INFORMAL LOGIC 

XY.2, Spring 1993 

Relevance, Conduction and Canada's Rape-Shield Decision) 

DEREK ALLEN University of Toronto 

Key Words: Canadian Supreme Court decision; 
Canada's 1982 rape-shield legislation; informal-
logic class; relevance; conductive argument; "ar-
gument from 'unchasteness"'; George Bowles's 
theory of propositional relevance; Trudy Govier's 
account of a conductive argument. 

Abstract: I examine a Canadian Supreme Court 
decision concerning the constitutionality of Canada's 
1982 rape-shield legislation, and suggest how ma-
terial from the decision might profitably be used in 
an informal-logic class in connection with the top-
ics of relevance and conductive argument. I also 
consider theoretical matters related to the decision: 
first I develop two analyses of what I call an argu-
ment from 'unchasteness' and connect them to 
George Bowles's theory of propositional rele-
vance; then I present Trudy Govier with a problem 
in response to which she might revise her account 
of a conductive argument in a way I describe. 

The Parliament of Canada recently en-
acted legislation to amend sections of the 
Canadian Criminal Code pertaining to the 
offence of sexual assault, The legislation 
came in the wake of a Supreme Court deci-
sion that struck down one of the country's 
two rape-shield laws. In Part I of this paper 
I examine that decision, in the belief that it 
promises to be of pedagogical and theoret-
ical interest within the informal-logic com-
munity. In Part 2 I offer suggestions as to 
how material from the decision might prof-
itably be used in an informal-logic class in 
connection with the topics of relevance 
and conductive argument. In Part 3 I turn 
to theoretical matters related to the deci-
sion, having to do with those same topics. 

1. The Decision 

Background: 

The rape-shield decision was in part a 
constitutional decision involving the Cana-

dian Charter of Rights and Freedoms. The 
Supreme Court was called upon to decide 
whether two sections of the Canadian 
Criminal Code, commonly known as the 
rape-shield laws, infringed certain Charter 
rights. The first section (s. 276) provided 
that in a sexual-assault trial evidence con-
cerning the sexual activity of the com-
plainant with any person other than the 
accused was not admissible in defence of 
the accused except under one or more of 
three narrowly defined conditions. The 
second section (s. 277) provided that in a 
sexual-assault trial evidence concerning 
the sexual reputation of the complainant 
was not admissible for the purpose of chal-
lenging or supporting the complainant's 
credibility. 

The rape-shield laws were enacted in 
1982. Their main purpose was to deal with 
a problem in the common law. The prob-
lem was that the common law allowed sex-
ua�-history evidence and sexual- reputation 
evidence to be presented in a sexual-as-
sault trial even if it was of little relevance 
and would mislead the jury, and allowed 
the jury to draw improper inferences from 
such evidence. For example, if Smith had 
been accused of raping Jones, and there 
were evidence that Jones had had consen-
sual non-marital sex with Brown, the com-
mon law would have permitted a jury to 
infer from this evidence that Jones was 
more likely to have consented to have sex 
with Smith on the occasion in question, and 
less credible as a witness, than she would 
have been had she been 'chaste'. The main 
purpose of the rape-shield laws was to 
abolish the old common-law rules that al-
lowed such inferences to be drawn. There 



106 Derek Allen 

were also three subsidiary purposes. One 
was to prevent judges and juries from be-
ing diverted by irrelevant and prejudicial 
evidence. A second was to encourage the 
reporting of sexual assaults. The third was 
to proteet the privacy of the complainant. 

The laws had the effect of limiting the 
defence strategies available to the accused 
in a sexual-assault triaL Consequently they 
had the potential to help determine wheth-
er the accused was convicted, imprisoned 
and thus deprived of liberty. But s. 7 of the 
Charter says that a person has the right not 
to be deprived of liberty except in accord-
ance with the principles of fundamental 
justice. So the question arose whether the 
potential deprivation of liberty that fol-
lowed from the rape-shield laws occurred 
in a manner that conformed to the princi-
ples of fundamental justice. A further and 
related question was whether the laws in-
fringed the right of an accused person to a 
fair trial, a right proclaimed in s. Il(d) of 
the Charter. A negative answer to the first 
question, however, would not by itself 
mean that the laws were unconstitutional, 
nor would an affirmative answer to the sec-
ond. For the Charter allows in its first sec-
tion that the rights it enumerates are 
subject to reasonable limits-in particular, 
"to such reasonable limits prescribed by 
law as can be demonstrably justified in a 
free and democratic society." Thus if the 
Supreme Court decides that a law violates 
a Charter right, it must go on to ask wheth-
er the law can nevertheless be 'saved' 
under s. 1 of the Charter. 

The Court delivered its decision on the 
constitutional validity of the rape-shield 
laws in August, 1991. All nine judges 
agreed that one of the laws-the sexual-
reputation law (s. 277)-was constitution-
aL But they disagreed about the other-the 
sexual-history law (s. 276): two of the 
judges held that the law was constitutional, 
seven that it was not. 

Madam Justice Beverley McLachlin 
wrote the majority opinion. Madame Justice 
Clair L'Heureux-DuM wrote an opinion 

partially dissenting. I shall now summarize 
parts of both opinions. beginning with 
McLachlin's. 

McLachlin: 

To be admissible in a trial, evidence 
must be relevant. This means that it must 
have probative value-it must count for or 
against some claim whose truth is at issue 
in the trial. But it is generally accepted 
throughout the common law world that 
Crown evidence against the accused in a 
criminal trial, even if it is relevant, may 
properly be excluded if, were it to be ad-
mitted, it would have a prejudicial effect 
exceeding its probative value-as it might 
in a jury trial if, for example, it would un-
duly arouse in the jury emotions of hostility 
or sympathy. When, however, prejudicial 
evidence is for the defence, the prejudicial 
effect it would have if admitted must sub-
stantially outweigh its probative value be-
fore a judge can exclude it. This is because 
a free and democratic society attaches 
great importance to the principle that an 
innocent person must not be convicted. 

McLachlin makes these points by way 
of background (66 Canadian Criminal 
Cases (3d) 389g-391 t). She then asks the 
following question: "can it be said a priori 
... that any and all evidence excluded by s. 
276 will necessarily be of such trifling 
weight in relation to the prejudicial effeet 
of the evidence that it may fairly be ex-
cluded?" (392h-393a) Her answer is no, 
and she cites supporting examples. The 
most striking is an American case (State v. 
Jaio, Oregon Court of Appeal 1976) in 
which a father discovered that his son and 
daughter were baving sexual relations. He 
stopped the relationship and subsequently 
his daughter accused him of the act. At his 
trial the father sought to present his discov-
ery as evidence supporting his defence that 
the accusation was "a concoction motivat-
ed by animus" (393g). Under s. 276, 
McLachlin claims, the evidence would 
have been excluded-"[n]otwithstanding 



its clear relevance" (393g) and (as she 
presumably believes but does not say) 
notwithstanding that its probative value 
would have substantially outweighed such 
prejudicial effect as it might have had. 

From this and other examples McLachlin 
concludes that s. 276 "overshoots the 
mark" (3951) and creates a real risk that an 
innocent person will be convicted. For this 
reason, she further concludes, the legislation 
infringes the Charter right to a fair trial. 

But can it nevertheless be saved under 
s. 1 of the Charter? McLachlin argues that 
it cannot, partly because it "strikes the 
wrong balance between the rights of com-
plainants and the rights of the accused" 
(4031). 

L'Heureux-Dube: 

L'Heureux-Dube begins her (partial) 
dissent by arguing that sexual assault is un-
like any other crime. It goes largely unre-
ported and has the lowest prosecution and 
conviction rates of all violent crimes. One 
reason why prosecution rates are low is 
that "very few cases that come to the atten-
tion of the police are classified as founded" 
(338d). Those that are so classified are 
cases that the police believe likely to result 
in convictions. And in deciding what cases 
are likely to result in convictions, the 
police rely on beliefs about what a model 
rapist is like and what a model rape victim 
is like-beliefs that reveal the influence 
within the judicial system of stereotypes 
about women and myths about rape. For 
example: there is the view that rapists are 
unknown to their victims; that rapists are 
mentally ill; that a raped woman will be 
visibly upset after the event; that women 
fantasize rape; that women are malicious; 
and the apparently self-contradictory view 
that a woman cannot be raped against her 
will. Views such as these are used by po-
lice, consciously or unconsciously, to filter 
out reported cases of sexual assault that are 
deemed unworthy of further attention. 

Relevance. Conduction & Rape-Shield 107 

This is not a matter of conjecture on 
L'Heureux-Dube's part: she cites academic 
studies that support the point. Studies also 
show that rape myths and stereotypes 
about women influence the thinking of 
judges and juries, which is one reason why 
conviction rates are low. For example, there 
was the infamous judge in Cambridge, On-
tario who averred in a 1982 case that 
"[w]omen who say no do not always mean 
no" (34Ih). And an American study (with 
which Canadian data agree) found that 
'''[j]urors are reluctant to convict the de-
fendant when any testimony about prior 
sexual history is introduced in support of the 
consent defense'" (343d; italics removed). 

But, according to L'Heureux-Dube, the 
very idea that such evidence is relevant in a 
sexual-assault case is based on myth and 
stereotype. This being so, the evidence ex-
cluded by s. 276 is "simply irrelevant" 
(364c). Suppose, however, that it is false 
that no relevant evidence is excluded by 
the provision. Then there is a further point 
to be made, namely that the excluded evi-
dence has an "extremely prejudicial effect 
on the trial of the legal issues" (364e), so 
that its exclusion is perfectly proper. 

L'Heureux-Dube summarizes her rea-
soning in remarks that may be reconstruct-
ed as the following conductive argument: 

Admittedly. s. 276 may prevent the accused 
in a sexual-assault trial from leading all rel-
evant evidence. But the sexual-history evi-
dence excluded by s. 276 "is either 
irrelevant or so prejudicial that its minimal 
probative value is overwhelmed by its dis-
torting effect" (372c). Thus, s. 276, in ex-
cluding the sexual-history evidence that it 
does exclude, does not violate the princi-
ples of fundamental justice (where these 
include the principle that an accused per-
son has the right to a fair trial). 

An Assessment: 

The issue that centrally divides the two 
judges is whether the prejudicial effect of 
the evidence excluded by s. 276 always 
outweighs its probative value. I believe 
that McLachlin succeeds in showing that it 



lO8 Derek Allen 

does not. What she does, in effect, is pro-
duce a convincing counter-example to the 
premise of L'Heureux-Dub6's summary 
argument-the claim that the sexual-history 
evidence excluded by s. 276 is "either ir-
relevant or so prejudicial that its minimal 
probative value is overwhelmed by its dis-
torting effect." I have in mind her example 
of the father charged with sexual assault by 
his daughter (State v. lalo). McLachlin is 
surely right that the evidence that s. 276 
would have excluded in that case, namely 
that the daughter had had a sexual relation-
ship with her brother, was, in the circum-
stances of the case, of "clear relevance" to 
the issue of whether her father sexually as-
saulted her; and surely its probative value 
would have substantially outweighed such 
prejudicial effect as it might have had. 

But consider this objection. Someone 
presented with the facts of the lalo case who 
believes the evidence relevant will do so 
because he or she believes that the father's 
explanation of why his daughter charged 
him with sexual assault is plausible if, as 
the father claims, the charge is false. His 
explanation is that his daughter reacted 
with animosity to his stopping her sexual 
relationship with her brother, and that it 
was out of animosity that she laid the 
charge. But the belief that this explanation 
is plausible if the charge is false 
presupposes a (false) sexist stereotype-a 
stereotype a!<in to one cited by 
L'Heureux-Dube, namely that "the femi-
nine character is especially filled with mal-
ice" (337a). Accordingly, someone 
presented with the laio facts who believes 
that the daughter's having had a sexual re-
lationship with her brother is relevant to 
whether her father sexually assaulted her 
will do so on the basis of a belief that 
presupposes a (false) sexist stereotype. 

Reply: To find that the father's expla-
nation is plausible if the charge is false, 
one need only make the non-sexist, and 
plainly true, assumption that animosity is 
an intelligible reaction on the part of any 
person, male or female, whose will is 

thwarted by another, especially in a matter 
as significant for personal well-being as a 
sexual relationship. Thus the belief that the 
father's explanation is plausible if the 
charge is false does not presuppose a (false) 
sexist stereotype. Accordingly, to believe 
on the basis of that belief that the evidence 
in question in laio is relevant is not to be-
lieve it relevant on the basis of a belief pre-
supposing such a stereotype. Hence the 
objection fails. I therefore continue to 
think that lalo is a convincing counter-ex-
ample to the premise of L'Heureux-Dube's 
summary argument. (Here, and above, I as-
sume that the sexual-history evidence in 
question would not have been admissible 
under one of the 'exception' clauses of s. 
276. McLachlin obviously makes this as-
sumption herself. Whether it is correct, I 
will not pause to consider.) 

2. Some Teaching Suggestions 

Relevance: 

In A Practical Study of Argument 
(Wadsworth, 1992; hereafter cited as 
PSA), Trudy Govier mentions several rea-
sons why "unwary audiences are often de-
ceived" (152) by arguments with irrelevant 
premises. One possibility that she does not 
mention, but that it would be desirable for 
students to be brought to recognize, is that 
arguer and audience may consider the 
premises relevant (or fail to see that they 
are irrelevant) because, consciously or un-
consciously, they subscribe to certain ster-
eotypes or myths. This is apparent from 
the problem that led to the legislation 
reviewed in Canada's rape-shield decision. 

From that problem it is a simple matter 
to construct an example that could be used 
with profit in an informal-logic class to make 
the poi'lt that, in a context of argumenta-
tion, highly consequential judgments of 
relevance may be accepted by arguer and 
audience on the basis of stereotype or 
myth. The example might go as follows: 



Smith is charged with raping Jones. At 
his trial, his lawyer cross-examines Jones 
on her sexual history and learns that she is 
'unchaste'-she has engaged in non-
marital consensual sex, with a person other 
than Smith. This evidence is admitted by 
the judge as being relevant to the question 
whether Jones consented to the sexual ac-
tivity that is at issue in the trial and to the 
question whether Jones is credible as a wit-
ness. From the evidence of Jones's sexual 
history the jury is therefore permitted to 
infer, and, in accordance with argumenta-
tion made by Smith's lawyer. does infer, 
that Jones is more likely to have consented 
to have sex with Smith on the occasion in 
question, and is less worthy of belief, than 
she would have been had she been 
'chaste' ,2 Judge and jury accept, then, that 
the evidence of Jones's sexual history is 
positively relevant to 

(I) Jones consented to have sex with Smith 
on the occasion in question, and to 

(2) Jones is not credible as a witness. 

The students would be told that while 
this example is hypothetical, it draws upon 
recent legal fact, namely that at common 
law in Canada prior to 1982 evidence that 
the complainant in a sexual-assault trial 
was 'unchaste' was considered relevant to 
the issues of consent and credibility in just 
the way that it is in the example. (On this 
point, see L'Heureux-Dube, 345d-347b.) 
There might then be discussion of whether 
such evidence is relevant to those issues 
(or, more specifically, whether the evi-
dence of Jones's 'unchasteness' is posi-
tively relevant to either (l) or (2))-
discussion leading to the conclusion that 
the example is one in which a judge and 
jury accept a certain judgment of relevance 
because they subscribe, consciously or un-
consciously, to an odious stereotype or 
myth, to the effect that women who have 
consensual sex outside marriage have, in 
the words of L'Heureux-Dube, "a dual pro-
pensity: to consent to sexual relations at 
large and to lie" (346c). 

Relevance, Conduction & Rape-Shield 109 

The following question might then be 
raised: even if Jones's sexual history, as 
reported in the example, isn't positively 
relevant to (1) or (2), can realistic circum-
stances nevertheless be pictured in which 
evidence that the complainant in a sexual-
assault trial had had non-marital consensu-
al sex with a person other than the accused 
would be relevant to the issue of whether 
she was raped by the accused? Following 
consideration of this question, the students 
would be presented with the facts of the 
lalo case and asked to decide whether the 
fact that the daughter had had a sexual re-
lationship with her brother is relevant. in 
the circumstances of the case, to whether, 
as the daughter charges, her father sexually 
assaulted her. They would be required to 
justify their answers to this question. Sev-
eral of those (and there would surely be 
some) who offered justifications for the af-
firmative answer would then be questioned 
to determine whether they did so under the 
influence of a stereotype; if they did, they 
would be asked to consider whether the 
justifications presupposed a stereotype.' 
Suppose it were agreed that they did not 
(as in my 'assessment' in Part I). It would 
then be appropriate to make the following 
point: a category of evidence (e.g .. sexual-
history evidence) and a type of factual 
question (e.g., did the accused commit the 
act that he or she is accused of having 
committed?) may be such that while in cer-
tain instances, or even normally, the belief 
that the former is relevant to the latter will 
be based on stereotype or myth, there are 
nonetheless instances in which evidence 
belonging to that category is relevant to a 
question of that type independently of ster-
eotype or myth.4 

Conduction: 

According to Govier, a conductive ar-
gument is an argument with the following 
characteristics: (a) "the pattern of support 
is convergent"S (PSA 355), except "in the 
limiting case where there is only one 



Ito Derek Allen 

premise" (Problems in Argument Analysis 
and Evaluation (Foris, 1987), 70; hereafter 
cited as Problems); (b) the "premises are 
put forward as separately being positively 
relevant to the conclusion" (PSA 355); (c) 
none of the premises entails the conclu-
sion, nor do they jointly entail it; thus a 
conductive argument is never a deductively 
valid argument (PSA 308, 357 n. 2; Prob-
lems 70); (d) "[clounterconsiderations may 
be acknowledged by the arguer", a coun-
terconsideration being a claim that is "neg-
atively relevant to the conclusion of an 
argument" (PSA 355). (A counterconsider-
ation is not a premise; rather, it "could be 
regarded as a kind of anti premise" (PSA 
310).) By 'conductive argument' (reason-
ing) I mean what Govier means. 6 

Govier remarks that conductive argu-
ments appear in many contexts, and men-
tions several (PSA 3 I 3). One, which she 
does not mention, is that of legal reason-
ing. The rape-shield decision is a case in 
point. For, as I observed above, the re-
marks in which L'Heureux-DuM summa-
rizes her reasoning on whether s. 276 
violates the principles of fundamental jus-
tice can be reconstructed as a conductive 
argument. And this is a natural reconstruc-
tion. Thus L'Heureux-Dube claims that 
"the argument that an accused is prevented 
[under s. 276] from adducing all relevant 
evidence going to innocence has little 
weight in this inquiry and must give way to 
other considerations" (372d). On my re-
construction, that "argument" is (what 
Govier calls) a counterconsideration, 
which is allegedly outweighed by a 
premise stating "other considerations". 

One of McLachlin's central arguments 
also admits of a conductive reconstruction. 
The argument reviews three justifications 
for s. 276 and in each case presents one or 
more reasons why the section nevertheless 
"overreaches" (395f). On the reconstruc-
tion I have in mind, these reasons are 
premises offered in support of the conclu-
sion that the wide reach of the section is 
unjustified, while each of the justifications 

figures in the argument as a counterconsid-
eration to that conclusion. 

On the other hand, in both L'Heureux-
Dube and McLachlin there is reasoning 
that is not conductive but that a student of 
Govier might be tempted so to construe. In 
L'Heureux-Dube the reasoning concerns 
whether "the effects of the measure Is. 
276] are so deleterious that they outweigh 
the importance of the objective" (379d). In 
McLachlin the reasoning concerns essen-
tially the same issue, namely, lithe balance 
between the importance of the objective 
and the injurious effect of the legislation" 
(403c-d). According to Govier, conductive 
reasoning frequently involves the weigh-
ing of pros and cons. Thus a student of 
Govier might be tempted by the talk of 
outweighing in L'Heureux-DuM and the 
talk of balance in McLachlin to interpret 
each judge's reasoning on the 'objective v. 
effect' issue as a conductive argument in 
which the injurious effect of the legislation 
counts against (or, as the case may be, for) 
the judge's conclusion on the matter, while 
the importance of the objective counts the 
other way. But this would be a mistake. 
Rather, what each judge does in effect is 
argue for a conclusion asserting that one 
factor outweighs another-in the case of 
L'Heureux-Dube that the importance of 
the objective outweighs the injurious effect 
of the legislation, in the case of McLachlin 
the converse. (I say 'in effect' because in 
each case the reasoning is elliptical.) And 
this is different from conductive reasoning 
that involves the weighing of pros and 
cons; for reasoning of that sort is based on 
the judgment that one factor, or set of fac-
tors, outweighs another, which is to say 
that some such judgment underlies the in-
ference to the- reasoning's conclusion-
rather than being itself the conclusion. 

Govier is of course right that conduc-
tive arguments frequently involve the 
weighing of pros and consJ But it is worth 
adding that arguments that involve the 
weighing of one factor, or set of factors, 
against another need not be conductive. 



Students might be brought to recognize 
this by being given the exercise of deciding 
whether either McLachlin or L'Heureux-
DubIS engages in conductive reasoning on 
the 'objective v. effect' issue. And this ex-
ercise might be supplemented by one re-
quiring them to reconstruct, in the form of 
a conductive argument, the remarks in 
which L'Heureux-DubIS summarizes her 
reasoning on whether s. 276 violates the 
principles of fundamental justice, and/or 
the reasoning in which McLachIin pro-
ceeds from a review of the justifications 
for s. 276 to the conclusion that the sec-
tion's wide reach is unjustified. 8 

3. Some Theoretical Matters 

L'Heureux-DubIS is strongly of the 
view that the area of the law concerned 
with sexual assault "has been particularly 
prone to the utilization of stereotype in de-
terminations of relevance" (356e). "[T]he 
concept of relevance has been imbued with 
stereotypical notions of female complain-
ants and sexual assault. That this is so is 
plain from the common law which held 
that evidence of 'unchasteness' was rele-
vant to both consent and credibility" 
(355g-356). The common law would thus 
have tolerated arguments in which evi-
dence that a female complainant in a sexu-
al-assault trial was 'unchaste' was offered 
in support of a proposition affirming that 
she was not credible as a witness or that 
she consented to have sex with the ac-
cused. An argument of this sort may be re-
constructed as what I shall call an 
argument from 'unchasteness', or as an ar-
gument that contains an argument from 
'unchasteness' as a subargument. In this 
section I develop two analyses of a hypo-
thetical argument from 'unchasteness', and 
connect them to a recent theory of proposi-
tional relevance. The discussion in which I 
do this leads me to present Govier with a 
problem in response to which she might re-
vise her account of a conductive argument 

Relevance, Conduction & Rape-Shield 111 

in a way that I describe. Relative to the re-
vised account, L'Heureux-DubIS's summa-
ry argument is no longer conductive, but 
the argument of McLachlin's reconstruct-
ed in note (8) still is, or so I argue. 

Relevance: 

L'Heureux-DubIS says that "[a]ny con-
nection between the evidence [of 'un-
chasteness'] sought to be adduced [in a 
sexual-assault trial] and the fact or matter 
of which [at common law] it was suppos-
edly probative must be bridged by stereo-
type (that 'unchaste' women lie and 
'unchaste' women consent indiscriminate-
ly), otherwise the propositions make no 
sense" (355h). This isn't right. There are 
two propositions, one of which is given as 
evidence for the other. And they "make 
sense" independently of their being 
bridged by stereotype. What L'Heureux-
DubIS should rather have said is that the 
view that one of the propositions (the evi-
dence proposition) is relevant to (or evi-
dence for) the other "makes no sense" 
unless the propositions are bridged by ster-
eotype. (By 'relevant' I mean here and, un-
less otherwise indicated, below 'positively 
relevant'.) Thus if the propositions are 

(4) Jill is an 'unchaste' woman, and 
(5) Jill lies, 

then the view that (4) is relevant to (5) 
"makes no sense" unless the two proposi-
tions are connected by a third proposition 
stating a stereotype, for example: 

(3) All 'unchaste' women lie. 

The connection is made in the follow-
ing linked argument: 

(A) (3) All 'unchaste' women lie. 
(4) Jill is an 'unchaste' woman. 

Therefore, 

(5) Jill lies. 

I call argument (A) an argument from 'un-
chasteness'. An argument from 'unchaste-
ness' (I stipulate for the purposes of this 



112 Derek Allen 

discussion) is a syllogism (by which I 
mean a two-premise linked argument with 
a single conclusion) containing a general 
premise (which may be universal or partic-
ular) attributing a certain characteristic, c, 
to all or most 'unchaste' women (which at-
tribution is a stereotype), a singular premise 
alleging of a certain woman, who is the 
complainant in a sexual-assault trial, that she 
is 'unchaste', and a conclusion attributing 
to that woman characteristic c. Such an ar-
gument may be a subargument of an argu-
ment whose main conclusion is, for exam-
ple, that the complainant in question is not 
credible as a witness. Thus argument (A) 
might be a subargument of an argument 
containing the premise that no one who lies 
is credible as a witness and the conclusion 
(deduced from this premise jointly with 
(5» that Jill is not credible as a witness. 

On one analysis of argument (A). (4) is 
an evidence premise and (3) is a warrant 
premise. I borrow the terms 'evidence 
premise' and 'warrant premise' from 
James Freeman,9 and understand them as 
follows. An evidence premise of an argu-
ment is a premise whose function is to af-
ford evidence for the conclusion (or for a 
conclusion) of the argument. A warrant 
premise of an argument is a premise that, if 
and only if it is true, warrants the judgment 
that some other premise of the argument 
(an evidence premise) is relevant to the 
conclusion for which it is offered as evi-
dence, by explaining why the premise is 
relevant to that conclusion. 10 What I shall 
caIl a warrant analysis of an argument in-
terprets the argument as containing at least 
one evidence premise and at least one war-
rant premise. Thus the envisaged analysis 
of argument (A), on which (4) is an evi-
dence premise and (3) a warrant premise, 
is a warrant analysis of that argument. 

Suppose that the analysis is correct. 
Then if (3) warrants the judgment that (4) 
is relevant to (5), then (4) is relevant to (5). 
For if (3) warrants the judgment that (4) is 
relevant to (5), it explains why (4) is 
relevant to (5), and if it explains why (4) is 

relevant to (5), then (4) is relevant to (5).11 
Now (3) warrants the judgment that (4) is 
relevant to (5) if and only if (3) is true. 
Thus, on the envisaged warrant analysis of 
argument (A), if (3) is true, then (4) is 
relevant to (5). 

Objection: Suppose that if (3) is false, 
(4) is not relevant to (5). Then it is not the 
case that (4) alone is relevant to (5)-even 
if (3) is true. But if the envisaged warrant 
analysis of argument (A) is correct, then if 
(3) is true, (4) alone is relevant to (5)-
even if (4) is not relevant to (5) if (3) is 
false. Thus that analysis is not correct. 

This objection is modelled in part on 
an argument made by George Bowles. In 
"Propositional Relevance" (Informal Logic 
XII.2, Spring 1990,65-77; hereafter cited 
as PR), Bowles defends a definition of 'rel-
evance' according to which one proposi-
tion, (p), is (positively or negatively) 
relevant to another proposition, (q), if and 
only if (p) makes (q) certain or probable or 
improbable or impossible. In the same arti-
cle, he considers the following objection: 

it is possible for one proposition to be rele-
vant to another without making the seeond 
certain, probable, improbable, or impossi-
ble. For instance, let 'p' be 'Aspirin tends 
to cause stomach bleeding' and 'q' 'You 
shouldn't take aspirin unless you really 
need it'. In this case, 'p' is relevantto 'q'-
it is a reason for although it does not 
make 'q' certain, probable, improbable, or 
impossible. (72) 

Bowles replies: 

'p's being a reason for, and relevant to, 'q' 
depends on the truth of a third proposition, 
'r'-e.g., 'Stomach bleeding is bad' or 
'You shouldn't take .anything that tends to 
cause stomaeh bleeding unless you really 
need it'. Forif 'r' were false-e.g., if stomach 
bleeding were neither good nor bad, so that 
there were no reason to avoid taking some-
thing that tends to cause it-, 'p' would not 
be a reason for, or relevant to, 'q'. Hence, it 
is not 'p' alone but the conjunction of 'p' 
and 'r' that is a reason for, and relevant to, 
'q'. And that conjunction does make 'q' 
probable if not certain. (72-3) 



Bowles argues, in part, as follows: 

(6) (p)'s being a reason for, and relevant to, 
(q) depends on the truth of (r) [i.e., on 
the fact that (r) is true]. 

Hence, 

(7) It isn't (p) alone but the conjunction of 
(p) and (r) that is a reason for, and rele-
vant to, (q). 

(7) is the conjunction of 

(7a) (p) alone isn't a reason for, and rele-
vant to, (q), and 

(7b) the conjunction of (p) and (r) is a rea-
son for, and relevant to, (q). 

(7a) is open to different interpretations. It 
may mean 

(7a') (p) alone-Le., (p)-isn't a reason for, 
and relevant to, (q). 

Or it may mean something like 

(7a") (p) alone-Le., independently of any 
third proposition-isn't a reason for, 
and relevant to, (q). 

Now obviously (7a') doesn't follow from 
(6). For even if (6) is true, it may be that 
(p) is a reason for, and relevant to, (q) just 
in case (r) is true, and that (r) is true. In this 
event, (p) is a reason for, and relevant to, 
(q), and so (7a') is false. Since it is obvious 
that (7a') does not follow from (6), Bowles 
surely does not interpret (7a) as (7a'). It is 
much more likely that he would interpret it 
as (7a"). For (7a") does follow from (6). 
However, (7a") is compatible with 

(8) (p) alone-i.e., (p)--is a reason for, and 
relevant to, (q), 

which claim is, of course, the negation of 
(7a'). And (8), in tum, is compatible with 
(7b). For if (p) is a reason for, and relevant 
to, (q), as (8) asserts, it may also be that the 
conjunction of (p) and (r) is a reason for, 
and relevant to, (q), as (7b) asserts. Thus 
even if Bowles has established the truth of 
(7b), a matter I will not pursue, as well as 
the truth of (7a) interpreted as (7a"), he has 
not shown that (8) is false. That is to say, 
for all Bowles has shown, it may be that 

Relevance, Conduction & Rape-Shield 113 

(p) is a reason for, and relevant to, (q)-
just as Govier thinks it is. 

Now the stated objection asserts that if 
the envisaged warrant anlysis of argument 
(A) is correct, then, if (3) is true, (4) alone 
is relevant to (5). This is true if '(4) alone 
is relevant to (5)' simply means '(4) is rele-
vant to (5)'. But it is not true if '(4) alone is 
relevant to (5)' means something like '(4), 
independently of any third proposition, is 
relevant to (5),; for an adherent of the anal-
ysis is at liberty to say that whether (4) is 
relevant to (5) depends on whether a 'third 
proposition~.g., (3)-is true. Thus the 
question to ask in assessing the objection 
is whether the inference from 

(9) if (3) is false, then (4) is not relevant to 
(5), to 

(10) (4) alone-i.e., (4)-is not relevant to 
(5) 

is sound. And plainly it is not-any more 
than the inference from (6) to (7a') is 
sound. For even if (9) is true, it may be that 
(4) is relevant to (5) just in case (3) is true, 
and that (3) is true. In this event, (4) is rele-
vant to (5). Thus, even if (9) is true, it may 
be that, in the relevant sense, (4) alone is 
relevant to (5). Thus the objection does not 
show that the envisaged warrant analysis 
of argument (A) is incorrect. 

On that analysis, to repeat, (4) is rele-
vant to (5) if (3) is true. The analysis does 
not, however, entail that (4) is relevant to 
(5) only if (3) is true (but that (3) warrants 
the judgment that (4) is relevant to (5) only 
if (3) is true). Thus the analysis does not 
entail (though it is compatible with the 
claim) that (3) is a presupposition of the 
judgment that (4) is relevant to (5). 

As to whether (4) is relevant to (5), the 
analysis does not affirm either that it is or 
that it is not. By contrast, my second analy-
sis of argument (A), which I shall now pro-
ceed to develop, affirms that (4) is not 
(positively) relevant to (5). 

Consider the following argument: 

(B) (p) Aspirin tends to cause stomach 
bleeding. 



114 Derek Allen 

Therefore, 

(q) You shouldn't take aspirin unless you 
really need it. 

On Bowles's theory of propositional rele-
vance, (p) is not (positively) relevant to (q) 
if (p) does not make (q) certain or probable 
(see PR 67, 69). Let us assume that, by this 
criterion, (p) is not (positively) relevant to 
(q), and that in argument (B) (p) is an evi-
dence premise. Then, relative to Bowles's 
theory of propositional relevance, there 
exists in argument (B) what I shall call a 
relevance gap. 

A relevance gap is a gap between an 
evidence premise of an argument and the 
conclusion for which the premise is of-
fered as evidence, and exists if and only if 
the premise is not (positively) relevant to 
that conclusion. 

Note that I have not claimed that there 
is a relevance gap in argument (B). Rather, 
I have claimed, on certain assumptions, 
that there is a relevance gap in the argu-
ment relative to a certain theory of propo-
sitional relevance. 12 Assuming that relative 
to that theory the gap exists, it is filled, rel-
ative to the same theory, if we add to the 
argument as a further premise proposition 
(r)-"You shouldn't take anything that 
tends to cause stomach bleeding unless 
you really need it"-provided that on that 
theory the conjunction of (p) and (r) is rel-
evant to (q). Adding (r) as a further 
premise to argument (B) gives us 

(C) (r) You shouldn't take anything that 
tends to cause stomach bleeding 
unless you really need it. 

(p) Aspirin tends to cause stomach 
bleeding. 

Therefore, 

(q) You shouldn't take aspirin unless 
you reall y need it. 

On what I shall call a relevance-gap analysis 
of argument (C), there is a relevance gap 
between (p) and (q), but the gap is filled by 
(r) because (according to the analysis) the 
conjunction of (p) and (r) is relevant to (q). 

My second analysis of argument (A) is 
a relevance-gap analysis. It asserts that 
there is a relevance gap in argument (A) 
between (4) and (5) but that the gap is 
filled by (3) because (according to the 
analysis) the conjunction of (3) and (4) is 
relevant to (5). 

We now have two analyses of argument 
(A). Corresponding to each is a different 
view attributing a certain logical function 
to the argument's general (stereotype-
stating) premise-and, by extension, to the 
general (stereotype-stating) premise of any 
argument from 'unchasteness'. To the war-
rant analysis of argument (A) corresponds 
the view that the logical function of the ar-
gument's general premise, (3). is to war-
rant the judgment that the argument's 
singular premise, (4), is relevant to the 
conclusion. (5). To the relevance-gap anal-
ysis of the argument corresponds the view 
that (3)'s logical function is to fill a rele-
vance gap in the argument, between (4) 
and (5). The former view does not, of 
course, entail that (3) does warrant the 
judgment that (4) is relevant to (5), and the 
latter view does not entail that (3) fills the 
alleged relevance gap; for a claim attribut-
ing a certain function to a premise of an ar-
gument does not entail that the premise 
succeeds in fulfilling that function. 13 

And plainly (3) does not fuflfill the 
function that the former view-the warrant 
view-attributes to it. For it fulfills that 
function only if it is true; but, at least when 
interpreted to mean that all women who 
have non-marital consensual sex habitually 
lie, it is false. 

For these same reasons Bowles would 
deny that (3) so interpreted (and it is this 
interpretation of (3) that I have in mind in 
this paragraph and the following) fulfills 
the function that the latter view-the rele-
vance-gap view-attributes to it-even if he 
agreed, as he obviously WOUld, that there is 
a relevance gap between (4) and (5). Or so 
his reply to Govier suggests. He holds the 
following position: in the 'aspirin' exam-
ple, (rrs being true is necessary for (p) to 



be relevant to (q); but (p) alone isn't rele-
vant to (q); rather, the conjunction of (p) 
and (r) is relevant to (q). But then his point 
must be that (rrs being true is necessary 
for the conjunction of (p) and (r) to be rele-
vant to (q). Evidently, then, he would say 
that, in the 'chastity-and-credibility' exam-
ple, the conjunction of (3) and (4) is not 
relevant to (5) unless (3) is true. But (3) is 
false. Thus Bowles would not agree that 
the conjunction of (3) and (4) is relevant to 
(5), and so he would deny that (3) fills the 
relevance gap he would grant exists in 
argument (A) between (4) and (5). 

However it is false that, in the 'aspirin' 
example, (rrs being true is necessary for 
the conjunction of (p) and (r) to be relevant 
to (q). This is false on Bowles's own theory 
of propositional relevance, which allows 
that one proposition may be relevant to an-
other even if the first is false. The conjunc-
tion of (p) and (r) is indeed relevant to (q), 
but this is independent of the fact that (r) is 
true. On Bowles's theory of propositional 
relevance, it is, rather, a consequence of 
the fact that (q) is "probable if not certain" 
given the conjunction of (p) and (r). If, 
similarly, (5) is probable or certain given the 
conjunction of (3) and (4), then, on Bow-
les's theory of propositional relevance, the 
conjunction of (3) and (4) is relevant to (5), 
despite the falsity of (3). In fact, (5) is cer-
tain given the conjunction of (3) and (4). 
Thus, on Bowles's theory of propositional 
relevance, the conjunction of (3) and (4) is 
relevant to (5). Hence, relative to that theo-
ry, (3) does fulfill the function that the rele-
vance-gap view attributes to it - assuming 
that, relative to that theory, there is a rele-
vance gap between (4) and (5); and plainly 
there is, for (4) does not make (5) probable 
or certain by Bowles's criteria. 14 

A relevance-gap analysis of argument 
(A) affirms both that there is a relevance 
gap in the argument, between (4) and (5), 
and that the conjunction of (3) and (4) is 
relevant to (5). This analysis, it is now ap-
parent, is compatible with Bowles's theory 
of propositional relevance. More generally, 

Relevance, Conduction & Rape-Shield 115 

that theory is compatible with a relevance-
gap analysis of any argument from 'un-
chasteness' whose singular premise does 
not make its conclusion probable or certain 
by Bowles's criteria, but whose conclusion 
is by those criteria made probable or cer-
tain by the conjunction of its premises; for 
in any such argument there is a relevance 
gap, relative to Bowles's theory, and the 
gap is filled, relative to that theory, by the 
argument's general (stereotype-stating) 
premise. (More generally still, Bowles's 
theory is compatible with a relevance-gap 
analysis of any syllogism containing an ev-
idence premise that does not make the con-
clusion probable or certain by his criteria, 
but whose conclusion is by those criteria 
made probable or certain by the conjunction 
of its premises.) 

On the other hand, Bowles's theory of 
propositional relevance is incompatible 
with a warrant analysis of an argument 
from 'unchasteness' whose singular 
premise does not make its conclusion 
probable or certain by his criteria-for ex-
ample, argument (A). For if the argument's 
singular premise does not make its conclu-
sion probable or certain by his criteria, it 
does not make the conclusion probable or 
certain by his criteria even if the argu-
ment's general premise is true. Thus, even 
if the argument's general premise is true, 
the singular premise is not relevant to the 
conclusion, relative to Bowles's theory of 
propositional relevance. But a warrant 
analysis of the argument will entail that the 
singular premise is relevant to the conclu-
sion if the general premise is true. (For it 
will entail that if the general premise is 
true, it explains (in the sense indicated in 
note (II) why the singular premise is 
relevant to the conclusion.) Thus a warrant 
analysis of the argument will be 
incompatible with Bowles's theory of 
propositional relevance. (So will be a war-
rant analysis of any syllogism containing 
an evidence premise that does not make 
the conclusion probable or certain by 
Bowles's criteria.) 



116 Derek Allen 

If, then, Bowles's theory is correct, a 
warrant analysis of an argument from 'un-
chasteness' is incorrect if the argument's 
singular premise does not make its conclu-
sion probable or certain by his criteria. But 
is Bowles's theory correct? I shall confine 
myself here to one observation, namely 
that, applied to the 'aspirin' example, the 
theory has a counterintuitive result, if, as 
Govier claims and Bowles does not dis-
pute, (p)-"Aspirin tends to cause stomach 
bleeding"-does not make (q)-"You 
shouldn't take aspirin unless you really 
need it"-probable or certain (by Bowles's 
criteria, I assume). For then, relative to 
Bowles's theory of propositional rele-
vance, (p) is not (positively) relevant to (q). 
But that (p) is not (positively) relevant to 
(q) is counterintuitive. For it is intuitively 
very plausible both that (p) is relevant to 
(q) if (p) is a reason for (q)and that (p) is a 
reason for (q)-that I am given a reason 
not to take aspirin unless I really need it if I 
am informed that aspirin tends to cause 
stomach bleeding. Hence it is intuitively 
very plausible that (p) is relevant to (q).15 

Conduction: 

Suppose that (p) is indeed relevant to 
(q). Then there is a third proposition which 
is true and whose truth (prs being relevant 
to (q) depends upon trivially-the proposi-
tion that (p) is relevant to (q). Furthermore, 
if (p) is relevant to (q) by a general criteri-
on of propositional relevance, CPR, then 
there is a third proposition (Le., a proposi-
tion in addition to (p) and (q» which is true 
and whose truth (prs being relevant to (q) 
depends upon criterially-a proposition 
whose conjunction with CPR entails that 
(p) is relevant to (q). (For example, if CPR 
asserts that one proposition is relevant to 
another that it makes probable, the propo-
sition is that (p) makes'(q) probable.) Now 
Bowles believes, recall, that (prs being 
relevant to (q) depends on the truth of a 
third proposition, (r)--e.g., "You shouldn't 
take anything that tends to cause stomach 

bleeding unless you really need it". But he 
would say that (p)'s being relevant to (q) 
depends upon the truth of (r) non-trivially 
and non-criterially. 

Govier, as we know, believes that (p) is 
relevant to (q). Suppose that she too would 
say (as she might) that (p)'s being relevant 
to (q) depends non-trivially and non-
criterially upon the truth of (r). (Hereafter 
when I use the verb 'depend' I shall omit 
the qualifying adverbial phrase 'non-trivi-
ally and non-criteriaIly', but in each case it 
is to be understood.) Then she would say 
that if (p) is, but (r) is not, a stated premise 
of an argument for (q) (as in argument 
(B», the (positive) relevance of premise 
(p) to conclusion (q) depends on the truth 
of a third proposition, (r), that is not a stat-
ed premise of the argument. But if she 
would say this, she would also say that if a 
stated premise of an argument is relevant 
to the conclusion (or to a conclusion) of 
the argument, its being relevant to that 
conclusion may depend on the truth of a 
further proposition that is not a stated 
premise of the argument. And I think it is 
clear that Govier would say this. Thus, in 
evaluating a certain conductive argument 
(about the American Revolution), she 
says: "the supporting [stated] premises are 
relevant to the conclusion provided we 
grant the assumption behind the argument: 
that a typical revolution involves moves, 
often violent, by the poorer classes to upset 
a social structure" (PSA 427). If that as-
sumption is true, the premises are relevant 
to the conclusion; indeed, since the argu-
ment is conductive, they are then separate-
ly relevant to the conclusion. Thus Govier 
would surely allow that if a stated premise 
of an argument is relevant to the conclu-
sion (or to a conclusion) of the argument, 
its being relevant to that conclusion may 
depend on the truth of a further proposition 
that is not a stated premise of the 
argument. 16 

But then it is possible to present Govier 
with a problem. Let (x) be a convergent ar-
gument with stated premises (a) and (b), 



which are "put forward as separately being 
positively relevant to the conclusion", (c) 
(PSA 355). Neither (a) nor (b) entails (c), 
nor do (a) and (b) jointly entail (c). But 
Govier would say, I shall assume, that (a) 
and (b) are separately relevant to (c). She 
would also say, I shall further assume, that 
(a)'s being relevant to (c) depends on the 
truth of a fourth proposition, (d), and that 
(b)'s being relevant to (c) depends on the 
truth of a fIfth proposition, (e). Question: 
would Govier say that (d) and (e) are un-
stated premises of argument (x)? She 
would certainly say that they are unstated 
assumptions of the argument. But in her 
view an unstated assumption of an argu-
ment is not necessarily an unstated premise 
of the argument (see Problems 92). By 
what I shall call Govier's missing-premise 
criterion, an unstated assumption of an ar-
gument is an unstated premise of the argu-
ment if its truth is "a necessary condition 
for properly inferring the conclusion from 
the stated premises" (ibid.).'7 Now it is 
proper to infer the conclusion of argument 
(x) from the stated premises only if (d) or 
(e) is true, or only if (d) and (e) are true, if 
(ars being relevant to (c) depends on the 
truth of (d) and (b)'s being relevant to (c) 
depends on the truth of (e). I am assuming 
that Govier would say that this relevance 
condition is satisfied. Thus her missing-
premise criterion would require her to say 
that at least one of propositions (d) and (e) 
is an unstated premise of argument (X).IS 

But besides stating the missing-
premise criterion, Govier outlines what I 
shall call a missing-premise procedure for 
deciding whether an argument has an un-
stated premise (Problems 102). The first 
step is to classify "the stated argument as 
being of some particular type". The next 
step is to ask whether the stated argument 
is "inferentially sound as an example of 
that type". [f it is not, but would be if any 
one of "a candidate set of supplementary 
premises were added", then a member of 
that set may properly be considered an un-
stated premise of the argument (unless the 

Relevance, Conduction & Rape-Shield 117 

argument is "a fallacy or a non sequitur"). 
Suppose it is decided in accordance with 
Govier's missing-premise procedure that 
proposition (r), say, is an unstated premise 
of a particular argument of type T. Then 
adding (r) to the argument as a further 
premise turns the argument into an inferen-
tially sound argument of that same type. It 
does not turn the argument into an inferen-
tially sound argument of a different type; if 
it did, then the decision that (r) is an un~ 
stated premise of the argument would not 
accord with Govier's missing-premise 
procedure: 'this point is crucial for the 
argument of the following paragraph. 

Govier would classify argument (x), on 
my description of it, as conductive. Now 
suppose that the conjunction of premise (a) 
and proposition (d) entails conclusion (c), 
and that so does the conjunction of premise 
(b) and proposition (e). (Call this the en-
tailment assumption.) Then if either (d) or 
(e) is added to argument (x) as a further 
premise, the resulting argument will not be 
a conductive argument but an argument of 
a different type-a deductively valid 
argument. 19 Thus it would be contrary to 
Govier's missing-premise procedure to de-
cide that (d), or that (e), is an unstated 
premise of argument (x). But by Govier's 
missing-premise criterion at least one of 
propositions (d) and (e) is an unstated 
premise of the argument, if, as I am assum-
ing Govier would say, (ars being relevant 
to (c) depends on the truth of (d) and (b)'s 
being relevant to (c) depends on the truth 
of (e) - even if (c) is entailed by the con-
junction of (a) and (d) and by the conjunc-
tion of (b) and (e). Thus we have a case 
in which Govier's missing-premise criteri-
on conflicts with her missing-premise 
procedure. 

The problem that the case presents 
Govier with relies on three stated assump-
tions. If any of the assumptions is aban-
doned, the problem is dissolved. The first 
assumption is that Govier would say that 
the stated premises of argument (x) are 
separately relevant to the conclusion. 



118 Derek Allen 

There is no reason to abandon this assump-
tion, for Govier admits the possibility of an 
argument whose stated premises are sepa-
rately relevant to the conclusion (cf. PSA 
31l)-for example, a convergent argument 
that is cogent as stated. The second as-
sumption is that Govier would say that 
(ars being relevant to (c) depends on the 
truth of (d) and that (b)'s being relevant to 
(c) depends on the truth of (e), where (a) 
and (b), but not (d) and (e), are stated 
premises of argument (x). There is no rea-
son to abandon this assumption either, for, 
as I have already argued, Govier would say 
that if a stated premise of an argument is 
relevant to the conclusion (or to a conclu-
sion) of the argument, its being relevant to 
that conclusion may depend on the truth of 
a further proposition that is not a stated 
premise of the argument. The third as-
sumption is the entailment assumption, 
which, on the face of it, is perfectly reason-
able. Thus the stated assumptions of the 
problem survive scrutiny. 

Govier might respond to the problem 
by revising her position on missing 
premises. Alternatively, she might respond 
by revising her account of a conductive 
argument-for example, by adding to it 
the condition that none of the stated 
premises of a conductive argument is such 
that it is relevant to the conclusion (or to a 
conclusion) of the argument only if a fur-
ther proposition is true whose conjunction 
with the prentise entails that conclusion. 20 

Let us now return to L'Heureux-Dube's 
summary argument. Relative to Govier's 
account of a conductive argument, 
L'Heureux-Dube's summary argument, as 
reconstructed earlier, is conductive. Minus 
its stated counterconsideration, the recon-
structed argument is this: 

(11) The sexual-history evidence excluded by 
s. 276 "is either irrelevant or so preju-
dicial that its minimal probative value 
is overwhelmed by its distorting effect". 

Thus. 

(12) S. 276, in excluding the sexual-history 
evidence that it does exclude, does not 

violate the principles of fundamental 
justice (where these include the princi-
ple that an accused person has the right 
to a fair trial). 

(I 1) is relevant to (12) only if it is true that 

(13) The exclusion [from a trial] of evi-
dence that is either irrelevant or so 
prejudicial that its minimal probative 
value is overwhelmed by its distorting 
effect does not violate the principles 
of fundamental justice. 

The conjunction of (11) and (13) entails 
(12). Thus, if Govier's account of a con-
ductive argument is revised by adding to it 
the condition stated above, then, relative to 
the revised account, L'Heureux-Dube's 
summary argument, as reconstructed earli-
er, will no longer be conductive. 

Finally, let us revisit the argument of 
McLachlin's reconstructed in note (8). 
Relative to Govier's account of a conduc-
tive argument, that argument is conductive. 
Minus its stated counterconsiderations, 
and minus the premise of its subargument, 
the reconstructed argument is this: 

(14) S. 276 categorically excludes evidence 
without permitting the trial judge to 
engage in the exercise of determining 
whether the possible prejudicial effect 
of the evidence outweighs its value to 
the truth-finding process. 

(15) S. 276 impairs the ability of the trier of 
fact to determine the truth of a report 
of sexual otfence. 

(\6) S. 276 goes further than required to 
protect privacy. 

Thus, 

(17) The wide reach of s. 276 is unjustified. 

Premises (14)-(16) each specify (what 
McLachlin considers) a negative conse-
quence of the wide reach of s. 276. Each of 
those premises is relevant to the conclu-
sion, (17), granted the truth of a proposi-
tion to the effect that 

(18) A criminal statute is of unjustified 
wide reach if and only if the conse-
quences of its wide reach are negative 
(harmful), at least on balance. 



The conjunction of (18) and any, or all, of 
premises (14)-( 16) does not entail (17), 
Hence none of those premises is relevant to 
(17) only if a further proposition is true 
whose conjunction with it entails (17), Thus, 

Relevance, Conduction & Rape-Shield 119 

if Govier's account of a conductive argument 
is revised by adding to it the condition stated 
above, then, relative to the revised account, 
the argument of McLachlin's reconstruct-
ed in note (8) will still be conductive. 21 

Notes 

I have benefited from comments made on an 
earlier draft of this paper by an Informal Logic 
referee. 

In saying that the jury is permitted to draw, 
and does draw, these inferences, I am adopting 
language quoted by L'Heureux-DuM, at 346e, 
and used by McLachlin, at 386c. (I adopted 
similar language in the second paragraph of 
the 'Background' section of Part I.) Compare 
sub-section I of s. 276 of the Canadian Crimi-
nal Code as amended by Parliament in 1992; 
the sub-section responds to the common-law 
problem that led to the enactment of the ver-
sion of s. 276 struck down in the rape-shield 
decision, and reads in part as follows: 

In proceedings in respect of an offence 
[pertaining to sexual assault] ... , evidence 
that the complainant has engaged in sexual 
activity, whether with the accused or with 
any other person, is not admissible to sup-
port an inference that, by reason of the sex-
ual nature of that activity, the complainant 
(a) is more likely to have consented to the 

sexual activity that forms the subject-
matter of the charge; or 

(b) is less worthy of belief. 

3 Suppose, for example, that one of the justifica-
tions is that the father's explanation of why the 
charge was brought is plausible if the charge is 
false. I might find the father's explanation 
plausible because I subscribe to the sexist ster-
eotype that "the feminine character is especial-
ly filled with malice". But the explanation (and 
hence the justification) does not presuppose 
this or any other sexist stereotype, for no such 
stereotype need be true if it is true that the 
explanation is plausible if the charge is false. 

4 It would also be worth considering whether the 
legal reasoning in my Smith/Jones example 
employs the right theory of relevance. I am in-
debted for this point to an Informal Logic refe-
ree, who wrote: "The ... example employs 
Keynes' theory of relevance, which says (very 

roughly) that x is relevant to y if it changes the 
probability of y .... according to that theory; 
the fact that the seven of diamonds has been 
withdrawn from a pack of playing cards and 
not replaced is positively relevant to the propo<--
shion that the next card drawn will be the ace 
of clubs, since the probability of that event is 
now higher than it would have been otherwise. 
Since in a trial it is desirable to find out wheth-
er it is (to a high degree) more likely than not 
that the defendant did what he was accused of 
doing-not merely whether it is more likely 
than it would have been otherwise-, perhaps 
the wrong theory of relevance was employed. 
Perhaps a theory should have been employed 
that admits something as positively relevant to 
a proposition only if it makes that proposition 
more likely to be true than false." Such a theo-
ry is, for example, that of George Bowles, 
cited below. 

5 Convergent support, as Govier understands it, 
is "[a] kind of support where premises work 
together in a cumulative way to support the 
conclusion, but are not linked. The bearing of 
one premise on the conclusion would be unaf-
fected if (he other premises were removed; 
however, the argument is strengthened when 
the premises are considered together, since 
more evidence is then offered" (PSA 59). 

6 Govier explains in Problems (Ch. 4) that the 
category of conductive argument was first de-
tined by Carl Wellman, Challenge and Re-
sponse: Justification in Ethics (Carbondale: 
Southern Illinois University Press, 1971). See 
also David Hitchcock, Critical Thinking: A 
Guide to Evaluating Information (Toronto: 
Methuen, 1983) 105-6, 130-37. 

7 I assume here that 'cons' are counterconsider-
ations understood a la Govier-hence that a 
'con' is not a premise but "a kind of anti-
premise" (PSA 310). 

S Obviously the suggested exercises would 
have to be accompanied by background 



120 Derek Allen 

infonnation. The 'objective v. effect' exercise 
would require background information that r 
have not yet supplied. L'Heureux-DubIS and 
McLachlin are engaged in what is known as a 
'section I analysis' when they consider the 
'objective v. effect' issue. The purpose of a s. I 
analysis is to detennine whether legislation 
that violates a Charter right can be saved un-
der s. I of the Charter. rt can, according to the 
test used by the Canadian Supreme Court, if 
and only if certain conditions are satisfied, one 
of which is that the importance of the objective 
of the legislation outweigh the legislation's in-
jurious effect, and by a substantial margin if 
the injury is serious. For L'Heureux-DubIS, as. 
I analysis is optional, since she tinds that s. 
276 does not violate the Charter rights in 
question. She argues that even if it did, it could 
be saved under s. I. 

Some references for the three exercises, all 
to 66 Canadian Criminal Cases (3d). 
L'Heureux-DubIS's reasoning on the 'objective 
v. effect' issue is at 379d-g; McLachlin's is at 
403d-g. L'Heureux-DubC's summary runs 
from 372b, sentence two, to 372e. McLachlin's 
"wide reach" reasoning is at 395g-396g, but 
the preceding paragraph (395d-t) should be 
included to provide context. 

Here is a conductive reconstruction of 
McLachlin's "wide reach" reasoning. Counter-
considerations are identified by the letter 'C 
followed by a numeral. 
Cl. S. 276 prevents judge and jury from being 

diverted by irrelevant evidence of other 
sexual conduct of the complainant which 
will unfairly prejudice them against the 
complainant. 

But, 
2. S. 276 categorically excludes evidence 

withol't pennitting the trial judge to en-
gage in the exercise of determining 
whether the possible prejudicial effect of 
the evidence outweighs its value to the 
truth-finding process. 

C3. S. 276 encourages the reporting of sexual 
offences. 

But, 
4. S. 276 impairs the ability of the trier of fact 

to determine the truth of a report of a sex-
ua� offence. 

CS. S. 276 protects the complainant's privacy. 
But, 
6. S. 276 fails to pennit an assessment of the 

effect on the complainant of the evidence 
in relation to the cogency of the evidence. 

So, 
7. S. 276 goes further than required to protect 

privacy. 
So, 
8. The wide reach of s. 276 is unjustified. 

Note: It is important to distinguish between a 
premise offered in support of the conclusion of 
a conductive argument and a claim to the ef-
fect that a counterconsideration mentioned in 
the argument does not show that the conclu-
sion is false. In the passage containing the rea-
soning I have reconstructed, McLachlin 
endorses two claims of the latter kind, namely: 
(a) to accept that persuasive evidence for the 
defence can be categorically excluded on the 
ground that it may encourage reporting and 
convictions is to say either that we assume the 
defendant's gUilt or that the defendant must be 
hampered in his defence so that genuine rap-
ists can be put down, neither of which views 
conforms to our notions of fundamental jus-
tice; (b) the constitutional right to a fair trial 
must take preeedence over easing the plight of 
the witness in case of conflict. McLachlin re-
gards (a) as a reason for denying that C3 
shows that the wide reach of s. 276 is justified, 
and (b) as a reason for denying that C5 does 
so. Neither claim is a premise offered by her in 
support of (8). 

9 James Freeman, Thinking Logically: Basic 
Concepts for Reasoning, Prentice Hall, Engle-
wood Cliffs, New Jersey, 1993,98. 

10 Compare Freeman: "a premise is a warrant or 
functions as a warrant just when it serves to 
explain why some other premise is relevant to 
the conclusion it is claimed to support" (op. 
cit., 97). 

II I here use 'explains' in a 'success' sense, such 
that if (p) explains why (q), then (p) is the (or a) 
correct explanation of why (q). In this sense of 
'explains', '(p) explains why (q)' entails '(q)'. 

12 The relevance gap that exists in argument (B) 
relative to Bowles's theory of propositional 
relevance (still assuming that, by Bowles's cri-
terion, (p) is not (positively) relevant to (q)) is, 
relative to that same theory, what Govier, fol-
lowing Robert Ennis, calls an inference gap 
(see Problems, pp. 94-96; Robert H. Ennis, 
"Identifying Implicit Assumptions", Syllthese, 
51: pp. 61-86). For, ex hypothesi, relative to 
that theory the inference from (p) to (q) is un-
sound, precisely because, ex hypothesi, rela-
tive to that theory (p) is not (positively) 



relevant to (q). Govier remarks that "[slome 
people see (inference] gaps where others 
don't, due to disagreements about the theory 
of argument" (Problems, pp. 95-96). Similarly, 
some people may see a relevance gap where 
others don't, because they subscribe to differ-
ent theories of propositional relevance. 

13 Compare the following remark: 'The logical 
role of premise (4) in argument (A) is to afford 
evidence for the conclusion, (5). But (4) is ir-
relevant to (5), hence does not afford evidence 
for (5), hence does not succeed in fulfilling its 
logical role in the argument: 

14 That is, by his criteria for when one proposi-
tion makes another probable or certain. 
Bowles says: '''p' makes 'q' certain if and only 
if the probability of 'q' conditional on 'p' is 
I", and "'p' makes 'q' probable if and only if 
the probability of 'q' conditional on 'p' is less 
than I but greater than 112" (PR 67). If so, then 
'p' need not be true to make 'q' certain or 
probable. I disagree, in the belief that a false 
proposition cannot make another proposition 
probably true or certainly true. If so, then 'p' 
may not make 'q' certain even if the probabili-
ty of 'q' conditional on 'p' is I, and 'p' may 
not make 'q' probable even if the probability 
of 'q' conditional on 'p' is less than I but 
greater than Y2. The phrases "'p' makes 'q' cer-
tain" and "'p' makes 'q' probable", in 
Bowles's above-quoted formulas. should be 
replaced, I believe, by the phrases '''p', if true, 
makes 'q' certain'" and '''p', if true, makes 'q' 
probable'''. respectively. 

15 Bowles might reply that I recognize (p) as a 
reason for (q) only because I recognize the 
truth of a third proposition, (r). and that this 
means that what is actually a reason for (q) is 
not (p) alone-Le .. (p)-but the conjunction of 
(p) and (r). Perhaps I do recognize (p) as a rea-
son for (q) only because I recognize the truth 
of (r). But, I claim, (p) is a reason for (q) even 
if I don't recognize the truth of (r) and so fail 
to recognize that it is. 

16 In the example given, Govier does not say 
(though she may believe) that the relevance of 
the argument's premises to the conclusion de-
pends on the truth of the stated assumption; 
what she implies is that its truth is sufficient for 
their relevance to the conclusion-non-trivial-
Iy and non-criterially sufficient, she would say. 

17 At any rate, remarks Govier makes on pp. 
92-93 of Problems suggest that this is her 

Relevance, Conduction & Rape-Shield 121 

mlssmg-premise criterion. But remarks she 
later makes (ibid., 93-94, 96-99) suggest that 
her missing-premise criterion might better be 
worded as follows: an unstated assumption of 
an argument that is not a principle or rule of 
inference, or a meta-linguistic assumption, on 
which the argument depends is an unstated 
premise of the argument if its truth is a neces-
sary condition for properly inferring the con-
clusion from the stated premises. Govier 
would say that, by this criterion. an argument's 
associated conditional is not a missing (unstat-
ed) premise of the argument: for in her view an 
argument's associated conditional is a "princi-
ple" of inference (tacitly) assumed by the ar-
gument (ibid., 96). 

18 Here (and below) I assume that neither (d) nor 
(e) is what Govier would consider a principle 
or rule of inference. or a meta-linguistic as-
sumption, on which argument (x) depends. 

19 A conductive argument is cogent, Govier be-
lieves, only if its premises "considered togeth-
er, in light of counterconsiderations 
provide adequate support for the conclusion" 
(PSA 314). The premises of a conductive argu-
ment, though "put forward as separately being 
positively relevant to the conclusion" (PSA 
355), are offered jointly in its support (see also 
PSA 48). Thus their conjunction is offered in 
its support. If a further premise is added that 
entails the conclusion, then the conjunction of 
the argument's premises will entail the conclu-
sion. and so the argument will be deductively 
valid, hence no longer conductive. Here is evi-
dence that for Govier a conductive argument is 
never a deductively valid argument: "one 
might ... have an argument with several dis-
tinct premises, each of which quite separately 
deductively entailed the conclusion. In such a 
case, the argument would be deductively valid 
and hence [my emphasis] would not be a con-
ductive argument, though it would exemplify 
convergent support" (PSA 357 n. 2; see also 
PSA 308-9.) In Problems, Govier writes: "We 
cannot define conductive arguments solely 
with reference to the convergent support pat-
tern, because if we were to do so, some deduc-
tive arguments would be conductive. . . . 
In a conductive argument no one premise 
deductively entails the conclusion. nor do 
the premises entail the conclusion when 
considered together" (70), 

20 Call this condition C I. Here is an alternative, 
C2: it is not the case that the conclusion of a 



122 Derek Allen 

conductive argument may be properly inferred 
from the stated premises only if a further prop-
osition is true that in conjunction with the stat-
ed premises entails the conclusion. Cl and C2 
are not extensionally equivalent. Thus let (y) 
be a convergent argument with stated premises 
(I), (m), (n), and conclusion (0). Premises (I), 
(m), and (n) are put forward as separately be-
ing positively relevant to (0). None of premis-
es (I), (m), and (n) entails (0), nor do they 
jointly entail (0). But assume that (n) is rele-
vant to (0) only if a further proposition, (w), is 
true, and that the conjunction of (n) and (w) 
entails (0). Then. by CI, (y) is not a conductive 
argument. But assume further that (I) and (m) 
are strong reasons for (0) and that nei ther (I) 
nor (m) is outweighed by any counterconsider-
ations to (0). Then (0) may be properly in-
ferred from the conjunction of (I) and (m) (or 
so Govier would say). Finally, assume that 
neither (I) nor (m) is relevant to (0) only if a 
further proposition is true whose conjunction 
with it entails (0). Then argument (y) is 
conductive even relative to the account of a 
conductive argument that results from adding 
C2 to Govier's account. Thus the addition of 
C2 to that account does not prevent argument 
(y) from being conductive (relative to the 

revised account), but the addition of Cl 
does. Thus C I and C2 are not extensionally 
equivalent. 

21 In ju?ging that (II) is relevant to (12) only if 
(13) IS true, and that each of (14)-( 16) is rele-
vant to (17) if (18) is true, I rely on intuition. I 
do not have in mind a particular theorv of 
propositional relevance-any more than Bow-
les has when, in reply to Govier's 'aspirin' ob-
jection, he says that (p)'s being a reason for, 
and relevant to, (q) depends on the truth of a 
third proposition, (r), on the ground that "if 'r' 
were false ... 'p' would not be a reason for, or 
relevant to, 'q'" (PR 72-73). Bowles cannot be 
relying upon his own theory of propositional 
relevance when he says this, for if he were he 
would be begging the question (of the adequa-
cy of his theory) against Govier; rather, he 
must take it to be intuitively clear that if (r) 
were false, (p) would not be a reason for, or 
relevant to, (q). 

DEREK ALLEN 
DEPARTMENT OF PHILOSOPHY 
UNIVERSITY OF TORONTO 
TORONTO, ONTARIO M5S IA] o