invalid, and vice versa.
    • Every formula that isn’t satisfiable (non-satisfiable formula) is unsatisfiable, and vice versa.
  2. On the other hand,

    • The negation of every valid formula is unsatisfiable, not merely invalid.
    • The negation of every unsatisfiable formula is valid, not merely satisfiable.
    • The negation of every satisfiable formula is invalid, not necessarily unsatisfiable.
    • The negation of every invalid formula is satisfiable, not necessarily valid.

    Examples:

    • $(\forall x\:x=x)$ is valid, while its negation $(\exists x\,x\ne x)$ is unsatisfiable.
    • $(\exists x\,x\ne x)$ is unsatisfiable, while its negation $(\forall x\:x=x)$ is valid.
    • Both $\big(\forall x\,P(x)\big)$ and its negation $\big(\exists x\,¬P(x)\big)$ are satisfiable and invalid.
  3. What is the negation of a tautology?

  4. Set Complement $(A^\complement=\text{“non-}A\text”.)$ versus Formula Negation $(\lnot \psi=\text{“not }\psi\text”.)$

ryang
  • 44,428
0

The notions and their negations will get clearer once you realize the meta-logical quantifications they involve, and how these quantifications systematically behave under negation.

All the definitions you listed quantify over interpretations: A statement is valid iff it is true under all interpretations, satisfiable iff it is true under some interpretation, contradictory iff it is valid under no interpretation, and so on.

In general, we have

  • "not all interpretations" = "there exists some interpretation such that not";
  • "not some interpretation" = "for all interpretations not";
  • "not no interpretation" = "there exists some interpretation such that".

So let's apply these equivalences to each of the definitions*:

-------------------------------------------------------------------------------------------------------------------------    
Notion          Definition                           Negation of definition                  Negation of notion    
-------------------------------------------------------------------------------------------------------------------------

tautological all i. true (= no i. false) not all i. true (= some i. false) contradcictory or contingent

contradictory all i. false (= no i. true) not all i. false (= some i. true) satisfiable (= unsatisfiable)

contingent some i. true and some i. false not (some i. true and some i. false) contradictory or tautological = (not some i. true) or (not some i. false) = no i. true or no i. false = all i. false or all i. true

satisfiable some i. true (= not all i. false) no i. true (= all i. false) contradictory

unsatsifiable no i. true (= all i. false) some i. true (= not all i. false) satisfiable (= contradictory)

valid (see tautological) (= tautological)

invalid not valid = not all i. true all i. true tautological

So we have

---------------------------------------------------------------------
Notion              Negation               can but doesn't have to be
---------------------------------------------------------------------
not tautological    = not valid         
                    = invalid
                    = contradictory        unsatisfiable (if contradictory),
                      or contingent        satisfiable (if contingent)

not contradictory = not unsatisfiable tautological, = satisfiable contingent, invalid

not contingent = contradictory unsatisfiable, invalid (if contradictory), or tautological satisfiable (if tautological)

not satisfiable = unsatisfiable invalid (must be) = contradictory

not valid = invalid
= not tautological = contradictory unsatisfiable (if contradictory), or contingent satisfiable (if contingent)

not invalid = valid satisfiable (must be) = tautological

The problem with the negations your first table is that your negation is to strong: The negation of "all interpretations" is just "not all interpretations", i.e. "there are some interpretations such that not", and not (as you did) "no interpretation". So the negation of "valid" is just "not true under all interpretations", which can be contradictory or contingent, and not "true under no interpretation", which would be contradictory. Likewise, the negation of contradictory (= false under all interpretations) is just "not false under all interpretations", i.e. "true under some interpretations", which satisfiable, and not the stronger statement "true under all interpretations", which would be tautological.

The diagram you made is correct, and explicates the misunderstanding as follows: Negation does not mean opposite. Negating a notion does not get you to the other extreme of the diagram, only to the complementary half, i.e. the entire part not covered by that notion: "not contradictory" gives you everything in the range of "satisfiable", not just the extreme "tautology". "not tautological" just gives you "invalid", not the opposite "contradictory". "Not satisfiable" is "contradiction", not "invalid", "not invalid" is "tautology", not "satisfiable", and lastly, if something is "not contingent" it must be in either "contradiction" or "tautology". That covers all the possible cases.

* Note that in the case of first-order logic, sometimes a distinction is made between a formula that is valid (true in all first-order interpretations) and one that is tautological (a first-order instance of a propositional tautology, i.e. one that has the form of a tautological propositional formula but with predicate logical formulas in the place of propositional variables). All tautological formulas are also valid, e.g. $\forall x P(x) \lor \neg \forall x P(x)$, but not all first-order validities are merely instances of propositional tautologies, e.g. $\forall x (x = x)$.

  • Ohh I just revised question to add a line in each of those table. Will take some time to read your answer. – RajS Sep 09 '19 at 16:10
  • 1
    Please dont feel annoyed, but I found the language of a answer a bit difficult to get hold of. So, I skipped to your tables. Very first row of your first table says Notion=tautological and Negation of notion=contradcictory or contingent. That is definitely not correct, right? I said that already in my above question referring to exact same question I asked more than an year earlier. Negation of tautology is contradiction (which is unsatisfiable), it cannot be contingent, right? Or am I misinterpreting your answer? – RajS Sep 09 '19 at 19:35
  • I just realized that you were asking two different questions in your posts, and my answer was about one question (and what I said there certainly is correct) while in this post you seem to be more interested in something different (for which what I wrote does not apply). I'll rewrite my answer clarify the difference and address both. – Natalie Clarius Sep 09 '19 at 19:47
  • also can you please point out (1) if I made any mistake in those two tables (2) if there is any piece of information (about relation between those terms) which those two tables dont contain – RajS Sep 09 '19 at 20:39
  • Thank you for the tables @lemontree, they are very helpful. I noticed that the first table treats "tautological" and "valid" as equivalent, whereas other authors draw a distinction bewteen the two by defining a first-order tautology as a WFF whose truth-functional form is a tautology and that the set of FOL tautologies is a strict subset of logically valid WFFs (i.e. those that are true under all interpretations). Would you know why this is? I'm not sure which definition to go with. – user51462 Sep 18 '22 at 07:23
  • @user51462 The distinction is not universally made, some treat the two as synonymous, but it is perhaps better to avoid any confusion. I'll add a note. – Natalie Clarius Sep 19 '22 at 19:25