<


 

Help
Index

 Maarten Maartensz:    Philosophical Dictionary | Filosofisch Woordenboek                      

 L - Logic - Extended Propositional

 

Extended Propositional Logic: A standard bi-valent propositional logic that is able to distinguish between "it is not true that" and "it is false that", by considering the former as weaker than the latter, which thereby also has the ability to distinguish statements that are undecided (undetermined, uncertain), by defining them as the statements that are neither true nor false.

The interesting fact is that indeed this can be achieved with a bi-valent propositional truth-functional logic that is a simple extension of standard classical propositional logic.

This is shown in what follows, using ordinary algebra, and building on the work done for Classical Propositional Logic.


1. Introduction:

That "it is not true that" and "it is false that" may be motivated in many different ways. Here I shall do it by considering the following two arguments, one relating intuitively to space, the other to time.

First, there is e.g. "it's not true there's extra-terrestial intelligence and it's not false either, because we have neither found it nor searched exhaustively through all the known universe". The point is not whether or not there is extra-terrestial intelligence (or what this would mean), but that one may in many cases plausibly deny that a proposition is true, on the ground it has not been established as true and deny that the same proposition is false, on the ground that it has also not been established as false.

Second, there is Aristotle's example, who knew that two fleets were encircling one another preparing for battle or escape: In such a case "it is not true that tomorrow there is a sea-battle and it is not false either, because what will happen is not yet decided in this and many similar cases", of so-called "future contingents".

The conclusion that is usually drawn from arguments like this is that to do justice to such cases one must reject the postulate of bivalence, and introduce a third truth-value, and perhaps a fourth, a fifth a.s.o.

That conclusion is not drawn here. Instead, we maintain the postulate of bivalence but introduce a new negation, so that now we have the possibility of saying "it is false that p" i.e. "-p" and "it is not true that p" i.e. "~p", and also have the possibility of saying "it is neither true nor false that p" i.e. "~p&~-p" i.e. in brief "?p", which may be read as "p is undecided" or "p is uncertain". (I tend to use "undecided" for cases of a spatial nature, and "uncertain" in temporal cases.)

And the basic reason for distinguishing between it is not true that p and it is false that p, and therewith the definable notion of p is undecided or p is uncertain is that there intuitively are possibilities that are neither true nor false but uncertain, in the sense that there are many domains that are not fully known, not fully decided, not fully certain for some reason.

Two important domains of this kind where this distinction applies are the past and the future, at least if one assumes that what's past is done and settled, and definitely so or definitely not so (whether or not that is known) while part of the future is neither done nor settled and neither definitely so nor definitely not so (and that is what we seem to know about the future).

However, once the basic distinction is drawn between the two negations, there are as many possible applications as there are appropriate domains one's statements are about where a distinction of the given form applies.

There are many possible ways of providing a semantics for these ideas, but it seems the clearest way is to follow the ideas used above to give a semantical foundation for CPL, and add to CPL the following postulates:


2. Syntactics of EPL

As for CPL plus: 
            "-" and "?" are unary operators.

EPL builds on CPL, and therefore little is needed in the way of added syntax.


3. Semantics for EPL:

For the semantics for EPL I choose here to do it algebraically, and in terms of four axioms:

As for CPL plus: 
            E+    [+p]+[-p]+[?p] = 1
            E--   [--p]=[+p]
            E?    [-?p]=[+pV-p]
            E&    [-(+p&+q)]=[-+pV-+q]

Each of these axioms can be motivated intuitively:

E+ parallels C+ by simply refining [~p] as [-p]+[?p];
E-- parallels the theorem [~~p]=[p];
E?  says what's the case if it is false that p is undecided/uncertain: then p is either true or false; and
E& gives the De Morgan law for strong denial, which we shall call negation.

Note that EPL is an extension of CPL: Semantically and syntactically, EPL = CPL + the above four postulates, apart from the necessary syntaxis for the new operators, that also conforms to CPL. We shall inquire below into the precise relation between EPL and CPL, and find that the present claim is correct and can be precisified.

As with CPL, one basic aim will be to derive truth-table definitions. To do so, here are some theorems with comments.


4. Theorems of EPL in 1 variable:

The first theorem is used to establish later theorems. It also incorporates several statements concerning consistency: [p&-p]=0 etc. that may be summarized as: a proposition and its negation or uncertainty are never both true. (The same holds for a proposition and its denial, both in CPL and EPL.)

T1 [p][-p]=[p][?p]=[-p][?p]=0
[p&-p]=[p&?p]=[-p&?p]=0
Contradictions
(1)

[p]+[-p]+[?p]=1      

E+
(2) [p][p]+[p][-p]+[p][?p]=[p] Algebra, (1)
(3) [p]+[p][-p]+[p][?p]=[p] Algebra and BV
(4) [p][-p]+[p][?p]=0 Algebra, (3)
(5) [p][-p]=[p][?p]=0 BV, Algebra. (4)

The last line follows from the line before it because if a sum of two quantities is 0 and both quantities are either 0 or 1 both must be 0 by BV (of the four possibilities 1+1, 0+1, 1+0, 0+0 only the last sums to 0).

The second equation on the first line is proved likewise, and the other equations follow by P&.

T2 establishes a fundamental fact about the negation of an uncertain proposition: it has the same value as the denial of the same uncertain proposition:

T2

[-?p]=[~?p]

Denial of uncertainty is
negation of uncertainty  
(1) [-?p]=[pV-p] E?
(2)

      =[p]+[-p]-[p][-p]

PV, (1)
(3)

      =[p]+[-p]             

T1, (2)
(4)       =1-[?p]                E+, (3)
(5)       =[~?p]                 P~, (4)

T3 establishes that the assertion that a proposition is uncertain has the same value as the assertion that the negation of the proposition is uncertain:

T3

[?p]=[?-p]

Uncertainty of truth is equivalent to uncertainty of falsity
(1) [~?-p]=~[-pV--p] E?
(2)

         = 1-([-p]+[--p]-[-p][--p])

PV, (1) (CPL keeps applying)
(3)          = 1-([-p]+[p]-[-p][p]) E-, (2)
(4)          = 1-[pV-p] PV, T1, (3)
(5)          = [~?p] E+, (4)

T4 establishes that the assertion that an uncertain proposition is itself uncertain is always false. (Of course, T4 can be restated as [-??p]=[~??p]=1.)

T4

[??p]=0

There are no true uncertain uncertain propositions
(1) [?p]+[-?p]+[??p]=1 E+
(2) [??p]=1-([?p]+[-?p]) Algebra, (1)
(3)        =1-([?p]+[~?p]) T2, (2)
(4)        =1-1=0 P~, (3)

T5 establishes that the value of any proposition equals the conjunction of the denials of the same proposition when negated and when uncertain:

T5

[p] =[~-p][~?p]=[~-p&~?p]
[-p]=[~p][~?p] =[~p&~?p]
[?p]=[~p][~-p] =[~p&~-p]

Fundamental EPL equivalences
(1) [~-p][~?p]=(1-[-p])(1-[?p]) P~
(2) =1-[-p]-[?p]+[-p][?p]  Algebra, (1)
(3) =1-[-p]-[?p]  T1, (2)
(4) =[p]  E+, (3)

This proves the first equation. The other two are proved likewise. Qed.

T6 shows denial can be wholly avoided and defined in terms of -,? and V. Incidentally, this shows that the classical denial used in CPL indeed is systematically ambiguous, at least from the point of view of EPL: saying something is not so generally is ambiguous between claiming it is false or it is uncertain.

T6

[~p]=[-pV?p]
[~-p]=[pV?p]
[~?p]=[pV-p]

Denials eliminable
(1) [~p]=1-[p] P~
(2)       =[-p]+[?p] E+
(3)       =[-p]+[?p]-[-p][?p] T1
(4)       =[-pV?p] PV

The first equation is proved as indicated, and the second likewise. The third already follows by T2 and E?.

T7 shows that a negated denial is like a double denial or double negation:

T7 [-~p]=[p] Negation of denial is affirmation
(1) [-~p]=[-(-pV?p)] T6
(2)         =[--p&-?p] E-, (1)
(3)         =[p&-?p] E--, (2)
(4)         =[p][-?p] P&, (3)
(5)         =[p](1-[?p]) T2, (4)
(6)         =[p]-[p][?p] Algebra, (5)
(7)         =[p] T1, (6)

So now we can write a truth-table for propositions with one variable. Note that it is considerably larger than for CPL, since there are more distinct cases to include. Also - and different in principle from the standard truth-tables for standard CPL - the left-most column gives the fundamental possibilities for p (thus avoiding the need for three truth-values) while the upper row gives the diverse propositions with "p" that are evaluated for each fundamental possibility.

Fundamental truth-table for EPL

Pos

"p"

"-p"

"?p"

"~p"

"~-p"

"~?p"

"-?p"

"?-p"

"??p"

"--p"

"~~p"

"-~p"

'p'

1

0

0

0

1

1

1

0

0

1

1

1

'-p'

0

1

0

1

0

1

1

0

0

0

0

0

'?p'

0

0

1

1

1

0

0

1

0

0

0

0

The point of the two styles of quotation-marks in the above table is to distinguish between the propositions that are evaluated, listed in the top row between double quotes, so that "p" may conveniently be read as "the proposition that p", and the possible facts that would make these propositions true or false, so that 'p' may be conveniently read as "the fact that p".

Note that both "p" and 'p' are linguistic notation, the first for the linguistic expression the double quotes embrace, the second for the possible fact represented by the linguistic expression the single quotes embrace. Also, the map that is encoded by the above table may be read in two ways: from single-quoted formulas to double quoted formulas, or from possible facts represented by single quoted formulas to ideas represented by double quoted formulas.

In the tables that follow both kinds of quotation marks will be left out as understood for reasons of space and clarity. (And the reader who is familiar with logic should realize that the present approach to truth-tables and semantics of propositional logic is in some ways subtly different from the received way, even though the resultant tables look much the same.)


5. Theorems of EPL in 2 variable:

T8 establishes a kind of general expansion law, corresponding to the PL- version [p]=[p&q]+[p&~q], which may be proved in a similar way:

T8

[p] =[ p&q]+[ p&-q]+[ p&?q]
[-p]=[-p&q]+[-p&-q]+[-p&?q]
[?p]=[?p&q]+[?p&-q]+[?p&?q]

EPL-Expansion
(1) [p]=[p]([q]+[-q]+[?q]) E+       
(2)     =[p][q]+[p][-q]+[p][?q] Algebra, (1)
(3)     =[p&q]+[p&-q]+[p&?q] T1, (2)

The first equation is proved as indicated.The other two are proved likewise. Qed.

T9 uses T8 to establish the EPL-counterpart of [p&q]+[p&~q]+[~p&q]+[~p&~q] =1, that may be proved in the same manner:

T9

[p&q]+[p&-q]+[p&?q]+
[-p&q]+[-p&-q]+[-p&?q]+
[?p&q]+[?p&-q]+[?p&?q]=1

EPL-Tautology

The proof uses T8 and E+. Qed.

Note that for CPL there are four distinct possibilities for any two distinct propositions, whereas for EPL there are nine distinct fundamental possibilities for any two distinct propositions (which also shows EPL cannot coincide with a 3-valued logic, since that distinguishes instead 2^3=8 distinct fundamental possibilities: EPL just involves a somewhat finer grid of logical distinctions, so to speak.)

Note also that by Bivalence and Algebra it follows as in CPL that precisely one of these possiblities will hold in any case any of these may hold or else none holds, if the proposition is a contradiction.

T10 shows that an uncertain conjunction has the same value as either conjunct being true and the other uncertain or both being uncertain. It also shows that ? in front of a conjunction can be worked inwards:

T10

[?(p&q)]=[p&?q]+[?p&q]+[?p&?q]

Uncertainty conjunction
(1) [?(p&q)] =1-[p&q]-[-(p&q)] E+       
(2) =1-[p&q]-[-pV-q)] E&, (1)
(3) =1-[p&q]-([-p]+[-q]-[-p][-q]) PV, (2)
(4)

=1-[p&q]-[-p&q]-[-p&-q]-
     [-p&?q]-[-q&p]-[-q&-p]-
     [-q&?p]+[-p][-q]

T8, (3)
(5) =1-[p&q]-[-p&q]-[-p&-q]-
     [-p&?q]-[-q&p]-[-q&-p]-
     [-q&?p]+[-p&-q]
P&, (4)
(6) =1-[p&q]-[-p&q]-[-p&?q]-
     [p&-q]-[-p&-q]-[-q&?p]
Algebra
(7) =[p&?q]+[?p&q]+[?p&?q] T9, (6)

T11 expands E- to a more precise statement. Note that after this point the truth-table for conjuctions can be written:

T11

[-(p&q)]=[p&-q]+[-p&q]+[-p&-q]+[-p&?q]+[?p&-q]

 

The proof is by T9, T10, P&, E- and E+.

Here is the truth-table for the basic conjunctions, which one may compare with one's intuitions, noting especially (1) that a conjunction is uncertain iff one conjunct is true and another uncertain or both conjuncts are uncertain and (2) that in fact the frontal "-" and "?" can be worked inside i.e. detached from the compound formula and attached to (a disjunction of conjunction of) its components like T10 and T11 show.

 

(p&q)

-(p&q)

?(p&q)

 p  q

1

 

 

 p –q

 

1

 

 p ?q

 

 

1

-p  q

 

1

 

-p -q

 

1

 

-p ?q

 

1

 

?p  q

 

 

1

?p -q

 

1

 

?p ?q

 

 

1

T12 parallels T10: An disjunction is uncertain iff either one disjunct is false and the other uncertain or both disjuncts uncertain:

T12

[?(pVq)]=[-p&?q]+[?p&-q]+[?p&?q]

Uncertainty disjunction
(1)

?(pVq) =~(pVq)&~-(pVq)

T5
(2)           =~p&~q&(~-pV~-q) PL
(3)           =~p&~q&~-p V ~p&~q&~-q PL
(4)            =?p&-q V ?p&?q V ?q&-p T5

T13 establishes the counterpart of EM, i.e. the dual of De Morgan's Law. After this point we also can write the EPL-truth-table for V.

T13

[-(pVq)]=[-p&-q]

De Morgan for negation
(1) [-(pVq)]=1-[pVq]-[?(pVq)] E+       
(2) =1-[p]-[q]+[p&q]-?(pVq) PV, (1)
(3) =1-[p&q]-[p&-q]-[p&?q]-[q&p]-
     [q&-p]-[q&?p]-?[pVq]
T5, (2)
(4) =[-p&-q]+[-p&?q]+
  [?p&-q]+[?p&?q]-?[pVq]
T9, (3)
(5) =[-p&-q]+[-p&?q]+
  [?p&-q]+[?p&?q]-
  [
?p&-q]-[?p&?q]-[?q&-p]
T11, (4)
(6) =[-p&-q] Algebra, (5)

Here is the truth-table for the basic disjunctions:

 

(pVq)

-(pVq)

?(pVq)

 p  q

1

 

 

 p -q

1

 

 

 p ?q

1

 

 

-p  q

1

 

 

-p -q

 

1

 

-p ?q

 

 

1

?p  q

1

 

 

?p -q

 

 

1

?p ?q

 

 

1


6. Theorems about basic uncertainties

In a sense, the semantical basis of EPL has been completed, for as in the case of CPL any formula may be replaced by an equivalent DNF (=Disjunctive Normal Form: A disjunction of mutually exclusive conjunctions). However, we are interested in more than the minimally adequate semantical basis, and want to know something about the behaviour of ?; how we may introduce implication and equivalence; and what we can say in the context of EPL about the problems we have raised about CPL.

So let's first consider ?.

First, let's prove two theorems that parallel T3: [?p]=[?-p]. These have the same import: For simple propositions, conjunctions and disjunctions, the uncertain true cases have the same values as the uncertain false cases:

T14

[?(p&q)] =[?-(p&q)]

 
(1) [?(p&q)]  = [p&?q]+[?p&q]+[?p&?q] T10       
(2) [?-(p&q)] = [?(-pV-q)] E-
(3)

              = [p&?q]+[?p&q]+[?p&?q]

E-, T12

And

T15

[?(pVq)]=[?-(pVq)]

 
(1) [?(pVq)] =[-p&?q]+[?p&-q]+[?p&?q] T12       
(2) [?-(pVq)]=[?(-p&-q)] T13
(3)

             =[-p&?q]+[?p&-q]+[?p&?q]

E-, T10

Next, while we have that [?p]=[?-p], we also have that denials when fronted with ? are always false:
 

T16

 [?(~p)]=0

 
(1) [?(~p)]=[?(-pV?p)] T6       
(2)

          =[--p&??p]+[?-p&-?p]+[?-p&??p]

T12
(3)           =0 T1, T4

We can restate T16 thus, stressing several tautologies in only one proposition. Indeed, in T17 the basic tautology of CPL is restated equivalently in several ways not available in CPL:

T17

[~?(~p)]=[-?(~p)]=1

EPL-tautologies in one proposition
(1)              =[~pV-~p]        
(2)              =[~pVp]  
(3)              =[~(p&-p)]  
(4)              =[~(p&~p)]  

Next, we can use T10 to derive the various cases of conjunctions of uncertain propositions. The first two are nearly immediate from T10:

T18

[?(-p&q)]=[-p&?q]+[?p&q]+[?p&?q]

T10

And

T19

[?(-p&-q)]=[-p&?q]+[?p&-q]+[?p&?q]

T10

The next two can also be read off from earlier results, but I'll give the proof of the first, showing the interplay between denial, negation, and uncertainty:

T20

?(p&?q)]=[?p&q]+[?p&-q] 

 
(1)

[?(p&?q)] = [~(p&?q) & ~-(p&?q)]

E?       
(2)              = [(~pV~q) & ~(-p V -?q)] E~, E-, (1)
(3)              = [(~pV~q) & ~-p & -?q)] E~, (2)
(4)              = [?p&q V ?p&-q] E?, (3)
(5)              = [?p&q]+[?p&-q] E+, (4)

And

T21

 [?(-p&?q)]=[?p&q]+[?p&-q]

T20

The last two cases involving uncertain conjunctions involving uncertain propositons conform to T4:

T22 [-?(?p&?q)]=1  
(1) [-?(?p&?q)]=[~?(?p&?q)] T4
(2)                =[~(~(?p&?q) & ~-(?p&?q)] T5, (1)
(3)                =[~((-?pV-?q) & (?p&?q)] T4, (2)
(4)                =[(?p&?q) V ~(?p&?q)] T4, (3)
(5)                =1 P+, (4)

And so

T23

1-[?(?p&?q)]=1

T22, T4

Note that there are three possibilities for a conjunction of two certain propositons to be uncertain; two possibilities for a conjunction of a certain and an uncertain proposition to be uncertain; while no conjunction of two uncertain propositions can ever be uncertain. This also may be stated as follows:

T24

[+(?p&?q) V -(?p&?q)]=1

A conjunction of two uncertainties is true or false

T12 can likewise be used to derive the various cases of disjunctions of uncertain propositions that mirrors the case for uncertain conjunctions in a way similar to De Morgan's Laws. The proofs are similar to the case of conjunction:

T25

[?(-pvq)]=[p&?q]+[?p&-q]+[?p&?q]

T12

T26 [?(-pv-q)]=[p&?q]+[?p&q]+[?p&?q] T12

T27 [?(pV?q)]=[?p&q]+[?p&-q] T12

T28 [?(-pV?q)]=[?p&q]+[?p&-q] T12

T29

[-?(?pV?q)]=1

T12

Here is a set of noteworthy consequences of T20, T21, T25 en T26:

T30

[?(pV?q)] =[?(-pV?q)]

 
(1)              =[?(p&?q)]        
(2)              =[?(-p&?q)]  
(3)

             =[?p&q]+[?p&-q]=[+(?pV-?q)]

 

This shows that uncertain conjunctions and disjunctions of a certain and an uncertain proposition amount to the same. An intuitive way of making sense of this in terms of time is that such disjunctions and conjunctions refer in one conjunction or disjunction to both the future and the non-future (past or present).

Here are the basic tables for conjunctions and disjunctions collected in one table:

 

(p&q)

-(p&q)

?(p&q)

(pVq)

-(pVq)

?(pVq)

 p  q

1

 

 

1

 

 

 p -q

 

1

 

1

 

 

 p ?q

 

 

1

1

 

 

-p  q

 

1

 

1

 

 

-p -q

 

1

 

 

1

 

-p ?q

 

1

 

 

 

1

?p  q

 

 

1

1

 

 

?p -q

 

1

 

 

 

1

?p ?q

 

 

1

 

 

1

 


7. Theorems about compounded uncertainties

T31 uses T10 and T11 to state when the possible values of an EPL-form in two variables are classical: Iff both their conjunction and disjunction is certain.

T31 [-?(p&q) &-?(pVq)]=
[p&q]+[p&-q]+[-p&q]+[-p&-q]
EPL to CPL-resolution
(1)

[-?(p&q)&-?(pVq)]=[-?(p&q)][-?(pVq)]

P&

(2)                         =[~?(p&q)][~?(pVq)] T2, (11)
(3a)

     =[[p&q]+[p&-q]+[-p&q]+
       [-p&-q]+[-p&?q]+[?p&-q]]

T10, (2)
(3b)

    * [[p&q]+[p&-q]+[p&?q]+
       [-p&q]+[-p&-q]+[?p&q]]

T12, (2)
(4)      =[p&q]+[p&-q]+[-p&q]+[-p&-q] T1, (3a), (3b)

Note this does give a fairly neat criterion and distinction: two propositions are both either true or false precisely if both their conjunction and their disjunction is either true or false. Of course, this also runs the other way: Of two propositions at least one is uncertain iff their conjunction or disjunction is uncertain.

In the proof of T31 the third and fourth line are one long multiplication printed over two lines for readability. This will be done below also without clarification. Also T1 is applied to quickly dispose of inconsistent conjunctions.

It helps intuitively to consider the other three similar cases.

I start with an uncertain conjunction but a decided disjunction: This turns out as equivalent with one of the two propositions true and the other uncertain

T32

[?(p&q) & -?(pVq)]=[p&?q]+[?p&q]

 
(1a) [?(p&q) & ?(pVq)]=  
  
[[p&?q]+[?p&q]+[?p&?q]]*
T9, T10, T12       
(1b)    [[p&q]+[p&-q]+[p&?q]+[-p&q]+
    [-p&-q]+[?p&q]]
 
(2)  =[p&?q]+[?p&q] T1, (1a), (1b)

Next, a decided conjunction but an uncertain disjunction of two propositions is equivalent with the falsity of one proposition and the uncertainty of the other:

T33

[-?(p&q) & ?(pVq)]=[-p&?q]+[?p&-q]

 
(1a) [-?(p&q) & ?(pVq)]=   
  [[p&q]+[p&-q]+[-p&q]+
    [-p&-q]+[-p&?q]+[?p&-q]]*
       
(1b)     [[-p&?q]+[?p&-q]+[?p&?q]]  
(2)  =[-p&?q]+[?p&-q]  

Finally, when both the conjunction and disjunction of two propositions are uncertain, both propositions are uncertain, and conversely:

T34

[?(p&q) & ?(pVq)]=[?p&?q]

 
(1a)

[?(p&q) & ?(pVq)]=[[p&?q]+[?p&q]+[?p&?q]]*

       
(1b) [[-p&?q]+[?p&-q]+[?p&?q]]  
(2) =[?p&?q]  

Here is a table summarizing T30-T32, showing some neat symmetries, that also show that combinations of conjunctions and disjunctions can neatly isolate the possibilities with no uncertainties, one uncertainty and two uncertainties.

It is also interesting to note - in the middle two columns - that in the most interesting case of one uncertainty, it does not follow which of the two propositions is uncertain:
 

 

-?(p&q)&-?(pVq)

?(p&q)&-?(pVq)

-?(p&q)&?(pVq)

?(p&q)&?(pVq)|

+p+q

1

 

 

 

+p -q

1

 

 

 

+p ?q

 

1

 

 

-p +q

1

 

 

 

-p -q

1

 

 

 

-p ?q

 

 

1

 

?p +q

 

1

 

 

?p -q

 

 

1

 

?p ?q

 

 

 

1


8. Theorems about implication and equivalence

Now let's briefly consider implication and equivalence in EPL, since implication and equivalence are the basic principles of inference (when inference is construed intuitively as did Leibniz: as concerned with inferring necessary consequences from given assumptions and substituting definitions (of equivalents) in known theorems).

Let's start with a simple and basic point about implication.

It is obvious that in EPL several styles of implication can be defined, notably the following two, where I shall write "I" for "implies": (1) [pIq]=[~pVq] or else (2) [pIq]=[-pVq]. So the question rises which of these two (and related other ones) is the intuitively correct rendering of implication, where "intuitively correct" means that the chosen definition should conform to one's intuitions about valid inferences.

Put in these terms, it is obvious that [pIq]=[~pVq] is to be preferred over [pIq]=[-pVq], because the latter if adopted would validate inferences of the form "pIq & ?p therefore q" (as ?p and -p exclude each other). It makes no sense to infer conclusions from uncertainties plus implications in this way, and so the basic implication is the one used in CPL.

The first truth-table with relevant information is this, with "T" written for "1" in some of the lines for clarity's sake:

 

(~pVq)

(~qVp)

(pBq)

+p -->+q &
-p --> -q &
?p --> ?q

+q -->+p &
-q --> -p &
?q --> ?p

(pEq)

 p  q

1

1

1

1

1

1

 p -q

 

1

 

 

 

 

 p ?q

 

1

 

 

 

 

-p  q

1

 

 

 

 

 

-p -q

1

1

1

1

1

1

-p ?q

1

1

1

 

 

 

?p  q

1

 

 

 

 

 

?p -q

1

1

1

 

 

 

?p ?q

1

1

1

1

1

1

Columns 1 and 2 provide the definitions corresponding to the [~pVq]=[p-->q] and [~qVp]=[q-->p]. If we call the lines on which a ? is occurring a non-classical line, and the other ones, i.e. 1,2,4 and 5 classical it will be seen that EPL adds 5 non-classical cases to the 4 classical ones, in all of which but one, namely [p&?q], [p-->q] holds, all simply by definition.

Now if we follow the definition of equivalence as in CPL we get [p iff q] = [(p-->q)&(q-->p)]. This leads to the cases [(-p&?q) --> (p iff q)]=1 and [(?p&-q) --> (p iff q)]=1, as column 3 shows, with "B" ("bi-implication") for "iff" for brevity.

This is not intuitive: although in both cases neither p nor q is true, it does not follow intuitively that in these cases therefore p and q are to come out as equivalent.

So here we reap a consequence of refining ~p to -pV?p: In CPL - and ? collapse to ~, and so both (-p&?q) and (?p&-q) only can come out as (~p&~q), which does imply (p iff q) intuitively - in CPL.

But in EPL matters are more refined and not quite as simple.

Columns 4, 5 and 6 provide the outlines for a solution, which resides in the fact that [p-->q&-p-->-q&?p-->?q]=[q-->p&-q-->-p&?q-->-p]=[p&q]+[-p&-q]+ [?p&?q]=1, abbreviated to "pEq" in column 6. Note that each of the columns 4, 5 and 6 amount to the same, i.e. that each of these amounts to pEq i.e. [p&q]+ [-p&-q]+[?p&?q]. Put as a theorem:

T35 [pEq] = [p&q]+[-p&-q]+[?p&?q]   
(1)

        = [p-->q & -p-->-q & ?p-->?q]

       
(2)         = [q-->p & -q-->-p & ?q-->?p]  

Note there is a parallel in CPL: In CPL [pEq]=[p-->q & ~p-->~q]. As before, EPL merely refines the analysis of denials.

Further relevant facts concerning pEq can be seen from the next table:

 

1

2

3

4

5

6

 

piffq

-piff-q

?piff?q

(pEq&-pE-q) E
(-pE-q&?qE?p)

(pEq&?pE?q) E
(-pE-q&pEq)

pEq

+p +q

1

1

1

1

1

1

+p -q

 

 

 

 

 

 

+p ?q

 

1

 

 

 

 

-p +q

1

 

1

 

 

 

-p -q

1

1

1

1

1

1

-p ?q

 

 

 

 

 

 

?p +q

 

1

 

 

 

 

?p -q

1

 

 

 

 

 

?p ?q

1

1

1

1

1

1

The theorem that can be read from this is to the effect that pEq is true precisely if p and q have the same signs. In other words: p and q are equivalent iff -p and -q are equivalent iff ?p and ?p are equivalent - as fits one's linguistic intuitions that equivalent propositions must have the same basic unitary operators (to which "~" does not belong):

T36

(pEq&-pE-q)=(-pE-q&?qE?p)

 
(1)                    =(pEq&?pE?q)        
(2)                    =(-pE-q&pEq)  
(3)                    =pEq  

This in turn suggests the following definitions of compatible, incompatible and contrary propositions:

Compatible(p,q) = def

= pEq = [+p&+q]+[-p&-q]+[?p&?q]

Incompatible(p,q)

= def = [+p&?q]+[-p&?q]+[?p&+q]+[?p&q]       
    = [?(-pV-q)]
    =?[-(+p&+q)]
    = [?(+p&+q)]
    = ?(pEq)
Contraries(p,q) = def = [+p&-q]+[-p&+q]
    = -(pEq)

Briefly, compatible propositions are equivalent i.e. have the same basic operator; incompatible propositions are uncertainly inequivalent or conjunctively uncertain in that one is certain and the other uncertain, which is the case precisely if their conjunction is uncertain; and contrary propositions are each others negations. Note that it follows from the definitions that, since the alternatives are exhaustive and exclusive that:

T37 Any two propositions are either compatibles or else incompatibles or else contraries.  

Again, one fundamental way of making sense of incompatible propositions is by noting that intuitively, when speaking of uncertainties and time, two propositions are incompatible if one refers to the contingent future and the other doesn't i.e. one is uncertain and the other certain. It is noteworthy that the non-incompatible pairs of propositions are precisely the classical ones plus the case both are uncertain.

Hence another way of dividing the 9 cases of pairs of propositions is: Both uncertain; both certain and equivalent or both certain and inequivalent; or mixed i.e. one certain the other uncertain.

 

See also: CPL, Logic, Mathematics, Propositional Logic,


Literature:

Maartensz, Körner, Zinoviev

The above semantics first appeared in my M.A.-thesis.

 Original: Nov 14, 2005                                                  Last edited: 12 December 2011.   Top