Académique Documents
Professionnel Documents
Culture Documents
DOI 10.1007/s11229-011-9908-6
Received: 26 August 2009 / Accepted: 15 April 2010 / Published online: 22 March 2011
Springer Science+Business Media B.V. 2011
Abstract The recovery of Aristotles logic during the twelfth century was a great
stimulus to medieval thinkers. Among their own theories developed to explain Aristotles theories of valid and invalid reasoning was a theory of consequence, of what
arguments were valid, and why. By the fourteenth century, two main lines of thought
had developed, one at Oxford, the other at Paris. Both schools distinguished formal
from material consequence, but in very different ways. In Buridan and his followers in
Paris, formal consequence was that preserved under uniform substitution. In Oxford,
in contrast, formal consequence included analytic consequences such as If its a man,
then its an animal. Aristotles notion of syllogistic consequence was subsumed under
the treatment of formal consequence. Buridan developed a general theory embracing
the assertoric syllogism, the modal syllogism and syllogisms with oblique terms. The
result was a thoroughly systematic and extensive treatment of logical theory and logical
consequence which repays investigation.
Keywords Aristotle Ockham Buridan Signification Syllogism Modality
Formal Material
1 Medieval logic
The medievals own original contribution to logic was stimulated by the recovery
of Aristotles logic in the Latin West during the twelfth century. Logic at the start
of that century consisted of what is generally known as the logica vetus: Aristotles
Categories and De Interpretatione and the works of Boethius. The logica nova, as it
Paper presented at the Workshop on The Philosophy of Logical Consequence at Uppsala, 31 October2
November 2008.
S. Read (B)
Department of Philosophy, University of St. Andrews, St. Andrews KY16 9AL, Scotland, UK
e-mail: slr@st-andrews.ac.uk
123
900
became known, consisted of the rest of the Organon, all of which was available in Latin
translation by around 1200. The Latin medievals own contribution, usually termed
the logica modernorum, consisted of: the theory of properties of terms (signification,
supposition, ampliation etc.); the theory of consequences; the theory of insolubles;
and the theory of obligations. It is usually taken, following De Rijk (1967), to have
been stimulated by the theory of fallacy, following recovery of Aristotles De Sophisticis Elenchis around 1140. It reached fulfilment in the fourteenth century, the most
productive century for logic before the nineteenth. In what follows, I will concentrate
on the fourteenth century, on the theory of consequence, and in particular on the works
of three significant logicians.
The first is Walter Burley, or Burleigh, born in Yorkshire, England, around 1275.
He was a Master of Arts at Merton College in Oxford University by 1301, and two
early logical works of his are Suppositions and Obligations, both dating from 1302.
He moved to Paris before 1310, and studied theology there until around 1326 or
1327. As was common, however, he continued to teach and write on Arts subjects
throughout this time, and his most famous logical treatise, De Puritate Artis Logicae
(On the essence of the art of logic) was composed in Paris in 1324, though in it
he is clearly aware of developments in England, in particular the ideas of William
Ockhams. Subsequently, he became a member of Richard de Burys circle. De Bury
was the Bishop of Durham, famous for having the largest library in England, who
gathered an intellectual circle round him of the best thinkers in England. Following
Edward IIIs accession to the throne of England in 1327, Burley became envoy to the
papal court. He wrote very many other works, including commentaries on Aristotle,
and died around 1344 or 1345.
William Ockham was born at Ockham (or Oakham) in Surrey in 1287 or 1288
and joined the Franciscans at their London house before 1300. He studied at London
and Oxford, lecturing in Oxford on the Sentences of Peter Lombard in 13171319. He
wrote his Summa Logicae in London around 1323 and shortly afterwards was called
to answer charges of heresy before the Pope in Avignon in May 1324. He escaped
from Avignon to the protection of the Holy Roman Emperor in Munich in May 1328.
He wrote many other works, including commentaries on Aristotle and on political
philosophy, and died in Munich in April 1347.
While the lead in logical, as in all other studies, was forged in Paris in the thirteenth
century, with Oxford following along, in the fourteenth century there was a distinct
split between Oxford and Paris. This started long before the 100 Years War between
England and France, which began in 1337, but it was certainly compounded by it. The
leading logician of the early fourteenth century in Paris was John Buridan, born in the
late 1290s near Bthune in Picardy in northern France. Buridan studied in Paris and
taught there as Master of Arts from the mid-1320s. Unusually, he remained in the Arts
Faculty throughout his career. He wrote his Treatise on Consequences in 1335 or soon
afterwards, and composed his Summulae de Dialectica (Compendium of Dialectic)
from the late 1330s onwards, in eight (or nine) treatises with successive revisions into
the 1350s. His Sophismata, often included as the ninth treatise of the Summulae, has
been much studied in the last 50 years. He wrote many other works, again, mostly
commentaries on Aristotle, and died around 1360.
123
901
2 Properties of terms
Before we turn to the theory of consequence, in the theories of the assertoric and modal
syllogism, we need to look at the medieval theories of properties of terms, in particular, signification, supposition and ampliation. In De Interpretatione 1 (as translated by
Boethius), Aristotle distinguishes three kinds of terms: spoken [words] are signs
(notae) of impressions (passionum) in the soul, and written of spoken. (16a34) Boethius, commenting on Aristotle in the sixth century, interpreted this as saying that
thing, concept, sound and letter are four: the concept conceives the thing, spoken
sounds are signs of the concept, and letters signify the sounds. In contrast, Augustine, a 100 years earlier, but far less influenced by Aristotle, decreed that words signify
things by means of concepts. This raises the question: do words signify concepts or
things? On the one hand, it was said that if words signified things, empty names would
be impossible; on the other, that when I say Socrates runs, I mean that Socrates, not
some concept of Socrates runs.
The thirteenth century insight was to recognise the concept itself as a sign. Ockhams
theory of mental language can be pictured like this:
Ockhams guiding principle was that everything is individualthere are no common natures; so nothing is universal except by signification (nominalism) and mental
acts (concepts) form a language which signifies naturally.
At Paris, Buridan had a rather different, more Aristotelian, theory of signification:
Buridan wrote:
Categorematic words signify things by the mediation of their concepts,
according to which concepts, or similar ones, they were imposed to signify.
123
902
123
903
theory is that, in it, p has the same meaning as not- p (Potter 2009, p. 281). However, he goes on:
p has the same meaning as not- p but opposite sense. The meaning is the fact. [p. 288] This is again the
acceptance of significates of propositions that Buridan rejected.
7 Sophismata, in Buridan (2001, Chap. 2); cf. Hubien (1976, I.1).
123
904
supposits; a universal affirmative is true when whatever the subject supposits for the
predicate also supposits for; and so on.
3 Consequence
Buridan wrote:
Now a consequence is a molecular proposition, for it is composed from several
propositions conjoined by the expression if or by the expression therefore or
something similar. (Hubien 1976, I.3)
Thus a consequence consists of several premises (antecedentia) and a conclusion (consequens). But it can also be taken as a single conditional proposition, with antecedent
and consequent. A consequence is by definition a valid consequence. E.g., if one proposition is antecedent to another, that means that the other does indeed follow from
it, and is succedent to it.
There are two main divisions of consequences. They first divided consequence into
formal and material consequence. These were treated differently in Oxford and in
Paris. For example, according to Buridan:
A formal consequence is one that holds for all terms retaining the same form,
or if you wish to speak carefully for which any equiform proposition which
might be formed would be an acceptable consequence. For example, That which
is A is B, so that which is B is A A material consequence is where not every
proposition of the same form is valid , e.g., A man runs, so an animal runs,
because it is not valid with these terms: A horse walks, so wood walks No
material consequence is evident except by reduction to a formal consequence by
the addition of some necessary proposition. (Hubien 1976, I.4)
There is a significantly different account of formal and material consequence in
England. A typical English view is expressed by Robert Fland, writing in Oxford
around 1350:
General rules are given in order to appreciate when an inference is formally valid.
The first is this: where the conclusion is formally understood in the premises.
For example, this inference is formally valid: There is a man, so there is an
animal because the conclusion animal is formally understood in the premise,
namely, man. (Spade 1976, p. 57)
Thus in the English tradition, formal consequence includes what Spade (1974, p. 78)
calls analytical validity. The examples of material consequence that are given in
English writers are limited to the strict implicational paradoxes:
from the impossible anything follows (ad impossibile sequitur quodlibet): e.g., If
a man is an ass, there is a stick in the corner.
the necessary follows from anything (necessarium sequitur ad quodlibet):8 e.g.,
If a man runs, there is a God.
8 Note the increasingly prevalent use of ad with the accusative in medieval Latin to replace the dative or
ablative.
123
905
The other main division was between simple and ut nunc consequence. In a simple
consequence, the premises can never be true without the truth of the conclusion. In
contrast, a matter-of-fact or ut nunc consequence (sometimes also translated as-ofnow) can have true premises and a false conclusion at some time, but not at present.
For example:
Every man is running
So Socrates is running
is valid ut nunc only while Socrates (exists and) is a man. Both simple and ut nunc
consequence allow suppressed premises, but only necessary truths may be suppressed
in simple consequence.
Buridan gives an intriguing example of ut nunc consequence:9
A white cardinal has been elected Pope
I see this man
So I see a deceitful man (homo falsus)
In 1335, the newly elected Pope, Jacques Fournier, was a member of the Cistercian
order of white monks. He was a fierce opponent of fourteenth-century innovations
in logic even before his election as Pope Benedict XII. Material consequence ut nunc
can be reduced to formal consequence in Buridans sense, to hold solely in virtue of its
form, by the addition of a contingently true premise (in this case, The white cardinal
is a deceitful man). Simple material consequences reduce to formal consequences by
adding a necessarily true premise.
An unknown author, writing in Paris probably in the 1340s (now known as PseudoScotus because his works were published with Scotus in a seventeenth-century
edition), challenged the modal criterion that a consequence is valid just when it is
impossible for the premise to be true and the conclusion false.10 Consider the argument: God exists
So this consequence is invalid.
According to the modal criterion, any consequence with a necessarily true conclusion
is valid. But the above consequence does have a necessarily true conclusion. For suppose it were valid. Then the premise would be true and the conclusion false, so it would
be invalid. It follows, by reductio ad absurdum, that its invalid. This proof of invalidity depends only on the assumption God exists, which (Pseudo-Scotus certainly
believed) is necessarily true. So that the consequence is invalid is also necessarily true.
Thus we have an invalid argument with a necessarily true conclusion. So the modal
criterion must be wrong.
Pseudo-Scotus proposed solution was to add an extra clause, provided how the
conclusion signifies is not incompatible with how the premise signifies. But PseudoScotus argument is a semantic paradox (what the medievals called an insoluble),
as recognised by many, e.g., Albert of Saxony, writing in Paris in the 1350s. For we
9 Hubien (1976, I.4); cf. Hubiens Introduction, p. 9.
10 English translation in Yrjnsuuri (2001, p. 227).
123
906
deduced its conclusion (Its invalid) from its premise (God exists), so it really is
valid (as well as invalid).11 According to Albert, every proposition signifies its own
truth. Having shown it invalid, we infer that things are as the conclusion signifies, and
so its true. But that does not follow: to show its true we have to show that things
are however it signifies. Whether Alberts solution is satisfactory is arguable. But any
decent solution to such paradoxes will rebut Pseudo-Scotus objection.
The doctrine that a consequence is valid if the premise cannot be true without the
conclusions being true was also challenged by Buridan (Hubien 1976, I.3): Every
man runs, so some man runs is valid, but it is possible for the premise to be true without the conclusion even existing, and so without it being true. One might, therefore,
suggest that consequence is valid if the premise cannot be true without the conclusions being true, when both are formed together. That will not work, says Buridan:
No proposition is negative, so no ass is running is not valid (for its contrapositive is
not valid), but the premise cannot be true, and so cannot be true without the conclusions being true. Buridan proposes that a consequence is valid if it is impossible that
however the premise signifies to be it is not however the conclusion signifies, when
they are formed together, provided that however it signifies is not taken literally, but
in the sense given at the end of Sect. 2 above.
4 The syllogism
We need first to rehearse Aristotles theory of the assertoric syllogism, composed of
non-modal propositions. A syllogism is a consequence with two premises, a major
premise and a minor premise, and three terms. The major premise, containing the
major term, is the first premise (the stock definition of the major term as the predicate
of the conclusion did not come until the sixteenth century). The middle term appears
in each premise but not the conclusion. Each proposition is of one of four forms, A-,
E-, I- and O-propositions, forming the Square of Opposition. A-propositions have the
form All A is B (Aa B), E-propositions the form No A is B (AeB), I-propositions
the form Some A is B (Ai B), and O-propositions the form Not all A is B, or
equivalently, Some A is not B (AoB). The medievals for the most part interpreted
A- and I-propositions as false if the subject, A, was empty, and E- and O-propositions
true in such a case:
11 Albert of Saxony (1988, p. 360). Cf. Currys paradox (Lb 1955, p. 118) in this form: If this proposition
123
907
where the middle term is subject of one premise and predicate of the
other;
Figure II
where the middle term is predicate of both premises; and
Figure III where the middle term is subject in both premises
What later became known as the fourth figure is here construed as an indirect form of
the first figure. We can display the three figures as follows:
Figure I (direct)
MP
SM
SP
Figure I (indirect)
MP
SM
PS
Figure II
PM
SM
SP
Figure III
MP
MS
SP
The famous medieval mnemonic then tells us which forms are valid (separating the
three figures by semi-colons):
Barbara Celarent Darii Ferio Baralipton
Celantes Dabitis Fapesmo Frisesomorum;
Cesare Camestres Festino Baroco; Darapti
Felapton Disamis Datisi Bocardo Ferison.
The vowels give the form of the premises and conclusion. Aristotles theory of the
assertoric syllogism was based on the dictum de omni et nullo: whatever is asserted
of a (quantitative) whole is asserted of its part, and whatever is denied of the whole is
denied of the part.12 These principles establish the validity of the direct moods in the
first figure: Barbara, Celarent, Darii and Ferio. All other moods can be reduced to one
of those four, the one beginning with the same consonant by the system of reductions
which are encoded in the consonant following each vowel in the mnemonic. For example, s instructs one to perform simple conversion on that proposition (see below),
m to interchange the premises, p to perform conversion per accidens and c to
perform reduction per impossibile. The rules of conversion are:
Simple conversion: Ai B implies Bi A and AeB implies Be A
Conversion per accidens: Aa B implies Bi A
Reduction per impossibile requires interchanging the negation of one premise with
the negation of the conclusion.
Here is an example of reduction, reducing Camestres (aee in Figure II) to Celarent:
12 Although not found explicitly in Aristotle, the dictum de omni et nullo is commonly traced to Prior
Analytics I 1 (24b2830): We say that one term is predicated of all of another whenever no instance of
the subject can be found of which the other term cannot be asserted; to be predicated of none must be
understood in the same way.
123
908
Pa M
SeM
Se P
SeM
MeS
Pa M
Pa M
m tells us to interSe P
the first s tells us to
Se P
change the premises
convert SeM simply
MeS
Me P
Pa M
Sa M
the second s tells us to PeS we now reletter to Se P which is Celarent, eae
convert Se P simply
make S and P
in Figure I.
subject and predicate
of the conclusion
The validity of Celarent therefore entails () the validity of Camestres.
In the fourteenth century, the syllogism was brought under a general theory of
consequence. Buridan, in particular, treats the syllogism as a special case of formal
consequence. Indeed, he introduces a novel approach to demonstrating the validity
of assertoric syllogisms. For example, he establishes the following theorems (Hubien
1976, III 4):
2. No valid syllogism has two negative premises
6. In every valid syllogism, the middle term is distributed at least once
8. Any term distributed in the conclusion must have been distributed in the premises.
Then he can go systematically through all combinations of premises and conclusion
and identify the valid syllogisms.
Another of Buridans novelties was to consider syllogisms in non-standard form,
e.g., with oblique terms:
No mans donkey (i.e., no donkey of a man) is running
A man is an animal
So some animals donkey is not running
In his translation of Buridans Summulae, Gyula Klima shows how the logical relations
hold analogously between propositions with oblique terms and modal propositions.13
E.g.,
Every man necessarily runs
is contrary to
Some man necessarily does
not run
just as
Of every man every donkey runs
is contrary to
123
909
Buridans theory can explain why this is valid. Its an enthymeme, not formally valid,
but valid simpliciter, that is, simply valid. The suppressed premise is the necessary
truth: Every horses head is a horses head. Then the argument has this form:
Every M is a P
Every horses head is an Ms head
So every horses head is a Ps head
123
910
and Ockhams solutions. First, Buridans. All modal propositions are either taken in
the composite sense (that is, de dicto), or in the divided sense (that is, de re). Nothing
follows from composite modal premises where the mode is possible or contingent.
From composite modal premises of necessity, the conclusion follows in the same
mode. The interesting modal syllogisms are those with divided modal premises, and
with combinations of assertoric and divided modals. Buridan claims that all divided
modal propositions ampliate their subject. Thus, e.g., Every A is necessarily B is
true iff everything which is or might be A must be B. Similarly, Some A is possibly
B is true iff something which is or might be A might be B. We can override this
ampliation by replacing the subject by a what is-phrase, e.g., Everything which is
A is necessarily B.
Paul Thom (2003, p. xiv, 17) sets out a useful notation to represent modal and
non-modal propositions. For terms a and b, let represent inclusion, | exclusion,
coincidence, a what is (a or) possibly a, and b what is necessarily b. Thus Every
A is necessarily B, for example, is represented as a b , on Buridans view that
the subject is ampliated to what is possible. Buridan set out a Modal Octagon, similar to the Square of Opposition, in which the relations between modal A-, E-, I- and
O-propositions fall in the divided sense.19 Using Thoms notation, we can represent
Buridans modal octagon as follows:
In a series of twenty-eight theorems, Buridan takes us through all the various combinations. E.g., Theorem 15:
In the first figure from an assertoric major [premise] and a minor [premise]
of necessity a conclusion of necessity does not follow, nor does an assertoric
conclusion follow except in Celarent. (Hubien 1976, IV 2)
In particular, Barbara XLL is invalid. Suppose God is, as it happens, creating:
123
911
Buridans account only validates Darii and Ferio LXL, where the major premise is
ampliated:
1. Darii: s m m p entails s p
2. Ferio: s m m | p entails s p
20 See, e.g., Priest and Read (1981).
123
912
6 Conclusion
Medieval logicians extended Aristotles theory of the syllogism to a general theory
of consequence. Within consequence, they distinguished formal from material consequence, and simple from ut nunc consequence. They developed a theory of properties
of terms, in particular, of signification, supposition and ampliation, which played a
central role in their theory of consequence. Much of their work on the modal syllogism was an attempt to correct and clarify Aristotles theory. John Buridans theory
of consequence and of the syllogism was the most sophisticated and extensive in its
theoretical organisation.
Acknowledgments This work is supported by Research Grant AH/F018398/1 (Foundations of Logical
Consequence) from the Arts and Humanities Research Council, UK.
References
Adams, M. (1987). William Ockham (Vol. 2). Notre Dame: University of Notre Dame Press.
Albert of Saxony. (1988). Insolubles. In N. Kretzmann & E. Stump (Eds.), Cambridge translations of
medieval philosophical texts, volume I: Logic and philosophy of language (pp. 337368). Cambridge,
MA: Cambridge University Press.
Anon. (1967). Ars Burana. In L. De Rijk (Ed.), Logica modernorum, Vol. II 2 (pp. 175213). Assen:
Van Gorcum.
Buridan, J. (2001). Summulae de dialectica. New Haven, CT: Yale University Press. G. Klima, Eng. Tr.
Cesalli, L. (2001). Le ralisme propositionnel de Walter Burley. Archives dhistoire doctrinale et littraire
du moyen ge, 68, 155221.
De Morgan, A. (1966). In P. Heath (Ed.), On the syllogism and other logical writings. London: Routledge
and Kegan Paul.
de Ockham, G. (1974). In P. Boehner et al. (Eds.), Summa logicae. St. Bonaventure: Franciscan Institute
Publications.
De Rijk, L. (1967). Logica modernorum (Vol. II 2). Assen: Van Gorcum.
Hubien, H. (1976). Iohannis Buridani: Tractatus de consequentiis. Louvain: Publications Universitaires.
Hughes, G. (1989). The modal logic of John Buridan. In Atti del Congresso Internazionale di Storia
della Logica: La teorie delle modalit (pp. 93111). Bologna: CLUEB.
Keele, R. (2006). Applied logic and mediaeval reasoning: Iteration and infinite regress in Walter Chatton.
Proceedings of the Society for Medieval Logic and Metaphysics, 6, 2337.
Lb, M. (1955). On a solution of a problem of Leon Henkin. Journal of Symbolic Logic, 20, 115117.
Potter, M. (2009). Wittgensteins notes on logic. Oxford: Oxford University Press.
Priest, G., & Read, S. (1981). Ockhams rejection of ampliation. Mind, 90, 274279.
Spade, P. (1974). Five logical tracts by Richard Lavenham. In J. ODonnell, Essays in honour of Anton
Charles Pegis (pp. 70124). Toronto, ON: Pontifical Institute of Mediaeval Studies.
Spade, P. (1976). Robert Flands consequentiae: An edition. Mediaeval Studies, 38, 5484.
Thom, P. (2003). Medieval modal systems. Farnham: Ashgate.
Wengert, R. (1974). Schematizing De Morgans argument. Notre Dame Journal of Formal Logic, 15,
165166.
Yrjnsuuri, M. (Ed.). (2001). Medieval formal logic. Dordrecht: Kluwer.
Zupko, J. (19941997). How it played in the rue Fouarre: The reception of Adam Wodehams theory
of the complexe significabile in the Arts Faculty at Paris in the mid-fourteenth century. Franciscan
Studies, 54, 211225.
123