The missing link revisited: The role of teleology in representing
legal argument
T.
J. M. BENCH-CAPON
Department
of Computer Science, The University of Liverpool, Liverpool, UK
E-mail:
tbc@csc.liv.ac.uk
Abstract. In this paper I recapitulate
the ideas of Berman and Hafner (1993) regarding the role of
teleology in legal argument. I
show how these ideas can be used to address some issues arising from
more recent work on legal
argument, and how this relates to ideas associated with the New Rhetoric
of Perelman. I illustrate the
points with a discussion of the classic problem of which vehicles should
be allowed in parks.
1. Introduction
The first morning of ICAIL
1993 in Amsterdam concerned argument. The first two
papers presented, Henry
Prakkens logical framework for modelling legal argument
(Prakken 1993) and Tom
Gordons paper on using dialogue games to model legal
argument (Gordon 1993), have
been enormously influential and there has followed
a great volume of work
articulating, developing and combining these approaches.
Just before lunch on that
morning a no less interesting paper with an entirely
different approach was
presented, Don Berman and Carole Hafners Representing
Teleological Structure in
Case Based Legal Reasoning: The Missing Link (Berman
and Hafner 1993).
Unfortunately this has not produced anything like the same
degree of subsequent research.
In this paper I want to recapitulate the ideas in that
paper and to re-examine them
in the light of what we have learnt about modelling
legal argument since 1993. In
doing so I hope to establish that the time is ripe to
pay some very serious
attention to their ideas.
_ This paper was originally written in December
1999. Since then it has received responses from
Henry Prakken and Giovanni
Sartor (in this volume), and has been developed in on-going work, both
by Henry Prakken and by myself
in collaboration with Giovanni Sartor. The most recent expression
of my views can be found in
Bench-Capon and Sartor (2001). In preparing this paper for publication
I have attempted to remain
faithful to my original views, and not to anticipate too much later
developments, although I have
taken the opportunity to correct some mistakes in the original version.
80 T. J. M. BENCH-CAPON
2. Hunting, shooting and
fishing
The paper presented in
Amsterdam (Berman and Hafner 1993) was based on three
cases, commonly used in
teaching American students law. In the first, Pierson v
Post, the plaintiff was hunting a
fox, on open land, in the traditional manner using
horse and hound when the
defendant killed and carried off the fox. The plaintiff
was held to have no right to the
fox because he had gained no possession of it. In
the second case, Keeble v Hickeringill, the plaintiff owned a pond
and made his
living by luring wild ducks to
the pond with decoys, shooting them and selling
them. Out of malice the
defendant used guns to scare the ducks away from the
pond. Here the plaintiff won.
In a third case, Young
v Hitchens,
both parties were
commercial fisherman. The
plaintiff spread a net, some half a mile in length, and
began to close it. When the
opening was no more than a few dozen feet wide, the
defendant sped into the gap,
spread his own net and caught the fish which had been
trapped by the plaintiff as he
closed his net. In this case the defendant won.
Berman and Hafner then
constructed an argument for the defendant in Young,
against the background of Pierson and Keeble. They say that they are
following
the approach used in HYPO
(Ashley 1990). They do indeed use the three-ply style
of argument developed in HYPO,
in which a party to the case cites a precedent,
the opponent responds and the
original party attempts to rebut the response. Their
representation of the cases,
however, differs somewhat from the original conception
of HYPO (e.g., Ashley and
Rissland 1988). Berman and Hafner use factors, which
are features of the case which
may be present or absent and which, if present, favour
either the plaintiff
(pro-plaintiff factor) or the defendant (pro-defendant factor).
HYPO in contrast used dimensions, which are features of the
case which can take
a range of possible values, ordered
according as to the extent to which they favour
a particular side. Consider
the question of whether the hunter had possession of
the animal. Seen as a factor,
we simply ask whether the animal was caught or
not. Seen as a dimension, we
can have a range of possibilities progressively more
favourable to the defendant,
running from caught
to no contact at all, and passing
through some intermediate
positions such as mortal
wounding, wounding, hot pursuit,
started, and seen. Using dimensions permits a
richer representation of the case
situation, and allows us to
avoid some awkwardness in choosing factors, such as
whether the factor should be caught (pro-plaintiff), or not caught (pro-defendant),
or perhaps both, as well as
not requiring us to make all or nothing decisions.
Using factors, on the other
hand, does make a useful simplification. In fact the use
of factors has in recent years
become rather more common than dimensions: the
description of cases in CATO
(Aleven 1997) and the reconstruction of case based
reasoning of Prakken and
Sartor (1998) both use factors rather than dimensions.
In the remainder of this paper
I shall use factors in this way, although I think
dimensions remain important
and interesting, and I hope to explore the difference
that using them makes in some
future work.
THE MISSING LINK REVISITED 81
Berman and Hafner can be seen
as identifying five factors. Two are proplaintiff:
that the plaintiff was
following his livelihood (LIVELIHOOD) and that
the land was owned by the
plaintiff (OWNLAND). Three are pro-defendant: that
the animal was not caught
(NOTCAUGHT), that the land was open (OPEN)
and that the defendant was in
competition with the plaintiff (COMPETE). Pierson
contains NOTCAUGHT and OPEN, Keeble OWNLAND, LIVELIHOOD and
NOTCAUGHT and Young NOTCAUGHT, OPEN and COMPETE.
Actually Berman
and Hafner speak of four
factors, one of which, the status of the land, takes two
values. This makes it look
rather like a dimension (although if it were a dimension
we might expect more than
these two values, and include, for example, the possibility
of the defendant owning the
land). I think it is more consistent to see five
factors, but to bear in mind
that OWNLAND and OPEN are mutually exclusive.
With these five factors, Pierson looks rather clear: the
plaintiff does not own
the land, did not catch the
beast and was looking for pleasure rather than business,
so only pro-defendant factors
are present. Keeble, although the plaintiff was
not in
possession of the ducks, had
the pro-plaintiff factors that the land was owned and
he was engaged in a commercial
pursuit, and we know that these were sufficient to
overcome the pro-defendant
factor NOTCAUGHT.
If we wish to make an argument
for the defendant in Young, we can propose
that Pierson is followed. The plaintiff can
reply by distinguishing on the grounds
that the plaintiff is making
his living, and cite Keeble
to show that
not having
captured the prey is not fatal
to the plaintiffs case. The defendant can now give a
rebuttal, since he can
distinguish Keeble, on the grounds that in Young the water
is not owned by the plaintiff,
and further add that the motive of the defendant was
not malice but business
competition. So the issues are identified, but the question
remains as to whether are
persuaded (or are able to persuade a judge) that the extra
pro-defendant factor and the
missing pro-plaintiff factor in Young are sufficient to
cause us to reject Keeble.
And this is where the
reasoning in this model runs out. Although we have
identified some candidate
cases, the question remains as to which should govern
Young, and why it should do so. To
find clues as to whether we can give reasons for
following Pierson rather than Keeble we need to examine the texts
of the decisions.
Importantly these texts refer
to the purposes
that the
judges saw as being promoted
by their decisions. Pierson was found in favour of the
defendant
For the sake of certainty,
and preserving peace and order in society. If first
seeing, starting or pursuing
such animals . . . should afford the basis of actions
. . . itwould provide a
fertile source of quarrels and actions. (Quoted in Berman
and Hafner 1993, italics
mine).
One judge dissented: for him
the pursuit and destruction of foxes was of suf-
ficient social value to be
encouraged and protected by law. Two points should be
noted: first that the reason
why capturing the animal matters is because only that
is considered sufficiently
clear evidence of a right to the animal, and second that
what seemed to be a clear case
containing only pro-defendant factors was in fact
82 T. J. M. BENCH-CAPON
disputed. Perhaps we should
have included in our analysis an extra, pro-plaintiff,
factor, relating to the social
value of the plaintiffs activity, to reflect that the case
was in fact decided by weighing competing social
values.1 In the second case,
Keeble, the social utility is
greater, and the evidence for this is that the plaintiff
could earn a living from
hunting ducks. Although the right to make ones living
undisturbed is alluded to, the
crucial point is that:
When decoys have been used . .
. in order to
take a profit for
the owner of the
pond . . . and whereby the
markets of the nation may be furnished; there is
great reason to give
encouragement thereunto (Quoted in Berman and Hafner
1993, italics mine).
In other words here (unlike in
Pierson) we can be sure that the
plaintiffs
activity is valuable because
we know he can make a living from it, and so people
are prepared to pay for his
activity. We can assuage our doubts with respect to
certainty because we have a
clear criterion for saying that the activity is one to be
encouraged. When we come to Young, the social benefit is
neutral the same fish
furnish the markets of the
nation whether they are delivered by Young or Hitchens.
Having removed this reason,
the appeal to the need for certainty can prevail, just
as it did in Pierson. Arguably, in addition in
this case the decision can also be
seen as encouraging vigorous
competition, which may have even greater economic
benefits.
Seen from this teleological
perspective, and informed by the reasons for the
rules as well as the rules
themselves we can see two things:
That we can come to a rational2 decision as to the case to
follow;
That apparent similarities and differences (the
open land in Pierson
and the
owned land in Keeble, and the engagement in making
a living in both Keeble
and Young) may be more or less useful.
Distinguishing on the ownership of the
land weakens the effect of Keeble for the plaintiff in Young, but does little to
positively promote a decision
for the defendant, unless we wish to argue that
potential presence of ducks on
a pond confers possession of the ducks on the
pond owner.
Berman and Hafner then proceed
to an analysis of an example of reasoning
performed by Brantings GREBE
system (Branting 1991). Here they clearly show
how relying on rules
manifested in cases without reference to the purposes of these
rules leads the reasoner
astray, through pursuing false similarities and differences
between the cases.
The last section of the paper
gives some suggestions for augmenting a representation
of cases in terms of factors
with the legal purposes which explain why each
of the factors favours the
plaintiff or the defendant. Now we can judge competing
arguments not only on the
importance of the factors themselves, but on the value
we accord to the purposes
from which they arise.
This both makes the arguments
more realistic, and the choice
between competing arguments less arbitrary.
THE MISSING LINK REVISITED 83
3. Theory construction and
theory coherence
I have recapitulated Berman
and Hafner (1993) at some length, both because it is
an excellent paper, and
because it provides a strong argument that working from
decisions without their
reasons will often leave us with no reason to prefer one
precedent to another, and can
even lead us into error. In this section I shall consider
one of the leading approaches
to modelling legal argument that has been developed
since 1993, and suggest that
it also is unable to resolve disputes because it ignores
the purpose of the rules it
employs.
In a series of papers, of
which we can take Prakken and Sartor (1998) as representative,
Henry Prakken and Giovanni
Sartor have articulated a model of legal
reasoning based on a logical
formalisation, and the notion of a dialogue game.
Their main aim is to allow
conflicting norms to be reconciled, and the essential
idea is that to win a case one
must put forward an argument which can be defended
against any attack. Attacks
may either be undercutting, attacking a premise in
the original argument, or
defeating, establishing the negation of the conclusion of
the original argument. An
attack can be defended against either by attacking the
attacking argument, or, in the
case of a defeater, claiming a higher priority for the
rule grounding the original
argument than that grounding the putative defeater. As
the debate proceeds, arguments
introduce new rules, and their coherence with other
rules is tested. What is
happening here is that we are constructing a theory (in the
logical sense), which is
considered coherent if its sentences are able to withstand
attack from within the theory.
This is excellent as an ex post reconstruction of the
reasoning, because here we can
use the decision itself to reveal priorities. It, is
however, less useful ex ante, since it is unclear where
these priorities come from.
In Prakken and Sartor (1998)
they make use of their account to reconstruct a
style of factor based reasoning,
such as we have seen above. Here each case gives
rise to three rules:
A rule of the form if conjunction of all
pro-plaintiff factors present then
plaintiff
A rule of the form if conjunction of all
pro-defendant factors present then
defendant
A rule expressing that one of these rules has a
higher priority that the other,
depending on the way the case
was decided.
In Bench-Capon (1999) I showed
how we could express these rules diagrammatically
as a partial order of factor
combinations. The idea is that we represent
all combinations of
pro-plaintiff factors and all combinations of pro-defendant
arguments. Together these
represent all possible pro-plaintiff and pro- defendant
arguments. The ordering will
express priorities between the arguments. We now
order these arguments on the
assumption that an argument with more factors is
stronger than an argument with
fewer factors. When we have a case we are able
to order the pro-plaintiff and
the pro-defendant arguments in that case, according
to the preference expressed in
the decision. These connections allow us to say
84 T. J. M. BENCH-CAPON
something about the relative
order on pro-plaintiff and pro-defendant arguments.
The theory is coherent if there are no cycles in the
resulting directed graph. Additionally,
we can say that the theory is complete if for every pro-plaintiff
argument
and every pro-defendant
argument we can see which is preferred.
Let us look at Pierson, Keeble and Young in this way, based on factors
adapted
from Berman and Hafner (1993).3
The pro-plaintiff factors I
shall use are:
(A) plaintiff was pursuing
his livelihood
(B) the plaintiff was on his
own land
The pro-defendant factors I
shall use are:
(C) plaintiff was not in
possession of the animal
(D) defendant was pursuing
his livelihood
(E) the land was open
We now use Prakken and
Sartors method decribed above to identify the rules used
in each of the three cases.
Pierson yields the rules:
(R1) plaintiff (no factors
in Pierson favour the plaintiff)
(R2) if C and E then
defendant
(R3) R2 > R1
Keeble yields the rules
(R4) if A and B then
plaintiff
(R5) if C then defendant
(R6) R4 > R5
In Young we have:
(R7) if A then plaintiff
(R8) if C and D and E then
defendant
And the point is to decide
which of R7 and R8 has priority.
The diagram of the resulting
partial order including the preferences R3 and R6 is
shown in Figure 1.
As we can see the priorities
gleaned from the previous cases are not helpful, or
at any rate not decisive, in
the case of Young
since we
cannot deduce the ordering on
the pro-plaintiff and
pro-defendant arguments in that case. The plaintiff must argue,
on the basis of Pierson and Keeble, that his pursuit of his
livelihood is sufficient
to overcome the additional
weight that the defendants case gets from the unowned
land and the fact that the
defendant was also pursuing his livelihood. As we know,
the plaintiffs case did not
convince, but the partial order does not allow us to
predict this. The problem is
that the theory remains coherent whether we add R7 >
R8 or R8 > R7. No cycle appears whichever
way we direct the Young
arc in Figure
1.
THE MISSING LINK REVISITED 85
Figure 1. Partial Order for Pierson, Keeble and Young.
But could we not broaden our
notion of coherence here to allow us to give a
rational underpinning to the
preference of one or other of these priorities? The
idea is to restore the missing
link of teleology by requiring theories not only to be
coherent logically, but also
to be coherent from the perspective of some ordering
of the desired values promoted
by the decisions. This idea also has jurisprudential
foundations in the New Rhetoric (Perelman and Olbrechts-Tyteca
1969; Perelman
1980).
4. Coherence and teleology
Let us start with a quotation
from Perelman (1980).
Since Descartes, however, only
mathematical demonstration, the compelling
proof, imposed by its evidence
on every reasonable human being, has been
judged worthy of attention. .
. . Arguments which justify our options, choices,
and decisions, are never as
compelling as demonstrative proofs; they are more
or less strong, relevant, or
convincing. . . . in argumentation it is always possible
to plead for or against,
because arguments which support one thesis do
not entirely exclude the
opposite one; this in no way means that all arguments
are of the same value. (p 150)
What Perelman is arguing here
is that while logic is important violation of
logical coherence requires a
position to be abandoned because it is simply untenable
it is not the only ground
for rational preference. Thus we can prefer an
argument for good reasons even when those reasons are
not coercive. Applied to
86 T. J. M. BENCH-CAPON
our situation we may say that
logical considerations should not be our only criterion
of coherence; we also need to
consider coherence from the perspective of purpose
and values.4
Let us now consider the animal
cases in this light,
We have 5 factors:
(A) plaintiff was pursuing
his livelihood
(B) the plaintiff was on his
own land
(C) plaintiff was not in
possession of the animal
(D) defendant was pursuing
his livelihood in competition with the plaintiff
(E) the land was open
We also find from the
decisions three values:
V1: certainty the need to
have clear law to discourage speculative litigation;
V2: promotion of a larger
Gross National Product to improve economic wellbeing;
and
V3: the sanctity of property
to allow people full enjoyment of what is rightly
theirs.
The last is implicit, but
underlies all the reasoning. In all three cases, only
because the plaintiff was not
in possession of the animal does the defendant have a
case at all. Now we can say
that values relate to factors as follows:
V1: relates to C
V2: relates to A and D
V3: relates to C and putatively relates to B
and E.
The reason why I say at this
stage that V3 only putatively relates to B and E. is
that wild animals of their
nature move around and do not respect ownership rights.
Thus the presence of a wild
animal on ones own land is perhaps not sufficient
to confer ownership, although
it probably does confer rights of (possibly not exclusive)
pursuit. Remember that in Keeble the defendant did not
trespass, since the
duck scaring was done from his
own land. I am not sure of the law here: Section
4(4) of the United Kingdom
Theft Act 1968 states that Wild creatures, tamed or
untamed, shall be regarded as
property; but a person cannot steal a wild creature,
but this does not entirely
clarify the matter for me. I do not think the issue is settled
from the three cases: I
address it later in the paper through the use of a hypothetical
case.
Now Pierson contains factors C and E. From
this we can see that in Pierson we
can only be concerned with the
values V1 and V3. Both these values are served
by finding for the defendant.
In Keeble, however, the factor A is
present and so we
need to consider another value,
V2. From the text of the decision, this appears to
be sufficient to win, and the
presence of B and absence of E is not relied on. The
relation of B and E to V3 is
thus not clarified by this case. These two decisions
suggest a possible ordering of
values: V3, V2, V1. Possession of the animal would
THE MISSING LINK REVISITED 87
Figure 2. Factor preferences based on
values.
be decisive because it
promotes V3, and does no harm to V1. Keeble shows V2 to
be more important that V1,
since although the facts considered alone might suggest
that B was the crucial factor,
with the decision relying on the preference of V3 to
V1, the text of the decision
makes very clear that V2 is the deciding issue. Other
orderings are possible, but
this is consistent with the decisions.
When we now turn to Young, we see that facts are
neutral with respect to both
V2, since D now applies, and
V3 since B is absent and C is present, and so we can
revert to a decision on the
grounds of V1, and follow Pierson.
The decision for the
defendant in Young, can therefore be explained.
If we take these values, and
the preference revealed in Keeble,
into account we
can get ordering pictured in
Figure 2.
Four lines have been inserted
to represent these value judgements:
We first put a line from []p to D&Ed,
to indicate that in the absence of
C, the defendant has no case:
where the plaintiff is in possession, no problems
of certainty arise, GNP is
unaffected, and so V3, the sanctity of property, is
unchallenged. This is assumed
to be implicit in the ruling of Pierson, which
states that the defendant had
not gained possession of the fox because he had
neither captured not mortally
wounded it.
Next we put a line from Ap to C&Ed.
If the plaintiffs livelihood is in
question and the defendants
is not, V2, GNP, holds sway. Thus A has priority
over anything not containing D
Third we put a line from C d to B p.
If GNP is unaffected, we say that
V1, certainty, rules, so that
C wins any case without A,
88 T. J. M. BENCH-CAPON
The fourth line, from C&D d to A&B p, is to indicate C also wins any
case with both A and D, since
where the case is neutral with respect to V2, V1
should govern.
This is one suggested
interpretation, one which downplays the relevance of B
and E to V3. This theory is
complete and enables any case (involving these factors)
to be decided. Moreover, we
can see that Pierson
is
effectively decided by (Cd)
> ([]p): E is unnecessary for the defendants case. Keeble is effectively decided
by Ap > Cd: neither the presence of B
nor the absence of E is necessary to
the plaintiffs case; and we
could predict that Young
will be
decided by C&Dd
> Ap: again we may say that E is irrelevant.
Now consider a hypothetical
case. Suppose, for example, that a person owned
a duck pond and a professional
hunter killed ducks on it and sent them to market.
Suppose the pond owner sues.
The factors in this case are thus B,C and D. The
above theory would suggest
that the pond owner should lose. This is because in
the above theory I am assuming
that B and E do not concern V3 with sufficient
certainty to override the
promotion of V2. Were, however, the case to be decided
in favour of the pond owner,
since B is the only pro-plaintiff factor, the doubts as
to whether or not B promoted
our highest value, V3, would be removed. Now that
we know that B promotes V3, we
can say that, since V3 is our highest value, B is
to be preferred to C and D.
Adding this preference gives the theory shown in figure
3, the modification being that
B can now be given the same effect as the absence
of C, that owning the land is
as good as being in possession of the animal. Note
that nodes containing B are
incomparable to nodes containing E, since the factors
cannot co-occur. Figure 3 thus
also represents a coherent (and complete) theory.
Both Figures 2 and 3 offer
complete and coherent theories: the difference between
them is whether we want to say
that B promotes V3 or not. This requires a case like
our hypothetical to supply an
answer: a decision for PondOwner in the hypothetical
case would render the theory
of Figure 2 incoherent.
The key point here is that the
cases in the example do not give us a decision as
to whether ownership of the
land confers possession of the animal, since that issue
favoured the winner in each of
our three cases anyway. The cases are thus neutral
between the theories in
Figures 2 and 3, although both suggest a decision for the
defendant in Young. However, if we have a case
like Pond-Owner, it is precisely this
point that we need to argue.
Note that we do not argue by starting from the rule of
Keeble and arguing that that neither
the fact that Pond-Owner was not pursuing his
livelihood and Duck-Hunter was
is sufficient to distinguish the case, but we must
argue that ownership of the
land confers possession of the animal and so promotes
V3. If this is accepted
victory for Pond-Owner will flow from our hierarchy of
values.
I shall now apply this
thinking to the classic example of open texture: the
problem of vehicles in the
park first raised by Hart.
THE MISSING LINK REVISITED 89
Figure 3. Theory on the assumption that
ownership of the land confers possession of the
animal.
5. Vehicles in the park
Before turning to vehicles in
the park I shall refer to a case told of by Recasens-
Siche and quoted in Perelman
(1980), which apparently took place in Poland at
the beginning of the century.
In a railway station a sign prohibited the use of the
stairway to people with dogs.
A peasant with a bear on a leash was annoyed to be
refused entry, since no one
could mistake his bear for a dog. The station master
was, however, adamant in his
refusal.
This is an example of an a fortiori argument: whatever reasons we
have for
excluding dogs we have also
for excluding bears. Some might regard this as an
argument from analogy: a bear
is sufficiently like a dog to be excluded. But a
bear is also unlike a dog it
is bigger, fiercer and generally more threatening. The
point is these differences are
all such as to give stronger reasons for excluding the
bear. If we construe the
argument as analogical, we need to recognise that what
makes the similarities and
differences relevant is an appeal to a value which the
rule is intended to promote,
in this case the comfort and peace of mind of other
passengers.
Now we turn to Harts example
of the vehicle in the park. A local park has
a by-law prohibiting vehicles.
The by-law poses two problems: what is to count
as a vehicle? For example are
roller-skates, skateboards, cycles, and lawnmowers
vehicles? And can we make
exceptions? Should we allow ambulances in if there
is an emergency, or a taxi in
exceptional circumstances, or a tank designed as a
memorial to local soldiers?
90 T. J. M. BENCH-CAPON
If we are going to use a
teleological perspective to answer these questions we
first need to construct a
model of parks which links parks and their activities to
values. I shall only sketch a
model here, and some of my modelling may well be
susceptible to (rational)
disagreement.
Let us suppose that the City
Fathers provide their parks with two main aims: to
provide a place for citizens
to relax in the fresh air, and to take exercise. I suppose
that their picture is of their
citizens sitting in the sunshine, or strolling through
pleasant surroundings. Certain
things are essential to achieve this: the park must be
a safe, clean place, and must
be an open space. Certain other things are conducive
to their aims: attractive
surroundings will encourage people to enter the park, and a
quiet atmosphere aids
relaxation. Other things are hostile to these aims: fast moving
objects make the park
dangerous, noise hinders relaxation and so forth. Anything
which is essential to the
values must be included; anything which entirely prevents
the realisation of the values
must be excluded. Other things can be considered on
their merits as to whether
they are more or less conducive or otherwise to the values.
Note that the relationship is
propagated: something conducive to a conducive
thing is itself conducive, and
something conducive to a hostile thing is hostile, and
the like.
When fitting out the park, the
City Fathers therefore provide lawns, benches,
paths, flower borders, some
statues and a lake. Also they provide a park keeper to
keep the place safe and clean.
We can now consider some
by-laws. What will guide us is the desire to provide
a coherent ordering of the
values. Should people be allowed on the grass? There is a
conflict between our values
here: allowing people on the grass promotes relaxation
since it provides more sitting
space and encourages sunbathing and dozing, but is
hostile in that it makes the
park less attractive, both because walking on the grass
is bad for lawns, and because
people lying around on them is generally untidy. The
decision here depends on the
view of the park and its clientele that the City Fathers
take: is it essentially to be
a green lung or a city showpiece? Let us introduce
a third value civic pride,
and suppose that if this is given pre-eminence, then
people will be kept off the
grass.
Next consider ball games. Ball
games are strongly conducive to exercise, but
create noise and disturbance,
are harmful to lawns, and potentially dangerous to
both flowers and people. If
people are excluded from the grass altogether, ball
games can be banned on an a fortiori argument. Otherwise we have a
conflict
between the two values of
exercise and relaxation: and on how these values are
rated for the particular park
in question our decision will depend.
Already we can see three
different models of park emerging according to how
the values are rated relative
to each other. We can have a recreation ground style
park where exercise is most
highly valued, a standard park where relaxation is
paramount, and a formal
garden where civic pride is most important.
Now consider dogs. Dogs
promote exercise, by encouraging their owners to
walk, but can be harmful to
lawns and flowers, can create noise and mess, and in
THE MISSING LINK REVISITED 91
the case of some dogs can
compromise safety. In the case of a formal garden we
might want to exclude them
altogether, but in the other cases dogs might be permitted,
although we might wish to
mitigate their most harmful effects by imposing
restrictions on their access
by insisting that they are kept on leads, and that their
owners are liable to remove
their fouling.
Now, having established some
context, we can turn to vehicles. Although today
we see cars as the
paradigmatic vehicle, recall that many of our parks (and their bylaws)
predate the internal
combustion engine. Probably therefore the correct central
case is a horse drawn vehicle.
Such vehicles promote none of our three values, and
are harmful to them, as a
source of noise, mess, and potential accidents. All of
these apply a fortiori to motor cars. But what of the
problem cases?
Roller skates, skate boards and cycles: These can be seen as
promoting exercise,
and are less polluting than
horse drawn or motor vehicles, but still
present the safety hazards,
and are disruptive of relaxation. My view is the
above argument suggests that
it would be coherent to class them as vehicles in
parks where civic pride and
relaxation are accorded sufficient weight as against
exercise, but not in the
recreation ground style park. So if the grass is out of
bounds, ball games are
prohibited and dogs must be kept on a lead, we would
expect roller skates and
skateboards to go also.
Lawnmowers: These vehicles are as noisy, polluting and
unsafe as any considered
so far. However, they can be
regarded as essential to the upkeep of
lawns, whether for display or
for ball games, and lawns are an important
contributor to all three of
our values. We would therefore expect them not
be banned as vehicles,
although we would expect their use to be restricted
to authorised park staff. This
also probably allows us to expect other vehicles
used for maintenance, even
quite large lorries, to be exempt from the ban.
Ambulances in an emergency: Here we have something which
clearly is a
vehicle, but one which banning
would compromise the safety of the park users,
an essential precondition for
achieving all of our values. Thus we would expect
our by-law to be waived here.
Taxis, in exceptional circumstances: Again these would probably
be permitted,
but the question as to what
constitute exceptional circumstances would depend
on the values governing the
particular park. A mayor might be driven to an
official ceremony in a
showcase park to promote civic pride, but would be
considered overbearing if he
was driven through a recreation ground. Taxis
might be permitted to give
disabled access to less formal parks, etc.
A Tank as Memorial: This will not exhibit any of
the harmful aspects of
vehicles, being stationary,
other than its intrinsic unsightliness. If we want to
ban it, we would be better
advised to argue on aesthetic grounds rather than
because it is a vehicle.
The above is intended to
suggest is that the questions look more difficult when
they are asked in isolation,
divorced from the context in which they might arise.
Consider roller skates: if
people have different conceptions of a park they might
92 T. J. M. BENCH-CAPON
well have different intuitions
about whether roller skates should be allowed. If,
however, we recognise that
that are different styles of park, shaped by the values
that they are intended to
promote, we can explain these different intuitions. The
style of park, and the values
it is intended to promote, is in practice revealed by the
context: particular features
of the park, and other regulations and decisions made to
organise the park. What we are
trying to do is to develop a coherent theory which
will explain the various
aspects of the park and its regulations, not in terms of
logical coherence there is
no logical contradiction between allowing ball games
and banning roller skates
but in terms of the various elements tending towards
the same ends. The last case,
the tank as a memorial, also illustrates another way
in which values can help: if
we know why a thing is banned we can see what
features of that thing cause
it to be banned. If those features are removed, as the
harmful features of vehicles
are absent in a static and disabled memorial tank, the
application of the rule in
that case becomes pointless pedantry.
6. Conclusion
In this paper I have tried to
revive the idea from Berman and Hafner (1993) that
teleology indeed provides a
missing link necessary to make sense of a body of
law. This aspect has been
neglected in recent years in favour of a purely logical
account of legal argument.
Useful as this is, it cannot tell the whole story: we have
to recognise that arguments
may have different value, even when either side of a
case can be argued and neither
side has a logically decisive argument.
Logic underwent a brilliant
development during the last century when, abandoning
the old formaulas, it set out
to analyze the methods of proof used
effectively bymathematicians.
. . . One result of this development is to limit its
domain, since everything
ignored bymathematicians is foreign to it. Logicians
owe it to themselves to
complete the theory of demonstration obtained in this
way by a theory of argumentation.
(Perelman and Olbrechts-Tyteca 1969, p.
10)
Note that logic is not
challenged within its own domain but is seen as failing
to provide guidance in many
areas that we can find important. This guidance can
come from teleology, from
notions of value, from consequences: in short from
the New Rhetoric. These ideas have already
been shown to be effective in the
domain of health education
(Grasso et al. 2000). I think we understand the logic
of legal argument quite well
now, and we need to recognise that we owe it to
ourselves to complete this
understanding, by adding in the purposes that our rules
are intended to promote. The
need was apparent in Berman and Hafner (1993), but
the programme is still to be
carried out.
THE MISSING LINK REVISITED 93
Acknowledgements
I hope I may be permitted to
make a short acknowledgement of the contribution of
Don Berman. I first
encountered him as an encouraging and supportive reviewer of
some of my very early work in
AI and Law. When I later met him I found him to
be open, friendly and always
excellent company. Don was always most helpful and
supportive to new researchers,
and a pertinent and constructive critic of the work
of others. His own work was
always insightful. Many people in the AI and Law
community owe much to Don, and
he is sorely missed.
I would also like to thank
Edwina Rissland and an anonymous referee for their
helpful comments on a previous
version of this paper.
Notes
1 If we were using dimensions
instead of factors, we might propose a dimension of motive ranging
from earning a living, through
social altruism and pleasure to malice. Alternatively, or additionally,
using a dimension which
recognised hot pursuit as a step towards possession would provide some
case for the plaintiff. Such a
dimension would also help the plaintiff in Young, since the strict test
used here, which would require
the fish to have been landed on the boat, could be softened to reflect
that the fish were under his
control in that they were trapped in an almost closed net.
2 By rational, I mean only that
reasons can be given, rather than that the decision is determined. The
social purposes recognised, and
the relative values ascribed to them can change over time, and vary
from one jurisdiction to
another. Moreover, the choice of factors has a very significant effect on the
outcome predicted by the model.
3 Here I use five factors, making
OWNLAND and OPEN distinct factors. I have also replaced
COMPETE by a factor true if the
defendant is pursuing his livelihood. I believe that this subsumes
COMPETE, and can be used as
well if two people are engaged in earning their livelihoods in
incompatible but not
competitive ways, as for example if one was a duck hunter and the other a
professional wildfowl painter.
It will also cover cases where the defendant is earning a living and the
plaintiff is not.
4 For a fuller discussion of
Perelmans ideas, see Bench-Capon (2001).
References
Ashley, K. D. and Rissland, E.
L. (1988). A Case-Based Approach to Modelling Legal Expertise.
IEEE Expert, pp. 7077.
Ashley, K. D. (1990). Modeling
Legal Reasoning. MIT Press: Cambridge.
Bench-Capon, T. J. M. (1999).
Some Observations on Modelling Case Based Reasoning with Formal
Argument Models. In Proceedings
of the Sixth International Conference on AI and Law, 3642.
ACM Press: New York.
Bench-Capon, T. J. M. (2001).
George C. Christie, The Notion of an Ideal Audience in Legal
Argument. Artificial Intelligence
and Law 9(1): 5971.
Berman, D. H. and Hafner, C. D.
(1993). Representing Teleological Structure in Case-Based Reasoning:
The Missing Link. In
Proceedings of the Fourth International Conference on AI and Law,
5059. ACM Press: New York.
Branting, L. K. (1991).
Reasoning with Portions of Precedent. In Proceedings of the Third
International Conference on AI
and Law, 145154. ACM Press: New York.
94 T. J. M. BENCH-CAPON
Gordon, T. F. (1993). The
Pleadings Game: Formalizing Procedural Justice. In Proceedings of the
Fourth International Conference
on AI and Law, 1019. ACM Press: New York.
Grasso, F., Cawsey, A. and
Jones, R. (2000). Dialectical Argumentation to Solve Conflicts in Advice
Giving: A Case Study in the
Promotion of Healthy Nutrition. International Journal of Human-
Computer Studies 53: 10771115.
Perelman, Ch. (1980). Justice,
Law and Argument. Reidel: Dordrecht.
Perelman, Ch. and
Olbrechts-Tyteca, L. (1969). The New Rhetoric: A Treatise on Argumentation.
University of Notre Dame Press:
Notre Dame.
Prakken, H. (1993). A Logical
Framework for Modelling Legal Argument. In Proceedings of the
Fourth International Conference
on AI and Law, 19. ACM Press, New York.
Prakken, H. and Sartor, G.
(1998).Modelling Legal ReasoningWith Precedents in a Formal Dialogue
Game. Artificial Intelligence and
Law 6(24):
231287.
Retirado de: http://www.kluweronline.com