Defeasibility in Legal
Theory
The idea of defeasibility remains highly controversial, as evidenced by
the contributions contained in a recent collection (Ferrer and Ratti
2012).
Carlos Alchourron, a leading legal logician, has opposed the ideal of
defeasible reasoning, arguing for a combination of systematic
interpretation and deduction: systematic interpretation should merge
rules and exceptions into a coherent whole to which deduction could be
applied (see Alchourron 1996a, 1996b). Other legal theorists, such as
Alexander Peczenik (Peczenik 2005, 115ff., Hage and Peczenik 2000) and
Neil MacCormick (1995), have on the contrary argued that defeasibility
plays a significant role in legal reasoning (see also Brozek .
It is no easy task to review the legal theorists’ approaches to
defeasibility, since such theorists have advanced different
understandings of defeasibility, which often do not comport with the
idea of defeasibility as nonmonotonic reasoning. Brozek (2014) has
pointed out different ways in which defeasibility is understood in
Ferrer and Ratti (2012a), a collection of approaches to defeasibility in
legal theory:
Ferrer Beltrán and Ratti consider, inter alia, the following
formulation: “a norm is defeasible when it has the disposition not to
be applied even though it is indeed applicable” (Ferrer Beltrán &
Ratti 2012b: 31). Frederick Schauer, in turn, claims that “the key idea
of defeasibility […] is the potential for some applier,
interpreter, or enforcer of a rule to make an ad hoc or
spur-of-the-moment adaptation in order to avoid a suboptimal,
inefficient, unfair, unjust, or otherwise unacceptable, rule-generated
outcome,” and concludes that “defeasibility is not a property of rules
at all, but rather a characteristic of how some decision-making system
will choose to treat its rules” (Schauer 2012, 81 and 87). Jorge L.
Rodríguez says that “when we express a conditional assertion, we assume
the circumstances are normal, but admit that under abnormal
circumstances the assertion may become false”, and—transferring this
characteristic of defeasibility into the domain of law—claims that
“legal rules [are defeasible since they] specify only contributory,
yet not sufficient, conditions to derive the normative consequences
fixed by legal system” (Rodríguez 2012, 88). […] Finally,
Riccardo Guastini claims that legal rules are defeasible since “there
are fact situations which defeat the rule although they are in no way
expressly stated by normative authorities in such a way that the legal
obligation settled by the rule does not hold anymore.” (Guastini 2012,
183)
All the foregoing formulations point to interesting aspects of legal
reasoning and to the practice of defeasible reasoning in the law. I
would argue, however, that they fail to provide convincing redefinitions
or clarifications of the notion of defeasibility. I have argued that
defeasibility applies to three objects:
-
Arguments . A defeasible argument is an internally valid
argument that may be defeated by counterarguments that do not
challenge the argument’s premises but rebut its conclusions or
undercut the link between its premises and its conclusion.
-
Inference . A defeasible inference is nonmonotonic, in the sense
that it makes it possible to derive conclusions that may no longer be
derivable if additional premises are added.
-
Conditionals . A conditional is defeasible when it has the
logical structure of a default, i.e., when it links a merely
presumptive (non-conclusive) consequent to its antecedent.
These three aspects are different faces of the same issue. A defeasible
argument \(\mathcal{A}\) consists in a nonmonotonic inference: if we
expand the argumentation basis from which \(\mathcal{A}\) is constructed
with premises that enable the construction of a defeater \(\mathcal{B}\)
to \(\mathcal{A}\), the conclusion of A will no
longer justified relatively to the expanded argumentation basis, and in
this sense, no longer derivable from it. Correspondingly, default
conditionals make it possible to construct defeasible arguments, i.e.,
nonmonotonic inferences: the results obtained through defeasible modus
ponens can be defeated by rebutters or undercutters.
According to this idea of defeasibility, a legal norm can be said to be
defeasible whenever all the following conditions are jointly
possible:
-
The norm is to be accepted (being valid and being generally applicable
in the special-temporal domain under consideration)
-
The norm’s antecedent is also to be accepted
-
The norm’s consequent is to be rejected.
As we have seen in the examples above (for instance, in Figure 8), a
defeasible norm \(N\) can be modelled as a default, i.e., in the logical
form “if \(P(\mathbf{x})\) then presumably \(Q(\mathbf{x})\)”, i.e.,
as
\(N\left(\mathbf{x}\right):\ P\left(\mathbf{x}\right)\Rightarrow Q\left(\mathbf{x}\right)\),
where \(\mathbf{x}\) is the list of the variables in the norm (and the
default would stand for the set of its ground instances). In fact., the
inferences (arguments) warranted by that norm—i.e., arguments having
the form
(\(P\left(\mathbf{a}\right),\ N\left(\mathbf{x}\right):\ P\left(\mathbf{x}\right)\Rightarrow Q\left(\mathbf{x}\right),\ therefore\ Q\left(\mathbf{a}\right))\)
where a is an individual case, namely, a list of values for variables
\(\mathbf{x}\) —can be rejected, given appropriate conditions, without
rejecting the norm or is antecedent. This would happen whenever the
premise for building a rebutter (a stronger norm having the
form\(\ N_{1}\left(\mathbf{x}\right):R\left(\mathbf{x}\right)\Rightarrow\neg Q\left(\mathbf{x}\right)\)
or an undercutter (a norm having the form
\(N_{2}\left(\mathbf{x}\right):R\left(\mathbf{x}\right)\Rightarrow\neg N_{1}\left(\mathbf{x}\right)\))
is available. This notion of the defeasibility also applies to the more
abstract view of the applications of a norm as involving a meta-level
warrant such as “If the norm ‘If
\(N\left(\mathbf{x}\right):\ \)P \(\left(\mathbf{x}\right)\)
then presumably Q \(\left(\mathbf{x}\right)\)’ is valid and
P \(\left(\mathbf{a}\right)\) is the case, then presumably
Q \(\left(\mathbf{a}\right)\)” (as in the model proposed by
Hage 1997). In the following I will speak of argument warranted by a
norm, to cover both models of norm-based reasoning.
If the coexistence of the three conditions above is impossible, then the
norm N at issue can be said to be strict or
indefeasible, and can be modelled as a material conditional,
having the form: “for all \(\mathbf{x}\), if \(P(x)\) then \(Q(x)\)”,
i.e. \(\forall x(P(x)\rightarrow Q(x))\). More plausibly an
indefeasible norm \(N\) could be represented through a universal strict
conditional “for all \(x\), if \(P(x)\), then necessarily \(Q(x)\)”,
i.e, \(\forall x(P(x)\twoheadrightarrow Q(x))\) (or possibly as a
strict rule, which may not allowing for contraposition, see Prakken
2010). The strict conditional, which is here denoted with the arrow,
\(\twoheadrightarrow\) expresses the idea that the correlation between
\(P(x)\) and \(Q(x)\) does not depend on the present factual situation
(on the actual world), but would rather hold in every possible factual
situation (the norm being unchanged). If we accept the indefeasible norm
N and also accept that its antecedent holds in any possible
context, we must accept that also the norm’s consequent holds in that
context. N could not be the object of exceptions in a strict
sense, namely, of provisions stating that the unmodified norm does not
apply when an impeding circumstance E(a) is established.
To avoid the effect of \(N\) to be triggered in circumstance
E(a) , we would have to substitute it with the new norm
\(\forall\mathbf{x}\ (P(\mathbf{x})\ \bigwedge\neg E(\mathbf{x})\ \twoheadrightarrow Q(\mathbf{x})\).
Because of this change the norm’s effect could be established in a case
only when both predicates \(P\) and \(\neg E\) are established in that
case. Rather than \(E\) being an impeditive fact capable of blocking the
application of the norm, \(\neg E\) would become a constitutive fact
that must be established, for that effect to be triggered.
Let us consider for instance a norm linking the causation of harm to the
obligation to compensate the victim. If the norm were defeasible, it
would mean that if any individual \(x\) culpably harms another
individual \(y\), then presumably \(x\) must compensate \(y\), and it
could be modelled in the logical
form\(:\ N\left(x,\ y\right):\ CulpablyHarms(x,y)\Rightarrow MustCompensate(x,y)\).
If the norm were indefeasible, it would rather mean that for all
individuals \(x\) and \(y\), if \(x\) harms \(y\) then necessarily \(x\)
has to compensate \(x\), and it could be modelled in the logical form:
\(\forall x,\ y(CulpablyHarms\left(x,y\right)\twoheadrightarrow MustCompensate\left(x,y\right))\).
Thus, from the perspective here developed, the defeasibility of a norm
pertains to its content, as expressible in its logical form, and
therefore is not affected by the fact that the norm may be declared
invalid: this may happen, under appropriate conditions, for both
defeasible and indefeasible norms. Similarly, the defeasibility of a
norm is not affected by the fact that the norm may be modified or
substituted through judicial interpretation or through legislation. Both
defeasible and indefeasible norms can be modified by new legislation or
case law. The difference rather pertains to the necessity of a
modification to introduce an exception:
-
If exceptions to a norm can be introduced without changing the norm
(without affecting its content or meaning), then the norm is
defeasible, regardless of whether exceptions are expressed or implicit
and whether they closed or open, and regardless of what authority and
procedure that is needed for introducing exceptions (for an analysis
of different kinds of exceptions, see Celano 2012). In particular, it
is irrelevant to the defeasibility of a norm, whether exceptions to it
can be introduced through judicial interpretation, or only through
legislation (or through new constitutional norms).
-
If the only way to legitimately exclude the application of a norm to
cases having feature \(E\) consists in changing that norm, extending
in its antecedent with the negation of E (\(\neg E\)),
then the norm is indefeasible.
In their analysis of the notion of defeasibility in the law Ferrer and
Ratti (2012b, 36) distinguish three cases: (1) the norm’s validity is
defeasible, in the sense that it depends on defeasible criteria, (2) the
norm is externally defeasible, in the sense that the “conditions of
applications contain implicit exceptions whose scope has not been
determined”, and (3) the norm’s normative content is defeasible in the
sense that it specifies operative facts that are contributory conditions
for the production of the norm’s legal effect. In particular, they say
that in the third case “the norm’ s antecedent contains
implicit exceptions which may not be exhaustively identified.” They
also affirm that in cases (1) and (2), it is not the norm itself which
is defeasible, but rather the criteria for its validity or application,
while the norm should be represented as a material conditional.
As Ferrer and Ratti (2012b, 36) rightly observe, only their third case
of defeasibility is really significant: the first two cases depend on
meta-norms on validity or application are defeasible in the third sense,
namely, according to their content. However, the way in which they
describe defeasibility by content differs from the approach here adopted
in three regards.
Firstly, it makes the implicitness of the exceptions to a norm a
necessary condition for the defeasibility of the norm. On the contrary,
I have argued that even explicit exception to a norm presuppose the
defeasibility of that norm, since they give rise to the pattern that
characterises defeasible argumentation: the absence of the exception is
not needed to construct an argument warranted by the norm, though that
argument can be defeated by arguments warranted by the exception.
Secondly, the characterisation of the antecedent of a defeasible norm as
providing contributory conditions for the norm’s conclusion fails to
capture the idea of defeasible connection between the antecedent and the
conclusion of that norm. As I observed in Section 3, a contributory
condition for a conclusion may fail to provide any presumptive support
for that conclusion. This is the case when a norm has a conjunctive
antecedent so that its application results in a linked argument (see
Figure 2). On the other hand, the antecedent of defeasible norm can be
described as a contributory reason for the norm’s conclusion. A genuine
contributory reason for a legal conclusion should indeed provide on its
own sufficient presumptive support to that conclusion, i.e., it should
match the antecedent of a legal default, and so enable the construction
of a separate defeasible argument. Separate defeasible argument sharing
the same conclusion contribute to a stronger convergent argument
supporting the same conclusion (see Figure 3 and Figure 4).
Finally. it is not clear to me why the issue of determining where a norm
is defeasible or not should be specifically addressed as a “matter of
interpretation”. It is a matter that pertains to the determination of
the logical structure of the norm at issue, an issue that could pertain
to interpretation or not depending on how one understand the notion of
interpretation, i.e., as concerning every ascription of meaning to a
text, or only the ascription of meaning meant to address some doubts
(see Dascal and Wroblewki 1986). Interpretation in the first sense
obviously covers the determination of every aspect of the content of a
norm having a textual source, and therefore it also covers also the
determination of its logical structure. In fact, to determine whether a
norm is defeasible or not, we have to consider —depending on whether
we are approaching the norm from a socio-legal or from doctrinal
perspective— (a ) the way in which the norm is comprehended and
used by the community of its users (those who endorse/follow/apply it)
or (b ) the way in which the norm should correctly be comprehended
and used by the same community. This determination would involve an
empirical assessment according to (a ) or a normative assessment
according to (b ). This assessment would be no less (and no more)
dependent on interpretation than the determination of any other aspects
of the norm’s content, such as the structure and the components of its
antecedent and its consequent. Finally, the issue of determining whether
a legal norm is defeasible would be utterly trivial if we were to adopt
—both empirically and normatively— the view that all legal norms are
defeasible in this abstract sense —i.e., the assumption that no strict
legal norm exists, as a matter of fact— following Leibniz’s
suggestion.
It seems to me that we need to distinguish clearly two aspect concerning
a norm’s defeasibility. The first aspect, to which we have referred in
this contribution by using the term “defeasible”, pertains the
intrinsic logical structure of a norm: is the norm meant to establish a
presumptive or a conclusive link between its antecedent and its
conclusion? Defeasibility so understood is a counterfactual property: to
say that a norm N, having antecedent P and conclusion
Q (I leave the variables implicit, for simplicity’s sake), is
defeasible just means that it is in principle possible to reject an
argument for Q warranted by N , while accepting both
N and P : we can imagine a system \(\mathbb{L}\) and a
factual constellation \(\mathbb{F}\), such that with regard to the
argumentation basis \(\mathbb{L\cup F}\) both N (unchanged) and
P are justified (e.g. being unchallenged), but the argument that
delivers Q on the basis of N and P is rebutted or
undercut, by arguments constructible from \(\mathbb{L\cup F}\).
Once we have determined that N is intrinsically defeasible, we
can address further issues. One issue concerns determining whether
N- warranted arguments can be defeated in the legal system
\(\mathbb{L}\) containing N , i.e., whether a rebutting or
undercutting counterargument to an \(N\)-warranted argument can be
mounted by using only norms in \(\mathbb{L}\), plus appropriate
operative facts (see Section 11). A different issue concerns determining
whether N could be defeated given the possible (permitted or
empowered) judicial modifications of the current legal
system L, namely, whether judicial
construction/interpretation could introduce in \(\mathbb{L}\) new norms
that enable the construction of defeaters against the application of
N . Obviously answering either of these issues may or will require
the interpretation of the legal system \(\mathbb{L}\) under
consideration (on the connection between the possibility that a norm is
defeated and interpretation, see Duarte 2011, 135).
Finally, the idea of a norm’s defeasibility as pertaining to the logical
structure of that norm, leads me to address one further claim by Ferrer
and Ratti, namely, the view that whenever a norm’s validity or
application is determined by defeasible metarules, the norm itself must
be indefeasible. Since a norm’s defeasibility only concerns the logical
structure of that norm, the fact that a norm is defeasible does not
exclude (nor require) that its validity as well as the domain of its
intended application are governed by defeasible criteria.
Inapplicability rules, however, may presuppose the defeasibility of the
norm that they address: a rule stating that norm N is
inapplicable under exceptional circumstances E , is usually meant
enable the construction of undercutters to N- warranted arguments,
namely, arguments having the following form: \(E\) is the case, if \(E\)
than \(N\) does not apply (does not warrant its conclusion), therefore
\(N\) does not apply: \(E,E\Rightarrow\neg N\), therefore \(\neg N\)
(see argument \(C\) in Figure 8).
In conclusion, I think that, notwithstanding the multifarious creative
ways in which legal theorists have framed the idea of defeasibility, it
would better to stick to the more limited and precise concept on which
other disciplines —such as logic, philosophy, and
computing— converge, namely, the view that defeasible reasoning is
nonmonotonic and that the antecedent of a defeasible norm provides only
presumptive support to the norm’s conclusion. The considerations that
have been presented as alternative analyses of the concept of
defeasibility should rather be rephrased are pertaining to the ways in
which (a) arguments applying defeasible legal norms can be rebutted or
undercut, or (b) existing legal norms, both defeasible or indefeasible
ones, can be abrogated or modified.