Online ISSN 2424-6158.

PROBLEMOS 2018 93
DOI: https://doi.org/10.15388/Problemos.2018.93.11753

GORGIAS’ ARGUMENT DOES NOT INCLUDE

ACTUAL CONDITIONALS
Miguel López-Astorga
Institute of Humanistic Studies “Juan Ignacio Molina”
Talca University
Av. Lircay s/n, 3460000 Talca, Chile
E-mail: milopez@utalca.cl

Abstract. It can be thought that Gorgias’ argument on the non-existence consists of three sentences, the
first one being an asseveration and the other two being conditionals. However, this paper is intended to
show that there is no conditional in the argument, and that the second and third sentences only appear
to be so. To do that, a methodology drawn from the framework of the mental models theory is used,
which seems to lead to the true logical forms of these last sentences as well.
Keywords: conditional, Gorgias, logical form, mental models theory, non-existence

Some translations of the argument on the ous and that they can be found from analyses
non-existence by Gorgias appear to indicate based on the very mental models theory.
that it has three sentences and that two of Thus, firstly, Gorgias’ argument will be
them (the second and the third) are condi- addressed. It will be commented on what
tionals. Probably, this is so because those exactly it provides, and then the problems
two sentences contain, in the original Greek that must be faced if its two last sentences
version given by Sextus Empiricus, the word are interpreted as conditionals from the
εἰ (“if”). However, this paper has two main point of view of classical logic will be
goals. On the one hand, it will be argued, described. As explained below, those prob-
with the help of the mental models theory lems are related to certain unwanted conse-
(e.g., Hinterecker et al. 2016; Johnson-Laird quences of understanding such sentences as
2010, 2012, 2015; Johnson-Laird & Byrne material conditionals.
2002; Orenes & Johnson-Laird 2012; Quel- Secondly, the theses and the most im-
has & Johnson-Laird 2017; Quelhas et al. portant resources of the mental models
2017; Ragni et al. 2016), that the two afore- theory that are needed to make it evident
mentioned sentences are not really condition- that the two mentioned sentences are not
als (at least if it is assumed that the suitable truly (material) conditionals will be pre-
interpretation of the conditional is the materi- sented. And, of course, after that, the way
al one). On the other hand, it will be attempt- this theory can reveal that that is really so
ed to show, with the help of works such as, will be pointed out.
for example, the one of López-Astorga Finally, after a brief description of the
(2017a), that their real logical forms are obvi- methodology used in works such as, for

81
example, that of López-Astorga (2017a), it conditional sense of the sentences would
will be shown how that very methodology, be preserved.
which is also based on the mental models But, as claimed, this can be a problem,
theory, can lead one to the actual logical at least if the conditional is understood as
forms of the second and third sentences in in classical logic, that is, in a material way.
Gorgias’ argument. Hence, the next section And this can be noted in an evident manner
deals with the original fragment in which if we consider possible logical forms for the
this last argument is expressed. previous sentences, since they can allow
deriving conclusions that appear to be, if
not inconsistent, at least different from what
Gorgias’ Argument
Gorgias seems to mean. Clearly, the first
on the Non-Existence
sentence causes no difficulty, as its form
As it is well known and said, the argument is simple. Let it be, for example, ““ the
has three sentences. One of them is undoubt- universal quanfier, “¬” the negation, and
edly an assertion. However, the other two a “B” a predicate denoting “to be”. Then, the
priori seem to be conditionals. This can be logical form of the first sentence can be, in
seen in a clearer way if the original text is first-order predicate logic, this one:
taken into account.
(1) ∀x (¬Bx)
It is really, as indicated, a fragment
given by Sextus Empiricus. It is exactly in The problems, as pointed out, begin with
Adversus Mathematicos VII, 65, and is to be the second sentence. If we assume now that
found, for example, as Fragment 3 in Tapia “→“ represents conditional relationship and
Zúñiga (1980). The Greek original version that “U” refers to “to be able to be under-
is as follows: stood by a human being”, this formula could
…ἒν μὲν καὶ πρῶτον ὅτι οὐδὲν ἔστιν, be attributed to it:
δεύτερον ὅτι εἰ καὶ ἔστιν, ἀκατάληπτον (2) ∀x (Bx → ¬Ux)
ἀνθρώπῳ, τρίτον ὅτι εἰ καὶ καταληπτόν,
ἀλλὰ τοί γε ἀνέξοιστον καὶ ἀνερμήνευτον Nonetheless, if this is so, it is clear that,
τῷ πέλας. in first-order predicate logic, (2) in turn
(“…firstly, nothing is; second, if something enables to derive a formula such as the
were, it could not be understood by a human following (where “a” is a constant):
being; thirdly, if it could be understood, of
course, it could be neither transmitted nor (3) Ua → ¬Ba
accounted for to a near person”.) And (3) is undoubtedly controversial,
Obviously, the presence of εἰ leads one since it provides that, if a (which stands
to think about conditionality in the second for any being here) can be understood by a
and the third points. It is true that, given that human being, then it is not (or, if preferred,
εἰ adjoins καὶ in those two cases, other trans- it does not exist). Thus, beyond the fact that
lations can be thought, for example, “even this idea does not appear to make sense, it
if” instead of just “if”. Nevertheless, there is obvious that it does not describe what
is no doubt that, in many of those possible Gorgias really wanted to say either. Hence
alternative translations, in principle, the a formula such as (2) is not the best logical

82
form that can be given for the second sen- But maybe the key is that (2) and (4) are
tence in this argument. not truly suitable logical forms of the second
Indeed, the conditional as understood in and third sentences in Gorgias’ argument on
classical logic allows drawing, in proposi- the non-existence. A possibility can be that
tional calculus, formulae such as ¬q → ¬p there really is no conditional relationship
from formulae such as p q, which in turn in them and that, therefore, a symbol such
means that, in first-order predicate calcu- as “→“ should not be used to formally
lus, it is absolutely correct to conclude (3) represent them. The mental models theory
from (2) (to check the reasons why, based allows exploring this last possibility and, for
on works such as those of Gentzen (1934, that reason, the next section is devoted to
1935), the general structure of the deriva- what that theory claims about the sentences
tions in propositional logic are also correct with “if”.
in first-order predicate logic (see, e.g.,
Deaño 1999)). However, this problem is The Mental Models Theory
compounded if we pay attention to the third and the Conditional
sentence, as exactly the same difficulty can
be observed in it. To note that, only three The mental models theory is a psycholog-
additional assumptions are necessary: that ical approach trying to show, among other
“∧“ is conjunction, that “T ” represents “to important aspects, the way people tend to
be able to be transmitted”, and that “A” think when faced with connectives such as
denotes “to be able to be accounted for to a those of classical logic. Nonetheless, as far
near person”. In this way, the logical form as the aims of this paper are concerned, it
of the third sentence can be: is necessary to consider only its account of
the conditional. According to it, when an
(4) ∀x (Ux → (¬Tx ¬Ax) individual thinks about a conditional such
as “if A then B”, he usually focuses on
Nevertheless, again, in the same manner
possibilities that are consistent with it. If
as (3) can be derived from (2), it is now pos-
all of them are identified, such possibilities
sible to draw this formula from (4) (where
are the following (see, e.g., Quelhas et al.
“∧“ refers to disjunction):
2017: 1004):
(5) (Ta ∨ Aa) → ¬Ua (6) A & B
The latter expresses that, if a can be Not-A & B
transmitted or a can be accounted for to Not-A & Not-B
a near person, then a human being cannot These are three possible scenarios: (i) A
understand it. So, apart from the fact that (5) and B happen, (ii) only B happens, and (iii)
is also hard to understand, since it is difficult neither A nor B happen. However, although
to accept that something that is transmitted this can remind a truth table of classical
or accounted for to other person cannot be logic, these scenarios are iconic, and to
understood by a human being, it presents, as support the definition of this last concept,
in the case of (3), an idea that does not seem the proponents of the theory often resort to
to correspond to the real Gorgias” thought. accounts such as that of Charles Sanders

83
Peirce (1931-1958). Johnson-Laird (2012), analyze theses proposed by ancient philoso-
for example, does that. Nevertheless, what phers. For example, López-Astorga (2017b)
is important in this way is that, because they argues that the thesis by Thales of Miletus
are iconic, the possibilities in (6) describe related to the fact that all of beings have
complete alternative scenarios with coher- souls, including the beings that are usually
ence and consistency, which means that, in thought not to have a soul, can be better
the real cases, some of them can disappear, interpreted, if it is linked to a set of possi-
be rejected, or be changed. A clear example bilities such as that of (7), that is, to a set of
in this regard can be the following, which is possibilities with just these two elements:
taken from Johnson-Laird and Byrne (2002: (A & B) and (Not-A & B). Nonetheless, the
663; see also, e.g., López-Astorga 2016: point here is that it can be useful to better
291, 2017b: 107): understand Gorgias’ argument as well, in
“If you are interested in seeing Vertigo, then particular, obviously, its second and third
it is on TV tonight”. sentences. If we consider the second one
firstly, we can note that, certainly, it does
This sentence does not admit the three
not refer to possibilities such as those of
possibilities in (6), but only
(6), but to a set with a structure similar to
(7) You are interested in seeing Vertigo the one of (7), that is, to this set:
& It is on TV tonight
(8) Something is & It cannot be under-
You are not interested in seeing
stood by a human being
Vertigo & It is on TV tonight
Something is not & It cannot be
The third possibility (Not-A & Not-B) understood by a human being
cannot be taken into account, because, from
The third possibility of (6) cannot be
what is said by the speaker, it is absolutely
accepted here, because, as pointed out, it
clear that Vertigo is on TV tonight, and
does not seem possible that something does
that fact cannot be denied (we cannot think
not exist and, however, it can be understood,
about a case of Not-B). Thus, given that the
and this is so independently of the fact that,
cases in which a conditional is true in a truth
as also indicated, this last idea does not
table of classical logic are those included
appear to be a thesis truly held by Gorgias.
in (6), which, as it is known, correspond
Furthermore, something similar happens
to those of the material interpretation of
with the third sentence in his argument.
the conditional, and in (7) the last one is
(6) is not the most suitable set for it but,
missing, it is evident that the previous
again, another one akin (at least in its struc-
example coming from Johnson-Laird and
ture) to that of (7):
Byrne (2002) is not actually a conditional,
even though the word “if” appears in it, at (9) Something can be understood &
least, if we understand the conditional just It can be neither transmitted nor
in a material way. accounted for to a near person
The name that Johnson-Laird and Byrne Something cannot be understood
(2002) give to that kind of sentence is “rel- & It can be neither transmitted nor
evance”, and it has already been used to accounted for to a near person

84
 Clearly, the third combination in (6) Let us think about (6). Its three possi-
must be rejected here too, and the reason bilities are linked by means of “&”, which
for that has also been mentioned above. It suggests that it can make sense to consider
is very difficult to think about a scenario them to be logical conjunctions. In fact,
in which something cannot be understood although they do not use the word “logical”,
and, at the same time, it can be transmitted the adherents of the mental models theory
or accounted for to a near person, and it explicitly state that the iconic possibilities
should be added to this, as claimed too, that represent conjunctions (e.g., Quelhas et al.
that does not seem to be an idea actually 2017: 1004). But, if we assume that those
supported by Gorgias either. possibilities are logical conjunctions, we
Therefore, it appears that the second and immediately obtain three formulae:
third sentences included in the argument (10) p ∧ q
by Gorgias that is being considered in this
(11) ¬p ∧ q
paper are not real material conditionals,
even if they have the word εἰ (“if”). But, if (12) ¬p ∧ ¬q
they are not material conditionals, which Thus, given that they are possibilities
logical forms could we attribute to them in or alternative scenarios, the next step is to
classical logic? Perhaps the mental models join them by means of disjunctions, which
theory can help us in this regard as well. leads us to this new formula:
(13) (p ∧ q) ∨ (¬p ∧ q) ∨ (¬p ∧ ¬q)
The Logical Forms of the Second
And finally it is only necessary to build
and Third Sentences in Gorgias’
the truth table for (13) in order to find the
Argument on the Non-Existence combinations in which it is true and it is
The way to find the actual logical forms of false. If that is done, we can see that, as it
such sentences can be clear if works such is obvious and reminded by López-Astor-
as that of López-Astorga (2017a) are taken ga (2017a: 36), (13) is true when either p
into account. Although the proponents of the is false or q is true, that is, in exactly the
mental models theory often ignore logical same cases as
forms and state that human reasoning is (14) p → q
made without resorting to them (see, e.g.,
Hence it can be thought that (14) is the
Johnson-Laird 2010 or Orenes and John-
suitable logical form here. Of course, other
son-Laird 2012), the main idea in these
formulae are possible, since, for example,
works is that the iconic possibilities to which the combinations in which (13) and (14)
sentences refer can be deemed as combi- are true are also, as it is known, the same
nations in a truth table of classical logic in in which a formula such as this one is true:
which the formula that is being looked for
is true. Thus, from such possibilities, we can (15) ¬p ∨ q
come to well-formed formulae in this last However, in any case, it is clear that this
logic. However, maybe an example can help can be a method to find underlying logical
to better understand this method. forms, and, for this reason, it appears that

85
it is worth using it to detect the real form propositional calculus and first-order pred-
of the relevance type sentences. If, as said, icate calculus, versions of (17), (18), and
the possibilities corresponding to this last (19) in first-order predicate logic can be
kind of sentence are (A & B) and (Not-A & given. Such versions can be, respectively,
B), based on what has been explained, it is the following:
evident that they allow constructing this
(20) ∀x ((Px ∨ ¬Px) → Qx)
formula:
(21) ∀x ((Px ∨ ¬Px) ∧ Qx)
(16) (p ∧ q) ∨ (¬p ∧ q)
(22) ∀x (Qx)
That is, a formula that is always true
when q is true, and that is only false when And, because, as shown with (8) and
q is so. Therefore, it can be thought that its (9), the second and third sentences in Gor-
actual logical form is as follows: gias’ argument on the non-existence are
instances of relevance, and it appears that
(17) (p ∨ ¬p) → q such sentences can be better expressed in
(Although without resorting to this the language of first-order predicate logic,
method based on the truth tables of classical at this point, it can be stated that possible
logic, a similar logical form is assigned to logical forms of them have been found.
relevance in, e.g., López-Astorga 2016). Indeed, (20), (21), and (22) can provide
Certainly, (16) and (17) are true and false possible formal structures for those sen-
in the very same cases, and the latter seems tences more suitable than (2) and (4). In
to express the sense of relevance better than particular, keeping the meanings of “B” and
(14), since what it states is that q is always “U” above, the logical forms for the second
true, no matter what happens with p, that is, sentence could be these ones:
whether or not p is true. (23) ∀x ((Bx ∨ ¬Bx) → ¬Ux)
Of course, more logical forms can be
(24) ∀x ((Bx ∨ ¬Bx) ∧ ¬Ux)
indicated for relevance, for example,
(25) ∀x (¬Ux)
(18) (p ∨ ¬p) ∧ q
Any of these logical forms eliminates
(19) q
the problems described for (2), and clearly
However, given that these four last reflects what Gorgias wanted to say with his
formulae – (16), (17), (18), and (19) – are second sentence: whether or not something
true if and only if q is so, it can be thought is, it cannot be understood by a human be-
that all of the forms that can be proposed ing. From this point of view, it is even more
for relevance, the mentioned ones being evident that that sentence does not indicate
just some possible examples, emphasize a material conditional relationship between
that, as said, what is truly important is that being and being able to be understood by
q obtains, and that the truth-value of p is a human being. The relationship expressed
irrelevant. by the sentence is obviously different from
But, on the other hand, based on works that one.
such as the one of Deaño (1999), which, As far as the third sentence is concerned,
as pointed out, describes the links between something similar can be argued. On the one

86
hand, keeping the meanings of “T” and “A” fragment presenting Gorgias’ argument on
too, its logical forms could be the following: the non-existence authored by Sextus Em-
piricus and cited above, that word is trans-
(26) ∀x ((Ux ∨ ¬Ux) → (¬Tx ∧ ¬Ax))
lated into Spanish by Tapia Zúñiga (1980:
(27) ∀x ((Ux ∨ ¬Ux) ∧ (¬Tx ∧ ¬Ax)) 1) as “aunque” (“although”) in both cases.
(28) ∀x (¬Tx ∧ ¬Ax) Certainly, it is not incorrect to translate εἰ as
“although” in certain contexts or in the case
(26), (27), and (28) also remove the of certain grammatical constructions, but
difficulties of (4) mentioned above, and there is no doubt that it is clearly relevant
appear to refer to Gorgias’ thought in a that Tapia Zúñiga uses the Spanish word
clearer way. The point is now that, whether equivalent to “although” in his translation.
or not something is understood by a human
being, what is indisputable is that it cannot
be transmitted and it cannot be accounted Conclusions
for to a near person. So, in the same way as As explained, for example, in López-Astor-
the case of the second sentence, this offers ga (2017a), regardless of the real intentions
even further support to the idea that there and goals of the mental models theory, its
is no material conditional relationship here analysis methodology is useful to identify
either. Certainly, at a minimum, there is no the actual logical forms of sentences. That
relationship of that kind between the ante- has been checked here by means of an ar-
cedent and the consequent that the sentence gument by the sophist Gorgias. However,
has in natural language (or, if preferred, in it is evident that that methodology can
his Greek original version), that is, between help us find the underlying logical form of
the fact of being able to be understood and any sentence in any language and in any
the fact of not being able to be transmitted context. And this is so because it takes into
and accounted for. account not only syntactic factors (the way
Therefore, Gorgias’ argument gives the words are combined in sentences), but
stronger evidence for the thesis supported also semantic and pragmatic factors (it pays
in papers such as that of Johnson-Laird attention to the iconic possibilities that can
and Byrne (2002), that is, that the fact that be really accepted).
a sentence in natural language includes the On the other hand, it can help understand
word “if” does not necessarily imply that and interpret what thinkers in ancient times
that sentence expresses a material condi- meant too. As said, this is not the first paper
tional relationship between its clauses. Of in that direction. In, for example, López-As-
course, that thesis is not new, but it seems torga (2017b), the mental models theory is
that the mental models theory shows why taken as a methodological resource as well,
other interpretations can be correct and how the aim being to deal with the fragment by
we can come to them. In this way, it can be Diogenes Laërtius in which it is stated that
very interesting to note, for example, that, Thales of Miletus assigned a soul even to
while the word εἰ, that is, the Greek word beings without a soul. Coincidentally, as
equivalent to the English word “if”, appears also pointed out, López-Astorga comes to
in the second and third sentences of the the conclusion that the sentence expressing

87
this last idea is also a relevance type sen- to a better understanding of ancient texts.
tence. However, maybe the most important The mental models theory can be the key
point is that analyses akin to the one offered framework in both lines, which are equally
here have already been carried out, and that interesting and deserve further exploration
hence they can make sense. to the same extent. Precisely, this paper
So, it seems that there are two relevant appears to have shown this last point, since
lines of study that are opening up. The first it has revealed that it is possible to discover
one is related to the search for an algorithm the actual logical form of sentences as puz-
of any kind to detect the true logical forms zling as those of Gorgias’ argument on the
corresponding to sentences. The second non-existence, and it has explained what
one is connected to the development of exactly this thinker wanted to express in a
an interpretative methodology to come very clear way.

REFERENCES
Deaño, A., 1999. Introducción a la lógica formal. and Types of Conditionals. Frontiers of Philosophy
Madrid: Alianza Editorial. in China 11(2): 279–297.
Gentzen, G., 1934. Untersuchungen über das López-Astorga, M., 2017a. Mental Models are
logische Schließen I. Mathematische Zeitschrift Compatible with Logical Forms. In: Interdiscipli-
39 (2): 176–210. nary Approaches to Semiotics, ed. A. López-Varela
Gentzen, G., 1935. Untersuchungen über das Azcárate. Rijeka: InTech, 31–42.
logische Schließen II. Mathematische Zeitschrift López-Astorga, M., 2017b. Thales of Miletus and
39 (3): 405–431. the Semantic Possibilities of his View of the Soul.
Hinterecker, T., Knauff, M., Johnson-Laird, P. N., Aisthema IV(1): 101–112.
2016. Modality, Probability, and Mental Models. Jour- Orenes, I., Johnson-Laird, P. N., 2012. Logic,
Models, and Paradoxical Inferences. Mind & Lan-
nal of Experimental Psychology: Learning, Memory,
guage 27(4): 357–377.
and Cognition 42(10): 1606–1620.
Peirce, C. S., 1931–1958. Collected Papers of
Johnson-Laird, P. N., 2010. Against Logical Form.
Charles Sanders Peirce, eds. C. Hartshorne, P. Weiss,
Psychologica Belgica 5(3/4): 193–221.
A. Burks. Cambridge: Harvard University Press.
Johnson-Laird, P. N., 2012. Inference with Mental
Quelhas, A. C., Johnson-Laird, P. N., 2017. The
Models. In: The Oxford Handbook of Thinking and
Modulation of Disjunctive Assertions. The Quarterly
Reasoning, eds. K. J. Holyoak, R. G. Morrison. New
Journal of Experimental Psychology 70(4): 703–717.
York: Oxford University Press, 134–145.
Quelhas, A. C., Rasga, C., Johnson-Laird, P. N.,
Johnson-Laird, P. N., 2015. How to Improve 2017. A Priori True and False Conditionals. Cognitive
Thinking. In: The Routledge International Handbook Science 41(55): 1003–1030.
of Research on Teaching Thinking, eds. R. Wegerif, Ragni, M., Sonntag, T., Johnson-Laird, P. N.,
L. Li, J. C. Kaufman. Abingdon and New York: 2016. Spatial Conditionals and Illusory Inferences.
Routledge, 80–91. Journal of Cognitive Psychology 28(3): 348–365.
Johnson-Laird, P. N., Byrne, R. M. J., 2002. Tapia Zúñiga, P. C., 1980. Fragmentos / Gorgias:
Conditionals: A Theory of Meaning, Pragmatics, and Introducción, traducción y notas de Pedro C. Tapia
Inference. Psychological Review 109(4): 646–678. Zúñiga. Mexico City: Universidad Nacional Autóno-
López-Astorga, M., 2016. Logic, Pragmatics, ma de México (UNAM).

88
GORGIJAUS SAMPROTAVIME NĖRA MATERIALIOSIOS IMPLIKACIJOS
Miguel López-Astorga
Santrauka. Iš pirmo žvilgsnio atrodo, kad Gorgijaus samprotavime apie neegzistavimą esama trijų sakinių, iš
kurių pirmasis – kategoriškas tvirtinimas, o likę du – sąlyginiai teiginiai. Šio straipsnio tikslas yra parodyti, kad
du pastarieji iš tikrųjų nėra materialiosios implikacijos sakiniai, jie tokie tik atrodo. Kad tą įrodyčiau, pasitelkiu
mentalinių modelių teoriją, kuri leidžia išryškinti tikrąją dviejų minėtų sakinių loginę formą.

Pagrindiniai žodžiai: materiali implikacija, Gorgijus, loginė forma, mentalinių modelių teorija, neegzistavimas

Įteikta 2017 12 02
Priimta 2018 02 05

89