Metaphysical Categories


Necessary Truth/Contingent Truth



Epistemological Categories (Categories of Justification or Knowledge)


A Priori/A Posteriori

        A Priori Justification is defined both negatively and positively

        Negative:  justification independent of experience

        Positive:  direct rational insight into necessary truth.


Semantic Categories (Categories of Proposition)



        Kant's account:  Subject contains the predicate.

        Fregean account:  Reducible to logic and definitions.

        Semantic account:  True in virtue of meaning.


What did Kant classify as analytic a priori?


What did Kant classify as synthetic a priori?








Challenges to A Priori Justification


        Quine's arguments against analyticity are really arguments against all a priori justification.


        Quine's Coherentist Challenge:  Examples from physics (F=ma), geometry (parallels postulate), logic (Law of the Excluded Middle), and epistemology (traditional definition of knowledge).


        Nozick's Naturalist's Challenge:  Evolution has produced powerful intuitions because they work well in this world.  No need for rational insight into necessary truth. 

        Examples:  Objects fall down if not supported.  Heavier objects fall faster than lighter ones. 

















BonJour's Fallibilist Foundationalist Defense of Rationalism (i.e., a priori justification)


Objections to the three conceptions of analyticity.  What is the basic objection?


Is BonJour a foundationalist about a priori justification?


Are the foundations infallible[WJT1] ?  The distinction between apparent and genuine a priori insight.


The Two-Part Role of A Priori Justification in Reasoning:

        (1) Premises

        (2) Inferential connections


Why does BonJour think that the second role is more important?  [Hint:  Why does he think it makes a priori justification indispensable?]














The Epistemic Indispensability of A Priori Justification for Reasoning



Let O be all of our empirical beliefs that can be justified by direct observation.


Let P be any other contingent proposition that we take ourselves to be justified in believing.


Consider the conditional:  If O then P.


We are not justified in believing that P unless we are justified in believing O and justified in believing that if O then P.


What could be the justification for believing that conditional?  It cannot be directly justified by observation.  If it is justified at all, it must be justified a priori.


CONCLUSION:  We cannot be justified in believing that P unless we have some a priori justification.



Questions:  Why does reasoning have to involve any beliefs about the premises and the conclusion?  Why can't we just infer P from O without any beliefs about the conditional at all?  How would a naturalist such as Nozick reply to BonJour?



The Problem of (Justifying) Induction



What is the problem of induction[WJT2] ?


(1) Observationally determinable features


(2) Standard inductive premise (concerning past observations)


(3) Standard inductive conclusion (concerning future observations




An Induction


(1) Observationally determinable features:  "Crow" and "black" can be determined by observation.


(2) Standard inductive premise:  All observed crows have been black.


(3) Standard inductive conclusion:  All crows observed in the future will be black.







Hume's Dilemma


Hume's Dilemma:  (1) Either induction can be justified by demonstrative reasoning or by experimental reasoning.

(2) It cannot be justified by demonstrative reasoning (Why not?)

(3) Any experimental justification would be viciously circular (Why?)


Conclusion:  There is no justification of induction.






















Potential Defenses of Induction


(1) Pragmatic Vindication


(2) Ordinary Language Justification


(3) A Priori Justification

        According to BonJour, what would an a priori justification of induction look like?


A fourth alternative added by Talbott:  Non-separable cognitive processes:  Processes whose reliability cannot be assessed without employing those very processes.  Induction may be a non-separable cognitive process.  What others are there? 


Hume would say that determinations of the reliability of non-separable cognitive processes are viciously circular.  Are they?  To be continued.


Not Plantinga P 213 or BonJour  [WJT1]96

 [WJT2]B 53