Journal of Scientific Exploration, Vol. 31, No. 3, pp. 379–386, 2017
0892-3310/17
EDITORIAL
Man is a rational animal who always loses his temper
when he is called upon to act in accordance with the dictates of reason.
—Oscar Wilde
I
’ve often noticed how debates within the SSE community sometimes
parallel debates in the political arena, perhaps especially with respect to
the passion they elicit and the intolerance and condescension sometimes
lavished on members of the “opposition.” Occasionally, of course, the
debates in the SSE are nearly indistinguishable from those in the political
arena—say, over the evidence for human-caused climate change. But what I
find most striking is how the passion, intolerance, etc.—perhaps most often
displayed by those defending whatever the “received” view happens to be—
betrays either a surprising ignorance or else a seemingly convenient lapse
of memory, one that probably wouldn’t appear in less emotionally charged
contexts. What impassioned partisans tend to ignore or forget concerns
(a) the tentative nature of both scientific pronouncements and knowledge
claims generally (including matters ostensibly much more secure than those
under debate), as well as (b) the extensive network of assumptions on which
every knowledge claim rests.
So I’d like to offer what I hope will be a perspective-enhancer,
concerning how even our allegedly most secure and fundamental pieces of
a priori knowledge are themselves open to reasonable debate. A widespread,
but naïve, view of logic is that no rational person could doubt its elementary
laws. But that bit of popular “wisdom” is demonstrably false. And if that’s
the case, then so much the worse for the degree of certitude we can expect
in more controversial arenas. Let me illustrate with a few examples.1
Consider, first, an empirical context in which some people have tried
to deploy a logical law. In philosophical discussions of the nature and
structure of the self, many writers invoke some version of the law of noncontradiction to argue for the existence of distinct parts of the self. This
strategy is at least as old as Plato and may be more familiar to JSE readers
in the form it took with Freud. Ironically, though, these arguments highlight
just how insecure this dialectical strategy is (for a more detailed account,
see Braude, 1995, Chapter 6).
Consider: In debates about the nature of multiple personality/dissociative
identity disorder (MPD/DID), many argue that because different alter
personalities/identities can apparently have different and even conflicting
380
Editorial
epistemic states, that there must be distinct parts of the self corresponding
to the conflicting states. So Kathleen Wilkes writes:
We break this law [of non-contradiction] as soon as we permit ourselves to
say that one and the same entity both knows and does not know that p, for
nothing can, at time t, be said to φ and not to φ. (Wilkes 1988:142)
Of course, to those without any philosophical axe to grind, cases of DID
might suggest that one can indeed be said to φ and not to φ at the same time.
Since that could easily be taken to suggest that the law of noncontradiction
has some hitherto unacknowledged limitation, and since one must always
be open to the possibility that logical laws have limitations of one sort or
another, let’s examine the status of the law which some dissociative and
other phenomena appear to violate.
Notice, first, that what logicians generally consider to be the law of
noncontradiction is either (a) the formal, syntactic law “~(A · ~A),” usually
rendered more informally as “not-(A and not-A),” or else (b) a claim in
logical semantics about truth-value assignments, namely, “no sentence can
be both true and false” (or alternatively, “the conjunction of any sentence
p and its denial not-p is false”). But the first of these is not violated by
dissociative conflicts, and the second is not even clearly a law.
Consider the syntactic law first. It concerns the form, rather than
the content, of strings of symbols within a formal system. It takes any
compound expression of the form “not-(A and not-A)” to be a theorem,
for any well-formed formula “A”. But strictly speaking, the law does not
pertain to sentences of any actual natural language. The syntactic law of
noncontradiction does nothing more than sanction a particular arrangement
of expressions within a certain set of formal systems. And although one can
easily determine which symbolic expressions are theorems, those logical
systems do not, in addition, offer a decision procedure for determining
which sentences in a natural language are true or false. On the contrary,
the relationship of formal to natural languages has to be both stipulated and
investigated. And ultimately, the utility of a formal system of logic has to be
evaluated empirically, by seeing whether or how well it applies to various
domains of discourse, for example by seeing whether the truth-values it
would assign to actual sentences matches our independent judgments about
what those truth-values should be.
In fact, formal logical systems don’t even specify which expressions
in a natural language count as legitimate instances of a simple (i.e.
noncompound) formula “A”, hence, which natural language expressions
are instances (or violations) of its theorems. Although logicians generally
Editorial
381
agree that the simple formulae of the systems should represent declarative
sentences, there’s considerable debate over which particular kinds of
declarative sentences are suitable. Interestingly, many would say that as far
as the purely formal laws of logic are concerned, “A” could stand even for
sentences whose truth-value or meaning are uncertain, such as “unicorns
are compassionate,” “the square root of 4 is asleep,” and “Zeus is insecure.”
But then it seems as if the uninterpreted formal law of noncontradiction
is simply irrelevant to the cases under consideration. At best, those cases
appear to challenge a semantic counterpart to the formal law, either
(NC1): The conjunction of any sentence p and its denial not-p is false
or
(NC2): No sentence can be both true and false
We needn’t worry at the moment about whether (or to what extent)
either of these versions of the law of noncontradiction is satisfactory. What
matters now is that even if the law of noncontradiction turns out to be a
viable principle of logical semantics, it may still have a variety of significant
limitations. In fact, the utility of formal logical laws varies widely, and the
interpretation of those laws has proven to be a notoriously tricky business.
As with all formal systems, no system of logic determines in which domains
(if any) its expressions may be successfully applied. Students of elementary
logic learn quickly that there are differences between the logical connectives
“and” and “or” and many instances of the words “and” and “or” in ordinary
language. Similarly, not all “if . . . then . . . ” sentences are adequately
handled by the material conditional in standard systems of sentential logic,
although that logical connective is undeniably useful in a great range of
cases. Moreover, varieties of nonstandard and “modal” logics have been
developed in attempts to represent types of discourse resistant to standard
logical systems.
But even more relevantly, in most standard systems of logic, the formal
law of noncontradiction, “not-(A and not-A),” is demonstrably equivalent to
the law of the excluded middle, “A v ~A” (i.e. “A or not-A”). Like the formal
law of noncontradiction, the law of the excluded middle concerns the form
rather than the content of expressions. It takes any compound formula of
the form “A or not-A” to be a theorem (or logical truth), no matter what
formula “A” happens to be. Now the semantic sibling of that syntactic law
is called the law of bivalence, which states that every sentence is either true
or false. But the law of bivalence has faced numerous challenges throughout
Editorial
382
the history of logic (in fact, since the time of Aristotle). Many people have
argued that it fails for sentences in the future tense and sentences whose
singular terms refer to nonexistent objects. Moreover, some logicians
consider these difficulties sufficiently profound to warrant the development
of logical systems that retain the syntactic law of the excluded middle but
reject the semantic law of bivalence (see, e.g., van Fraassen 1966, 1968,
Thomason 1970). Now granted, these same logicians don’t also reject
the semantic version of the law of noncontradiction. Nevertheless, their
reservations concerning bivalence should give us pause (especially in light
of the caveats noted above regarding the limitations of formal systems
generally). The debate over bivalence illustrates an important point, namely,
that the relative impregnability of a formal logical law may not be inherited
by its semantic counterpart (i.e. one of its interpretations). But at the very
best, it’s only the semantic counterpart of non-contradiction that rests at
the center of the Platonic/Freudian arguments for parts of the self. And in
fact, as far as Plato’s argument for the parts of the soul is concerned, the
argument turns on an even more exotic interpretation of non-contradiction.
See Braude (1995) for details.
But before we leave this topic, it’s important to note that
(NC1): The conjunction of any sentence p and its denial not-p is false
and
(NC2): No sentence can be both true and false
are likewise problematical, and probably more so than most JSE readers
appreciate. First of all, (NC1) has numerous counterexamples familiar to
students of logic and the philosophy of language. For example, it seems to
fail for sentences such as the aforementioned “unicorns are compassionate,”
“the square root of 4 is asleep,” and “Zeus is insecure,” which seem to
lack truth-value. Many people (but, notably, not all) would say that when
a sentence lacks truth-value, the conjunction of that sentence and its denial
also lacks truth-value.
The somewhat more common (NC2) has similar problems. Most
notoriously, perhaps, it fails for the self-referential sentence “this sentence
is false,” as well as for kindred expressions that don’t seem even remotely
suspicious inherently. For example, it fails for the innocent “the sentence
on page 42 is false,” when that sentence happens to be the only sentence on
page 42. If these sentences have any truth-value at all, it seems as if they
will be both true and false.
Editorial
383
Furthermore, (NC2) apparently fails for quite mundane present-tense
sentences. For example, “Socrates is sitting” may be true at one time and
false at another. Of course, one standard response to such cases would be to
claim that the sentence “Socrates is sitting” contains an implicit reference
to its time of production, so that it’s not really the same sentence that’s
true at one time and false at another (i.e. those nonsimultaneous sentences
would allegedly differ in meaning or express different propositions). For
reasons too complex to be explored here, it seems to me that this particular
maneuver creates more problems than it solves. Indeed, I’ve argued that the
standard Aristotelian notion of contradictories (stated in terms of opposing
truth-values) fails conspicuously for a tensed natural language, and that
tensed contradictories can have the same truth-value (see Braude (1986)
for a discussion of these issues). Although I recognize that my position is
most definitely a minority view, I submit that there are additional serious
reasons here for challenging the straightforward application of (NC2) to a
real natural language, hence, for questioning its inviolability outside of the
highly artificial or overly simplified linguistic situations to which logical
laws apply easily. In any case, this nest of issues illustrates again the kinds
of concerns involved in evaluating the apparently uncertain status of what
are considered to be our most cherished logical principles.
Please note that my point is not that the semantic law of noncontradiction
is useless as a philosophical tool. And the moral is not simply that logical
laws (like formal laws generally) may not hold in all domains (although
that’s certainly true and relevant here). Rather, the point is also that logical
laws hold in real life only for sentences we regard as acceptable (or
legitimate) and appropriate, or as understood in certain ways rather than
others. But these interpretations and classifications of linguistic entities are
practical decisions, made as part of a much larger network of interrelated
philosophical commitments. Accordingly, those decisions don’t stand or fall
in isolation from others in various areas of philosophy and logic. In fact, they
will continually be open for reassessment in light of apparent difficulties
arising at numerous points in our overall system of commitments.
One further example reinforces that last point; it concerns what many
regard as a fundamental principle about what philosophers typically call
numerical identity. Many people have argued that it’s an indisputable
rational principle that everything is identical with itself. However, it turns
out that the concept of numerical identity is not so straightforward.
To see this, consider first the expression
(x)(x = x)
Editorial
384
usually interpreted as “anything x is such that it’s identical to itself,” or more
colloquially, “everything is self-identical.” The acceptability of this alleged
law of identity is not something we can decide by considering that law
alone, and it’s certainly not something that’s immune from debate among
reasonable and well-informed persons. Regarded merely as a theorem of a
formal system, it has no meaning at all; it’s nothing more than a sanctioned
expression within a set of rules for manipulating symbols. But as an
interpreted bit of formalism, it’s acceptable only with respect to situations in
which we attempt to apply it. And perhaps more interesting, it’s intelligible
only as part of a larger network of commitments. That is, what we mean
by “everything is self-identical” depends in part on how we integrate that
sentence with other principles or inferences we accept or reject.
To see this, consider whether we would accept as true the statement
(1) Zeus = Zeus
To many people, no doubt, that sentence seems as unproblematically
true as the superficially similar
(2) Steve Braude = Steve Braude
However, in many systems of deductive logic containing the rule of
Existential Generalization (EG), from the symbolization of (1), namely,
(1ʹ) z = z
we can infer
(3) (Ǝx) x = z
which we typically read as
(4) Zeus exists.
And of course, many people consider that result intolerable.
Not surprisingly, philosophers have entertained various ways of dealing
with this situation. One would be to taxonomize different types of existence
and interpret the rule of Existential Generalization as applying only to
some of them (for example, prohibiting its application to cases of mythical
or fictional existence). Another approach would be to get fussy about the
concept of a name. We could decide that “Zeus” is not a genuine name
Editorial
385
and that genuine names (like “Steve Braude”) pick out only real existent
individuals, and not (say) mythical or fictional individuals. (Readers might
be especially surprised to learn that some people have actually endorsed the
view that we should not consider “Hamlet” or “Zeus” to be names if they
pick out fictional or mythical characters.) In any case, both these approaches
concede certain (but different) sorts of limitations to standard predicate
logic and the way or extent it connects with ordinary discourse. Others
prefer to tweak the logic directly, either syntactically or semantically. For
example, some people simply reject the rule of Existential Generalization
and endorse a so-called (existence) free logic. Alternatively, some retain EG
but adopt a substitutional interpretation of the quantifiers “(x)” and “(Ǝx)”,
so that instead of reading (3) as
(3ʹ) There is (or exists) some x such that x is identical with z (Zeus)
we read it as
(3ʺ) Some substitution instance of “x = z” is true.
The latter, they would say, is acceptable and carries no existential
commitments.2
Now the reader needn’t understand all these options. The moral,
however, should be clear enough. All these approaches raise concerns
about what should be regarded as a thing in certain contexts. The statement
“everything is self-identical” is not as clear or indisputable as one might
think, and even more important in the present context, it’s not simply true
no matter what. Its truth (and indeed, meaning) turn on a number of other
decisions as to which other principles or inferences are acceptable, and that
whole package of decisions can only be evaluated on pragmatic grounds.
Moreover, it’s perfectly respectable to decide that some solutions to this
conundrum are appropriate for some situations and that other solutions are
appropriate for others. We’re never constrained to select one solution as
privileged or fundamental.
The reason why I’ve gone on at such length about these matters is that
they should serve as a cautionary note to those who all too easily display
intolerance and condescension in empirical (or political) debates. It’s
completely clear that reasonable and informed people can disagree (and have
disagreed) over the nature and status—and, indeed, the meaning—of what
we take to be fundamental logical laws. Of course, scientific (and political)
debates rest not only on logical assumptions but on various empirical,
methodological, and other conceptual assumptions as well. So presumably
Editorial
386
they’re even more contentious and vulnerable to reasonable challenges than
disputes over the foundations of logic. But then one would expect to find
even more room there for reasonable and informed disagreement. Ideally,
then, one would expect participants in empirical debates to be particularly
open-minded, tolerant, and respectful of opposing views. So the next time
you find yourself tempted to dismiss or deride with a disdainful flourish
someone with whom you disagree over a matter of science (or politics), I
encourage you to remember how venerable and substantive are the serious
debates over the very foundations of our conceptual framework.
Notes
1
2
I’m indebted to Aune (1970) for much of what follows.
For more on free logic, see Lambert (2004), Morscher & Hieke (2001),
and van Fraassen (1966). And for an accessible review of many of the issues concerning nonexistent objects, see Reicher (2016).
—STEPHEN E. BRAUDE
References Cited
Aune, B. (1970). Rationalism, Empiricism, and Pragmatism: An Introduction. New York: Random
House. [Reprinted: Ridgeview Publishing, 2003]
Braude, S. E. (1986). You can say that again. Philosophic Exchange, 17:59–78.
Braude, S. E. (1995). First Person Plural: Multiple Personality and the Philosophy of Mind (revised
edition). Lanham, MD: Rowman & Littlefield.
Lambert, K. (2004). Free Logic: Selected Essays. Cambridge: Cambridge University Press.
Morscher, E., & Hieke, A. (Editors) (2001). New Essays in Free Logic: In Honour of Karel Lambert.
Dordrecht: Kluwer Academic Publishers.
Reicher, M. (2016). Nonexistent Objects. In The Stanford Encyclopedia of Philosophy (Winter 2016)
edited by E. N. Zalta.
https://plato.stanford.edu/archives/win2016/entries/nonexistent-objects/
Thomason, R. H. (1970). Indeterminist time and truth-value gaps. Theoria, 36:264–281.
van Fraassen, B. C. (1966). Singular terms, truth-value gaps, and free logic. Journal of Philosophy,
63:481–495.
van Fraassen, B. C. (1968). Presupposition, implication, and self-reference. Journal of Philosophy,
65:136–152.
Wilkes, K. V. (1988). Real People: Personal Identity without Thought Experiments. Oxford: Oxford
University Press.