Notre Dame Journal of Formal Logic

Volume 43, Number 3, 2002

Paraconsistency Everywhere

Greg Restall

Abstract Paraconsistent logics are, by definition, inconsistency tolerant: In a

paraconsistent logic, inconsistencies need not entail everything. However, there
is more than one way a body of information can be inconsistent. In this paper
I distinguish contradictions from other inconsistencies, and I show that several
different logics are, in an important sense, “paraconsistent” in virtue of being
inconsistency tolerant without thereby being contradiction tolerant. For exam-
ple, even though no inconsistencies are tolerated by intuitionistic propositional
logic, some inconsistencies are tolerated by intuitionistic predicate logic. In this
way, intuitionistic predicate logic is, in a mild sense, paraconsistent. So too
are orthologic and quantum propositional logic and other formal systems. Given
this fact, a widespread view—that traditional paraconsistent logics are especially
repugnant because they countenance inconsistencies—is undercut. Many well-
understood nonclassical logics countenance inconsistencies as well.

1 Inconsistency and Contradiction

“Paraconsistent” means “beyond the consistent” (Arruda [3], Priest and Routley
[13]). Paraconsistent logics tolerate inconsistencies in a way that traditional logics
do not. In a paraconsistent logic, the inference of explosion
A, ∼A ` B
is rejected. This may be for any of a number of reasons (Priest et al. [10]). For
proponents of relevance (Anderson and Belnap [1], Anderson et al. [2]) the argu-
ment has gone awry when we infer an irrelevant B from the inconsistent premises.
Those who argue that inconsistent theories may have some logical content but do not
commit us to everything have reason to think that these theories are closed under a
relation of paraconsistent logical consequence (Meyer and Martin [9], Routley et al.
[15]). Another reason to adopt a paraconsistent logic is more extreme. You may

Received February 13, 2002; accepted February 5, 2003; printed December 19, 2003
2001 Mathematics Subject Classification: Primary, 03B53; Secondary, 03A05
Keywords: paraconsistent logic, intuitionistic logic, quantum logic
c
2003 University of Notre Dame

147
148 Greg Restall

take the world to be inconsistent (Priest [12]), and a true theory incorporating this
inconsistency must be governed by a paraconsistent logic.
However, not all inconsistencies are straightforward contradictions. As a simple
example, consider the set {A ∨ B, ∼A, ∼B}. It is as inconsistent as can be, yet it
contains no contradictory pair of formulas. This set is inconsistent, and it is classi-
cally unsatisfiable without containing an explicit contradiction. Of course, we can
note that to members of the set, ∼A and ∼B together entail ∼(A ∨ B) and this is
the negation of a formula in this set. Therefore, we might say that the set contains an
implicit contradiction without containing an explicit one. The fact that some incon-
sistent sets are not themselves explicit contradictions motivates a closer look at the
definition of paraconsistency. Let’s specify what it is for a consequence relation to
be paraconsistent in the following two ways:
1. A consequence relation ` is contradiction tolerant if and only if for some
formulas A and B, the contradictory set {A, ∼A} does not entail B; that is,
A, ∼A 6` B.
2. A consequence relation ` is inconsistency tolerant if and only if for some
inconsistent set X and some formula B we have X 6` B.
A contradiction tolerant consequence relation is also inconsistency tolerant—as the
set {A, ∼A} is inconsistent. However, it is not at all obvious that the converse holds.
Perhaps there are inconsistency tolerant consequence relations which are not contra-
diction tolerant. Such consequence relations are the focus of the rest of this paper.

2 Inconsistency Tolerance
To judge whether or not a relation is inconsistency tolerant we must know when a
set is inconsistent. This makes a judgment about inconsistency tolerance depend on
a judgment about what could count as an inconsistency. This might appear to make
the notion of inconsistency tolerance more problematic than contradiction tolerance,
which has the virtue of being much more straightforward to check for. However, this
appearance is misleading. While it seems straightforward to check for the presence
of an explicit contradiction in a given set, this requires at least some judgment. In
particular, you must know what counts as a negation in the logic in question. For
example, the classical modal logic S5 is contradiction tolerant, if we take ♦¬ (com-
bining possibility with the Boolean negation of classical logic) to be the negation
in question. We might argue over whether or not ♦¬ deserves to be called “nega-
tion,” and this argument is similar to an argument over whether or not a set deserves
to be thought of as inconsistent. Determinations of inconsistency tolerance require
an account of consistency, and determinations of contradiction tolerance require an
account of negation.
There are a number of different possibilities for characterizing inconsistency. Let
me consider some here.
1. Inconsistency as unsatisfiability: A set X is inconsistent if and only if X ` A
for each A.1
2. Inconsistency as contradiction entailing: A set X is inconsistent if and only
if there is some A such that X ` A and X ` ∼A.
Fact 2.1 These two characterizations of inconsistency agree if the consequence
relation ` satisfies explosion, transitivity, and the structural rule of contraction and
if negation is present.
 Paraconsistency 149

Proof Suppose we have X ` A and X ` ∼A. Then transitivity applied to X ` A

and A, ∼A ` B gives X, ∼A ` B, and transitivity again, with X ` ∼A gives
X, X ` B. Contraction, then, supplies X ` B. B was arbitrary, so X is unsatisfiable:
it entails every formula whatsoever. Conversely, if X ` B for every B, then for any
A, X ` A and X ` ∼A (at least if negation is in the language in question). 
In the context of a logic rejecting explosion, inconsistency as unsatisfiability is a
much stronger requirement than contradiction entailment. In fact, in many para-
consistent logics (such as first-degree entailment [1] or Priest’s propositional logic
LP [11]) no finite set of formulas is unsatisfiable, but many entail contradictions. So
these notions can come apart.

3 Inconsistency Tolerant Logics

With these distinctions at hand, we can now begin to consider what some have taken
to be a decisive failing of paraconsistent logics. For some opponents of paraconsis-
tency, paraconsistent logics are especially bad because they are inconsistency toler-
ant: they take as “possible” things which are genuinely impossible (Slater [16]). In
the semantics for paraconsistent logics, valuations (or worlds, or setups, or situations,
or what-have-you) allow inconsistencies to be true, and no sense can be supplied to
this notion. The paraconsistentist countenances A and ∼A being true (while at the
very same time some other B is not true) but the critic cannot see what it is for A and
∼A to both be true together.
Now it is not my place to endorse this reasoning (Restall [14]). It begs the question
against the paraconsistentist, if it is an argument at all and not merely an expression
of an inability to understand. My point here is that this objection, if it is any good
at all, applies equally to logics which are not paraconsistent in the traditional sense.
That is, inconsistency tolerance is just as bad as contradiction tolerance. For the
objection that there is no sense to be made of the joint truth of A and ∼A applies
just as well to any other inconsistent set. So, for the rest of this paper, I will look at
how this objection applies in two cases: intuitionistic predicate logic and orthologic,
as these are both inconsistency tolerant logics. To fruitfully continue this discussion
we ought to settle on a notion of inconsistency, and for simplicity, I will choose this
notion:
Inconsistency as classical unsatisfiability: A set X is inconsistent if
and only if X ` K A for each A, where ` K is classical logical conse-
quence.
This is not problematic, because the opponents of paraconsistency most often en-
dorse classical consequence. We will see that their objections, if sustainable against
paraconsistent logics, ought to apply much more generally to other nonclassical log-
ics too. But before discussing our two main examples, let us show that the notion
of paraconsistency as tolerating classical inconsistency is not a completely trivial
notion. It does not include every nonclassical logic.
Non-Example 3.1 Intuitionistic propositional logic is not tolerant of classical in-
consistency. That is, if X ` K A for each A, then X ` J A for each A too, where ` K
and ` J are classical and intuitionistic propositional consequence, respectively.
Proof Suppse X ` K A for each A. Then it follows that X ` K ⊥ where ⊥ is
some contradiction. By compactness, X 0 ` K ⊥ where X 0 is a finite subset of X.
Take B to be the conjunction of X 0 , and then B ` K ⊥. It follows that ` K ∼B.
150 Greg Restall

Now if a formula is provable in classical propositional logic, its double negation is

intuitionistically provable. Therefore ` J ∼∼∼B. But ∼∼∼B is intuitionistically
equivalent to ∼B, so ` J ∼B. As a result, B ` J ⊥ and hence X 0 ` J ⊥ and X ` J ⊥.
That is, X is intuitionistically unsatisfiable. 
So even though intuitionistic propositional logic is strictly weaker than classical
propositional logic, this weakness does not apply when it comes to proving incon-
sistency. If X ` K ⊥ then X ` J ⊥ too. However, this does not apply in the case of
predicate logic, as we will see.
Example 3.2 Intuitionistic predicate logic is tolerant of some classical inconsisten-
cies. That is, there are sets X where X ` K A for each A, but X 6` J A for each
A too, where ` K and ` J are now classical and intuitionistic predicate consequence,
respectively.
Proof In order to give a concrete example, I will present the Kripke semantics for
intuitionistic propositional logic. I will attempt to do with as little technicality as
possible. Introductions for intuitionistic predicate logic are available elsewhere (van
Dalen [17] and [18], Dummett [6], Fitting [7]). For us, an interpretation for the lan-
guage of intuitionistic predicate logic will consist of a domain C of constructions,
partially ordered by a relation v of inclusion, such that for each c ∈ C, D c is the
domain of objects constructed by C. If c v c 0 (c0 is a stronger construction than c)
then we must have Dc ⊆ Dc0 : anything constructed by c is also constructed by c 0 .
An infinite sequence α assigning an element of Dc for each variable in the language
is said to be an assignment fit for c. As is customary, we are interested in varying
assignments one variable at a time. In our case, α(x := d) is the assignment which
agrees with α about the values of each variable except for x, to which this new as-
signment gives the value d. The final element in an interpretation is the relation
of forcing (or constructing, or proving) between a construction, together with an as-
signment fit for that construction, and a formula (possibly containing free variables).
So an interpretation relation is a quadruple hC, v, D, i. The assignment relation
must satisfy these inductive clauses.
1. c, α A ∧ B if and only if c, α A and c, α B.
2. c, α A ∨ B if and only if c, α A or c, α B.
3. c, α A ⊃ B if and only if for any c 0 w c, if c0 , α A then c0 , α B.
4. c, α ∼A if and only if for any c 0 w c, c0 , α 6 A.
5. c, α ∃x A if and only if for some d ∈ Dc , c, α(x := d) A.
6. c, α ∀x A if and only if for any c 0 w c and any d ∈ Dc0 , c0 , α(x := d) A.
An entailment X ` A holds according to a particular interpretation hC, v, D, i if
for every c ∈ C and every α appropriate for c, if c, α B for every B ∈ X then
c, α A. An entailment X ` A holds in intuitionistic predicate logic if and only if
it holds in every interpretation.
I will not tarry to discuss the significance of these clauses here: suffice to say that
they are well motivated by the Brouwer, Heyting, Kolmogorov (BHK) interpretation
of constructions, and the resulting logic is weaker than classical logic. However, to
my knowledge, no one has claimed that the logic is so weak as to interpret impos-
sibilities which cannot be understood. The constructive account of the connectives
makes sense, given constructive motivations. However, it is not difficult to find clas-
sical inconsistencies tolerated in models for intuitionistic predicate logic.
 Paraconsistency 151

Here is a simple example of an inconsistency tolerance. We will examine an

interpretation verifying that
∼∀x(F x ∨ ∼F x) 6` B.
The interpretation is straightforward. The set C of constructions is the infinite set
{c0 , c1 , c2 , . . .}, ordered with ci v c j if and only if i ≤ j . Each construction has the
same domain Dc j = {0, 1, 2, . . .} at each construction. Finally, let’s set F(i ) true at
c j if and only if i ≤ j . (More precisely, c j , α F x if and only if α(x) = i and
i ≤ j .) This means that at each stage c j , F is true of the objects 0 up to j but not true
of j + 1, j + 2, and the rest. So, for every point ci , there is an object i + 1 such that
ci 6 F(i +1) but ci 6 ∼F(i +1). So, for each construction, ci 6 ∀x(F x ∨∼F x). So
nowhere in the model is ∀x(F x ∨ ∼F x) true, and it follows that ∼∀x(F x ∨ ∼F x)
is true everywhere. But ∀x(F x ∨ ∼F x) is a classical tautology, and its negation
∼∀x(F x ∨ ∼F x) is a classical inconsistency. Yet we have found an interpretation
in which it is true. Intuitionistic predicate logic tolerates this inconsistency. 
Models like this are of independent interest. The smooth worlds of constructive
infinitesimal analysis rely essentially on these strong counterexamples of the law of
the excluded middle (Bell [5]). It is essential to this program of analysis that classical
inconsistencies like these be tolerated. (In fact, they are not only tolerated: they are
true in the intended models.)
For someone committed to classical consequence, thinking that any possibility
is closed under classical predicate consequence, the smooth worlds of intuitionistic
analysis are genuinely impossible. They are just as impossible as the impossible
worlds of the paraconsistent logician. They do not include outright contradictions,
but they do include propositions which cannot be true and are no more palatable than
the inconsistencies of more traditional paraconsistent logics. If paraconsistent logics
are to be rejected, then so intuitionistic predicate logic ought to be rejected alongside
them.
Example 3.3 Both lattice logic with orthonegation (more simply, “orthologic”)
and quantum logic, which extends lattice logic with the orthomodular law
(A ∧ (∼A ∨ (A ∧ B)) ` B) tolerate classical inconsistencies.

Proof Lattice logic is a straightforward account of conjunction and disjunction

which avoids the inference of distribution: A ∧ (B ∨ C) ` (A ∧ B) ∨ (A ∧ C).
In other respects, conjunction and disjunction behave normally. Conjunction is the
greatest lower bound (with respect to the ordering of entailment) and disjunction is
the least upper bound (on that same ordering). The most orthodox way to extend
lattice logic with negation is to add an orthonegation. The resulting logic we will
call orthologic. An operator ∼ is an orthonegation in a lattice logic when it satisfies
the double negation rules
A ` ∼∼A ∼∼A ` A
and the bound rules
A ∧ ∼A ` B A ` B ∨ ∼B.
These are the most orthodox negation rules imaginable. Were we to add them to the
logic of distributive lattices, the result would be classical propositional logic. The
context of general lattices, however, provides more leeway. Let’s consider a simple
152 Greg Restall

nondistributive lattice model for orthologic. The following diagram is a Hasse di-
agram for a six-element lattice. The lines in the diagram represent the ordering of

>

a b c d

PSfrag replacements
⊥

Figure 1 A nondistributive lattice

entailment: ⊥ is the lowest element in the order. Next come a, b, c, d which are pair-
wise incomparable. The greatest element in the order is >. Conjunction is defined
as greatest lower bound and disjunction as least upper bound. So the conjunction of
any two different elements from a, b, c, d will be ⊥ and their disjunction will be >.
This lattice is not distributive, because a ∧ (b ∨ c) is the element a ∧ > which is a,
while (a ∧ b) ∨ (a ∧ c) is ⊥ ∨ ⊥, which is ⊥, and a does not entail ⊥ because a
comes strictly higher than ⊥ in the ordering.
We can make this a model for an orthonegation by choosing the interpretation
for ∼ carefully. We must take ∼⊥ = > and ∼> = ⊥. The negations of a, b, c, d
must also be values from a, b, c, d. The negation of a may be any from b, c, d (but
it cannot be a, for the bound laws must be satisfied). Once we make the choice, the
negation of this element must be a. So without loss of generality, take ∼a to be c.
Then the other negations are fixed: ∼b must be d, for we must have ∼∼b = b, and
this rules out a or c for ∼∼b. So ∼d must be b. (It follows that there are exactly
three orthonegations on this lattice, corresponding to the three choices possible for
∼a.)
The orthomodular law A ∧ (∼A ∨ (A ∧ B)) ` B (which holds in all lattices
of subspaces of Hilbert spaces—which arise in the interpretation of quantum logic)
holds on this lattice. So this lattice is a model of quantum logic too.
This lattice gives us the following counterexample, showing orthologic and quan-
tum logic are both tolerant of classical inconsistency.
A ∧ (B ∨ C) ∧ ∼ ((A ∧ B) ∨ (A ∧ C)) 6` ⊥.
As discussed before, a ∧ (b ∨ c) takes the value a, while (a ∧ b) ∨ (a ∧ c) takes ⊥,
which means that ∼ ((a ∧ b) ∨ (a ∧ c)) is >. So a ∧ (b ∨ c) ∧ ∼ ((a ∧ b) ∨ (a ∧ c))
is a, and a does not entail ⊥ in this lattice.
I will end this discussion by recasting the counterexample in a frame model for
quantum logic. These models (due to Goldblatt [8]) stand to orthologic and quantum
logic as Kripke frames stand to intuitionstic logic.2 A compatibility frame, for our
purposes here, will be a nonempty set P of points, together with a symmetric and
reflexive binary relation C to model negation. Conjunction and negation are modeled
on a compatibility frame as you would expect.
1. x A ∧ B iff x A and x B.
2. x ∼A iff for each y where xC y, y 6 A.
 Paraconsistency 153

A conjunction is true at a point just when the conjuncts are true there. A negation
is true at a point just when its negand is not true at any compatible points. Now this is
not enough to model orthologic. For one thing, we have no guarantee that the double
negation laws hold. For another, we have not said how we are to model disjunction.
The naïve interpretation, setting x A ∨ B if and only if x A or x B, will do us
no good, as it will validate distribution. Thankfully, we can solve both problems in
one go, as Goldblatt noticed. I will explain how by way of an example compatibility
frame with four points {0, 1, 2, 3} such that each point is compatible with all points
other than its opposite (found by adding 2, modulo 4). In a diagram we can present
C by arrows, to get this:

1 2

0 3

Figure 2 A compatibility frame

Now consider a proposition true at 0 only. If we represent propositions as sets of

points (the points at which they are true) the proposition “is” {0}. Consider where its
negation ∼{0} might be true. It is not true at 0, as 0 is compatible with itself. It is
not true at 1, as 1 is compatible with 0. Neither is it true at 3, since 3 is compatible
with 0. However, it is true at 2, since 2 is not compatible with 0. So ∼{0} is {2}—it
is true at 2 only. The same reasoning shows that ∼{1} is {3}, ∼{2} is {0}, and ∼{3} is
{1}. Now consider ∼{0, 1}, the negation of a proposition true at both 0 and 1. This
cannot be true at either 0 or 1 (as 0C0 and 1C1) but neither is it true at 2 or 3, for
2C1 and 3C0. So ∼{0, 1} is {}, the empty set. But this is the case for any proposition
true at two or more points. For any point in our model is compatible with every point
except one. It will always manage to be compatible with some member of a set with
two or more elements. So the negation of every set with two or more elements is
{}. As a result, the double negation laws fail with these propositions: as an example,
note that ∼∼{0, 1} = ∼{} = {0, 1, 2, 3}.
It is not difficult to show that for every proposition X on a compatibility frame,
X ⊆ ∼∼X (symmetry of C does the work here). Propositions X for which
∼∼X = X are called closed. The closed propositions on our example frame are
the empty proposition {}, each one element proposition—{0}, {1}, {2}, and {3}—and
the full proposition {0, 1, 2, 3}. If we demand that sentences be interpreted on a
compatibility frame only at closed propositions, then ∼ is an orthonegation.
It remains to define disjunction. It could be done indirectly, taking A ∨ B to be
defined as ∼(∼A ∧ ∼B). Or we could define it directly in the following way:

x A ∨ B iff x ∈ ∼∼([[A]] ∪ [[B]]).

154 Greg Restall

That means that A ∨ B is true at x if x is a member of the closure of the set of points
where A or B are true. So the disjunction of {0} and {1} in our frame is the closure
of {0, 1} which is the entire set {0, 1, 2, 3}.
Now note that in our frame we have exactly six closed propositions. This lattice
of proposition is isomorphic to the six element lattice shown in Figure 1. An isomor-
phism maps {} to ⊥, {0, 1, 2, 3} to >, {0} to a, {1} to b, {2} to c, and {3} to d. Each of
the logical connectives (conjunction, disjunction, and negation) are preserved by this
isomorphism. The frame provides a concrete model of the lattice we have already
seen.
This frame also provides another way to view the classical inconsistency toler-
ated in models of orthologic and quantum logic. In this frame no point allows a
contradiction—the reflexivity of the compatibility relation sees to that—but the clas-
sical inconsistency A ∧ (B ∨ C) ∧ ∼ ((A ∧ B) ∨ (A ∧ C)) is tolerated. In the case
where A, B, and C are true at 0, 1, and 2, respectively, A ∧ (B ∨ C) is true at 0,
because A is true at 0 and B ∨ C is true everywhere. However, (A ∧ B) ∨ (A ∧ C)
is true nowhere, so its negation ∼ ((A ∧ B) ∨ (A ∧ C)) is true everywhere. Why
is this classical inconsistency tolerated here? It is not purely because negation is
interpreted nonclassically. Negation is as classical as one can hope for in a nondis-
tributive lattice. The classical inconsistency is tolerated because of the interpretation
of disjunction. B ∨ C is true at more than the places where either B is true or C is
true. It is true everwhere. This allows A ∧ (B ∨ C) to be true somewhere, despite the
fact that A ∧ B and A ∧ C are true nowhere. This allows the impossible to happen:
not simply that A ∧ (B ∨ C) is true and (A ∧ B) ∨ (A ∧ C) isn’t true: that would not
be enough for inconsistency tolerance—recall the example of intuitionistic proposi-
tional logic which is not inconsistency tolerant at all. Some classical inferences fail,
such as ∼∼A ` A. Kripke frames may have points where ∼∼A is true and A is not.
This is not enough for inconsistency tolerance, for we do not yet have a classical
inconsistency true at these points. Similarly, the presence of A ∧ (B ∨ C) and the
absence of (A ∧ B) ∨ (A ∧ C) is not enough to show inconsistency tolerance. What
we need, and what we have here, is the presence of A ∧ (B ∨ C) and the presence
of the negation ∼ ((A ∧ B) ∨ (A ∧ C)). This provides us with a classical inconsis-
tency, an example of something which cannot happen according to classical logic,
but which is allowed in models of orthologic and quantum logic. 

4 Conclusion
These examples have brought to light a general phenomenon of which the example
of paraconsistent logics is just a single species. Many different nonclassical logics
(but not all of them) tolerate classical inconsistencies. If this toleration is taken
to be a failing of paraconsistent logics, then the same must apply to intuitionistic
predicate logic, orthologic, and quantum logic. If, on the other hand, we can make
sense of inconsistencies in these cases, the fact that inconsistencies are tolerated in
paraconsistent logics is not a failing. Rather, it shows that these logics are in good
company.

Notes

1. I call this feature unsatisfiability because it is suggestive of its reading in model theory.
If X ` A if and only if every model satisfying X also satisfies A, then X ` A for every A
 Paraconsistency 155

when X is satisfiable in no model at all. Of course, X might also be satisfied in a model

provided that this model satisfies every statement whatsoever, but any model such as this
is of no use in determining the difference between valid and invalid argument forms, and
so it can safely be ignored for these purposes.

2. Bell [4] gives a philosophical analysis of Goldblatt’s semantics for orthologic, in which
the two-place compatibility relation is interpreted as proximity.

References

[1] Anderson, A. R., and N. D. Belnap, Entailment: The Logic of Relevance and Necessity,
vol. 1, Princeton University Press, Princeton, 1975. Zbl 0323.02030. MR 53:10542.
147, 149

[2] Anderson, A. R., N. D. Belnap, and J. M. Dunn, Entailment: The Logic of Relevance
and Necessity, vol. 2, Princeton University Press, Princeton, 1992. Zbl 0921.03025.
MR 94b:03042. 147

[3] Arruda, A. I., “Aspects of the historical development of paraconsistent logic,” pp. 99–
130 in Paraconsistent Logic: Essays on the Inconsistent, edited by G. Priest, R. Routley,
and J. Norman, Philosophia Verlag, 1989. Zbl 0694.03018. MR 92m:03036. 147

[4] Bell, J. L., “A new approach to quantum logic,” The British Journal for the Philosophy
of Science, vol. 37 (1986), pp. 83–99. Zbl 0616.03037. MR 88k:03131. 155

[5] Bell, J. L., A Primer of Infinitesimal Analysis, Cambridge University Press, Cambridge,
1998. Zbl 0904.26014. MR 99h:00002. 151

[6] Dummett, M., Elements of Intuitionism, Clarendon Press, Oxford, 1977.

Zbl 0358.02032. MR 58:16194. 150

[7] Fitting, M., Intuitionistic Logic, Model Theory and Forcing, Studies in Logic and the
Foundations of Mathematics. North-Holland Publishing Company, Amsterdam, 1969.
Zbl 0188.32003. MR 41:6666. 150

[8] Goldblatt, R., “Semantic analysis of orthologic,” The Journal of Philosophical Logic,
vol. 3 (1974), pp. 19–35. Zbl 0278.02023. MR 55:5398. 152

[9] Meyer, R. K., and E. P. Martin, “Logic on the Australian plan,” The Journal of Philo-
sophical Logic, vol. 15 (1986), pp. 305–32. Zbl 0614.03006. MR 87i:03025. 147

[10] Priest, G., R. Routley, and J. Norman, editors, Paraconsistent Logic. Essays on
the Inconsistent, Analytica, Philosophia Verlag, Munich, 1989. Zbl 0678.00003.
MR 92m:03036. 147

[11] Priest, G., “The logic of paradox,” The Journal of Philosophical Logic, vol. 8 (1979),
pp. 219–41. Zbl 0402.03012. MR 80g:03007. 149

[12] Priest, G., In Contradiction. A Study of the Transconsistent, vol. 39 of Nijhoff

International Philosophy Series, Martinus Nijhoff Publishers, Dordrecht, 1987.
Zbl 0682.03002. MR 90f:03007. 148

[13] Priest, G., and R. Routley, “Introduction,” pp. xix–xxi in Paraconsistent Logic: Essays
on the Inconsistent, edited by G. Priest and R. Routley and J. Norman, Philosophia
Verlag, Munich, 1989. Zbl 0697.03014. MR 92m:03036. 147
156 Greg Restall

[14] Restall, G., “Paraconsistent logics!” Bulletin of the Section of Logic, vol. 26 (1997),
pp. 156–63. Zbl 0926.03004. MR 1488294. 149

[15] Routley, R., V. Plumwood, R. K. Meyer, and R. T. Brady, Relevant Logics and their
Rivals. Part 1: The Basic Philosophical and Semantical Theory, Ridgeview Publishing
Company, Atascadero, 1982. Zbl 0579.03011. MR 85k:03013. 147

[16] Slater, B. H., “Paraconsistent logics?” The Journal of Philosophical Logic, vol. 24
(1995), pp. 451–54. Zbl 0830.03002. MR 96e:03035. 149

[17] van Dalen, D., “Intuitionistic logic,” pp. 225–339 in Handbook of Philosophical
Logic, vol. 3, edited by D. M. Gabbay and F. Günthner, Reidel, Dordrecht, 1986.
Zbl 0875.03038. 150

[18] van Dalen, D., “The intuitionistic conception of logic,” pp. 45–77 in European Review
of Philosophy, edited by A. Varzi, CSLI Publications, Stanford, 1999. Zbl 0995.03004.
MR 1798403. 150

Acknowledgments
This research is supported by the Australian Research Council through its Small Grant
program. Thanks to Charlie Donohoe and Richard Walker for comments on a draft of
this paper.

Philosophy Department
The University of Melbourne
Victoria 3010
AUSTRALIA
[email protected]