Journal of Philosophy, Inc.

On Presumption
Author(s): Edna Ullman-Margalit
Source: The Journal of Philosophy, Vol. 80, No. 3 (Mar., 1983), pp. 143-163
Published by: Journal of Philosophy, Inc.
Stable URL: http://www.jstor.org/stable/2026132 .
Accessed: 03/08/2014 18:36

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal
of Philosophy.

http://www.jstor.org

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 THE JOURNAL OF PHILOSOPHY
VOLUME LXXX, NO. 3, MARCH 1983

P5
ON PRESUMPTION*
resumptionshave to do with assumptionsmade ahead of
time,in advance. The conceptis suggestive,I think,of a sup-
position not fullyjustified,yetnot quite rash either.There is
in presumptiona sense of an unquestioned taking forgranted,but
at the same time of some tentativeness,overturnability.Given this
fertilesoil of gentlycontrastingconnotations,it is hardlysurpris-
ing that philosophers have not altogethershunned the use of this
notion.' But theiremploymentof it is neithersystematicnor criti-
cal, and the notion itselfhas not so far been the focus of proper
philosophical attention.I shall in this paper give it the attentionI
thinkit deserves.
The clarificationof the epistemicclaims thatpresumptionshave
on us-or ratheron our set of beliefs-is undoubtedlypart of any
adequate account of the notion of presumption.But this will not
be the orientationof the explication I shall offer.I shall look,
rather,to the role that the notion of presumptioncan be made to
play within the theoryof action. That is to say: ratherthan treata
presumption as an assumption made in advance of some theoreti-
cal venture,I shall treatit as an assumption made in advance of
practical deliberation.Furthermore,ratherthan view presumption
as a logical prerequisiteforthe launching of a theoreticalinquiry,
* Sidney Morgenbessersaw the earliest draftof this paper, and his incisive com-
mentshelped me redirectmy thinkingon the subject.A subsequent versionwas read
to membersof the philosophy departmentat PrincetonUniversityin January,1979.
I wish to thank L. Jonathan Cohen, Derek Parfit,Isaac Levi, J. R. Lucas, Avishai
Margalit,and Joseph Raz for helpful conversations,and P. F. Strawsonand J. L.
Mackie for theircommentsin writing.Part of the researchfor this work was done
during a sabbatical yearat Oxford which was made possible by a Rothchild Foun-
dation grant.
'Among thearticlesusing 'presumption'in theirtitles:JamesW. Lamb, "Knowl-
edge and JustifiedPresumption," this JOURNAL, LXIX, 5 (Mar. 9, 1972): 123-127;
Louis I. Katzner, "Presumptions of Reason and Presumptionsof Justice," ibid.,
LXX, (Feb. 22, 1973): 89-100; J. E. Llewelyn, "Presuppositions, Assumptions and
Presumptions,"Theoria, xxviii(1962): 158-172.
0022-362X/83/8003/0143$02.00 ? 1983 The Journalof Philosophy, Inc.

143

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
144 THE JOURNAL OF PHILOSOPHY

I shall view it as a rational prerequisiteforarrival at a varietyof

decisions about action.2
I. PRESUMPTIONS IN THE LAW
Explication is usually guided by thepre-systematic, everydayusages
of the notion under consideration.In the presentinstance,how-
ever,it seems to me that the ordinary-languageanalysis of the no-
tion of presumption(or such cognatesas 'presumably','a presump-
tive such-and-such') will not get us very far. Guidance in the
presentcase is to be sought ratherin the realm of the law. Within
the frameworkof the law and, more specifically,within the law of
evidencepresumptionsare both extensivelyused and made thecen-
ter of much theorizing.3It will be rewarding,therefore, to devote
some space to unravelingthenatureof legal presumptions,in spite
of thefactthattheoreticaldiscussionsabout theirstatusare not free
fromcontroversy and thatthereis no unanimityof opinion among
lawyerseven about theirmode of functioning.It is not my inten-
tion to tryto arriveat an exhaustivesummaryof the issue of pre-
sumptionsin thelaw, nor to scrutinizeall theuses of thisnotion in
the legal literature.My intentionis different:
to locate thehard core
of the use of presumptionswithin the law, in order that this may
serveas a startingpoint forthe task of philosophical explication.
Accordingly,at various stages along the way, the explication will
be halted momentarilyand a point will be checked in the legal
realm so as to clarifythe directionin which the explication is to
proceed. It should be stressed,then,thatthe referenceto the law is
forpurposes of guidance and illustrationonly. My aim is to import
the notion of presumptionfromthe law into philosophy, not to
exportphilosophical advice forthe use of lawyers.
Here, first,is a sample list of legal presumptions:that a child
born during lawful wedlock is legitimate;thata personwho, with-
out reasonable explanation, has not been heard fromfor at least
seven years is dead; that a marriageregularlysolemnizedis valid;

2This mightbe contrastedwith Nicholas Rescher'streatmentof presumptionsin

his Methodological Pragmatism(New York: NYU Press, 1977). His frameworkis not
practical deliberation,but what he refersto as the "cognitive venture,"wherepre-
sumptionsare conceivedof,roughly,as workinghypotheses.
verytexton the Law of Evidence containsa chapteron presumptions.Among
the eminentauthors of such textsare Phipson (1892, 8th edition 1942),Jones(1896,
5th edition 1958), Wigmore(1904, 3rd edition 1940) and Stephen (1925).
The locus classicus of theorizingon presumptionsis thechapterdevotedto them
in James Bradley Thayer's A PreliminaryTreatiseon Evidence at Common Law,
Boston, 1898. Also of special significanceare two articlesby Edmund M. Morgan:
"Some ObservationsconcerningPresumptions,"Harvard Law Review, LXIV (1931):
906-934,and "InstructingtheJuryupon Presumptionsand Burdenof Proof,"ibid.,
XLVII (1933): 59-83.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 ON PRESUMPTION 145
that a child under fourteenyearsof age has no criminal intention;
that if A buys propertyfromB and directsit to C, C was intended
to be a trusteeforA; thatwhen in a common disasterthe death oc-
curs of two or more persons related as to inheritance,the younger
(or healthier)person survivedlonger. Also: thata person accused of
crime is innocent; that everyperson is sane; that a person intends
the natural consequences of his or her actions.
Consider the firstpresumptioncited above. "There is a presump-
tion thata child born in wedlock is legitimate(i.e., thatits fatheris
the mother's husband)." In stating this, what is the law actually
saying? Let us consider what consequences follow, if any, from
these two premises:
(1) There is a presumption that a child born in wedlock is
legitimate.
(2) Adam (a particularchild) was born in wedlock.
More specifically,are thereconditionsunder which theconclusion:
(3) Adam is legitimate (i.e., Adam's father is his mother's
husband).
can be validly drawn from(1) and (2)? Or perhaps the conclusion
should be
(4) There is a presumptionthatAdam is legitimate.
But then we shall want to know what it is that (4) is saying, and
how it is related to (3). Does it actually say anythingabout Adam?
It may perhaps be feltthat (4) is not assertinganythingat all, or at
any rate not primarily,but thatits importlies elsewhere.
The law itselfhas some things to say about what it is saying,or
perhaps doing, by means of its presumptionstatements.Here are
just two authoritativepassages fromthe law of evidence:
A presumption meansa ruleoflaw thatcourtsand judgesshall draw
froma particularfact,or fromparticularevi-
a particularinference
dence, unless and until the truthof such inferenceis disproved
(Stephen).
A presumption maybe definedto be an inferencerequiredbya ruleof
law drawnas to theexistenceof one factfromtheexistenceof some
otherestablishedbasic facts.-Itis a truepresumptionof factin the
sensethatanotherfactis assumedfromestablishedbasicfacts.It is a
presumption of law in the sensethata rule of law requirestheas-
sumptionto be made(Jones).4
4There are also several furtherdistinctionsin the law concerningpresumptions,
notably (a) betweenpresumptionsof law and presumptionsof fact;(b) betweenre-
buttable presumptionsand irrebuttable,or conclusive, presumptions.The firstis
commonly taken to be ratherconfusedand confusing,a presumptionof factbeing
perhaps an altogetherredundantnotion (see Thayer's attackon thisdistinction,op.
cit., pp. 339-342). The second is interesting,but will not occupy us here.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
146 THE JOURNAL OF PHILOSOPHY

The way a specificpresumption,thatof death upon the expira-

tion of seven yearsof unexplained absence, is put to work is illus-
tratedthus:
. I . ifthey[thejury]findthefactofabsenceforsevenyearsunheard
from,and findno explanatory factstoaccountforit,thenbya ruleof
law theyare to takefortruethefactofdeath,and are to reckonupon
it accordinglyin makingup theirverdictupon the whole issue
(Wigmore, ?2490).
Above all, it is the notions of taking somethingfor true and of
"reckoningupon it" which interestme in this explicativeventure.
The picturethatbegins to emergeis this. Suppose thedescentof
a certainman's estateis at stake. The factin issue beforethe triers
of fact (judge or jury) may be whetheror not Adam is this man's
legitimate heir (or, mutatis mutandis, whetheror not this man,
long absent, is still alive, or whetheror not this man died before
his wifein theairplane crash thatkilled themboth). In theabsence
of evidenceon thisissue, or in case of conflictingevidence,how are
the facttriersto proceed?Considerationsof statistical,or of "prior"
probability,even wheretheycan be appealed to, will clearlynot do
as substitutesforparticularevidencein each particularcase. How-
ever, the law sometimesintervenesby laying down rules, in the
formof presumptions,which effectthe inference(or "inference")
fromcertain basic factsalready established to the existenceof the
factin issue, as long as no evidence (or no sufficientevidence; see
below) to the contraryis produced. These rules supply, wherethey
apply and pending rebuttal,a ready-madeanswer,prescribedrather
than ascertained,to the factualquestion involved. "When an infer-
ence derivesfromthe law some arbitraryor artificialeffectand is
obligatory upon judges and juries, that inferenceis a true pre-
sumption" (Jones?11). I would go one step furtherand claim that
thereis not only an elementof arbitrarinessor artificialityin pre-
sumptions, but also an element of bias. Given that thereare two
possible answersto thefactualquestion underconsideration,either
"yes" or "no", the presumptionrule is partial towardone of them
and favorsit in advance over the other. What we have here is not
the proverbial situation of gauging, preferablyblindfold,which
side of an evenlybalanced scale turnsout to tip thebalance. Rather,
we are deliberatelyputting the thumb on one side of the scale to
begin with.5
5See Barbara D. Underwood's paper under the suggestivetitle "The Thumb on
the Scales of Justice:Burdensof Persuasion in Criminal Cases," Yale Law Journal,
LXXXVI, 7 (1977): 1299-1348.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 ON PRESUMPTION 147

Beforewe are readyto move on fromthe law, then,a rough pre-

liminaryanswer to our question concerningtheimportof sentence
(4) suggestsitself:forpurposes of coming to a verdictand provided
thatno (sufficiently strong)evidence to the contraryis in, the trier
of factis (instructedby the law) to take Adam's legitimacyfortrue.
II. THE EXPLICATION
Sentenceslike (1) will be referredto as presumptionformulas.The
presumptionformulawill be representedby the string'pres (P,Q)'.
'P' stands for the presumption-raising fact (the lawyerscall it the
basic fact)-in our example, being born in wedlock. 'Q' standsfor
thepresumedfact-in our example, being fatheredby the mother's
husband. Capital lettersindicate thatgenericdescriptionsof states
of affairsare involved; lower-caseletterswill stand forparticular
descriptions.
The presumptionformulais read, then,as 'P raises thepresump-
tion that Q',6 or, alternatively,as 'There is a presumptionfromP
that Q'. When it is said that it applies in a certain concretein-
stance, this should be taken to mean that the presumption-raising
(generic)factis instantiatedin thatconcreteinstance.
I proceed now to interpretthepresumptionformulain termsof a
presumptionrule. The presumptionrule expressedby the formula
'pres (P,Q)' is directedto any person who is engaged in a processof
practical deliberationwhose resolutionmateriallydepends,among
other things,on an answer to the factual question of whetherq is
or is not the case. Such persons may be referredto as the rule sub-
jects. The rule is this:
Giventhatp is thecase,you(= therulesubject)shallproceedas ifq
weretrue,unlessor untilyouhave(sufficient)
reasonto believethatq
is notthecase.
(a) "Proceed. ..: thenatureoftherule
What sortof a rule is a presumptionrule? What does it actually
instructits subjectsto do?
Note firstthat the presumptionformulais propositional in na-
ture ("P raises the presumption that Q"); it is ostensiblyabout
facts.However,I submit thatit is concernednot so much with as-
certainingthe factsas with proceeding on them,as its rule inter-
pretation brings out. Presumption rules belong in the realm of
praxis, not theory.Their point is to enable us to get on smoothly
with business of all sorts,to cut throughimpasses, to facilitateand
expedite action. But thereis no specificaction thata presumption

6Note: 'raises' not in the sense of "increases" but of "creates,""bringsabout."

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
148 THE JOURNAL OF PHILOSOPHY

rule charges its subjectswith. It instructsits subjects to hold a cer-

tain proposition as trueso as to have a foothold(as it were) forac-
tion. Put somewhat differently, the instructionis this: given p,
make q a premisein therestof thepertinentpiece of yourpractical
reasoning.
What are we to make of the quasi-jussive mode ("you shall pro-
ceed as if . .") in which thepresumptionrule is cast: is it, or is it
not, a mandatoryrule?
In the law it is: the triersof factare required to draw the infer-
ence fromthe P-factto the Q-fact.Outside the law, with respectto
the ratherlooser and more amorphous frameworkof practical de-
liberationin general,theansweris less clear-cut.I tendto thinkof
presumptionrules as offeringa way out and, therefore, tend to re-
gard the question of their status-i.e., whethertheyprescribeor
permit,license, or enjoin-as slightlybeside the point. What mat-
ters,in my view, is that theyentitlethe deliberatorsto make-and
to ground their subsequent course of action in-an assumption
theyare otherwisenot justifiedin making.
At the same time,as with many rightsand norms,waiving such
entitlementsor failing to exercise them may be censurable. The
veryofferof a way out creates the expectation that it will be re-
sortedto in the appropriatecircumstances.Thwartingsuch expec-
tations may well count as violation of a norm and be subject to
disapprobation.
(b) '. . . as if q": an inference?
Are presumptionrules concernedwith inference?Is it theirpoint
that the presumedfactis inferred,in some sense or under certain
conditions,fromthe presumption-raising fact?
To avoid confusion,let me draw attentionto a distinction.Quite
apart fromwhetheror not presumptionrules are about inference,
thereis associated with them a rule of inference.It is the formal
rule governing the operation of the 'pres' operator: from(i) pres
(P,Q) and (ii) p, it follows that(iii) pres q [compare sentences(1),
(2), and (4) above].
This underscoresthe following featureof the operator: that it
functionseitheras a two-placeoperator,its two variables standing
for generic descriptionsof states of affairs(the presumptionfor-
mula), or as a one-place operator,its single variable standingfora
particulardescriptionof a stateof affairs(the presumedfact).The
connection betweenthesetwo functionsis at the same timeunder-
stood as well. Note, too, that a 'pres q' formula in effectpresup-
poses a 'pres (P,Q)' formula.That is, if it is thecase that,forsome

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 ON PRESUMPTION 149

q, pres q, then thereis a stateof affairsrepresentedby 'p' such that

both p and the appropriatepresumptionrule expressedby the for-
mula 'pres (P,Q)' obtain.
So, the conclusion of the rule of inferenceassociated with pre-
sumption rules is to the effectthat a certainfact is presumed. But
this is not to be confusedwith an inferenceto the presumedfact.
And the question with which we startedwas whetheror not what
presumptionrules are about is inferenceto thepresumedfact(s).In
answer to this question I submit that the presumptionrule in-
volves no commitmentto, nor guaranteeof, the truthvalue of the
presumedfactq. It makes no claims upon its subjects'cognitiveor
epistemic systems.The rule entitlesone to hold q as true for the
purpose of concluding one's practicaldeliberationon the impend-
ing issue; it neitherrequiresnor entitlesone to believe that q.
In light of theseconsiderationsI shall avoid talkingof presump-
tion rules as being about, or prescribing,inference.Rather,I shall
say of the presumptionrule expressedby 'pres (P,Q)' that,with a
view to its subjects' pertinentpractical purposes, it sanctions for
them the passage from p to q. Somewhat more succinctly,the
phrase to be adopted is thata presumptionrule sanctionstheprac-
tical passage fromp to q.
(c) "Unless or until . .": the rebuttalclause
The ordinary-languageword 'presumption' is suggestive of
somethingtentative,contestible,reversible.There are various terms
used, or usable, in this connection:a presumptionmay be said to
be rebutted,overcome, overridden,reversed,defeated,displaced,
nullified,and more.I shall single out thefirstof theseand speak of
the "unless or until" clause of thepresumptionrule as the rebuttal
clause.
The question beforeus now is, What is it fora presumptionto
be rebutted?More specifically,we shall want to know what the re-
buttal consistsin and what it takesto bringit about.
Recall the rule: given thatp, you shall proceed as if q unless or
until you have (sufficient)reason to believe that q is not the case.
This rule sanctions the practical passage fromp to q while at the
same timeacknowledgingthepossible falsityof q. The rule should
be understoodas settingsome sortof mechanismin motion. It sets
its subjects on a certaincourse of action, namely,thatof proceed-
ing to act on the assumption that q. This course can be blocked
only if (this is theperfectivesense conveyedby the 'unless') or once
(this is thecontinuatesense conveyedby the 'until') therule subject
has (sufficient)reason to believe thatq is not the case. When this

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
150 THE JOURNAL OF PHILOSOPHY

happens, thepresumptionis rebutted;so long as it doesn't,thepre-

sumption standsand is operative.
Note: it is the presumption-that-q which is rebutted,not thepre-
sumption rule itself.Given a presumptionrule expressed by the
formula 'pres (P,Q)' and a certainrule subject A, the presumption
that q is rebuttedforA if the circumstancesare such thatp is the
case and A has (sufficient)reason to believe that not-q: the "un-
less" clause having been fulfilled,the proceed-as-if-qinjunction
lapses. (The presumptionrule, though strictlyspeaking not rebut-
table, is neverthelessrevisable.7)
The question of what it takes to bring about the rebuttalof a
presumptionconcernsthe allocation of duties,and corresponding
benefits,by the presumptionrule. It has to do with the twin no-
tions of the burden of proof and the benefitof the doubt. Let us
once again take our cue fromthe law. In litigation,where a pre-
sumption rule applies, if and as long as no (sufficient)evidence-
that-not-qis in, the triersof factare to proceed as if q were estab-
lished to be the case. With legal presumptions,then,it is evidence
that rebuts them,and it takes the production of evidence to bring
the rebuttalabout. Now since in an adversarysystemevidencecan
be produced only byone or theotherof the litigants,thelaw has to
be clear about who is charged with the task of producing such re-
buttingevidence. The answer is indeed straightforward. To theex-
tentthat the unrebuttedpresumptionservesto furtherthe case of
one of theparties,referred to as the "presumptionproponent,"it is
up to the otherparty,referredto as the "presumptionopponent,"
to attemptto rebut it. This is so regardlessof who it was who es-
tablished the presumption-raisingfact to begin with, or, for that
matter,regardlessof where the over-all burden of persuasion lies.
(The law distinguishes furtherbetween the so-called "persuasive
burden" and the evidential burden, but this need not detain us
here.)
So, in addition to the substantiveaspect of the presumption
rules, viz. the sanctioning of the practical passage fromp to q,
theserules also have a procedural aspect, thatof casting a burden
of producing counterevidenceupon the presumption opponent.
The presumptionproponentbecomesentitledto something:to the

'Any presumptionrule is revisable,even thoserelatingto so-called "conclusive,"

or irrebuttablepresumptions.In "Analyticityby Way of Presumption" (Edna Ull-
mann-Margalit and Avishai Margalit, Canadian Journal of Philosophy, xii, 3
as well as the
(1982): 435-452) the distinctionbetweenrevisabilityand rebuttability,
notion of conclusivepresumption,are used to shed some new lighton the notion of
analyticity.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 ON PRESUMPTION 151

fact-triers'proceedingon q; the presumptionopponent is charged

with something:with the burden of showing that not-q. And the
entitlementholds as long as the burdenhas not been discharged.
How can this be related to the wider and looser frameworkof
practical deliberation, where no procedures for "going forward
with evidence", and no pair of disputing parties,are part of the
picture?
In the foregoingdiscussion it was alreadysuggestedthatrebuttal
in this frameworkconsists in the deliberator'shaving (sufficient)
reason to believe thatq is not the case. That is, it is reasons forbe-
lief which rebuta presumption,and it takes the deliberator'scom-
ing to have themto bring the rebuttalabout. It may not be idle to
note at thispoint thatit is not theexistenceof reasonsforbelieving
that not-q, but ratherthe deliberator'shaving them,which counts
so faras rebuttalis concerned.Otherwise,theplain and "objective"
factthat q is not the case would by itselfrebutthe presumption.
It follows that here,in contrastwith the legal context,the pre-
sumptionrule's proceduralaspect of castinga certaindutyon some
specifiedperson or partyis ratherdiffuse.Anyoneor anythingthat
provides the deliberatingagent with the appropriatereason forbe-
lief rebuts the presumptionfor him (or for her). It may, but it
needn't be, anotherperson who convinces the deliberatorthatq is
not the case. It is in generaljust up to the deliberatorto check that
no available counterindicationhas been overlooked beforehe or
she is both entitledand enjoined to proceed as if q; and any such
(sufficientlyweighty)counterindicationthat turns up may rebut
the presumption,therebyhalting the mechanism that was set in
motion by thepresumptionrule. There is, then,some mild version
of the principle of total evidenceat work here. But thereis still the
question of the lengthto which one is to go in activesearchof such
counterindication,how far away from the region of the imme-
diatelyat hand and into theregionsof the reasonablyaccessible or
potentiallyobtainable. This question, however,cannot be given a
general answer. It can be answered only relative to the pressures
and constraints,notably time constraints,that the deliberatoris
under. (Anotherfactorhere is the strengthof the presumption,to
be taken up presently.)Indeed, it may turnout that the verypoint
of some presumptionrules (e.g., the conversational)is not that of
coming to the aid of a person whose process of deliberationgets
stuck,but ratherto anticipateand precludethedeliberationprocess
altogether,by providingtheagent with a baseline foraction which
is to be abandoned just in case some counterindicationis more or
less thrustupon him.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
152 THE JOURNAL OF PHILOSOPHY

Be thatas it may,it is importantto note theasymmetry thatgoes

to the heart of the matter:whereas the presumption(that q) and
the injunction correspondingto it (to proceed as if q) are triggered
in the absence of certainreasons forbelief,it is only thepossession
of certainreasons forbeliefwhich rebutsthe presumptionand an-
nuls the injunction. Whereone has reasons forbeliefsufficientfor
the grounding of action, thereis no deliberationproblem; it is to
pave the way to action in defaultof such reasons thatpresumption
rules come about.
(d) The strengthof the presumption8
The presumptionthat q is rebuttedforA when A has sufficient
reasons to believe thatnot-q. How much is sufficient?
The "sufficient"qualifier is, I contend,a place holder. It should
be replaced by an indication of the weightof the reasons forbelief
required for the rebuttal.Correlatedwith this measure will be an
index of the strengthof the presumption. Thus, a presumption
rule relatingto a strongpresumptionwill be expressedby the for-
mula 'pres. (P,Q)', read as "P raisesa strongpresumptionthatQ."
Spelled out, it will say, "Given thatp, you shall proceedas if q un-
less or until you have a conclusive reason to believe that not-q."
Similarly,thepresumptionrule relatingto a presumptionof an in-
termediatestrengthwill be wordedin termsof good or prima facie
reasons forbelief,and the rule relating to a weak presumptionin
termsof some reason forbelief.9
The situation underlyingthe workingof a presumptionrule re-
lating to a weak presumptionis this. Take a rule subject who is
constrainedto act and whose choice of course of action depends on
whetheror not he believes q to be the case. Suppose now that he
has no reason for belief either way, i.e., he is in a state of igno-
rance regardingtheanswer to thefactualquestion q or not q. Such
persons cannot act on the balance of reasons since the proverbial
scales are for them empty and in equipoise. But, given that a
(weak) presumption rule applies to their deliberation situation,
theyare to proceed on q. One may think,forexample, of the va-
rietyof interpretative presumptionsthatcome to our aid in resolv-
8I am indebtedto Ronald Dworkin forillumination on the topic of this section.
9This is an extension of the grading of legal presumptionsaccording to their
strength.The presumptionsof legitimacyand of innocenceare strongpresumptions
in Anglo-Americanjurisdictions,requiringconclusive counterevidencefortheirre-
buttal. The presumptionof sanityis sometimesconsideredweak in the sense thatit
lapses as soon as some counterevidenceis in [see Durham v. United States,214 f. 2d
862 (D.C. Cir. 1954), 741-749]. Most legal presumptions,however,are intermediate
in degreeof strength.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 ON PRESUMPTION 153

ing practical deliberationswhich turn on the ascriptionof inten-

tions and motivationsin the case of a piece of human behavior,
either verbal or nonverbal. Prominentamong them is the Inter-
pretativePresumptionof Cooperation, which enjoins the hearerto
interprethis converser'sutteranceas if it is an appropriatecontri-
bution to theacceptedpurpose or directionof the talkexchange be-
tweenthem,unless or until he has some reason to believeit is an in-
appropriatecontributionthereto.'0
No such lack of reasons forbeliefunderlies the workingof pre-
sumption rules relating to the strongerpresumptions.The situa-
tion in which the rule subjects find themselvesmay broadly be
characterizedas one of doubt regarding the factual question of
whetherq is or is not thecase. They may well have some reasons to
believe q not to be the case, and thesereasons may even be weight-
ier than the reasons theyhave, if any, to believe the opposite. So
thereis in principle a possibilityof acting on the balance of rea-
sons here. Yet if these reasons are perceived as not sufficiently
weightyforthepersonconcernedto proceedto act on them,if there
isn't sufficientconviction, thereremains a deliberationproblem.
Now insofaras a presumptionrule applies to thissituation,it may
operate so as to sway thedeliberatortowardacting on q. The situa-
tion here may be picturednot as one of evenlybalanced scales, the
presumptionrule operating so as to tip the balance, but ratheras
one where the presumptionrule operatesso as to tilt the unevenly
balanced scales towardthe otherside.
As a case in point, consider the situation where the futureaca-
demic employmentof a junior facultymember is to be decided
upon at the end of a certain probationaryperiod. If the person
either excels or obviously fails there is no deliberationproblem.
But if the person succeeds somewhat,theremay well remain a de-
liberation problem. It can be resolved,however,if the institution
adopts a clear presumptivetenurepolicy, e.g., a policy that treats
the academic job as belonging to the person hired as qualified for
it, unless he or she proves incapable of meeting the professional
standards of the universitywithin the specified probationary
period."

'lImplicit in this is the proposal, which I develop elsewhere(see note 23 below),

to recastGrice's well-knownCooperative Principle (in H. P. Grice, William James
Lectures,Harvard 1968, 2nd lecture)as a principle governingthe interpretation of
utterancesratherthan as a principlegoverningtheproductionof utterances;as such
it becomes a Presumptionof Cooperation.
" See Margaret Atherton, Sidney Morgenbesser and Robert Schwartz, "On
Tenure," Philosophical Forum, x, 2-4 (Winter-Summer1978-79): 341-352.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
154 THE JOURNAL OF PHILOSOPHY

But this is not quite accurateenough. If a presumptionrule ap-

plies in a certain instance, it is there prior to the deliberation
process and may at times preemptit altogether.The situation is
more correctlyto be pictured,then,as involving scales which, be-
cause of the presumptionrule and the bias inherentin it, are tilted
(toward the q side) to begin with, and where the balance can be
reversedonly when a certainweightis put on theotherside: when
one has some reason to believe not-q in the case of a weak pre-
sumption,good (prima facie) reasons to believenot-q in thecase of
a presumptionof intermediatestrength,or downrightconclusive
reason to believe not-q in the case of strongpresumptions.
It is this image of some fancied scales being atilt prior to any
weighing which is conveyedby the 'pre-' of 'presumption'.And it
is the strengthof thepresumptionwhich determinestheweightre-
quired forreversingthe balance. As forthe question of the factors
thatdeterminethedifferential strengthof presumptions,thesehave
to do with the relativestrengthof the considerationsin which the
justificationof each presumptionis grounded,as well as with the
"work" it is expected to do. There are no generalizationsthatcan
be made here, except perhaps for the tentativeobservation that
strongpresumptionscan hardlybe expectedto be encounteredout-
side of the frameworkof the law. We turn now to the issue of
justification.
III. THE JUSTIFICATION OF PRESUMPTIONS
The central question up to now was, What are presumptions(or
presumption rules)? The question to be taken up now is, Why
presumptions?
There are in point of fact two justificatorytasks involved. The
firstconcerns the justificationfortherebeing a presumptionrule,
some presumptionrule, ratherthan none: Why are there,or why
should therebe, presumptionrules?The second concernsthejusti-
fication of the specific presumptionespoused by a presumption
rule: Why this presumptionratherthan some alternative?
These two tasksare obviously connected.Only if the firstcan be
metsatisfactorily is thereany point in embarkingupon the second.
And also this: the factthat the second can, in certaincases, be an-
sweredconstitutespart of the answer to the first.
(a) Whypresumptionrules?
A presumptionrule, we recall, comes to the aid of a deliberating
agent when he or she is called upon to act, when the choice of the
course of action to be takenhinges in a materialway on whethera
certain stateof affairsobtains, and when the agent is in a stateof
ignorance or doubt concerning the answer to the question. The

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 ON PRESUMPTION 155

significantfactorin the descriptionof the situation is thattheper-

son concernedis constrainedto take action, some action, beforehis
or her deliberationcan be terminated:the time to act precedesthe
rationalresolutionof thedeliberationprocess.'2(Note thatthismay
comprisecases whereone is constrainedto act, or perhaps to react,
not just beforeconcluding one's deliberations but even without
having startedto deliberate.)In particular,the wait-and-seeoption
allowing for suspended judgment is supposed to be ruled out in
this typeof situationon eitherlogical or practicalgrounds.
This is a descriptionof a deliberation process which, whether
because inconclusive, aborted, or altogetherpreempted,is unre-
solved. Unresolveddeliberations,to be sure,need not be unresolva-
ble. But fromthestandpointof thedeliberatingagents,constrained
as we imagine themto be by timepressureas well as by such "ex-
traneous" factorsas emotional stress,distraction,and languor,
resolvability-in-principle is all but beside the point. The agents
may be in need of some means of extrication.Now a varietyof
rules, or strategies,or second-orderreasons for action, have been
suggested and explored by writerson the subject. I cannot here,
however,delve into that."3The pertinentpoint I wish to make as a
starteron the justificationissue is that presumptionrules should
be regardedas coming to satisfyjust such a need: theyfunctionas a
methodof extrication,one among several,fromunresolveddeliber-
ation processes.What theydo is supply a procedurefordecision by
default.
In order to grasp the rationale of presumptionrules qua means
of extrication,I suggest that we revertonce again to the law and
look at the working of the presumption of innocence, which in
more ways than one formsa class of its own.
A criminal trial is a process at the end of which the triersof fact
must reach an unambiguous verdictfor or against the defendant,
howeverinconclusive the evidence beforethem may be. There is a
point beyond which no more evidence is, or can be, introduced,

12
In his "Reasons for Action, Decisions and Norms," in Joseph Raz, ed.,
Practical Reasoning (New York: Oxford, 1978),pp. 128-143,Raz says: "It should be
rememberedthata decision is reachedonly when the agent (1) reachesa conclusion
as to what he ought to do and (2) formsthe beliefthat it is time to terminatehis
deliberations" (134). He is thereconcernedto emphasize that the firstcondition is
not enough. What I wish to point out is thatoftenthe second condition is met but
the firstis not; it is in this contextthatvarious means of extricationhave a role to
play.
13 Iwould just like to mention in this connection the treatmentof the notion of
picking as a means of extricationfromunresolveddeliberationproblemsof a rather
special and restrictednature,in Edna Ullmann-Margalitand Sidney Morgenbesser,
"Picking and Choosing," Social Research,XLIv, 4 (Winter1977): 757-785.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
156 THE JOURNAL OF PHILOSOPHY

and thereis a time at which the facttriers'deliberationsmust ter-

minate. If the evidence is conclusive and leaves no doubt in their
minds as to thedefendant'sguilt or innocence,thereis no problem.
But where doubt persists,should theydecide by mere preponder-
ance of evidence,howeverslight?And what if the scales are evenly
balanced?
This situation paradigmaticallycalls fora genericmeans of ex-
trication.It is not to functionas a surrogatefor the deliberation
process in each particularcase. It is, rather,to providea generaldi-
rectionforsolution forthe unresolveddeliberationprocesses.The
defendant-any defendant-is either to be acquitted for lack of
proofof guilt or else is to be foundguiltyforlack of proofof inno-
cence. Structurallyspeaking, eitherof these two counterpresump-
tions will do; both supply a procedurefordecision by default.(In-
deed, the old modes of trial-by ordeal, by wager of law, or by
battle-were controlledby a presumptionof guilt: a man charged
with a criminal offensewould be punished unless he managed to
clear, or ratherto "clear" himself.)The question whetherto adopt
one of these two counterpresumptions-and,if so, which-already
slides us into the second justificatory task.
Taking stock:presumptionrules functionas, and are thusjusti-
fied qua, means of extricationfromunresolveddeliberationpro-
cesses. They are called forin situationswhich,when describedgen-
erically,presenta recurrentpatterninherentto which is a decision
problem such that those required to act can be anticipated to be
stuck with an inconclusive or an aborted deliberation process.
They are thus called forwherearbitraryor haphazard decisions are
otherwiselikely to be made.
But themereneed forextricationdoes not in itselfsufficetojustify
the institutionof a presumptionrule. What such a rule in effect
offersis a policy of extricationbased on a systematicbias favoring
one of theavailable alternative-types over theother(s).And it is the
independentjustifiabilityof such a biased solution which is crucial
forthe institutionof a rule espousing it to be justified.
A presumptionrule may be seen, then,as replacingarbitrariness
with somethinglike rational prejudgment;although plainly pre-
judging an issue, it may neverthelessbe defendedas rational14in
the following twofoldsense: (i) in any particularinstancethe pre-

14 An altogetherdifferent approach to the question of the rationalityof presump-

tions is offeredby Louis I. Katzner(op. cit.), who addresseshimselfto the question
whetherthe presumptivistprincipleof formaljustice is rationalin thespecificsense
of whetheror not it is necessarilytrue.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 ON PRESUMPTION 157

sumption it relatesto is open to rebuttal;(ii) the bias it promotesis

independentlyjustifiable.
There remain to be discussed the grounds on which a specific
presumption,i.e., the biased solution promotedby a presumption
rule, may be justified,once the need forsome presumptionrule in
the pertinenttypeof situationhas been established.
(b) Whythespecificpresumption?
The justification of a specific, though generic, presumption-
that-Q is, I suggest,a blend ot two major typesof consideration,
sometimessupplementedby a thirdancillaryone.15They are these:
(i) inductive-probabilistic
considerations
(ii) value-relatedconsiderations
(iii) proceduralconsiderations
Considerations of inductive logic and probability,in a broad
sense, have to do with the likelihood of Q, given P. Broadlyspeak-
ing, theyunderlie most presumptions.Most people unexplainably
absent forupward of seven yearsare in factdead; most people liv-
ing togetheras man and wifeare in factman and wife;mostpeople
mean what theysay (mostof the time); theevidenceof our sensesis
trustworthy in most cases; a descriptiveword is mostlycorrectly
applied to an object displaying certainantecedentlyspecifiedfea-
tures;'6and so on. When the presumed fact-typeQ is routinelya
concomitantof thepresumption-raisingfact-type P, it would seem
to be a waste of time and energy-sometimes even to the point
where the verypurpose of the action to be taken is defeated-to
begin an investigationas to whetherq is indeed thecase in any in-
stanceof theoccurrenceof p. It would seem to make good practical
sense to proceedon theassumptionthatq so long as no indications
to the contrarycrop up.
So with presumptionrules relatingto presumptionsthataccord
with the normal balance of probabilitythe chance of an error(i.e.,
proceedingon q while in factq is not thecase) is reduced.The na-
ture and functionof the inductive-probabilisticconsiderationsin
justifyinga specificpresumptionvis-A-visits counterpresumption
is thus straightforward enough. The point to be emphasizedabout
these considerations,however,is this: theycannot by themselves
provide the ground on which a presumption is justified and,
hence, on which the institutionof a presumptionrule is justified.

I5 On the interplayof theseconsiderationswith regardto legal presumptionssee

Thayer (1898) and Morgan (1931), op. cit.
16For more on the issue of presumptionsof referencesee Ullmann-Margalitand
Margalit op. cit.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
158 THE JOURNAL OF PHILOSOPHY

If the sole rationale of the proceed-as-if-qinjunction of a presump-

tion rule expressedby the formula 'pres (P,Q)' were the factthatP
probabilizes Q, then we would have on our hands just a piece of
practical advice in accordance with some canons of inductiverea-
soning, not a presumptionrule properlyso called. Indeed, to the
extent that pre-systematicor nontechnical uses of the term 'pre-
sumption' merelyconvey the notion of some probabilityconnec-
tion, I considerthe termto be devoid of its special significance.For
a presumptionrule to count as such, its justifyingconsiderations
fromtherealm of inductivelogic and probability,to theextentthat
theyexist,have to combine with otherjustifyingconsiderations.
Moreover,considerationsof the second, normativetypemayeven
outweighthoseof thefirstand lead to theespousal of a presumption-
that-q in spite of the fact that the odds may possibly favornot-q.
Thus, consider the presumptionof innocence: it is not even clear
in thatcase whethertherelevantprobabilisticconsiderationrelates
to the entireclass of human beings or to the much narrowerclass
of persons charged with criminal offenses.This is a trickyques-
tion. But the point is that,even if the relevantclass is taken to be
that of persons charged and even if it turnsout that most of the
people in this class are actually guilty of the crimes attributedto
them, the presumption in favor of proceeding as if the person
charged is innocent may neverthelessbe retainedand defended;it
may flyin the face of the probabilisticconsideration.Or take the
case of the presumptiveprinciple of equality, in situationsfalling
under the broad heading of the administrationof distributivejus-
tice.'7 The proponent of this presumptionwill hold on to its in-
junction, i.e, to proceed as if the people to be treatedwere similar
in all relevantrespectsunless or until relevantdifferences among
themare shown, regardlessof whetherpeople are in factlikelyor
unlikely to be thus similar.
So the justification of presumptions may, and perhaps com-
monly is, couched in inductive-probabilisticterms;but such con-
siderationsare neithernecessarynor sufficientto justifythe pre-
sumption rules relatingto them."8
17
For discussions of thepresumptivistprincipleof equality (or ofjustice) see, e.g.,
S. I. Benn and R. S. Peters,Social Principles and the Democratic State (London:
Allen & Unwin, 1959), p. 111, and The Principles of Political Thought (Glencoe,
Ill.: The Free Press, 1965),pp. 127/8;Joel Feinberg,Social Philosophy (Englewood
Cliffs,N.J.: Prentice-Hall,1965),pp. 100-102; Katzner,op. cit.
18 Consider this statementby David Lyons [Forms and Limits of Utilitarianism
(New York: Oxford,1965), p. 124], who joins issue with Marcus Singer and claims:
"A presumptionagainst lying,forexample, is fullycompatible with mostinstances
of lying,or lying in most kinds of cases, not being wrong. We might say thatpre-
sumptionsas such have no qualitative implications."

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 ON PRESUMPTION 159

It is the justificationof presumptionsin normativetermswhich

toucheswhat I take to be thecore of theconceptof presumption.If
the firsttypeof considerationhad to do with the chance of error,
this normativetypeof considerationhas to do with the acceptabil-
ityof error.Presumptionrules operate in situationswhereactions
have to be decided upon in the light of insufficientinformation
and often under external pressuresand constraints.Errors (i.e.,
proceeding on q when not-q is in factthe case and vice versa) are
bound to occur. There is no question of avoiding errors;at best
thereis the question of reducingtheirnumber.But a different sort
of question is whetherone type of error is to be preferred,on
grounds of moral values or social goals, over the other(s). Evalua-
tiveconsiderationsmayexistwhich justifya systematicand generic
bias in favorof erroneouslyproceeding on Q ratherthan errone-
ously proceeding on not-Q, given that P and given lack of suffi-
cient reasons in the circumstancesto believe eitherq or not-q to be
the case.
Thus, in his Treatise on Judicial Evidence JeremyBentham'9
says that "in doubtfulcases [the judge should] consider the error
which acquits as morejustifiable,or less injurious to the good of
society,than the errorwhich condemns." Hence the presumption
of innocence. It is, then,as a correctivedevice that this presump-
tion,as well as others,may be thoughtof: as regulatingin advance
the directionof errors,whereerrorsare believed to be inevitable.
Similarly,take Louis Katzner'sview with regardto the two rival
presumptions,against treatingpeople differently/similarly in any
respectuntil groundsfordistinction/similarity have been shown:
The onlypossiblebasis foroptingforone of themratherthanthe
otheris whichstateof affairsone would rathersee-that in which
someof thosewhoare similarare treated or thatin which
differently
some of thosewho are different are treatedsimilarly.-[I]tcomes
downto a questionofgoals or values(cf,note1; emphasisadded).20
Consider now the case of the various so-called "conversational"
presumptions,like thepresumptionof truthfulness, or sincerity,or
Grice's Cooperative Principle (see fn 10 above). It is easily realized
thatspending timeand effort, in each and everyinstancein which
19Extractedfromthe manuscriptby M. Dumont (London: J. W. Paget, 1825),pp.
197/8.
20Joel Feinberg argues that to presume equal treatment"would be to make a
presumptioneverybit as arbitraryas thepresumptionin favorof unequal treatment
in the absence of knowledge of the relevant similaritiesand differencesof the
persons involved" (Social Philosophy, p. 102). He concludes thatneitherof the two
opposing presumptionsis independentlyjustifiable,on moral grounds (he ignores
all otherconsiderationsaltogether).

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
160 THE JOURNALOF PHILOSOPHY

one's interlocutorhas utteredsomething,to findout to one's satis-

factionwhetheror not it is indeed trueor sincereor whetheror not
it indeed contributesappropriatelyto the purpose of the talk ex-
change in which one is engaged, beforeone proceedsto respond to
thatutteranceis quite likelytantamountto underminingthepartic-
ular piece of conversationunderconsideration,if not interpersonal
communication in general. It is along such lines, then,of the gen-
eral interestin thesmoothnessof talk exchanges,thatthejustifying
considerations for there being conversational presumptionrules
(some presumptionrules,ratherthan none) are to be couched. But
which presumptionsshall theybe? Can a specificpresumptionin
this area be independentlyjustified?
It may or may not be truethatmost utterancesof indicativesen-
tencesby most human speakersin situationsof communicationare
true,or are utteredin sincerity.Again, it may or may not be true
that most utterancesof most speakers constituteappropriate con-
tributionsto the (accepted,mutuallyrecognized)goal or direction
of the talk exchange in which theyare engaged. To theextentthat
thesethingsare true,therewill be inductive-probabilistic consider-
ations in favorof the presumptionsin question. But regardlessof
the question whetheror not a given utteranceis likely to be true
(etc.), thereis thefurtherquestion of which-if any-of thefollow-
ing alternativesis morallyor socially "better":to proceed(respond,
react) erroneouslyon the assumption of truth(sincerity,appropri-
atenessof contribution),or to proceed erroneouslyon the assump-
tion of falsity(insincerity,
inappropriatenessof contribution).Now
it is my contentionthat if thereis no commitmentto some value
according to which one type(or "zone") of errorsis judged as pref-
erable to the other,then talk of presumptionin this area is inap-
propriate.2"And it is furthermycontentionthatsuch a value judg-
ment is indeed congenial to anyone who accords primacyto such
notions as the dignityof persons in one's thinkingabout the fun-
damentalsof ethics.This evaluativeconsiderationcomes,of course,
on top of the factorof social desirability,if not necessity,of main-
21
Quine's Principle of Charity,the way he formulatesit, is in this sense not a
presumptiveprinciple: "Assertionsstartinglyfalseon the face of themare likelyto
turnon hidden differences of language" [Wordand Object (Cambridge,Mass.: The
MIT Press, 1960), p. 59, my emphasis]; "One's interlocutor'ssilliness, beyond a
certain point, is less likely than bad translation"(ibid.). It seems to me, however,
that this principle can be viewed as a presumptiveprinciple to the extentthat the
charitableinterpretations it calls forare perceivednot just as more likelyto be true
but also as conformingbetterwith one's moral conception of one's fellowmen and
women; to the extent,that is, that giving creditwhere creditis not due is judged
more pardonable, and of a higher moral order,than denyingcreditwhere it is in
factdue.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 ON PRESUMPTION 161
tainingand sustaininginterpersonalcommunication,which is fur-
theredby the conversationalpresumptionsof truth,sincerity,and
appropriatenessof contributionand not by any of theircontrary
presumptions.The normativeconsiderationwe are dealing with,
in sum, can be viewed as two-tiered:it has to do (i) with the ques-
tion of which sortof erroris morallyor socially more acceptable,
and (ii) with the moral or social evaluation of the regulativeeffect
on people's behaviorof thepresumptionrule's being institutedand
operative.
There is yet another consideration intermingledhere, which
ought neverthelessto be kept analyticallyseparate.It may perhaps
be labeled the determinatenessconsideration.22Sometimesit is the
case that,once the need forsome presumptionrule has been recog-
nized, thereis littlereal choice in the matterof which presumption
it should espouse. For example, the presumptionin favorof equal
treatmenteven where thereis ignoranceas to the characteristics of
the persons to be treatedis determinate,whereas the counterpre-
sumption, in favorof unequal treatmentin the absence of knowl-
edge of the relevantsimilaritiesand differences among the persons
involved is so indeterminate(or, as one mightsay, merelydetermi-
nable) as to be quite useless. The same mightseem to apply to the
presumptionof sincerity(or that fluentspeakers mean what they
say): any contrarypresumptionwould be determinableratherthan
determinate,and as such could hardly be of use as a guide for
action.
In such cases, whereit is primarilythedeterminateness consider-
ation thatdecides which specificpresumptionshall be espoused by
the presumptionrule, regardlessof whetheror not it is supported
by inductive-probabilistic considerations,the followingstipulation
needs to be made: where no separate normativeargumentis avail-
able according to which erroneouslyproceeding on q is judged
morallyor socially superiorto erroneouslyproceedingon not-q, it
ought at least to be the case that no such normativeargumentis
available to justifyany contrarypresumption.
A considerationdistinctfromthosedealt with thus far,and often
appealed to in the law, is that of procedural convenience. In the
case of a railway passengerwho is injured by wreckor derailment,
the presumptionis that the railway company was negligent.Mor-
gan (1931, p. 931, cf. note 3) conjecturesthat this presumptionis
grounded in a combination of all threetypesof justifyingconsid-
erations:in the normal balance of probability,in a normativejudi-

22I am indebtedto the late John Mackie forthis point.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
162 THE JOURNAL OF PHILOSOPHY

cial policy directedtoward protectingthe public, and in the pro-

cedural considerationof the comparativeconveniencewith which
the parties can be expected to produce pertinentevidence. In gen-
eral, the procedural considerationhas to do with the question of
what presumptionwill be the most useful to adopt as an initial
step in theprocessof deliberation,what will "help the game along
best"-quite apart from the question whetherthe conclusion to
which theadoption of thisrule points is likelyto be true,as well as
from the question whetherthereexists a standard by which one
sortof erroris judged as more acceptable than the other.
Obviously,iftheanswergivento theproceduralquestion conflicts
with the answer given to the other two questions, the procedural
considerationwill be outweighed.But in thosecases wherethereis
no conflict,whereit is in line with the answers to the otherques-
tions, or indeed where it is the only clear and solid answeravaila-
ble, it counts.
I would like, finally,to returnto the discussion raised above (sec-
tion IIb) about the natureof theinferenceinvolvedin presumption
rules and to offerthe followingobservation.Underlyingthenotion
of presumptionrules therelies a conception of reasoning peculiar
to the practical sphere and different both fromdeductiveand from
inductivereasoning. In deductivereasoning,once the premisesare
specified,arrival at the conclusion is a matterof derivation,in ac-
cordance with some rules. In inductivereasoning,once the prem-
ises are specified,thearrivalat theconclusion is, roughlyspeaking,
a matterof calculation, in accordancewith some canons. Withpre-
sumption rules, once the premises(presumption-raisingfacts)are
specifiedand provided thecontextis practicalratherthan theoreti-
cal, arrival at the conclusion (the presumedfact) is in principle a
matterof a prior, generic decision. It is the decision to sanction
the practical passage fromP to Q. This decision may be motivated,
to a lesseror to a greaterdegree,by thecanons governinginductive
reasoning,but not exclusivelyby them.As we have seen, it mustbe
motivated also by certain evaluative considerations which are
primarilyconcernedwith the differential acceptabilityof the rele-
vant sortsof expectederrors:thefactthatone sortof erroris judged
to be, in the long run and all things considered,preferredon
grounds of moral values or social goals to the alternativesort(s)
constitutesan overridingreason for the decision underlyingthe
presumptionrule. The presumptionrule itself,which such a (ge-
neric) decision underlies,thenenables thedeliberatingagentsto ar-
rive,when necessary,at a (specific)decision by defaultin each con-
creteinstance to which it applies.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS 163
This concludes my systematicpresentation.Clearly,this is but a
bare skeleton. The question whetherthese drybones can live will
be answeredonly when some fleshand sinews-in the formof de-
tailed illustrationsand case studies-are broughtupon them.This,
however,is a task foranothertime.23
EDNA ULLMANN-MARGALIT
The Hebrew Universityof Jerusalem

BOOK REVIEWS
Nonexistent Objects. TERENCE PARSONS. New Haven, Conn.: Yale
UniversityPress, 1980. xiii, 258 p. $19.00.
Putative objects like Hamlet and Jane Eyre have always been puz-
zling, for,despite theirunreality,we seem able to thinkand talk
about them in serious and successfulways. Meinong and others
have wanted to acknowledgesuch objects, but, at least since Rus-
sell's attack on Meinong, the philosophical orthodoxyhas been
that these "objects" are not objects at all. And the orthodoxhave
urged thatour apparentreferences to such entitiesshould somehow
be paraphrased away. Yet doubts have persisted,for we do seem
oftento deal with theseentitiessuccessfully,and satisfactory para-
phraseshave proveddifficultto achieve. In thesecircumstancesvar-
ious philosophers have seen hopes fora more adequate treatment
of such objects in the new ideas in logic thatweredeveloped in the
1960s. One such philosopher was Terence Parsons, who saw in
such ideas the possibility of reviving something like Meinong's
own theory.In NonexistentObjects Parsons now providesa com-
prehensive statementof the version of that theorywhich he has
been workingout in recentpublications.
Nonexistent Objects falls into threeparts: an initial sketch of
Parsons' theory(Introductionand Chapters One to Three); a pres-
entation of a formalsystemembodying the exact version of that
theory(Chapters Four and Five); and an account of various appli-
cations of, and problems for,that version of the theory(Chapters
Six to Nine). The applications in question include the topic of ob-
jects in fiction,the issue of whetherParsons himselfis a nonexist-
ent object (!), and the interpretationof Leibniz and Plato. Parsons
writesinformallyand clearly,and he has tried-for the most part
23 This task is partiallycarriedout in my forthcoming"Some Presumptions".
0022-362X/83/8003/0163$00.90 X 1983 The Journalof Philosophy, Inc.

This content downloaded from 89.160.155.250 on Sun, 3 Aug 2014 18:36:26 PM

All use subject to JSTOR Terms and Conditions