Beruflich Dokumente
Kultur Dokumente
From this distinction, Sosa begins to highlights two arguments, both shared by foundational
pictures of justification, which end in disproving foundationalism itself.
o A propositional attitude gives us at best indirect contact with reality, hence it doesnt
shield us from the possibility of error. This means that a propositional attitude cannot
provide a foundation for knowledge, hence, any mental state incorporating a
propositional attitude is no foundation.
Remark: It seems that the first three points can be applied to Cartesianism, while
the fourth is different. I think therefore I am is definitely a propositional
attitude, derived from something indubitable, and is foundational at least for
establishing the subject. For the truth of the objects we will need something
different (ultimately, God), but it seems therefore that we have direct contact
with the reality of the cogito even though it is expressed through a propositional
attitude (it is effectively an existential attitude)
o If the basic mental state incorporates no propositional attitude, we appear to have no
support for any hypothesis from it. This amounts to saying we cannot state the content
of our knowledge-claims or observations. If the mental state is propositionally empty,
we have no foundation for knowledge in a mental state.
o It does not matter at the end whether a mental state incorporates a propositional attitude,
as in both cases, no foundation for knowledge is to be found. Mental states grant no
foundational support for knowledge.
This is the coherentist argument against foundationalism. Sosa investigates the possibility that
propositional attitudes can be supported by something else than themselves (which is incidentally
something shared by the coherentist as well). What if justification is based on practical or
consequential (utilitarian) concerns?
o Remark: This argument is spurious. What we have here is not a foundational, a priori
justification, but an a posteriori validation of a propositional attitude. Justification by
consequence runs against foundationalisms tenets in the same way coherentism does
(and could be seen as a kind of pragmatic justification)
What if the belief that something is red has its origin in the direct perception, in ones own visual
experience of something appearing as red? In this case we would have a belief resting upon
something non-propositional. A mental state as something non-propositional and foundational at
the same time.
o Remark: this is the crux of Foundationalism. It relies in the end on something brute, a
physical fact that has to be recognized without being reduced to either a propositional
attitude or an inference but whose content can be ultimately grasped as a mental state.
How this actually happens (from the evidential experience of something red to
recognizing that this is really red) seems left mysterious (cf. Sellars and the necktie
example). This is an example of the Myth of the Given.
What if, furthermore, this mental state is propositional after all and it is not infallible? Why
should it be rejected as a foundation? Here Sosa analyses again the situation in terms of
consequence of a certain action (intended in parallel as the consequences in adopting a certain
belief as foundational) as giving a foundation for the justification of adopting a certain
action/belief. For Sosa, if this is true then the anti-foundationalist is on the wrong track and it is
furthermore a distinctive intellectualist trait that taints its concept of justification.
o Remark: The problem is, once more, how this a posteriori justification can be intended
as a foundational belief in the sense the foundationalist would want it to be. It appears
not to be indubitable, nor immediate. It is not even a reference to some brute fact, as its
consequence are at best contingent; if such consequence were indeed something
necessary (i.e. self-evident), shouldnt this kind of necessity already occur at the terminal
node of the pyramid? It seems difficult for a justification obtained by the consequence of
a belief to be terminal in the way the foundationalist model requires (and indeed, if the
pain or seeing red, we can easily see that higher-order beliefs obtain justification from
these fundamental states because they have already some kind of content expressed in
them. If I know that being hit on the head entails a headache, and Im hit on the head
with a bat, I have a justified belief to know that Im in pain. However, why is Sosa
supposing that in coherentist systems, the place of a certain proposition doesnt bear
anything at all on its actual content? If coherentism is only a matter of a web of beliefs
bestowing justification upon a certain belief P, it means that the content of those beliefs is
already Given, hence actually opening up a space for a foundational outlook on these
beliefs themselves. It appears to put in place a linear conception of justification regarding
the content of single beliefs, while obtaining some other kind of justification (left
somewhat vague) from its position in the coherent set. This seems therefore only a
foundational system that requires the beliefs in the system to cohere with each other,
which is already a requirement of whatsoever belief system we might adopt
(Mathematics, with its avoidance of contradictions, is a pretty apt example). This would
indeed make manifest a fundamental inadequacy in coherentism. However, this picture
doesnt withstand the fact that the set of coherent beliefs bears also on their content, with
the position of a certain proposition P influencing the content of the beliefs that support it
while in turn supporting and bearing on the content of these same beliefs in return. P is
therefore not merely logically implied by P, something that betrays again a linear
conception of justification, but is logically implied, and its content too, by all of the other
propositions it is connected with, and by its position too. The mere position of the
proposition in the web of beliefs is not enough to confer justification on it, rather, its
position in the web of coherent beliefs is due to its having a certain meaning, and its
having a certain meaning is due to its being connected to the other beliefs of the system.
This betrays a further interesting point: Sosa views coherentism as threatening an infinite
regress, due to the fact that to justify a belief another belief is needed. However, to view
this picture in these terms is precisely to see it in a linear, foundationalist manner. The
problem with coherentism, both regarding justification and content of beliefs is a
problem of circularity. This is why coherentist pictures such as Davidsons often adopt a
kind of causal externalism, where is the causal history of the particular belief that allows
it to take initially part in a web of beliefs. These pictures might be shown to rely too on
some kind of foundational given (as McDowell has argued for in Mind and World), but
this will suffice to show that Coherentism about justification cannot be meaningfully be
detached from coherentism about content (holism), without falling into some kind of
linear foundationalism at some stage, and that the issue with coherentism is not an issue
of regress, but rather a problem of circularity, entailing that coherentist pictures of
justification should also employ a genealogical account aiming at explaining how we
acquired the beliefs that form the raw material for the coherent web of beliefs.
The threat of the regress is for Sosa an indication that we have to look for beliefs that do not
require other beliefs for their being justified. This is however something that arises out of the
supposed impossibility of an actual infinite regress. Foundationalism needs to show why an
infinite regress is something vicious, and why foundationalism has to rule out a chain of
justification without end in order to obtain foundational beliefs. This is the doctrine of epistemic
foundationalism.
Sosa Begins to analyse objections to an endless regress of justification.
o Finitude of Human mind Objection: For Sosa this is a mere presumption and furthermore
our problem is the nature of justification in itself, not to see whether it can actually be
ever achieved by human minds. Is an infinite regress ruled out by the very nature of
justification?
o It is ruled out in principle because otherwise justification would never end: Sosa argues
that if the end is merely a temporal matter, an eternal being could very well obtain that
coherentist position (albeit the relation would have more members than just two
propositions), but the foundationalist has a need to show where the justification
arises in the first place, and is exactly this requirement that drives
foundationalism into the perils of an infinite regress. A coherentist position
instead does not need to disprove an infinite regress, because it doesnt share the
linear conception of justification upon which the infinite regress is created (as
explained before, the problem with coherentism is ultimately the Trope of
Circularity)
o Objection of the appeal to the successor: This objection seems to be the second problem I
highlighted in the remarks over here. A mere appeal to the successor of a proposition P in
the chain of justification wont be enough to grant justification to the proposition P itself.
Justification seems to be floating somewhere between P and its successor, leaving vague
its anchoring point. Sosa tries to avoid this problem by making a distinction between
actual and potential regress, however it is not clear how this distinction dispels the threat
of a regress. Sosa seems to think, rightfully so, that even a merely potential regress is
enough to disrupt the chain of justification. If a single proposition were to be justified,
then its predecessor would be justified, but the question of whether (and most
importantly how) that single proposition is actually justified is left unanswered, therefore
none of its predecessors are known to actually obtain justification.
o Objection against this distinction: For Sosa, an actual regress has as its members only
justified beliefs, while a potential regress doesnt need to. For Sosa we can hold an
infinite chain of actual beliefs that are justified and that terminate with a belief at its head
that is justified in virtue of its position. However, if it were not justified due to its
position would actually settle the question in the negative. The possibility of a potential
regress with a belief at its head that is not justified doesnt show the impossibility of an
actual regress with every of its member being justified
Remark: The observation already stated apply also here. What does it mean for a
belief to be justified in this linear setting, if a foundation is not achieved? It is an
arbitrary cut-off just to say that the successor of a proposition P is all that is
needed to grant justification to P. Where does the justification reside? If it reside
in Q then we need to be sure whether Q is justified. The problem is that in going
through a regress we never achieve the justification we were seeking unless we
arbitrarily assume a certain cut-off point. Furthermore how can a belief at the top
of a chain be actually justified if the original instantiation of the justification is
out of reach in this sense?
Sosa proceeds in arguing that the possibility of an actual infinite regress shouldnt be seen
immediately as tarnishing the justificatory enterprise for the foundationalist. A route that doesnt
involve ruling out the regress is to accept the fact that what we need are justified beliefs at the top
of the chain, at the top of a terminating regress. A single belief can lead to both an actual regress
and a terminating regress, as in the case of the set of even natural numbers. This could be a
terminating regress for a powerful enough intellect.
o Remark: apart for the observations already carried out, the example relying on numbers
seems to appear intuitive enough to show us that a terminating regress is an actual
possibility even in those cases where the same justified belief could incur in an endless
regress. However, using natural numbers is trickier than Sosa supposes. The definition
and justification of beliefs about natural numbers is actually grounded in mathematics on
the Peano axioms, which in turn give rise to Peano Arithmetics when addition and
multiplication are added to the axioms. These are fundamental propositions, hence the
regress to the infinite is avoided in a completely different way, namely by turning to
fundamental assumptions and definitions regarding the nature of the numbers. This move
terminates the regress employing a classical foundationalist route: the validity of the
foundations themselves however is not discussed (Goedels theorems prove their being
either incomplete or incoherent, cutting the ground under mathematical axiomatic
foundationalism). What if laymen do not use this definition in order to have justified
beliefs about numbers? If we employ a concept in a commonsense way as in this case, we
are just using an incomplete, imperfect concept of number, good enough for pratical and
everyday purposes, but that would not stand epistemological scrutiny regarding their
ultimate grounds.
Sosa recognizes that the foundationalist would not be thrilled to go for this route, the very
possibility of an actual infinite regress is enough for him to avoid infinitism altogether
How we answer to our need for foundations if an endless regress is not ruled out? Sosa
introduces at this point a distinction between two forms of coundationalism. Formal
foundationalism and substantive foundationalism: the former defines the general conditions under
which a certain normative or evaluative property phi applies, whereas the latter is a particular
way of doing so.
o Remarks: This definition has some issues: formal foundationalism can hardly be defined
classical foundationalism tout-court, as it only specifies when normativity obtains the
content and authority it has, independently from more fundamental assumptions. This
definition could be useful when some anti-foundationalist positions, such as
Wittgensteins, are branded as kinds of foundationalism just because they allow for the
application of some normative property. However, if, as Sosa argues, foundationalism
and coherentism are both kinds of substantive foundationalism, isnt this definition too
vague? Every normative proposal, even relativism, could be understood as a kind of
foundationalism which would be odd to say the least. If the distinction is merely
formal/general and substantive/particular as it appears to be by the inclusion of
coherentism in the latter grouping, why labelling it as foundationalism in the first
place?
Classical foundationalism has three main tenets: infallible, indubitable beliefs are justified, every
belief obtained through deductive reasoning from these basic beliefs is justified, every justified
belief is either a basic belief or deducted from one. This is an example of formal foundationalism
(why not substantial? Coherentism or relativism would not need to share any of these tenets)
This kind of foundationalism is not really attractive anymore and has been deemed as impossible.
Philosophy has however lost sight of the different types of foundationalism as exemplified by the
formal/substantial distinction.
o Remark: hasnt this happened because the distinction is too inclusive and vague?
Everything can be labelled foundationalism, so if we are to argue for an antifoundationalist position the distinction doesnt actually help this task.
In a passing note, Sosa makes a more fundamental distinction that characterizes foundationalism
proper: conditions that epistemically justify a belief are to be specifiable in nonepistemic terms.
We saw an hint of this position when Sosa analysed propositional and non-propositional attitudes
towards basic features of experience such as having a perception of something red.
Foudnationalism holds that ultimately, epistemic beliefs are to be justified in virtue of something
nonepistemic.
o Remark: we should clarify here that non-epistemically basic for a foundationalist means
something immediate, indubitable, whose content is not established by recourse to a
proposition about it. A given, brute fact (Sosa doesnt state this explicitly, but following
his various definitions of foundationalism in the paper, this position can pretty easily be
stated)
Epistemic justification is something that supervenes on something else that is neither normative
nor evaluative. Something is epistemically justified to its having a nonepistemic basis in
perception, inference, memory, etc. Sosa bases this on an analogy with what is regarded as good
in ethics
Remark: why are perception, inference or memory valid bases in themselves for
granting the non-epistemic basis? Why should they grant us something non-epistemic in
the first place? Usually an argument about this is made from the supposed non-epistemic
experience of colours; however perception of colours isnt something trivially nonepistemic. In philosophy of perception a lively debate focuses on whether even these
basic facts belonging to perception have a conceptual, linguistic, or inferential property.
Relying on perceptually basic (nonepistemic) properties, upon which justification
supervenes, in order to explain why a certain belief arising from perception is
epistemically justified doesnt actually explain the connection between non-epistemic
facts and epistemic justification. How is the non-epistemic content delivered? What does
it mean that epistemic normativity supervenes on it? Does this mean that already at the
non epistemic level we can have a propositional attitude about its content? If so, why
should it still be considered as epistemic? If we allow for a gap between brute, basic,
non-propositional facts and epistemic justification, we are left puzzled on how either side
can bridge this gap without becoming ultimately the other side? This is a distinctive
Cartesian problem (addressed the most by Hegel, Wittgenstein, Sellars and McDowell).
Sosa takes into account however that there are higher grades of formal foundationalism, in which
the conditions for epistemically justified beliefs can be specified recursively and by a simple,
comprehensive theory.
o Remark: This recursive specification is left here a bit vague. It seems, from what Sosa
has just said, that we can reference to perception (or memory, or inference) in order to
ground epistemically justified beliefs based on the contents of perception (or memory, or
inference). This appears to be circular reasoning, and a kind of circular reasoning usually
found in externalist theories (Appeal to the reliability of perception in order to explain
perceptual content which in turn explains why perception is reliable).
In this sense classical foundationalism is a higher grade formal foundationalism, because it
delivers a recursive simple theory in which epistemic justification supervenes with infallibility
and indubitability. However, Sosa argues that even coherentism should be allowed within this
definition, provided that coherentist justification supervenes on non-evaluative or normative
facts.
o Remark: this is, again, problematic, because coherentism doesnt need this nonnormative requirement. Classical Foundationalism needs it because otherwise it would
lose its foundational grounds which connects beliefs to something immediate, certain,
indubitable and infallible (hence non-normative. Something that is really such-and-such).
However, why should coherentism share this proposal? If we picture coherentism about
justification as having some linear, noncoherentist ultimate connection with something
immediate, indubitable, infallible, etc., how does this coherentism differ ultimately from
foundationalism? Isnt this kind of coherentism just a more complex and structured
foundationalist variant, a pyramid with Relations between beliefs that cohere and that
terminate by referring to some non-epistemic, non-evaluative brute fact? Even more so, if
coherence between beliefs is the only thing that characterizes coherentist justification,
isnt this a feature shared by classical foundationalism as well? The beliefs that make up
the various layers of the pyramid do need to be coherent with each other. This should
mean that coherentism is something more than just coherence between beliefs, a web of
justified beliefs. In coherentism, every single belief supports every other belief.
Therefore coherentism about justification doesnt need to supervene only on something
non-normative. All coherentism need for justification is normativity between mutually
supporting beliefs. This again raises the problem for coherentism of how it can relate to
an outside world but coherentism can work for explaining justification perfectly well by
merely employing normative, epistemic beliefs.
o
Coherentism is for Sosa therefore a formal foundationalism as it explains that a belief is justified
iff it has a place within a system of beliefs that is coherent and comprehensive
o Remark: as already observed, if this is all that coherentism requires, classical
foundationalism is itself a kind of coherentism (a non-coherent pyramid of
foundationalism would grant support to hardly anything). This must mean that
coherentism must require something else, such as the mutual support in beliefs for their
mutual justification. This means that coherentism has a distinctive holist property that
foundationalism has not, and that furthermore the content (or meaning) of beliefs in a
coherent system does not need to be ultimately tied to something non-normative, nonevaluative, brute, given, but it is the holistic web of justifications and mutual support that
explains why a belief has the (justificatory) role and the content it has in a coherentist
proposal.
For Sosa, epistemologys goal is to explain how justification supervenes on something nonepistemic, non-normative, non-evaluative. Sosa realizes that if coherentism share this goal, this
includes strong restrictions on the notion of coherence. Coherentism must rely on explanations
that are free of normative or evaluative content. Therefore coherentism and substantive
foundationalism turn out to be varieties of a deeper fountionalism as they both develop a formal
foundationalist theory that explains how justification supervenes on the non-epistemic.
o Remark: Sosa seem to realize that if Coherentism shares these goals then much of its
opposition to foundationalism is in fact misplaced, and coherentism has actually much
more in common with foundationalism than usually recognized. However this picture of
coherentism is both puzzling and problematic. Why doesnt Sosa realize that, if all that
made up coherentism seems to vanish into foundationalism, then this should be a hint
that coherentism require something more? (such as what we explained in the previous
remarks)
ADDENDUM: we should probably here remark again that the non-epistemic the
foundationalist is seeking is something indubitable and infallible in the classical variant,
while in more modern and modest proposals it is still something immediate, nonattitudinal, a brute matter of fact. A non-epistemic basis such as pragmatic, practical or
teleological reasons wouldnt work for the foundationalist, as they rely either on
something evaluative of normative, such as an inter-subjective setting of rule-following.
However, the coherentist could easily accept these features of non-epistemic reasons,
provided that coherentism is not just what Sosa defined it to be.
o