REFLECTIVE EQUILIBRIUM AND JUSTIFICATION
Daniel Little
Southern Journal of Philosophy 22 (1984)
Take it as given that
the purpose of justification in moral philosophy is to provide rational grounds
for preferring one moral theory over another as a basis for making moral
judgments. This is not the only
function which justification may serve; establishing moral truths and building
consensus around moral values are alternatives on either side of this
conception. But
it is the sense of justification with which I will be concerned here. This is a moderate conception which is
identifiably epistemic and yet not committed to the existence of moral
truths. The central problem of
moral epistemology on this view is to provide an account of the rational
grounds on the basis of which one may discriminate between competing
theories. Two extreme positions
are possible here: one which holds that a method of justification exists for
choosing between competitors which depends only upon the nature of rationality
(Kant’s view); and another which holds that there are no rational grounds at
all for choosing between competing and equally consistent moral systems (moral
skepticism).[1]
In what follows I will
be concerned with an idea about justification which lies between these two
extremes, and which is most fully developed in the writings of John Rawls.[2] This is the method of reflective
equilibrium: the idea that moral theories are justified through a process of
deliberation in which we consider a wide range of beliefs and judgments as a
system, and evaluate competing moral theories in terms of their coherence with
this system. The attraction of
this model is that it avoids the strong Kantian claim that moral theories may
be established unconditionally, but also seeks to avoid moral skepticism. My purpose in what follows is simply to
assess the degree to which arguments of this sort succeed in permitting
rational choice between theories.
I will conclude that the method Rawls describes is in fact very weak, since
it depends heavily on the substantive moral presuppositions with which one
begins. Instead of justifying a
moral theory it seems largely to be a technique for systematizing and
organizing one’s antecedent moral convictions. Given that Rawls’s system is one of the most
methodologically sophisticated to appear in recent years, this conclusion casts
a substantial shadow over the project of providing an objective justification
for moral theories in any form.
Reflective
equilibrium and coherence arguments. The central idea at work here is the conception
of moral philosophy as a process of seeking “reflective equilibrium”: that is,
a process through which the philosopher attempts to arrive at a coherent moral
theory which is compatible with a wide range of pre‑theoretical
commitments. This process of
justification requires a continual interaction between the formulation of
theoretical principles and a consideration of pre‑theoretical commitments. We formulate a general moral principle,
then test it against our pre‑theoretical commitments. Do its concrete consequences correspond
to the judgments we would actually make? Does it satisify conditions which we
pre‑theoretically believe to validly characterize all acceptable moral
theories? If the theory conflicts at either stage with our moral convictions,
then we have to decide whether to modify the theory or abandon the
convictions. This process is to
continue until theory and commitments are in harmony with each other. Rawls characterizes this final stage as
the condition of reflective equilibrium.
We may refer to this process as one of “reflective deliberation.”
This conception
postulates that the preferred moral theory is the one which best serves to
organize and explain “considered judgements” and moral convictions. More specifically, Rawls’s view is that
we have a whole range of moral commitments, from the level of concrete
judgments like “it is unjust for one person to hold all land as private
property,” to formal judgements like “moral theories ought to be general,
universal, and public” to substantive principles like “it is fair to submit
people to principles they themselves would have chosen. “ These commitments are substantially
pre‑theoretical, in that everyone (not just moral philosophers) shares
them, or analogous ones.
Accordingly, a moral theory is justified to the degree that it serves to
systematize and explicate the full range of these moral commitments, and to
extend our commitments in cases where our pre‑theoretical judgment is
uncertain.
Rawls uses this process
of reflective deliberation to support his formulation of the premises of
justice as fairness: the idea that the correct theory of justice is that which
rational persons would choose to regulate their public affairs, and the particular
description which Rawls puts forward of the circumstances of choice (the
original position). Having arrived
at these ideas, Rawls argues that they strongly support the two principles of
justice over other possible candidates.
But this argument from the original position to the two principles of
justice is intended chiefly as a vivid model of the moral ideas we bring to the
problem of justice; it is chiefly intended to provide a way of coherently
expressing and extending those ideas.[3]
Thus
the deductive argument from the description of the original position is not
itself a primary source of justification for Rawl’s theory of justice. It depends wholly on the deliberative
justification available for the general program of justice as fairness.[4]
The process of
reflective deliberation described here depends heavily on a coherence
epistemology: the idea that justification depends upon the power of a theory to
smoothly incorporate a wide range of diverse moral commitments into a coherent
moral conception.
A
conception of justice cannot be deduced from self‑evident premises or
conditions on principles; instead, its justification is a matter of the mutual
support of many considerations, of everything f~tting together into one
coherent view. (TJ, p. 21).
Is there any
specifically moral import to coherence arguments? Some philosophers have argued
that the coherence methodology entails a sharp distinction between ethics and
empirical knowledge. For example,
Ronald Dworkin argues that the use of coherence arguments entails that we must
accept a constructivist view of moral beliefs, according to which moral truths
are constituted by
the persons who subscribe to them.[5]
He
reaches this conclusion through the observation that coherence considerations
lead us sometimes to revise basic judgements, sometimes moral principles; and
he asks how this could be justified if there were independent moral
truths. This argument is unsound,
however; for a coherence epistemology in empirical science entails just the
same revisability of natural law statements, without entailing that observers
constitute those laws. The
coherence epistemology is a method of justification, not a thesis concerning
the ontological status of the laws arrived at. Consequently, this approach does not entail constructivism.
David Lyons doubts the
justificatory import of the coherence approach in “Nature and Soundness of the
Contract and Coherence Arguments”.[6]
He
holds that since the coherence approach depends upon the considered judgments
of the moral philosopher, the principles it supports are no stronger than
ordinary moral common sense. “One
may still doubt whether a coherence argument says anything about the validity
of such principles; for pure coherence arguments seem to move us in a circle,
between our current attitudes and the principles they supposedly manifest.”[7]
On
Lyon’s view, therefore, the coherence argument serves merely to demonstrate the
presuppositions of common morality, and not at all to demonstrate the
correctness of the principles which it discovers. However, Lyons’ argument is too broad‑gauge; if it
were correct it would apply equally to the use of coherence arguments in
empirical science as well. But
Quine, Goodman, and others have made out a strong case for the viability of a
non‑foundationalist epistemology.[8]
Consequently,
if a coherence‑based argument is defective in moral reasoning, its defect
must derive from the content of the argument and not the general epistemology
on which it is based.[9]
Against these views, I
hold that the coherence epistemology is uncontroversial from the moral point of
view. It is a thesis about the
nature of rational theory‑assessment generally, and is as well suited to
moral as to empirical reasoning.
(This is not to say that the coherence method is uncontroversial, but
only that the arguments which may be offered against it are largely independent
of the particular features of moral reasoning.) The chief import of Rawls’s
choice of the coherence epistemology is its implicit rejection of a foundationalist
theory of knowledge, according to which some class of statements is
epistemically basic and unrevisable.
Rawls instead borrows a chapter from Goodman and Quine and depicts moral
reasoning as a mutual adjustment of beliefs at every level, in which no belief
is immune from revision. This
position rejects the model which represents knowledge as being founded on some
set of privileged statements (perception reports or considered moral
judgements), and instead represents the acquisition of knowledge as a process
in which theoretical beliefs sometimes dictate the rejection of more immediate
judgments. In the moral case, this
stance entails rejecting rational intuitionism, or the view that moral theories
may be founded on a privileged set moral intuitions. This stance is thus one shared with general epistemology.[10]
If
this treatment is correct, then the coherence epistemology ought not be the
center of discussion in considerations of Rawls’s theory of justification;
rather, the central question is the strength of the theories and judgments
which are the content of the process of reflective deliberation. For it must be emphasized that the
coherence argument does not itself confer justificatory weight on its conclusion;
it is a general form of argument rather than a source of justification. In order to know whether the argument
has justificatory weight, it is necessary to know what the content of the
argument is. Merely to conclude
that “T encompasses all our beliefs in the smoothest and most consistent way”
does not serve to justify T.
Rather, we must add that the bulk of our beliefs are credible to some
degree.
In
order to evaluate Rawls’s justificatory arguments, therefore, it is necessary
to consider the strength of the beliefs and judgments which serve as the source
of credibility within the process of deliberation described here. This process of deliberation may be
carried out at two levels of generality, depending on the sorts of beliefs
which are introduced into the process.
First (in what Rawls calls “narrow reflective equilibrium”) the process
may take account only the “considered judgments” with which the philosopher
begins: the concrete moral judgments in which we have the greatest degree of
antecedent confidence. Here the
task of deliberation is to construct a moral theory which organizes and
explains these considered judgments in the most coherent fashion possible. Second, the process may take account of
a wide range of philosophical and empirical beliefs as well as considered
judgments. This version of the
process (which Rawls describes as one pursuing “wide reflective equilibrium”)
is much more general than the former, in that it attempts to link the moral
theory eventually produced to the full range of moral and empirical commitments
with which we begin.
A quasi‑inductive
argument. In
his conception of narrow reflective equilibrium, Rawls proposes that it is
possible to “test” a moral theory in terms of its consonance with our
antecedent moral convictions (in analogy with the testing of an empirical
theory against its consequences for experience). That is, we may evaluate a moral theory by working it out in
detail and seeing the degree to which its distant implications match the judgments
we are inclined to make in specific cases. “There is, however, another side to justifying a particular
description of the original position.
This is to see if the principles which would be chosen match our considered
convictions of justice or extend them in an acceptable way.” (TJ, p. 19) More fully, Rawls
postulates that we all have a psychologically real “moral sense” or moral
competence, which leads us to make moral judgments in a variety of actual
cases. These judgments form the
range of relatively fixed points to which moral theory must generally
conform. Any moral theory which
widely and grossly disagrees with the spectrum of our considered judgments must
be judged inadequate; our initial moral opinions are presumed to have some
preliminary standing to which the moral theory must conform.
An
especially crisp formulation of this idea occurs in “The Independence of Moral
Theory”: “People profess and appear to be influenced by moral conceptions. These conceptions themselves can be
made a focus of study; so provisionally we may bracket the problem of moral
truth and turn to moral theory: we investigate the substantive moral
conceptions that people hold, or would hold, under suitably defined
conditions.”[11] A similar view is
expressed in A Theory of Justice: “Now one may think of moral philosophy at
first. . . as the
attempt to describe our moral capacity; or in the present case, one may regard
a theory of justice as describing our sense of justice. .
. “ ( TJ, p. 4)
This
model of moral philosophy borrows from Noam Chomsky’s description of the nature
of linguistic theory.[12]
Its
premise is certainly correct: we do in fact make a vast number of moral
judgments without being explicitly aware of their relation to more general
moral principles. So the project
of constructing a moral conception which best organizes and supports those
judgments is a reasonable one.
Rawls is also correct in supposingthat theories of this sort may only be
evaluated on the basis of their success at organizing the “behaviors” to which
they correspond: the actual moral judgment‑making practices which the
theory is intended to account for.
Through the process of deliberation which Rawls describe. we may arrive at a theory of the moral
competence of the subject: a set of moral principles which best represents S’s
preliminary considered judgments, in much the way that a linguist arrives at a
generative grammar to represent judgments of grammaticality.[13]
Thus it
is reasonable to say that a moral theory may in some sense be “tested” by
examining its fit with actual practices.
However,
this linguistic analogy by itself does not allow us to describe this process as
one supplying moral justification.
Rather, this analogy suggests that the process described here is a
purely empirical one: the theory which emerges is a theory of a psychological
capacity (the capacity to make moral judgments), and is true or false as a
description of the psychological makeup of the agent. But nothing in this process as described up to this point
provides any moral justification for believing that the resulting theory is
correct: the fact that T best systematizes S’s moral judgments does not serve
to show that T is morally correct.
Thus the linguistic analogy represents moral philosophy only as a division
of psychological theory.
This
process of deliberation might seem to provide a basis for moral justification,
however, if we shift metaphors and conceive of it as a form of moral
“induction”. The fact that “T is
the best available theory of S’s moral competence” would gain standing as a
moral justification if we were to suppose that S’s moral judgments are usually
morally correct. On this account,
considered moral judgments are analogous to observation reports in empirical
science. They have an initial
moral credibility which can be transferred to the theories which moral
philosophers construct. If we make
this assumption (analogous to the assumption in science that our perceptions
are generally veridical) then the following quasi‑inductive argument has
weight:
i. S’s considered judgments are by and large morally correct.
ii. The principles of justice best explicate those considered judgments.
iii. Therefore the principles of justice are at least approximately correct.
This arguments
represents a case in which a class of judgments is presumed to have some prima
facie credibility,
and that credibility is passed on to the theory which best explains the
judgments.
This interpretation of
the moral competence model of justification depends upon a strong analogy
between considered judgments and observation reports in empirical science. The antecedent credibility of
considered judgments is transmitted to the moral theory. But the disanalogies between these two
sorts of beliefs seem more striking than their similarities. We have reasons for being confident in
observation reports which do not pertain to considered judgments.
First, our knowledge of the causal
basis of perception gives us reason to believe that empirical observations are
likely to be true. Using physical
theory we can vindicate our reliance on observation reports as the epistemic
basis of empirical knowledge. By
contrast, considered moral judgments plainly do not have this property. Nothing we know about the psychology of
moral belief formation gives us any reason for supposing that the beliefs which
emerge from that process are likely to be correct. Rather, accidents of familial, religious, and educational
experience seem to have a major influence on the particular constellation of
beliefs which eventually express our moral competence.
A
second and related point has to do with differences in the degree of theory‑dependency
characteristic of these two sorts of beliefs. Notwithstanding post‑positivist critiques of the
observation‑theoretic distinction, it is reasonable to conclude that
observation reports in empirical science are theory‑dependent to only a
small degree: the observations made by holders of different empirical theories
are generally relatively invariant.[14]
By
contrast, considered moral judgments are inseparable from the moral theories
which they are supposed to evaluate.
This is so because moral judgments are not primitive beliefs; rather,
they generally reflect a process of reasoning through which the agent weighs
relevant principles and facts in order to arrive at a concrete judgment. The resulting judgment therefore
depends heavily upon the backround moral theories of the agent.
Third,
considered judgments seem to be no more certain than the moral theories which
they are intended to evaluate. In
fact the existence of widespread disagreement at the level of concrete moral
judgments is one of the sources of our impulse to arrive at more general
statements of moral principle. In
large part we want a justified moral theory in order to resolve disagreements
at the level of considered judgments.
Thus the analogy between considered judgments and observation reports is
unpersuasive in view of the controversial disagreements to be found among the
moral competences of different individuals and groups. Whereas observation reports serve to
validate or refute empirical theories, concrete moral judgments are themselves
often validated or refuted by appeal to more general moral theories.
These
considerations suggest that the competence argument is only weakly
justificatory in the moral sense defined above. It depends upon the approximate correctness of our moral
competence. This appeal would be
strongly justificatory only if we could treat considered judgments as
relatively credible. But
considered judgments are not epistemically similar to perceptual reports; they
lack the form of immediacy and reliability which observation reports possess,
in virtue of which they function as an empirical control on physical
theory. We therefore hav’e little
reason for confidence in the correctness of our considered judgments; but
narrow reflective equilibrium makes the resulting moral theory only as strong
as the considered judgments themselves.
It is thus reasonable to hold that the primary import of the process of
narrow reflective deliberation is empirical rather than moral. The theory which results from this form
of deliberation is a good description of existing moral competence, but the
process provides only a very weak normative basis for evaluating the moral
theory against other moral theories.
This form of reasoning thus falls under empirical moral psychology
rather than normative moral theory.[15]
Inter‑theoretical
support. Let
us turn now to Rawls’s concept of “wide reflective equilibrium.” According to
this approach, “. . .
one is to be presented with all possible descriptions to which one might
plausibly conform one’s judgments together with all relevant philosophical
arguments for them. “ ( TJ,
p. 49) The correct moral theory is
that which would be selected through this highly general form of
deliberation. This method requires
that we treat our considered judgments, philosophical theories, and proposed
moral theories as a system, and modify various parts in order to improve their
consistency, until no further changes are required. In this form the considered judgments are open to
substantial revision.
In a series of articles
Norm Daniels has elaborated on the method of wide reflective equilibrium. He represents wide reflective
equilibrium as “an attempt to produce coherence in an ordered triple of sets of
beliefs held by a particular person, namely, (a) a set of considered moral
judgments, (b) a set of moral principles, and (c) a set of relevant background
theories,’[16] He maintains that the
addition of (c) makes the problem of justification more tractable, insofar as
the background theories provide some support for the moral principles which is
independent of the considered moral judgments.
On this account, the
method of wide reflective deliberation gets its justificatory bite from the
fact that we may bring theories from other areas to bear on the choice between
competing moral theories, and those background theories may possess independent
credibility. Daniels cites Rawls’s
theories of the person, of procedural justice, and of the role of morality in
society as examples of the sorts of background theories he has in mind; other
examples would include the empirical theories we hold of human psychology,
social dynamics, and the like, on the basis of which we consider the likely
stability of proposed moral arrangements.
Supposing that those theories have some degree of credibility which
derives from considerations distinct from those available for the moral theory
itself, and that these theories offer grounds for discriminating between competing
moral theories, we will be able to justify one moral theory over another in
terms of their relative compatibility with the background theories.
This approach represents
a different analogy with justification in natural science: in this case, not inductive
support, but inter‑theoretical support. Moreover, Daniels’ reconstruction is well‑suited to
providing a form of justification of the sort needed here (to provide reasons
for preferring one moral theory to another). If we have reason to believe the background theories and if
they support one moral theory over another, then we have reason to accept the
favored moral theory. In other
words, the method of wide reflective equilibrium seeks to embed disagreements
about moral theory within a wider framework in which rational argument may get
some traction.
This move represents
progress, however, only on the assumption that philosophical theories are
somehow easier to justify than moral theories in the first place. And in fact Daniels claims that “disagreements
about [background] theories may be more tractable than disagreements about
moral judgments and principles.
Consequently, if the moral disagreements can be traced to disagreements
about theory, greater moral agreement may result.”[17]
Two
points may be made about this view.
First, it may be noted that Rawls seems to reject this move in “The
Independence of Moral Theory”, where he maintains that substantive moral
questions are largely independent of background philosophical theories
(e.g. theories of the person,
etc.). “Now my thought is this:
much of moral theory is independent from the other parts of philosophy.”[18]
And
second, a certain amount of skepticism may greet Daniels’ expectation that
disagreements over theories of the person or of the role of justice in society
are less controversial, or more easily resolved, than disagreements about moral
principles. It seems as reasonable
to suppose that broadening the discussion from moral judgments and principles
to moral judgments, principles, and philosophical theories, has simply
broadened the possible sources of irresolvable disagreement.
Let us focus the latter
criticism more sharply. To what
extent are there strong forms of justification available in support of
philosophical theories (e.g. theories
of the person)? Thejustifications offered for such theories are apparently as
inconclusive as the justifications offered for moral principles in the first
place. Some candidate theories of
the person may be ruled out on the grounds of internal inconsistency; but among
the surviving class we can only discriminate on grounds of consistency with
other radically uncertain beliefs.
To make one relevant example: Rawls critizes utilitarianism on the
ground that its associated theory of the person is defective. It treats persons as mere receptacles
of happiness rather than morally significant units. “Utilitarianism does not take seriously the distinction
between persons.” (TJ, p. 27) But
a committed utilitarian may be as little persuaded by Rawls’s theory of the
person as he is by the full theory of justice. The arguments Rawls would give to support this background
theory seem neither stronger nor weaker than those available for the full
theory of justice. In this
instance, then, retreat to the background theory of the person represents no
justificatory gain.
To broaden the argument
in the way that Daniels proposes, therefore, seems not to represent an
epistemic gain, or a gain in the rational confidence we may have in the
correctness of a given meral theory.
This failure derives immediately from the fact that we have no greater
warrant in holding typical philosophical theories than we do in holding the
moral theories we are trying to justify.
What Daniels’ move does allow, however, is a wider basis on which to
achieve consensus among particular discussants. If it happens that the philosophical theories appealed to
are in fact accepted on all sides of a particular moral debate (whatever their
objective warrant), then this move may permit the gaining of consensus where
none was possible previously.
Consider discussants Smith and Jones who disagree over moral theories T
and T’. Suppose T and T’ are
equally consistent with all considered judgments shared by Smith and Jones, so
both discussants may continue to hold their distinct views without
inconsistency. But when we
introduce a philosophical theory P which Smith and Jones both accept, it may
emerge that P supports T and is inconsistent with T’. In this case Smith and Jones must come to agree on the correctness
of T, on pain of inconsistency.
Thus appeal to background philosophical theories may permit us to
achieve consensus over a moral theory; but (unless we have independent reasons
for believing these theories to be correct) such an appeal does not increase
the rational warrant we may attach to the moral theory.
Let us briefly consider
Daniels’ idea that the background theories in question should have independent
credibility. How shall we define the requirement of
independence here? Daniels chooses an analogy with natural science, in the
distinction between accidental generalizations and laws of nature. His criterion is that “we should
require that the background theories in (c) be more than reformulations of the
same set of considered moral judgments.
. . The background theories should have a
scope reaching beyond the range of the considered moral judgments used to test
t’he moral principles.”[19]
He thus
construes the background theories in analogy with theories of natural science:
each is tested by reference to concrete judgments (experimental findings and
considered moral judgments), and an accidental generalization is distinguished
from a real law in that the latter has test circumstances which are disjoint
from the class it is advanced to encompass. In the moral case, this criterion would require that the
background theories be supported by a set of considered moral judgments
disjoint from those currently under consideration (referred to as (a) in Daniels’
formulation). Daniels asserts that
Rawls’s background theories are independent in the sense described here. “His central level III theories of the
person and of the role of morality in society are probably not just
recharacterizations or systematizations of level I moral judgments.”[20]
This criterion seems to threaten collapse of Daniels’ three‑part account
of wide reflective equilibrium, however.
The class (a) ought ideally to include allavailable considered moral
judgments, in which case the class constraining (b) is identical with that constraining
(c). In this case there is no gain
in justificatory force.
These remarks suggest
that the method of wide reflective deliberatio.. does little to enhance the justificatory resources available
to moral theory. Rather, it relates
the problem of justifying moral theories to that of justifying philosophical
theories. This is at best a
lateral move, however, since philosophical theories are themselves notoriously
controversial, and the question of justification of such theories is equally
unresolved. To
what extent, then, do arguments based on the method of wide reflective
equilibrium serve to justify particular moral theories (that is, to provide
rational grounds for preferring one theory over another)? Only to the degree that
all discussants share (or can be brought to share) the philosophical theories
appealed to in the argument. The
justification offered here is relative to a set of background assumptions, with
the result that if these are not satisfied then the argument has no
justificatory grip.
Assessment. Let us take stock. To what extent does the process of
reflective deliberation provide a basis for comparative evaluation of
alternative moral theories? There are three cases to consider. First, against a given set of
considered judgments and background theories, two moral theories may be
evaluated in relation to their respective abilities to encompass these
beliefs. Here the coherence
argument has some bite, since it permits a ranking of competing theories in
terms of their fit with background beliefs. In the case in which all discussants share a significant
number of considered judgments and background theories, the method of
reflective deliberation can have justificatory force; it can provide reasons
for the discussants to prefer one theory over another.
Second,
it may be that two competing moral theories are equally consistent with the
background beliefs, but inconsistent with one another. In this case the rhethod ranks them
equally and provides no basis for choosing between them. This possibility seems to be part and
parcel of the coherence epistemology.
This is Quine’s point about the underdetermination of physical theory by
empirical evidence.[21]
Given
an incoherence within a theoretical system, there will generally be alternative
and non‑equivalent ways of resolving the contradiction. So we may suppose that
the process of deliberation might have led to a family of non‑equivalent
moral theories. Each is a coherent
system given the totality of background beliefs, and consequently the coherence
epistemology offers no basis for choosing between them. This observation suggests a strong
conclusion: indeterminacy of moral theory given all possible background
beliefs. If this is a real
possibility, then it is possible for all discussants to share the bulk of their
background beliefs and considered judgments, but still disagree irresolvably
about moral theory. This
possibility undercuts the hope of finding a uniquely best moral theory, and
hence the idea of moral truth.
Third,
the method provides no basis at all for comparing moral theories deriving from
different sets of background beliefs.
If two persons have irresolvable disagreements at the level of their
background beliefs, then each will find a reflective equilibrium around
different moral theories, and the method will be unable to rank these
alternatives. This
possibility suggests that Rawls’s system of argument has strongly relativist
implications: the correct moral theory depends upon the background beliefs with
which one begins, and these may differ in important ways from one community to
another.
Thus
the method of reflective deliberation is a weak form of justification at
best. This is so because the
requirement of consistency upon which it depends is itself a weak constraint,
and because the constraining beliefs involved in the process are themselves
highly uncertain. We may well
imagine that there are alternative moral theories which are equally good on
coherence grounds as reconstructions of a fixed set of background beliefs. And among these alternatives the method
of reflective deliberation provides no ranking whatsoever. More radically, the
method gives no way to begin to resolve moral disagreements among discussants
having substantially different starting beliefs and judgments.
We
concluded above that the coherence epistemology itself is equally appropriate
in ethics and science. We have now
seen, however, that there is a crucial difference in the epistemic standing of
science and moral philosophy. This
difference derives from the availability of a source of independent control in
the coherence argument. A chief
part of the objectivity of science stems from the relative independence of
perception from background theory.
When we add this feature to the process of seeking a coherent system of
belief, we get a process of knowledge‑formation which is arguably
objective and truth‑attaining.
But arguments offered above suggest that there are no analogous controls
in moral reasoning. Considered
judgments and general philosophical theories seem like the only available
candidates for such an independent control; but neither will perform that
function. Considered judgments
cannot perform this function because they derive from our background moral
theory. And background philosophical
theories are even less suited to function as an independent control on theory
formation, since they are themselves only weakly controlled by rational
argument.
There
is thus a crucial disanalogy between the two areas: empirical theory is constrained
by evidence which is in a rough and ready sense independent of the theoretical
conclusions we eventually arrive at, whereas moral deliberation is not. But if there is no class of beliefs
which have an antecedent credibility independent of their relations to other
statements, then how does any epistemic warrant enter the system of moral
deliberation? We must conclude that the method of reflective deliberation
provides only a very weak scheme of justification; it depends heavily and
unavoidably on the particular considered judgments and background beliefs with
which we begin, and these are themselves unjustified. The most this process can do is to bring to the surface the
principles which underlie the moral practices of the existing community, and
subject those practices to a limited form of rational criticism.
This
low estimate of the epistemic force of the method of reflective deliberation
may help to explain what is otherwise an extraordinary fact about Rawls’s
“Kantian Constructivism in Moral Theory”:[22]
the fact that this essay turns sharply away from the idea of moral
epistemology. In these lectures
Rawls maintains that “tine ‘real task’ of justifying a conception of justice is
not primarily an epistemological problem.
The search for reasonable grounds for reaching agreement rooted in our
conception of ourselves and in our relation to society replaces the search for
truth.”[23]
For the
process of reflective deliberation we have analyzed here provides a much
better basis for the problem of consensus‑building than it does for
establishing correctness. The fact
that the process of deliberation depends upon shared background beliefs has an
obvious political counterpart: it will only be possible to build a democratic
society if there is a substantial degree of agreement about basic values and
principles. Within the context of
such agreement, the process of reflective deliberation provides an attractive
model of a process through which agreement might be reached.[24]
[1] Kant provides an example of a highly objectivist moral epistemology: “[Everyone] must concede that the ground of obligation here must not be sought in the nature of man or in the circumstances in which he is placed, but sought a priori solely in the concepts of pure reason.” (Foundations of the Metaphysics of Morals (Bobbs‑Merrill, 1959, p. 5) “From what has been said it is clear that all moral concepts have their seat and origin entirely a priori in reason.” (ibid., p. 28) “[Moral laws] are valid as laws only insofar as they can be seen to have an a priori basis and to be necessary.” (The Metaphysical Principles of Virtue (Bobbs‑Merrill, 1964), p. 14) Rawls’s recent Dewey lectures (“Kantian Constructivism in Moral Theory,” Journal of Philosophy 77:9 (1980)) present a view which depends upon a consensus‑building conception of justification, which will be discussed briefly in the final paragraph of this article.
[2] A Theory of Justice (Cambridge: Harvard, 1971) (referred to as TJ in the text).
[3] “One should not be misled, then, by the somewhat unusual conditions which characterize the original position. The idea here is simply to make vivid to ourselves the restrictions that it seems reasonable to impose on arguments for principles of justice, and therefore on these principles themselves.” ( TJ, p. 18)
[4] In treating Rawls’s theory of justification I will not discuss his use of rational choice theory. This reasoning might seem to justify the theory of justice directly by showing that it is “the only choice consistent with the full description of the original position.” Rawls describes this style of argument as “a kind of moral geometry.”(TJ, p. 121) However, this application of rational choice theory is embedded within the assumptions defining the original position. These assumptions must themselves be justified on moral grounds, and their justification depends upon the method of reflective deliberation. It thus seems fair to conclude that the use of rational choice theory plays only a minor role in the justificatory argument.
[5] Ronald Dworkin, “The Original Position” in Reading Rawls (ed. Norman Daniels) (New York: Basic Books, 1977), pp. 16‑53.
[6] David Lyons, “The Nature and Soundness of the Contract and Coherence Arguments” in Reading Rawls, pp. 141‑167.
[7] Reading Rawls, p. 146.
[8] Quine’s treatment of the coherence methodology may be found in “Two Dogmas of Empiricism” (From a Logical Point of View 2nd ed. (New York: Harper, 1963), pp. 42ff., and Word and Object (Cambridge: MIT Press, 1960), Chapter One.
[9] I will reach something like Lyons’ conclusion below; but the reasoning will not depend on weaknesses of the coherence epistemology generally, but rather the uncertainty of the chief forms of belief which constitute the content of such arguments in moral philosophy.
[10] C.F. Delaney discusses the relation between Rawls’s theory of justification and a general coherence theory of epistemology in “Rawls on Method,” The Canadian Journal of Philosophy Supp. v. III (1977), pp. 157ff.
[11] “The Independence of Moral Theory,” Proceedings of The American Philosophical Association v. 48 (Nov. 1975), p. 7.
[12] “We thus make a fundamental distinction between competence (the speaker‑hearer’s knowledge of his language) and performance (the actual use of language in concrete situations).... The problem for the linguist ... is to determine from the data of performance the underlying system of rules that has been mastered by the speaker‑hearer. . . . A grammar of a language purports to be a description of the ideal speaker‑hearer’s intrinsic competence.” Noam Chomsky, Aspects of the Theory of Syntax (Cambridge: MIT, 1965), p. 4.
[13] For further discussion of the linguistic analogy see Norman Daniels’ “On Some Methods of Ethics and Linguistics,” Philosophical Studies 37 (1980), pp. 21‑36.
[14] W. H. Newton‑Smith gives a sophisticated defense of a rationalistic interpretation of scientific knowledge in his recent The Rationality of Science(Boston: Routledge & Kegan Paul, 1981). A crucial part of his argument hinges on his defense of the relative fixity of observation through change of theory. “We must not make the fallacious inference that, as any observation report no matter how well‑corroborated may be rejected by appeal to theory, all observation reports might be rejected by appeal to theory. For we can reject particular reports involving O‑terms only if we have strong evidence (which will rely on other observational reports) for a theory.... While we cannot have absolute faith in any particular reports or a particular range of reports, we are entitled to have general faith in the low‑level O‑reports we are inclined to make.” (p. 28) See his detailed criticisms of Kuhn and Feyerabend in chapters 5‑7.
[15] This conclusion is in line with Rawls’s view that an important part of moral philosophy is the description of alternative moral conceptions. Thus “one may think of moral philosophy at first . . . as the attempt to describe our moral capacity. “ ( TJ, p. 46)
[16] Norman Daniels, “Wide Reflective Equilibrium and Theory Acceptance in Ethics,” The Journal of Philosophy (1979), p. 258. (For a similar treatment of Daniels’ view see also his “Reflective Equilibrium and Archimedean Points,” Canadian Journal of Philosophy 10:1 (March 1980).)
[17] Daniels, “Wide Reflective Equilibrium and Theory Acceptance in Ethics,” p. 262.
[18] Rawls, “The Independence of Moral Theory,” p. 5.
[19] Daniels, “Wide Reflective Equilibrium and Theory Acceptance in Ethics,” p. 259.
[20] Daniels, “Wide Reflective Equilibrium and Theory Acceptance in Ethics,” pp. 260‑261.
[21] Quine holds that theoretical beliefs are in principle underdetermined by empirical evidence. “If by some oracle the physicist could identify outright all the truths that can be said in common‑sense terms about ordinary things, still his separation of statements about molecules into true and false would remain largely unsettled.... Conceivably the truths about molecules are only partially determined by any ideal organon of scientific method plus all the truths that can tee said in common‑sense terms about ordinary things.” Word and Object, pp. 21‑22. This underdetermination follows from the fact that it is possible to adjust a set of theoretical beliefs in non‑equivalent ways in order to bring the system into harmony with experience. “The total field [of beliefs] is so underdetermined by its boundary conditions, experience, that there is much latitude of choice as to what statements to reevaluate in the light of any single contrary experience. “ (“Two Dogmas of Empiricism,” p. 42.)
[22] Rawls, “Kantian Constructivism in Moral Theory,” pp. 515‑572.
[23] Rawls, “Kantian Constructivism in Moral Theory,” pp. 518‑19.
[24] An earlier version of this paper was presented to the New York State Philosophical Association (the Creighton Club) in October, 1981. I gratefully acknowledge the useful criticisms received on that occasion.