2. The Fallibilism of Kant’s Architectonic Gabriele Gava 1. Introduction At various points in the Critique of Pure Reason, Kant claims that the arguments and proofs he offers in his inquiry must be able to provide apodictic certainty. He maintains that this certainty is immediately connected with the necessary status of the propositions he defends and with the a priori justification he offers. Taking these contentions into consideration, it is easy to conclude that Kant attributed an infallible status to his philosophical claims. This seems to be confirmed from this passage contained in the preface to the first edition of the Critique: As far as certainty is concerned, I have myself pronounced the judgment that in this kind of inquiry it is in no way allowed to hold opinions, and that anything that even looks like an hypothesis is a forbidden commodity, which should not be put up 70
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
2. The Fallibilism of Kant’s Architectonic
Gabriele Gava
1. Introduction
At various points in the Critique of Pure Reason, Kant claims that the
arguments and proofs he offers in his inquiry must be able to
provide apodictic certainty. He maintains that this certainty is
immediately connected with the necessary status of the
propositions he defends and with the a priori justification he
offers. Taking these contentions into consideration, it is easy
to conclude that Kant attributed an infallible status to his
philosophical claims. This seems to be confirmed from this
passage contained in the preface to the first edition of the
Critique:
As far as certainty is concerned, I have myself pronounced the
judgment that in this kind of inquiry it is in no way allowed to
hold opinions, and that anything that even looks like an
hypothesis is a forbidden commodity, which should not be put up
70
for sale even at the lowest price but must be confiscated as soon
as it is discovered. For every cognition that is supposed to be
certain a priori proclaims that it wants to be held for absolutely
necessary, and even more is this true of a determination of all
pure cognitions a priori, which is to be the standard and thus even
the example of all apodictic (philosophical) certainty. (CPR A
xv)1
In this passage, Kant regards the claims to a-priority,
necessity, and certainty as inescapably connected. Insofar as he
argued that we have a priori cognitions that are necessarily valid
in possible experience, it is thus quite unsurprising that he
also maintains that his arguments for sustaining this view are
certain and infallible.2 In this respect, Kant was a man of his
time and for him claiming a-priority and necessity implied 1 References to Kant’s work will be given according to the standard edition (1900-), using the abbreviation KGS and indicating volume and page number. References to the Critique of Pure Reason (CPR) will use A and B to refer respectively to the paging of the first and the second original editions. English translations are given according to the Cambridge Edition of the Works of Immanuel Kant.2 Kant famously maintains that the identification of necessary propositions can only rest on a priori cognitions (CPR B 3-4). In the Critique of Pure Reason he also stresses that a priori cognitions of necessary propositions are the only basis of certainty (CPR B 5), while in the Jäsche Logic he concedes that we can also have empirical certainty (KGS 9:70-1). In both contexts, however, he maintains that a claim to a priori knowledge about necessary propositions requires certainty in order to be justified.
71
claiming certainty and infallibility.3 These aspects of Kant’s
thought are evidently in contrast to the fallibilism advocated by
all the pragmatists. Accordingly, they have cast serious doubts
on the appropriateness of reading pragmatism as developing some
Kantian ideas, or, alternatively, on the possibility of taking
Kant as anticipating pragmatist insights.
Nowadays, the entailment relationship among claims to a-
priority, necessity, and certainty (or infallibility) has been
put into question,4 and the plausibility of fallible a priori
justification (claiming necessity) has been defended.5
Accordingly, it has been argued that it is not contradictory to
regard some claims about the necessary status of some
propositions as both a priori justified and fallible. This is so
because we can identify cases of valid a priori justification which
3 In a similar way Watkins and Willaschek (unpublished manuscript) and Pasternack (2014) maintain that Kant endorsed an infallibilist account of justified claims to knowledge (not limited to a priori claims). They argue that for Kant justified claims to knowledge entail the consciousness of the truth of the justified claim. Watkins and Willaschek also argue that this account ofjustification is compatible with a form of fallibilism, according to which we can be wrong about the grounds we have. It is possible that we believe ourselves to have a ground when in fact we do not. However, in this case we would still know that if we had that ground, we would have known with certainty the proposition which would have been supported by that ground. 4 See for example: Haack 1979, Warenski 2009.5 The most relevant defenses of fallible a priori justification are: Bonjour 1998, Casullo 2003.
72
do not exclude the possibility of our being mistaken about the
necessary truth of the propositions in question. In a similar
way, I will here argue that if we closely consider Kant’s account
of reason and its powers, as well as his account of philosophy
and philosophical argument, it makes much more sense to see Kant
as proposing a strong (at least in his own eyes), but fallible,
defense of the a priori and necessary status of some synthetic
claims about experience. Whilst thus Kant, unlike many, but not
all, pragmatists, certainly considers it possible to obtain a
priori knowledge about necessary propositions, his views about
reason and philosophy suggest that he could coherently associate
a fallibilist account of a priori justification for claims to
knowledge with his contentions about the possibility of a priori
knowledge. This fallibilism concedes that there are cases in
which we could be justified in claiming to have a priori knowledge,
even though we in fact have not it. If this is true, there might
be much more continuity between Kant’s views on justification and
those of the pragmatists.
This interpretation of Kant might strike some as implausible
regarding Kant, and inconsistent in its own right. As a quick
73
answer to these worries, I offer two brief remarks. First, Kant
is clear in stressing that our reason has a natural tendency to
deceive itself with respect to its powers of obtaining necessary
and a priori truths concerning transcendent objects. Accordingly,
Kant claims that human reason “is already dialectical on account
of the tendency of its nature” (CPR A 849/B 877).6 This account
of reason seems to require a more modest perspective on its
contentions. Second, the compatibility of fallibilism with claims
to a-priority and necessity can be made a bit less implausible if
we consider for a moment mathematics and geometry. We normally
regard mathematical claims as being a priori justified claims
concerning the necessary status of some mathematical
propositions. It is however a historical fact that claims in
mathematics and geometry are fallible. Maintaining that the sum
6 Of course one can argue that Kant, in the Critique of Pure Reason, identifies the philosophical tools that are required in order to avoid this self-deception and thus be certain, and apodictically certain, about the claims we advance through reason. However, reason is evidently non transparent to itself, and this seems to agree with the recognition of the possibility, no matter how slight it might be, of being mistaken. Westphal (2004) argues at various points that Kant endorsed a fallibilist account of justification, while he links this position to an externalist perspective. By contrast, in this paper I will attribute to Kant an internalist view on justification. Since Westphal bases his interpretation of Kant’s fallibilism on other passages than those here taken into consideration, it would be interesting to see how all these passages should be read together. This seems to be particularly appropriate ifwe want to avoid attributing to Kant two opposed theories of justification.
74
of the angles of a triangle is necessarily 180° could be
considered justified, and indeed justified a priori, in a context
in which non-Euclidean geometry was still yet to come.7 The claim
that the proposition was true could be regarded as a priori
justified even though the proposition is false. This is just to
give at least provisional plausibility to the idea that one can
coherently maintain that an argument for the necessary character
of some propositions is both a priori justified and fallible.
In this paper I will defend the view that we can find some
hints that Kant allowed room for fallibilism concerning his
philosophical edifice, his architectonic of reason, even though
he certainly wanted to identify necessary and a priori justified
principles.8 When Kant explicitly addresses the status of his
philosophical claims he maintains that they are necessarily true,
7 It seems also reasonable to maintain that this contention was justified in that context even if we consider the issue from our own perspective, that is, a perspective in which we know that the contention is false. An alternative would be to consider the contention unjustified, both in our historical context and before the development of non-Euclidean geometries. This would setthe bar for a priori justification very high and would render it very difficult to find criteria for determining if we are in fact justified in claiming something or if it just seems that we are justified.8 In Gava 2011 I have argued, from a different perspective, that transcendental philosophy can endorse fallibilism. In that context, I maintained that when we claim that a proposition is a priori necessary, we must not claim that it is in principle indubitable. By contrast, a claim to infallibility requires exactly this latter contention.
75
that they are a priori justified, and that they are consequently
absolutely certain. This seems to be based on a common assumption
that Kant shared with his rationalist predecessors. For him, a
claim to knowledge, and especially to rational knowledge,
requires certainty (KGS 9:66). Still, if we have to make sense of
Kant’s account of reason as a fallible faculty, and of his
identification of philosophy with an archetype (Urbild) which we
have not reached yet (CPR A 838/B 866), it seems plausible to
read his arguments as allowing room for fallibilism, while
claiming to provide a strong a priori justification for supposedly
necessary propositions. I will begin in section 2 by introducing
Kant’s criteria for justified claims to knowledge, especially as
they are presented in the section “On Having Opinions, Knowing,
and Believing” of the first Critique (CPR A 820-31/B 848-59). In
section 3, I will then illustrate Kant’s modest account of reason
and philosophy, giving particular attention to the “Discipline of
Pure Reason” and the “Architectonic of Pure Reason.” In section 4
I will identify what I see as a sustainable account of a priori
fallible justification for claims to knowledge. To finish, in
section 5 I will develop a more modest criterion for the
76
justification of our claims to knowledge starting from Kant’s
account of conviction (Überzeugung).
Before moving to the next section, let me make some
preliminary distinctions. Throughout this paper I will use the
form “justified claims to knowledge” rather than the form
“justified beliefs,” for two main reasons. First, this is meant
to avoid confusion in my discussion of Kant’s concept of belief,
which, according to him, is a justified assent that cannot be
taken as a claim to knowledge. The notion of belief currently
used in philosophical inquiry seems to be much broader than
Kant’s and include claims to knowledge as a special case.
Therefore, second, the reference to “justified claims to
knowledge” is intended to make clear that the main interest of
this paper is epistemic justification. I follow Chignell (2007,
34) in defining epistemic justification as identifying the
conditions of rational acceptability for assents that are
candidates for knowledge and thus at least claim to be knowledge.
On the other hand, non-epistemic justification identifies the
conditions of rational acceptability for assents that are not
candidates for knowledge.
77
2. Justifying Claims to Knowledge
In the section of the “Canon of Pure Reason” entitled “On Having
Opinions, Knowing, and Believing” (CPR A 820-31/B 848-59), Kant
identifies knowledge as a kind of assent or taking-to-be-true
(Fürwahrhalten). Kant defines assent in general as “the subjective
validity of judgment” (CPR A 822/B 850), meaning by that the
propositional attitude we have with respect to a particular
proposition. He then identifies three stages or kinds of assent
which are: having an opinion, believing, and knowing. Having an
opinion is an assent that is both subjectively and objectively
insufficient, while believing is only subjectively sufficient,
and knowing is both subjectively and objectively sufficient (CPR
A 822/B 850). What does Kant mean by that? What are subjective
and objective sufficiency? What does this imply for the criteria
used to decide if a claim to knowledge is justified or not?
It seems plausible to read subjective sufficiency as
identifying some grounds that are sufficient to produce the
acceptance of a proposition in a particular subject.9 These
9 For a similar reading see: Pasternack 2011, 202; Watkins and Willaschek (unpublished manuscript).
78
subjective grounds can be of different types. They can simply be
what we take to be objective grounds for our assent,10 or they can
be practical grounds that do not aim at objective validity,11 but
that can nonetheless be relevant grounds of assent for non-
epistemic justification. On the other hand, objective sufficiency
identifies intersubjectively valid grounds,12 for which we are
conscious that they entail the truth of the proposition we assent
to.13 A sufficient subjective ground is thus a ground that is 10 This seems to be the relevant sense for the kinds of assent Kant calls opinion (Meinung) and knowledge (Wissen), where thus subjective sufficiency seems to refer mainly to the psychological state of the subject.11 This sense is relevant for belief (Glaube). In this case subjective sufficiency has both a psychological and a normative component, insofar as thesubjective grounds are not simply those that we take to be objective, but are grounds that have justificatory force from a practical point of view. Accordingly, both Chignell (2007, 49, 53) and Stevenson (2003, 87-8) argue that Kant uses two very different understandings of sufficient subjective grounds for opinion and knowledge, on the one hand, and for belief, on the other. This is certainly true. However, these two different understandings canbe read as two cases of a broader category, that is, as two cases in which a ground is sufficient to produce an assent in a particular subject. 12 “If it [the assent, or taking-to-be-true, my note] is valid for everyone merely as long as he has reason, then its ground is objectively sufficient” (CPR A 820/B 848).13 Accordingly, in the Jäsche Logic Kant identifies knowledge with certainty (KGS9:70), and he claims that certainty “is combined with consciousness of necessity,” while uncertainty “is combined with consciousness of contingency or the possibility of the opposite” (KGS 9:66). Given that Kant will later argue that we can have certainty on empirical propositions, here necessity andcontingency should not be taken as expressing characteristics of the propositions themselves, but rather of the relationship between our grounds and the propositions they are supposed to justify. Thus, if our grounds make certain that a contingent proposition is true, our assent can be combined withthe consciousness of the impossibility of the opposite, even though the proposition in question is in itself empirical and contingent, and thus, from a metaphysical perspective, the opposite would be possible. Saying that the
79
sufficient to produce the acceptance of a proposition from the
point of view of a particular subject,14 while an objective ground
is a ground that not only can be intersubjectively recognized as
a valid ground, but also warrants (i.e. provides certainty) that
the proposition is true.
According to this description of subjective and objective
grounds, having an opinion is a doxastic state where I do not
have grounds that are sufficient either to produce subjective
acceptance, or to provide an intersubjectively valid epistemic
justification for acceptance. This does not mean that in
opinions we do not have any grounds at all, but only that they
are not sufficient to produce acceptance or justify a claim to
knowledge. When I have an opinion, I recognize that the objective
grounds that I have are sufficient to make the truth of a
proposition more probable than that of its opposite, but are
grounds entail the truth of a proposition should thus be taken as applicable to the justification of both contingent and necessary propositions. With reference to the passage quoted in this note, it should also be highlighted that Kant’s emphasis on consciousness points toward an internalist account of justification.14 An assent based on sufficient subjective grounds can be either justified ornot. It is not justified if these grounds are only grounds that we take to be objectively sufficient. This would be a case of persuasion (Überredung) for Kant. It is justified if our subjective grounds are sufficient to produce acceptance and, at the same time, we recognize that these grounds are not candidates for objective sufficiency.
80
insufficient to epistemically justify acceptance of the former.
On the other hand, believing is a state where I have sufficient
subjective grounds, that is, I have grounds that are able to
produce an acceptance of the proposition in question when I
consider the matter from my own perspective, but I have not
sufficient objective grounds and I am conscious of that. This
means that a subjective ground which is a ground that I take to
be objective and is not cannot be a relevant ground in a state of
belief, insofar as for Kant in this state I must be able to
recognize that my ground is not a candidate for objective
sufficiency. A subjective ground that I take to be an objective
one and is not would generate persuasion (Überredung), and not
belief (CPR A 820/B 848).15 Since it consciously lacks objective
grounds, the justification of beliefs is therefore non-epistemic.
Finally, knowledge is an assent that is able to produce
acceptance (and has thus sufficient subjective grounds), and is
based on objective grounds that can be intersubjectively
15 Recall that Kant’s concept of belief is not our own. Belief is not simply the attitude of considering something to be true, but, roughly, can be equatedwith a justified belief that is not a candidate for epistemic or theoretical justification. It would probably make much more sense to connect the contemporary concept of belief to Kant’s concept of assent or taking-to-be-true. See: Chignell 2007 and Stevenson 2003.
81
recognized and that warrant the truth of the proposition we
assent to, while we are conscious that they do so.
As is evident from this reconstruction, Kant approaches the
concept of knowledge by means of the kind of justification a
claim to knowledge requires, where the justification that is
needed must be able to warrant the truth of the proposition in
question and we must be conscious of this ability. This means
that Kant’s account of justification for claims to knowledge is
pretty demanding and requires infallibility. In order to be
justified in claiming to know that p, we must possess grounds
that warrant the truth of p and we must be conscious of that.
Accordingly, following Kant, we cannot have cases of justified
claims to knowledge which do not entail knowledge. As stated in
the Jäsche Logic, this requirement for justified claims to knowledge
applies to both a posteriori and a priori inquiries and Kant recognizes
claims to knowledge that can be justified by either empirical or
rational grounds (KGS 9:70-1).16 Here, the fact that, for Kant, we
16 If both a posteriori and a priori justification require certainty, it is a priori justification that is the best candidate for reaching this latter standard. Thus, when it is possible to choose between rational and empirical grounds to sustain our claims to knowledge, we should always prefer the former: “[w]e cannot have rational certainty of everything, but where we can have it, we must put it before empirical certainty” (KGS 9:71).
82
can only recognize the truth of contingent propositions on
empirical grounds should not generate confusion. The
infallibility of a claim to knowledge concerns the relationship
of our grounds with the truth of the propositions they are
intended to support and does not affect the contingent or
necessary status of the proposition itself. Thus, given the
grounds that I have, I might be in a position to be certain that
a particular proposition about a contingent fact is true, even
though, from a metaphysical perspective, the proposition in
question is not necessarily true, insofar as what it expresses is
not necessarily the case. Therefore, unlike what Kant suggests in
the first Critique, in his Lectures on Logic he acknowledges that
certainty is not necessarily only associated with a-priority and
necessity, but can be reached also in the case of the
justification of claims to knowledge concerning contingent
empirical propositions.17
17 If we want to apply the distinction between necessity and contingency to the discussion of fallibility and infallibility, we can say that in the case of infallibility the truth of the claim we intend to justify is necessarily implied by our grounds, be they empirical or rational. By contrast, in the case of fallibility, the truth of the claim in question remains “contingent” given the grounds we have. Obviously, this is quite different from saying thatfallibility and infallibility correspond to the contingent or necessary character of the propositions themselves.
83
However, if both a priori and a posteriori justified claims to
knowledge require infallibility, Kant seems to think that in
cases of a priori epistemic justification no other form of assent
can be justified than knowledge. This means that we are either
able to justify a claim to knowledge, or we cannot have any other
form of doxastic assent, like opinion, or belief.18 This is clear
for example in the following sentence: “[i]n judging from pure
reason, to have an opinion is not allowed at all. For since it
will not be supported on grounds of experience, but everything
that is necessary should be cognized a priori, the principle of
connection requires universality and necessity, thus complete
certainty, otherwise no guidance to the truth is forthcoming at
all” (CPR A 822-3/B 850-1).
From this brief reconstruction, it is evident why Kant,
according to his own account of a priori justification, described
the claims he advanced in his philosophy as being infallible.
Accordingly, when he explicitly addresses the status of the
18 Kant argues that both opinion and belief are inadequate in inquiries aimingat a priori knowledge. Opinion is inadequate because a priori grounds must be able to show that a proposition is necessary. Kant seems here to confuse the necessity of the proposition itself with the necessity with which it should follow from our grounds. On the other hand, belief, in its Kantian sense, is astate of assent that is misplaced in inquiries aiming at knowledge.
84
contentions in his theoretical philosophy, he maintains that they
provide an infallible a priori epistemic justification of some
necessary truths (concerning mathematical and empirical
cognitions). On the other hand, it seems that he regarded the
claims in his practical philosophy as involving two kinds of
infallible justification: an epistemic infallible a priori
justification of necessary moral principles,19 and a non-
epistemic, practical, but infallible, a priori justification of
claims concerning God and the soul (CPR A 823ff./B 851ff.). In
this latter kind of justification, our aim is a state of belief
based on moral grounds, and not knowledge.20 In the following I
will focus on the kind of justification required for claims to a
priori knowledge. I will thus avoid considering the kind of
19 “It is just the same with the principles of morality, since one must not venture an action on the mere opinion that something is allowed, but must knowthis” (CPR A 823/B 851).20 The infallibility of moral beliefs seems to be compatible with regarding the belief as possibly false from a theoretical perspective. Accordingly, in the Jäsche Logic Kant identifies belief as a subclass of uncertain assent, where we have “consciousness of the contingency or the possibility of the opposite” (KGS 9:66). However, belief is uncertain only from the point of view of epistemic justification based on objective grounds, where I am at the same time certain that the belief cannot be proved either true or false on the basis of those grounds. From the point of view of practical, non-epistemic justification, belief provides certainty, “which is necessary only in a practical respect given a priori, hence an assent of what I accept on moral grounds, and in such a way that I am certain that the opposite can never be proved” (KGS 9:67).
85
justification Kant attributes to his claims to belief in his
practical philosophy. In this respect, it is clear that when Kant
explicitly considers the status of the claims he advances in his
theoretical philosophy he maintains that they are apodictically
certain and infallible.
3. Kant’s Fallibilism in the “Discipline” and the “Architectonic”
Kant’s ascription of certainty and infallibility to his claims to
a priori knowledge seems to contrast with his modest description of
reason and his modest account of the capacity to build
philosophical systems. I will address these two issues in turn.
Kant’s account of reason is modest in at least two important
respects:21 1) reason is a faculty that is not transparent to
itself, and the claims that it can justify are not immediately
evident; 2) reason is described as a faculty that we have in
common, as something requiring intersubjective agreement.
As to the first point, a central theme in the
“Transcendental Dialectic” and the “Discipline of Pure Reason,” a21 For a modest reading of Kant’s conception of reason see: O’Neill 1992. By “reason,” I here refer to Kant’s broad concept of reason, the one referring toall our higher faculties of cognition and including reason in the narrow sense, as the faculty of inference, and understanding (cf. CPR B 169, A 835/B 863).
86
theme that is relevant for understanding the whole Critique of Pure
Reason, is that human reason has a natural tendency to develop
illusory transcendent ideas, which, while giving the impression
of being necessarily valid and indubitable, are in fact sources
of deception and insoluble conflicts in speculative metaphysics.
From this point of view, reason presents us “an entire system of
delusions and deceptions” (CPR A 711/B 739). This is why we need
a critique of our capacity to cognize objects through reason and
ultimately a discipline which erects “a system of caution and
self-examination out of the nature of reason and the objects of
its pure use, before which no false sophistical illusion can
stand up but must rather immediately betray itself” (CPR A 711/B
739). Moreover, even if we consider the valid cognitions that are
justifiable through theoretical reason, as mathematical
cognitions and a priori cognitions concerning experience are, the
way in which they are so justifiable is not immediately clear.
This does not mean that we should doubt the validity of
mathematical cognitions or of a priori cognitions concerning
possible experience. By contrast, their validity can be taken for
granted for Kant (CPR A 710-1/B 738-9). However, what is not
87
immediately clear are the sources that lie at the basis of these
cognitions.22
As to the second point, in the “Discipline of Pure Reason”
Kant maintains that reason “has no dictatorial authority.”
Accordingly, a claim of reason “is never anything more than the
agreement of free citizens, each of whom must be able to express
his reservations, indeed even his veto, without holding back”
(CPR A 738-9/B 766-7). Reason is not a private faculty by means
of which I can easily obtain an evident and certain justification
of my claims to knowledge. On the contrary, reason is essentially
a public affair and when I follow its commands I need to seek the
agreement of others, just as to listen to their objections and
counterproposals. “This lies already in the original right of
human reason, which recognizes no other judge than universal
human reason itself, in which everyone has a voice; and since all
improvement of which our condition is capable must come from
this, such a right is holy, and must not be curtailed” (CPR A
22 In both cases our capacity to have a priori cognitions rests on the possibility to connect a priori concepts and intuitions, even if relevant differences are at stake, as it is expressed in the Kantian distinction between rational cognitions from the construction of concepts and rational cognitions from concepts (cf. CPR A 713ff./B 742ff.).
88
752/B 780).23 The lack of self-transparency and the public
dimension of reason contrast with the absolute certainty and
infallibility that Kant ascribes to claims to a priori knowledge by
reason. If reason can be deceptive and if we must test the
validity of our claims in a domain that is essentially public,
the self-ascription of absolute certainty for our claims seems to
be misplaced.
We can find a confirmation that this is the case if we
consider for a moment Kant’s modest account of philosophy in the
“Architectonic of Pure Reason.” In this chapter, Kant first
identifies a general condition of scientificity, that is,
architectonic unity, and then he points out the way to attain
this standard in philosophy. He describes architectonic unity as
a collection of cognitions, where the relationships between the
parts is not the result of an accidental combination, but is
dependent on an idea which anticipates the form of the whole (CPR
A 832/B 860). This idea of the whole must be given a priori by
reason.24 However, even if the idea must be given a priori by
23 The recognition of this public character of reason is totally different from a form of relativism as O’Neill (1992, 305) emphasizes.24 It might seem odd to claim that the idea of the whole of a science should be given a priori by reason. After all the sciences are historical entities. As
89
reason, due to the lack of self-transparency of reason, the idea
which lies at the basis of a science often remains hidden and
unclear for a very long time, and in the meantime we arrange the
notions of the science in a way that does not reflect the
fundamental idea. In other words, we can be mistaken about the
way in which we represent a science and the fundamental
relationships among its doctrines.
Nobody attempts to establish a science without grounding it on an
idea. But in its elaboration the schema, indeed even the
definition of the science which is given right at the outset,
seldom corresponds to the idea; for this lies in reason like a
seed, all of whose parts still lie very involuted and are hardly
recognizable even under microscopic observation. (CPR A 834/B
862)
a remedy to this perceived oddity, one can notice that in this passage Kant isreferring to rational sciences, that is, to sciences based on rational a priori cognitions. Moreover, the a-priority of the idea should not be understood in astrict sense, as an idea that is identifiable independently of particular inquiries. Rather, the idea of the whole of a science can be interpreted as being a priori in the sense that, in considering the system of cognitions of a particular science, we should see the various parts of that system as occupying a place that follows from our very idea of the whole. That is, the idea of the whole must be considered as if it were prior to the parts. On thispoint see: Gava 2014.
90
Thus, reason can fail to recognize the fundamental structure of a
science which, in the end, is based on principles of reason that
are however hidden and unclear to reason itself. Philosophy, as a
particular discipline that seeks to attain scientificity, also
experiences this difficulty in recognizing its fundamental idea.
Accordingly, in the “Architectonic,” Kant presents the history of
philosophy as a series of failed attempts to clarify the idea on
which a coherent system of philosophy should be built. As a
remedy to this history of failures, Kant tries to identify the
idea which should be placed at the basis of philosophy, thus
allowing philosophy to become architectonic and scientific.
According to him, philosophy should be constructed following its
“cosmic concept” (CPR A 838-9/B 866-7), where the idea that gives
unity to the science is the orientation toward the “essential
ends,” and ultimately the “final end,” of human reason. These
ends are considered by moral philosophy (CPR A 838-40/B 866-8).
Establishing what philosophy according to its cosmic concept
exactly is, and determining what Kant considered the essential
ends of reason to be, is not an easy task, but it is not one that
we need to address here. What is relevant for our discussion is
91
that Kant regarded the idea of an ultimate and complete system of
philosophy based on a fundamental idea of reason as an
“archetype” (Urbild) (CPR A 838/B 866) which we must try to
approach, but which we cannot ever be sure to have achieved. Kant
seems to think that it is possible to match this archetype,25 but
even if we do so (or if we think we do), we cannot be totally
sure of our success and our views on the scientific character of
philosophy must continue to be subjected to the criticism of
public reason.
The need to submit our views on philosophy to the judgment
of public reason is clear if we analyze for a moment Kant’s
distinction between learning philosophy and learning to
philosophize. This distinction makes clear that even in the case
in which our views on philosophy in fact reflect the ultimate
system of philosophy, nonetheless we could never be completely
sure that this is the case and we should always submit those
views to the rational judgment of ourselves and other human
beings. Accordingly, Kant maintains that we cannot learn 25 “In this way philosophy is a mere idea of a possible science, which is nowhere given in concreto, but which one seeks to approach in various ways until the only footpath, much overgrown by sensibility, is discovered, and the hitherto unsuccessful ectype, so far as it has been granted to humans, is madeequal to the archetype” (CPR A 838/B 866).
92
philosophy, but we can only learn to philosophize (A 837/B 866),
meaning by that that we cannot approach philosophy as the
assimilation of a complete and definitive system of doctrines.26
To begin with, Kant claims that we cannot learn philosophy
because for him there was not a philosophy to learn yet, since,
at least according to his diagnosis of the philosophy of his
time, philosophy as an ultimate and coherent system of cognitions
based on an idea of reason was still only an archetype.27 However,
Kant claims that even if this archetype were achieved, the
appropriate way to approach the philosophical system which
realized it would be to submit it to the judgment of one’s own
reason. Thus, we would not have to learn the system as a static
set of doctrines, but we would have to approach it by
philosophizing on the views it presented, being able to
rationally recognize those views as ours. Philosophy can either
be learned “historically,” that is, as a given series of
teachings that are part of a particular philosophical system, or
26 On the distinction between learning philosophy and learning to philosophizesee: Hinske 1998, ch. 3; La Rocca 2003, 225-7, 235-42.27 “Until then [that is, until we have matched the archetype, my note] one cannot learn any philosophy; for where is it, who has possession of it, and bywhat can it be recognized? One can only learn to philosophize” (A 838 B 866).
93
“rationally,” that is, as a system of principles and concepts
that we recognize as valid after we have subjected it to the
judgment of our own reason (CPR A 835-7/B 863-5). The right way
to do philosophy however is to do it rationally, and, from this
perspective, even if there were an ultimate and complete system
of philosophy, we would always need to subject it to the
criticism of our own reason.28 In other words, to do philosophy
appropriately we must not apprehend a particular system (we must
not learn a philosophy), even if it were the ultimate and
definitive one, but we should always rationally evaluate the
views we are scrutinizing, trying to develop them further (we
must then learn to philosophize). This description of philosophy
and of the correct way of approaching its claims is strongly
related to Kant’s public account of reason. Accordingly, Kant
argues that it would be pretentious to claim that we are in
possession of the ultimate system of philosophy, the one that
everybody could learn as a definitive and immutable system. He
maintains: “[i]t would be very boastful to call oneself a
28 “But even granted that there were a philosophy actually at hand, no one wholearned it would be able to say that he was a philosopher, for subjectively his cognition of it would always be only historical” (KGS 9:25).
94
philosopher in this sense and to pretend to have equaled the
archetype, which lies only in the idea” (CPR A 839/B 867). This
seems to imply that our claims to knowledge in philosophy should
always be subjected to the scrutiny of public reason. Kant seems
to partially recognize the consequences of this account of reason
and philosophy for his criteria for justified claims to knowledge
when he stresses that “[w]ith knowledge one still listens to
opposed grounds” (KGS 9:72). This means that even when one is
justified in claiming knowledge, one should always submit one’s
views to the judgment of other human beings.
In following this reconstruction of Kant’s account of reason
and philosophy, it seems that Kant, when he explicitly considers
the justification required by claims to knowledge, sets the bar
too high for the powers he himself assigns to our reason. This
fact does not entail a skeptical result. On the contrary, it is
possible to develop a modest account of a priori justification
that, while maintaining that our a priori claims to knowledge can
be justified, leaves room for fallibilism concerning these
claims. This is what I will try to do in the next section. After
95
that, I will find some hints toward this fallible justification
in Kant’s concept of “conviction” (Überzeugung).
4. Fallible A Priori Justification
Kant’s reason is thus not transparent to itself and is
essentially public. At least at a first glance, it seems possible
and consistent to consider the justified claims advanced by this
reason as being fallible, even though these claims ascribe
necessity to some propositions through a priori justification. But
how should we understand fallibilism in this context?
By fallibilism I mean an attitude toward our justified
claims to knowledge which regards them as not excluding the
possibility that the claims in question are false and that our
justification for them is defeasible. By infalliblism I mean an
opposite attitude. That our claims to knowledge concerning
empirical objects can be considered justified and fallible at the
same time seems something that can easily be conceded. To turn a
Kantian example to our needs, we could consider the claim stating
that the sun rotates around the earth as justified in a context
in which Copernican astronomy was unavailable. However, the claim
96
was obviously fallible, insofar as it is false and it has been
defeated by better grounds of justification.
Maintaining that a claim to knowledge which identifies a
supposedly necessary proposition can be both justified a priori and
fallible seems to be more controversial. The fact that a priori
justification does not rest on empirical data and that it
normally claims necessity for the propositions it justifies makes
it quite easy to see a priori justification as necessarily
involving infallibility.29 The basic thought seems to be
expressible in the following question: how can we claim that a
proposition is a priori necessary if we cannot demonstrate that our
justification for this claim is infallible? This seems to be an
intuitive and common way to consider the relationship between
claims to a priori necessity and infallibility.
The thesis that a priori justification of claims concerning
necessary propositions entails infallibility involves the
association of two different senses of necessity: the necessity
of the proposition we claim to know and the necessary 29 The entailment relationship among claims to a-priority, necessity, and infallibility has been attacked from various perspectives. Saul Kripke (1980),for example, defended the possibility of contingent a priori propositions, whileBonjour (1998, chs. 1, 4) and Casullo (2003, ch. 3) have argued that a priori justified beliefs can be fallible and revisable.
97
relationship existing between our a priori justification and the
truth of the proposition in question. Accordingly, an
infallibilist account of a priori justification for claims to
knowledge would involve the following idea: we are a priori
justified in claiming that a proposition p is necessarily true
only if the truth of the necessary proposition is de facto entailed
by our a priori justification and we know it. This account of a priori
justification set the bar very high for a priori justified claims
to knowledge. According to this view, we have justification for a
priori claims to knowledge only if those claims are equivalent to
knowledge, where the knowledge of p is dependent on the knowledge
that p is de facto entailed by our a priori justification. But how
can we establish that there is this de facto necessary relationship
between our a priori justification and the truth of the necessary
proposition in need of justification? How can we determine that
we in fact have knowledge of the existence of this entailment
relationship and not only a fallible justification for claiming
to know that there is this entailment relationship? The answer
seems to be that we need a second order infallible justification
of the claim that our a priori justification entails the truth of
98
the necessary proposition in question.30 But can we really hope
for this second order infallible justification? If we consider
the refutation of basic claims of Euclidean geometry for example,
this infallible justification seems unavailable even in those
cases that seemed to be the best candidates for infallibility,
that is, mathematics and geometry. An a priori justification (for
example a mathematical demonstration) we regard as entailing the
truth of a proposition could simply contain errors that were
inaccessible to our considerations, or it could be based on
premises that turn out to be false. Alternatively, it could
simply be misleading. Accordingly, we cannot identify criteria
that infallibly indicate that the truth of a proposition
necessarily follows from our a priori justification, and this is
exactly the kind of second order infallible justification that
infallible accounts of a priori justification for claims to
knowledge need. If this is true, we cannot be absolutely certain
of knowing that the truth of a proposition is de facto entailed in 30 Alternatively, we could endorse an externalist position: in this case we would be justified in advancing a claim to knowledge only if our a priori justification de facto entailed the truth of the claim in question, without taking into consideration if we are conscious of this de facto relationship or not. This is of course a plausible proposal, but it does not seem to go in thedirection of Kant’s own views, insofar as Kant, at various points, claims thatwe must be conscious of the grounds we have.
99
our a priori justification for it. Maintaining that this kind of
justification is required for a priori claims to knowledge would
render it impossible to identify any claim that satisfies these
criteria. Moreover, this would have counterintuitive consequences
concerning those claims that we regard as a priori justified in
some historical contexts, but turned out to be false or
defeasible. If we recall the example I have used in the
introductory section, it seems reasonable to maintain that the
claim that the sum of the internal angles of a triangle is always
180° was justified before the development of non-Euclidean
geometries. If this is true, the claim was justified a priori, but
of course the justification did not entail the truth of the claim
in question and it was also defeasible.31
A more plausible account of a priori justification for claims
to knowledge is possible if we endorse a fallibilist standpoint.
From this point of view, we would regard differently the
supposedly necessary relationship existing between our a priori
justification and the truth of the necessary proposition in
31 One could argue that the claim was in fact unjustified, because the a priori grounds we had in fact did not entail the truth of the claim in question. However, as I have already suggested, this would render very hard the task of finding the criteria to decide if our a priori claims are justified or not.
100
question. Accordingly, a fallibilist account of a priori
justification for claims to knowledge does not require actual
knowledge, and more specifically actual infallible knowledge, of
an existing entailment relationship between our a priori
justification and the truth of the necessary proposition we aim
to justify. On the contrary, a priori justification for claims to
knowledge would only involve the following idea: we are a priori
justified to claim that a proposition p is necessarily true only
if our a priori justification can justifiably be regarded as entailing the
truth of the necessary proposition in question. According to this
perspective, we do not need to provide a second order infallible
justification of the claim that our a priori justification entails
the truth of the proposition in question. When we consider the
relationship between our a priori justification and the truth of
the proposition in question, we need only to be fallibly
justified in regarding the latter as entailed by the former. This
perspective is of course compatible with the absence of an actual
entailment relationship between our a priori justification and the
truth of the proposition in question.
101
But what are the criteria to decide if an a priori claim to
knowledge is fallibly justified or not? I have suggested that
criteria to identify valid a priori justifications are unavailable
if we endorse an infallibilist perspective. By contrast, we are
in a much better position with a fallibilist standpoint, insofar
as from this point of view we do not need to identify criteria
which guarantee that there is actual knowledge of a de facto
necessary relationship between our a priori justification and the
truth of the proposition in need of justification. On the
contrary, we only need to identify criteria which indicate that
we are fallibly justified in regarding our a priori justification
as entailing the truth of the proposition in question, while it
is in fact completely possible that this alleged entailment
relationship does not apply. Accordingly, we can be considered a
priori justified in claiming that a proposition is necessarily true
if we can reasonably expect our claim to be intersubjectively
recognized as valid. This fact highlights an important
characteristic of fallible a priori justification for claims to
knowledge: a priori fallible justification is historically
102
contextual.32 Ancient geometers were justified in claiming that
the sum of the internal angles of a triangle was always 180°, but
contemporary geometers are not. Accordingly, a claim can be
considered a priori justified if, in our historical context, we can
rationally expect our a priori justification for that claim to be
intersubjectively recognized as entailing the truth of the claim
in question. The rationality of this expectation is thus decided
by the recognizability of our justification in a particular
historical context, which decides what is the set of basic
knowledge and presuppositions which is required to a particular
subject. The relevance of the historical context for a priori
justification of claims to knowledge should not be confused with
the claim that a priori knowledge is historically contextual. When a
fallibly a priori justified claim to knowledge actually amounts to
knowledge, that is, when our justification for the claim not only
32 This historical contextualism might suggest that a priori justification is notat all a priori after all. Somebody could argue that if our justification is dependent on the kind of reasons that can be accepted as a priori sufficient forsustaining a claim to knowledge in a particular historical context, this meansthat those reasons are not in fact a priori. However, this historical contextualism simply acknowledges that the domain of a priori reasons is not immediately clear and transparent for us. By contrast, our access to these reasons is always perspectival and influenced by what we can expect to be considered a valid and sufficient a priori reason for a claim to knowledge in a particular historical context. In other words, the reasons and the grounds arenot historical in themselves, but our access to them is.
103
can be regarded as entailing the truth of the proposition we
claim to know in our historical context, but actually entails it,
our claim to a priori knowledge would amount to knowledge and would
have validity regardless of the historical context.
To summarize, an infallibilist account of a priori
justification for claims to knowledge requires not only that our
a priori justification for claiming that proposition p is
necessarily true de facto entails the truth of p and we know it,
but also that we have a second order infallible justification for
claiming that our a priori justification entails the truth of p. On
the contrary, a fallibilist account of a priori justification for
claims to knowledge only requires that the truth of p can justifiably
be regarded as entailed by our a priori justification. In this way, our
fallibly justified claim that there is such an entailment
relationship is compatible with its actual absence. The criterion
that decides if we are fallibly justified in claiming that our a
priori justification entails the truth of a supposedly necessary
proposition is the rationally expectable acceptance of our claim
by other rational inquirers in our historical context.
104
5. Kant on Conviction
In section 2 we have seen that when Kant explicitly addresses the
conditions for justified claims to knowledge, he maintains that
justified claims to knowledge need to be based on both subjective
and objective sufficient grounds, where the latter are
intersubjectively valid grounds for which we are conscious that
they entail the truth of our claims. This form of justification
is required for both claims to a priori and to empirical knowledge.
I have suggested that this account of epistemic justification is
too demanding for Kant’s own description of reason as a faculty
that is not transparent to itself and that is essentially public.
Even though Kant’s explicit treatment of the justification
required by our claims to knowledge endorses a form of
infallibilism, we can use his distinction between conviction
(Überzeugung) and knowledge for developing an account of epistemic
justification, and indeed of a priori epistemic justification, that
better suits his account of reason and philosophy that I have
presented in section 3.
Besides the distinction between having an opinion,
believing, and knowing, Kant identifies a further distinction to
105
differentiate between different forms of assent, that is, the
distinction between conviction (Überzeugung) and persuasion
(Überredung). Conviction and persuasion should not be considered
as two further members of the former group, though. Rather, they
identify a different distinction which overlaps with the former
in ways that are not immediately clear.33 Here I do not want to
consider the complex relationship between these two distinctions.
By contrast, I want to focus on the relationship between
knowledge and conviction.
The key feature to distinguish conviction from persuasion
is intersubjective recognizability.34
The touchstone of whether the assent is conviction or mere
persuasion is therefore, externally, the possibility of
communicating it and finding the assent to be valid for the
33 For example, Chignell (2007, 49) and Stevenson (2003, 82) represent very differently the way in which persuasion should be placed within the general scheme.34 In his contribution to this volume Marcus Willaschek associates conviction with subjective validity. Kant himself makes this identification at CPR A 822/B 850, but in this context he is not discussing the difference between conviction and persuasion. When Kant addresses this latter distinction he associates conviction with intersubjective validity. It thus seems that in thepassage to which Willaschek refers Kant is using another sense of conviction than the one that is relevant here.
106
reason of every human being. (CPR A 820-1/B 848-9, translation slightly
modified)
Accordingly, Kant describes conviction as an assent that “is
valid for everyone merely as long as he has reason” (CPR A 820/B
848). On the other hand, persuasion “has only private validity,
and this assent cannot be communicated.” (CPR A 820/B 848).
Intersubjective recognizability is here a means to identify
assents which are based on sufficient objective grounds (CPR A
820/B 848). Kant thus emphasizes different aspects of sufficient
objective grounds, when these are considered as conditions of
either conviction or knowledge respectively. While the
distinctive feature to recognize sufficient objective grounds for
conviction is intersubjective acceptability,35 the distinctive
feature to recognize sufficient objective grounds for knowledge
35 Kant’s analysis of conviction is sometimes confusing, because he identifiesa kind of conviction based on sufficient subjective grounds (and lacking sufficient objective grounds). He calls this kind of conviction practical (KGS9:72), and he associates it with a moral belief, which he sometimes defines asa conviction based on sufficient subjective grounds (cf. CPR A 829/B 857). Thecriterion to decide if what seems a practical conviction is a case of persuasion or not cannot rest on public recognizability because, resting on subjective grounds, a practical conviction cannot be intersubjectively accepted. From these remarks, it is clear that Kant uses two different understandings of conviction, with different conditions of validity. In this section I am only concerned with convictions based on sufficient objective grounds.
107
is infallibility or certainty (they are grounds that de facto
entail the truth of the proposition we assent to and we know it).
It is reasonable to think that for Kant these different ways of
describing sufficient objective grounds ultimately identify the
same thing. From this point of view, universal agreement could be
obtained only for those grounds that infallibly sustain the truth
of the claim they support. In this way, sufficient objective
grounds for conviction and sufficient objective grounds for
knowledge would amount to the same thing for Kant. Kant himself
suggests this, when in the Jäsche Logic he associates logical
conviction with knowledge and objective certainty (KGS 9:72).
Still, it seems that between persuasion, as a form of assent
that has only private validity, and knowledge, as a form of
assent that has universal validity because it is based on grounds
for which we know that they actually entail the truth of the
proposition we assent to, it is possible to recognize a whole
spectrum of possibilities with growing degrees of intersubjective
recognizability. Within this spectrum, there are surely claims to
knowledge that are justifiably regarded as candidates for
universal recognizability, because, in a particular historical
108
moment, they are generally seen as universally valid and as based
on grounds that entail their truth. Again, here Euclidean
geometry can help us in clarifying the issue. Some claims in
Euclidean geometry that we now regard as false could be
considered as candidates for universal recognizability, because,
in a particular historical moment, they were generally seen as
universally valid and as based on grounds that entailed their
truth. As this example shows, we can distinguish two kinds of
claims here. On the one hand, a claim can reasonably be regarded
as a candidate for universal validity, because, in a particular
historical moment, we can reasonably expect it to be publicly
recognized as valid and as based on grounds that entail the truth
of the claim in question. On the other hand, a claim can actually
amount to knowledge because it is based on grounds that entail
the truth of the claim in question and these grounds are thus
universally recognizable as sufficient grounds for that claim. Of
course, in this case the validity of our justification would not
be restricted to a particular historical context. It would be
actually universal. Kant does not seem to recognize the
difference between these two situations. For him, a claim to
109
knowledge has either only private validity or it is universally
recognizable as valid insofar it is based on grounds that entail
its truth and we are conscious of this.36 As we have seen in
section 2, on this account, we are justified to claim knowledge
only if we in fact have knowledge on the basis of infallible
grounds.
In section 4 I have suggested that it would be impossible to
identify criteria that would tell us when we in fact have this
kind of infallible justification and Kant seems to recognize this
fact in his own account of philosophy and reason. Recall that
Kant describes the idea of an ultimate and complete philosophical
system as an archetype that, until his times at least, we were
not able to achieve. He also claims that even if we reached this
archetype, we would still need to continue to test its universal
validity by submitting our claims or the claims of others to the
judgment of our public reason. This suggests that an absolute
certainty is unreachable by our philosophical claims, and even if
we reached the archetype, that is, even if we obtained a system
36 Opinion and belief are kinds of assent that are not relevant in this respect. Both cannot be considered claims to knowledge, insofar as in both cases we recognize to lack sufficient objective grounds.
110
of interconnected claims to knowledge where our a priori
justification for these claims entailed their truth, we could not
decide the universal validity of these claims on our own, and we
would always have to wait for the judgment of other rational
beings, without being able to establish universal validity once
and for all. This means that infallibility is too high a
condition for our justified claims to knowledge, and especially
for our justified claims to knowledge in philosophy. In this
respect, it makes more sense to develop further the Kantian
distinction between conviction and knowledge in order to
differentiate between justified claims to knowledge, and
justified claims to knowledge that actually amount to knowledge.
Accordingly, a conviction, which I here interpret as a
justified claim to knowledge, would be a claim that can
reasonably be regarded as a candidate for universal validity,
because, in a particular historical moment, we can reasonably
expect it to be publicly recognized as valid and as based on
grounds that entail the truth of the claim in question. This is a
pretty high requirement for justified claims to knowledge,
because they must be able to be seen as candidates for universal
111
validity in a particular historical moment and they should be
publicly recognizable as based on grounds which entail the truth
of the claims in question. However, this requirement leaves open
the possibility that a justified claim to knowledge is false and
thus revisable. On the other hand, knowledge, which I here
interpret as a justified claim to knowledge that actually amounts
to knowledge,37 would be a claim that is based on grounds that
actually entail the truth of the claim in question and are thus
universally recognizable as sufficient grounds for that claim.
However, if, in doing philosophy and other kinds of inquiry, we
always need to submit our claims to the judgment of public reason
even when we are in possession of knowledge, it means that we do
not have the criteria to distinguish if we have a justification
which actually entails the truth of the claim in question, or
simply a justification which we are justified in regarding as
entailing the truth of the claim in question, insofar as it is a
candidate for universal validity. If we cannot actually determine
with certainty if we are in possession of knowledge, that is, if
37 This seems to be in contrast with Timothy Williamson’s account of knowledge, which regards the latter as a sui generis mental state that is not analyzable in terms of other concepts, like justification and belief (cf. Williamson 2000). I do not have the space to take up this issue here.
112
we are in possession of a justification that actually entails the
truth of our claims, it means that we must regard that idea of
knowledge as the archetype we must always try to attain, but that
we cannot ever be completely and absolutely certain to have
achieved.
6. Conclusion
In this paper I have shown how Kant’s claims concerning the
infallible and certain status of the theses he defends in his
theoretical philosophy are based on his demanding account of
justification for our claims to knowledge. According to this
account of justification, a justified claim to knowledge is a
kind of assent based on grounds for which we are conscious that
they entail the truth of the proposition we assent to. Kant
argues that both empirical and a priori claims to knowledge should
meet this demanding criterion of justification. However, what is
distinctive about a priori epistemic justification is that it
cannot justify forms of assent different from knowledge.
According to Kant, we cannot have an a priori epistemically
113
justified opinion.38 In a priori inquiries aiming at knowledge, we
can either claim knowledge or nothing else. Since for Kant, as we
saw in section 2, a justified claim to knowledge has to entail
actual knowledge, it is quite unsurprising that he maintains that
the theses he defended in his theoretical philosophy are
infallible and certain.
However, in section 3 I have argued that this infallibilist
account of epistemic justification is too demanding for Kant’s
own account of reason and philosophy. Kant describes reason as a
faculty that is not transparent to itself and that can be
deceptive. Moreover reason is essentially public and requires us
to always submit our claims to the judgment of other rational
beings. These features of reason are reflected in Kant’s account
of philosophy in the “Architectonic of Pure Reason,” where Kant
claims that nobody can maintain to have achieved the ultimate
system of philosophy. By contrast, a philosophy scholar should
38 For Kant, we are justified in holding an opinion when we have some objective grounds for our assent, even if we lack sufficient objective groundsfor a claim to knowledge. He maintains that a state of opinion can only be justified in empirical inquiries. In a priori inquiries a ground can either showthe necessity of the proposition we assent to, or it cannot be considered an objective ground at all (cf. CPR A 822/B 850).
114
always expect her views to be rationally scrutinized by other
inquirers.
In section 4 I have introduced an account of fallible a priori
epistemic justification that better suits this characterization
of reason and philosophy. According to this account, a claim to
knowledge concerning a supposedly necessary proposition is a priori
justified when the a priori grounds we have can reasonably be
regarded as entailing the truth of the proposition in question.
This justification is compatible with the absence of an actual
entailment relationship and thus presents an account of a priori
justification which regards the latter as fallible and revisable.
To finish, in section 5 I have used Kant’s distinction between
conviction and knowledge in order to find some hints toward a more
modest account of epistemic justification in his own philosophy.
According to this interpretation, a conviction is a justified
claim to knowledge insofar as it is a claim that can reasonably
be regarded as a candidate for universal validity, because, in a
particular historical moment, we can reasonably expect it to be
publicly recognized as valid and as based on grounds that entail
the truth of the claim in question. We can apply this account of
115
conviction to a priori claims to knowledge, where the distinctive
feature of the latter is that they are based on a priori grounds.
If we understand conviction in this way, Kant could well have
argued that his philosophical claims, in order to be considered
justified a priori, should have been able to be considered
convictions and thus worthy of public recognition. However, this
would have been compatible with a fallibilist standpoint on those
views that recognized the fallibility of our reason and the
necessity to continuously submit our views to the judgments of
others.39
References
Bonjour, L. 1998. In Defense of Pure Reason: A Rationalist Account of A Priori
Justification. Cambridge: Cambridge University Press.
Chignell, A. 2007. “Kant’s Concept of Justification.” Noûs,
41(1): 33-63.
39 I would like to thank Marcus Willaschek and Robert Stern for their commentson previous drafts of this paper.
116
Gava, G. 2011. “Can Transcendental Philosophy Endorse