There are many recipes one could use to cook up a coherent a priori/a posteriori distinction. In this essay, I look at three so-called ‘recipes’ which I think do not quite work, before proposing my own, mixing elements from each. I begin by looking at Laurence BonJour’s (2013, 178) and Michael Devitt’s (2013, 184) competing theories of the a priori/a posteriori distinction. I argue that both of their theories are united by the fact that they surreptitiously address the epistemic regress problem, and furthermore, that their solutions leave much to be desired. I consider another solution to the epistemic regress problem, also relating to the a priori/a posteriori distinction, endorsed by theorists such as Crispin Wright, Martin Davies (2004b, 167), and Albert Casullo (2015, 2701) which I call the entitlement solution, but I see problems with this too. I then propose a conception of the a priori/a posteriori distinction, composed of three key moves from each approach, which I see as balancing their benefits and drawbacks, coherently answering the epistemic regress problem and drawing the lines of the a priori/a posteriori distinction.
2. Bonjour Versus Devitt
BonJour and Devitt offer two different theories of the a priori/a posteriori distinction, with BonJour arguing for the necessary existence of a priori justification in the form of ‘rational insight’ (2013a, 179), while Devitt argues that all can be accounted for with a posteriori justification in a Quinean picture (2013a, 185), and that the nature of a priori insight is deeply mysterious (2013a, 190). What follows is a brief account of a five-paper exchange between the two.
BonJour’s arguments revolve around his ‘prima facie’ (Kornblith, 2000, 68) case for the a priori, in which it must exist to provide the justification for the justified non-sensory beliefs he argues we do hold, which could not be justified by sensory experience. Specifically, he argues, that given that we do hold justified non-sensory beliefs (concerning the unobserved past, present and future, metaphysical beliefs, and unobservable entities described by science (BonJour, 2013a, 181)), and that these beliefs cannot be directly justified by experience, but are justified, BonJour argues that this justification must be non-sensory, and therefore, a priori. He asks us to consider an inferential rule in the form of a conditional (2013a, 181), in which the antecedent is all of our sensory experience, and the consequent our non-sensory beliefs, which somehow go beyond the mere sum of our sensory experience:
That this conditional is true, he argues, could not be provided by sensory experience, because all sensory experience is included in the antecedent, leading to the attempted justification being circular. Furthermore, given that we do hold justified non-sensory beliefs that go somehow beyond the mere conjunction of our direct experiences, the inferential rule must be justified somehow. BonJour therefore argues that this justification can only take the form of “logically prior” (2013a, 181), non-sensory, a priori rational insight. In certain cases, according to BonJour, we just see that certain conclusions follow from certain premises.
Devitt has two aims in his response to BonJour, firstly seeking to undermine the need for a priori justification, and secondly to demonstrate its obscurity as a concept (2013a, 197), and makes five ‘moves’ to do so. Devitt’s first move is to endorse the Duhem-Quine (2013a, 186), ‘web of belief’ view, in which beliefs are interlinked, with beliefs on the edge of the web of belief being more sensitive to being revised in light of contradictory sensory experience, than those ‘insulated beliefs’ closer to the core. Seemingly ‘a priori’ truths, such as those found in maths and logic, are merely those that are deep within the web of belief, and so very rarely revised in response to sensory experience, though they ultimately may still be. In this sense, Devitt functionally denies the set-up of BonJour’s conditional, denying that we have non-sensory beliefs that ‘go beyond’ and are not answerable to sensory experience. Secondly, Devitt invokes a pragmatic conception, arguing that we need not know exactly why our inferential rules are justified, but merely know they lead to “empirical success” (2013a, 188), akin to someone not knowing exactly how a boat was built, only that it floats. Thirdly, responding to BonJour’s charge of circularity regarding sensory justification of our inferential rules, Devitt argues that we should endorse Braithwaite’s conception of ‘rule-circularity’ versus ‘premise-circularity’ in which an argument which “asserts the goodness of the rules used in that very argument” (2013a, 189), fails to be reprehensible in the same way that a premise-circular argument, which presupposes its own conclusion as a premise, is. Fourthly, Devitt fires back, arguing that even if we accept BonJour’s a priori insight, his inferential system, which includes all of one’s inferential rules, also faces the charge of circularity. Devitt argues (though not in these exact terms) that a priori justification is inferential-rule-generating, in the sense that if inference via a priori insight is permitted, this ‘generates’ an inferential rule, and these rules then equally need to be justified. If ‘the very system that asserts that a priori justification is valid’ is itself confirmed by a priori justification (and what else could it be), then BonJour’s proposal is equally as circular as his (2013a, 190). Fifthly, and finally, Devitt argues that BonJour’s conception of extra-physiological rational insight that somehow provides insight into worldly facts lacks an explanation of how this could possibly be so, relying on seemingly occult non-naturalistic input mechanisms (2013a, 193) (cf. Jenkins, 2012, 181).
BonJour replies to Devitt’s attacks. The primary thrust of his counter-argument against Devitt’s assertion “that the existence of empirical knowledge is not in question” (2013b, 195), is to double down on his claim that it very much is in question whether one could construct a coherent epistemology based solely on a posteriori justification (2013b, 195), reinforcing his previous claim that, to him, we need a priori rational insight to provide the links between sensory data and non-sensory beliefs. Secondly, in response to Devitt’s abductive, pragmatic ‘it just works’ argument, BonJour argues that this argument is equally ‘rule-generating’, and it is not yet established on what basis we can epistemically appeal to such abductive arguments (2013b, 196). Thirdly, BonJour responds that Devitt’s invocation of ‘rule-circularity’ in situations in which the rules themselves are the very subject under consideration, seems just as worrisome as premise-circularity (2013b, 196). Fourthly, responding to Devitt’s argument of circularity, BonJour attempts to wriggle free from such claims by arguing that a priori rational insights are to be considered “non-propositional” (2013a, 179), and “atomistic” (2013b, 196), and therefore not requiring further justification, avoiding circularity. He closes by doubling down on his previous claims, that we must have this intuitive ‘striking as us true’ rational insight, if any of our non-sensory beliefs are to be considered justified at all (2013b, 197).
Devitt, in his follow-up, addresses some of BonJour’s objections. In response to BonJour’s arguments about a rule-circular system, and the trouble with justifying those rules, Devitt argues that a system in which each epistemological rule is justified by a different rule, and none justify themselves, would be non-problematic (2013b, 198). Responding to BonJour’s claims that his abductive argument is equally ‘inferential-rule-generating’ and therefore justification-begging, Devitt reiterates his previous arguments that BonJour’s system, which includes a priori justification as valid, is also inferential-rule-generating, and therefore equally faces this problem (2013b, 198). Finally, Devitt makes a similar move to BonJour, attempting to secure his theory against the need for further justification by switching to an externalist position, in which an in inquirer need not be able to provide internal justification for their claims or epistemological rules, their inferences simply have to be justified (2013b, 199), by some external metric.
In the final paper of the exchange, BonJour pushes back on problems he sees with rule-circularity, and with Devitt’s proposed shift to an externalist concept of justification. Devitt’s previous argument, that one could have a system in which each rule is justified by a different rule, and none-self justify, BonJour argues, seems to be circular in the same objectionable manner in which a self-justifying rule is (2013c, 200), in that someone unconvinced of the validity of the inferential system is forced to accept it, or at least one of the rules, ‘wholesale’, before they have any purchase at all. Secondly, he attempts to double down on the atomistic nature of a priori justification, such that it “only relies on the specific insight that is relevant to it” (2013c, 200) and therefore does not generate a general rule licensing a priori justification that also needs justification. Finally, BonJour ends by pushing back on Devitt’s move to externalism, arguing that this move is of no use for an inquirer who is attempting to reason, and has no access to the external metric by which their inferences actually are justified (2013c, 200).
We therefore seem to be hoisted by a horn on either proposal; both sides seem to have valid and unresolved complaints against the other. It seems we either a) accept BonJour’s ‘we just know’ account, in which we accept that we have some non-sensory insight to account for our non-sensory beliefs, b) accept Devitt’s ‘it just works’ account, in which the fact that our inferential system seems to be working justifies itself circularly or c) figure out some way of balancing the positives and drawbacks of the two. We shall proceed with option c).
3. The Epistemic Regress Problem
I argue that both BonJour and Devitt are motivated in their argumentation for their a priori/a posteriori distinctions by the epistemic regress problem, the problem that justification appears to be valid only insofar as it is further justified, ad infinitum (cf. Boghossian, 2001, 639). The problem of epistemic regress occurs as a result of holding the following intuitive premise:
Justificatory Inheritance: Belief p is justified by reason1 if and only if reason1 is also justified by some reason2. Reason2 is justified if and only if reason2 is justified by some reason3…
As a result of this, beliefs ‘inherit’ the justification of their ‘parent’ reason, which themselves inherit the justification of their parent reason, the grandparent reason for the original belief. This continues ad infinitum, leading us on a search for initial bedrock, or determination of justification, before we can determine the justified status of any given belief.
It will be useful in what follows to consider the following layout of premises:
- We have justified sensory and non-sensory beliefs.
- All beliefs require justification. To any belief, the regressive question “What justifies that?”, may be asked, regressing one layer back in justification.
- When premise 2) applies to justificatory reasons themselves, we end up in a situation of infinite epistemic regress.
It should be noted that both BonJour (2013a, 179) and Devitt (2013a, 188) (2013b, 199) briefly refer to the problem of epistemic regress, and then quickly move past it, as though by naming it they are permitted to consider it banished, Rumplestiltskin-esque. Despite both of them considering it ‘dealt with’, they both continue to make the exact moves that lead to the problem; circling the others entire inferential system and saying “Yes, but what justifies that?”, where that is either a priori rational insight or abductive arguments. This circling of one’s entire inferential framework, and asking what justifies it, is precisely the regressive question which leads to the problem of epistemic regress, as it seemingly can be asked no matter what justification is given.
4. BonJour and Devitt’s Solutions
This problem set up leads us to two natural and related solutions, which I argue that both BonJour and Devitt employ at different times. Firstly, we might resolve the regress by providing a justification that justifies itself, and allows for a justified bedrock from which all further justification may then be inherited. This therefore continually answers and satisfies the regressive question, in that the answer to what justifies reason1 is itself reason1. I will call this approach satisfying the regress. Secondly, we might resolve the regress by providing a justification that is somehow immune to being asked the regressive question, “What justifies that?”. I will call this approach resisting the regress.
These approaches are visible in BonJour’s argumentation firstly when he argues that rational insight may provide the foundational bedrock for our non-sensory beliefs (2013a, 181), and we might presume that the rationale for why this inferential rule is justified is also in fact justified by further rational insight, and so on, thereby satisfying the regress. This move, however, introduces the circularity that BonJour wishes to critique, and thus he makes the move to argue that a priori rational insight should be considered non-propositional (2013a, 179) and atomistic (2013b, 196), thereby immune from the regressive question, and therefore resisting the regress, attempting to secure a priori insight as foundational justification.
Devitt, on his side of the fence, straightforwardly endorses rule-circularity (2013a, 189), thereby attempting to satisfy the regress by arguing that our inferential system provides the justification for itself. Furthermore, his shift to externalism (2013c, 200), arguing that an inference simply needs to be justified, instead of an inquirer being able to continually state their justifications, is a way of excusing that inquirer from the burden of the regressive question, and therefore resists the regress.
Both of their pictures, I argue, capture important intuitions, but as a whole, are unsatisfying. On BonJour’s side, it does seem correct that our non-sensory beliefs require justification, but his assertion that there therefore must exist mysterious quasi-perceptual rational insight seems like wishing justification into existence. Furthermore, his assertion that rational insight should be considered non-propositional, when content certainly seems propositional, seems like ad-hoc handwaving avoidance of the regressive question. On Devitt’s side, it does seem that there is something strange about rational insight, and it seems that our inferential system does currently yield empirical success, however, his endorsement of rule circularity means that any case made for someone who does not already take the inferential rules under discussion to be valid will be utterly unconvincing. Opposed to BonJour’s handwaving, Devitt’s rule-circularity seems like a willed ignorance of a lack of justification. Furthermore, BonJour’s critique seems to hit home regarding his shift to an externalist metric of justification being useless to an inquirer, who is attempting to discern whether the inferential rules they are using are valid or not. The a priori/a posteriori distinction that we will later provide, then, will seek to capture these positive notions of BonJour’s and Devitt’s proposals, whilst counterbalancing their negatives.
5. The Entitlement Solution
There is a third and final solution I will discuss, before I get to a-cooking; The Entitlement Solution. This solution, endorsed by theorists such as Crispin Wright, Martin Davies (2004b, 191) and Albert Casullo (2015, 2719) pertains to the a priori/a posteriori distinction, and seeks to solve the problem of epistemic regress by denying the second premise of our set up; ‘that all claims require justification’. As such, when we:
- Have no sufficient reason to believe that a claim is untrue,
and
- Have much to gain in a pragmatic or empirical success sense, from acting as though the
claim is true (Wright and Davies, 2004b, 191),
we are entitled to bold-facedly take whatever axiom we require to be true, to “trust” (Wright, 2004a, 174) that it is. This thereby provides us with our foundational axiom, resisting the regressive question “What justifies do you have for that?”, because we answer straightforwardly, “We do not have any, though we have much to gain by acting as though we do”.
I bring up this solution in the same vein as BonJour’s and Devitt’s, in that it captures the worthy intuition of pragmatic success being a consideration, and that it seems like we start from a justified position, but at the seemingly too-high cost of denying that all claims require justification.
Furthermore, I argue it is incoherent, in that the idea ‘choice of axiom’ appears to me to be in total contradiction. For if we are truly choosing between two foundational axioms, at the core of our inferential system, we ought to be stuck like Buridan’s donkey, with nothing to break the tie; we will have no justification by which to choose one axiom over the other. The moment we do make a choice, however, it appears that we smuggle in a more foundational form of justification. This is exactly what occurs in the entitlement solution, with the sneaky introduction of the notion of ‘gain’ licensing taking something as true when “we may gain by doing so” (2004b, 191), in the form of pragmatic or empirical success. This notion is used to justify, used as a tie-breaker, allowing us to determine which axiom we should ‘take’. This, however, is inferential-rule-generating. Our critique now takes the same form as BonJour’s did previously, attacking Devitt’s appeal to empirical success and abductive arguments without further justification.
A defender of the entitlement solution might be tempted to solve this by claiming that we are equally entitled to take abductive ‘empirical success’ arguments as justified too. However, if we make this move, we end up in a position synonymous with Devitt’s circular ‘empirical success’ argument. We are arguing that we are entitled to believe rules such as “We ought to believe that which leads to pragmatic success” is a true inferential rule, because doing so will lead to pragmatic success. This leaves the original question, whether we in fact “Ought to believe that which leads to pragmatic success” open, and is especially problematic for one who initially disagrees, and is forced to ‘buy-in’ by accepting the notion wholesale.
For these reasons, the entitlement solution is in a difficult spot. On one hand, it seems right to accept that which makes empirical, practical, pragmatic success possible. On the other hand, it seems even more objectionable than BonJour’s ‘wishing justification into existence’ to admit with a straight face that one does not have anything justifying their inferential chain at root. Furthermore, attempts to ‘take’ abductive appeals to empirical success have been shown to lead to the same problems as Devitt’s proposal, circular justification, which from the outside, appears to be no justification at all. We therefore find ourselves hoisted upon a third horn. Fortunately, with these three not-quite solutions on the board, we are all set to get a-mixing, combining their best elements into what I hope is a balanced a priori/a posteriori conception.
6. The Proposed Solution
To take stock: On BonJour’s solution, we got an explanation for the justification of non-sensory beliefs, but we got this through asserting their justification, and therefore ‘wishing’ a priori rational insight into being. On Devitt’s we got an appeal to the justification of empirical success whilst maintaining the principle that all beliefs require justification, though to maintain this, self-justifying circularity had to be endorsed. Finally, on the entitlement solution, we saw a similar capturing of the intuitions of endorsing that which ‘works’, though through an attempt to deny the principle that all beliefs require justification. If we can pull it off, therefore, our proposed a priori/a posteriori distinction ought to 1) address the problem of epistemic regress 2) explain non-sensory beliefs 3) capture our intuitions of appealing to empirical success 4) avoid circularity and 5) maintain the principle that all beliefs require justification.
6.1 Foundationalism
Faced with the problem of epistemic regress, the first move is to capture at least some solid justified ground. Of the two solutions previously canvassed, then, the most straightforward solution appears to be satisfying the regress by appealing to a foundationalist notion, in which beliefs which are immediately conferred by our sensory experiences are justified. Looking at a belief such as “There is an apple on the table”, BonJour’s critique strikes at the leap from the sensory experience of an apple being on the table, and the metaphysical claim that there actually is “an apple on the table”. There is not a straightforward truth transfer relationship between the two, indeed, the metaphysical claim could be false, and the apple a trick of the light.
Instead, taking heed from Descartes and Husserl, the belief “I am having an experience of an apple on a table” seems much more robust, self-justifying and impossible to be mistaken. In this sense, what justifies the claim is the having of the very sense experience that the claim is about. Even if the apple were a trick of the light, it would still be the case that one would not be mistaken about their having the experience of the apple on the table. Therefore, when the regressive question “What justifies that?”, is asked of the belief “I am currently having x sensory experience” and the answer is “I am currently having x sensory experience”, this therefore satisfies the regress, as further attempts to ask “What justifies that?”, will merely be met with the same answer. A foundationalist solution, captures at least some solid ground. So long as our inferential chains termite in such sensory experiences then, we ought to consider them justified with a solid bedrock. As we shall see, however, this is harder than it sounds.
6.2 Jettisoning Non-Sensory Beliefs in Exchange for Quinean Posits
Whilst foundationalism secures us some foundation, as we have seen, it certainly does not secure very much. Indeed, on the above argument, even something as simple as “There is an apple on the table” is rendered as a non-sensory, metaphysical belief, and therefore apt to be asked the BonJourian question, “What justifies the leap from sensory to non-sensory?”.
As such, we here have a choice. We can either:
The BonJourian Move: Accept axiomatically that we have justified non-sensory beliefs, and go hunting for some form of non-sensory justification.
The Quinean Move: Deny that we have non-sensory beliefs, insist that all beliefs terminate in sensory experience, and provide some explanation for the illusion of non-sensory beliefs.
Of the two of them, the Quinean move leaves us with ultimately less trouble than going hunting for non-sensory justification. Under this conception, we are to consider all the non-sensory beliefs BonJour mentions as requiring explanation, such as those concerning the unobserved past, present, and future, metaphysics, and any unobserved entities and processes, as posits, or ‘best guesses’ as to what explains the sensory experience that we are experiencing. As such, even metaphysical statements about the physical world, such as “There (actually) is an apple on the table”, Quine would regard as a ‘best guess’ that provides a causal and predictive backstory to our sensory experience (1951, 41). On our view then, this is to say that every chain of inference that does not directly terminate in our current sensory experience is to be regarded as a posit, as it necessarily differs in epistemological pedigree. This is not to say that they are useless, but to acknowledge the existence of fallibility within our posits, whereas our current sensory experience is ‘given’ to us. Therefore, by providing an explanation for seemingly non-sensory beliefs, by explaining them as Quinean posits or ‘best guesses’, we are not required to hunt for, or invent, a form of non-sensory a priori justification as BonJour is.
One might argue that the shift from non-sensory beliefs to Quinean posits does not escape the BonJourian question; How do we make the leap from our sensory experience to confirming/‘being in favour’ of any posit which goes beyond it? There are two ways of reading this objection. Firstly, on a strong reading of BonJour’s objection, in which sensory experiences confirms a posit, the objection misfires. By downgrading the epistemological pedigree of non-sensory beliefs to posits, we are not claiming that non-sensory beliefs can be confirmed by sensory experience, not asserting a strict truth-transfer relation, and therefore the strict BonJourian question does not apply. Secondly, however, a weak reading of the BonJourian question might ask how it is that sensory experience can ever count in favour of certain posits over others. Certainly we will want to say that not all posits are equal, and the BonJourian question will ask in what ways sensory experiences can weigh on what non-sensory posits we ought to accept, when it seems like they ought to be incommensurable epistemic currencies. I think the best answer to this objection is a coherentist, reductionist one. Certain posits, or best guesses as to why we are having the sensory experience of an apple on the table (e.g. “There’s actually an apple on the table” / “It’s a hologram” / “It’s a trick of the light”) will better cohere with, and be able to predict/explain a wider range of sensory experiences, will be stable over a wider range. In this sense, posits or ‘best guesses’ are rendered in the same epistemic currency as sensory experience, are in some manner reducible to them, because we are liable to reject those posits if they fail to predict our sensory experiences. Therefore, so far we are able to explain the illusion of non-sensory beliefs, without searching for non-sensory a priori justification, relying only on a posteriori sense data, of which our posits, are our hypothetical and predictive ‘best guesses’.
6.3 The A Priori Normativity of Justification
Are we not just full-blooded Quineans, then? I think not. In this section, I will argue that we take ourselves to be axiomatically justified, in an important normative sense, before we have done any investigation, and then go searching for that justification. Furthermore, I will argue that this normativity represents a BonJourian non-sensory belief, and to the extent that we can know normative facts, we must also commit to some a priori, non-sensory knowledge.
The previous sections have concerned themselves with which of our inferences were to be justified, if anything was. But is anything justified, in the end? It seems to me that there is an important pragmatic and normative sense of epistemological justification, that is revealed by the question “What is different about a world in which our concept of ‘justification’ is working properly versus one in which it is not?”. Namely, I argue it is that justified inferences are worthy of coordinating our behaviour around. To the extent that a belief is perceived as justified is the extent to which it is considered appropriate to build it into our plans for action for the future. In this sense, we might think all justified inferences and beliefs contain some spark of normativity, in that they are conditionally relevant to our future plans, and are guiding points on our map, which will inform how we act according to our interests.
But are we justified at all? Is the entire set of our beliefs, sensory experiences and inferences normatively relevant, worthy of guiding our behaviour? I take it that it obviously is, but that this can only be given a priori, axiomatically. We take it to be true, before any investigation, that ‘our best guess’ is good enough for guiding our behaviour, and indeed, what else could we do, so long as we are pre-committed to act. This normative implication, that our best guess is good enough, furthermore, seems contingent. We can perhaps imagine a Buddhist practitioner that manages to totally severe themselves from the normative implication of any facts; to them, there is nothing that implies in any sense that action is called for. That to us, certain facts do appear to be pre-loaded with normativity, and indeed, if we commit to the existence of any normative facts that we are able to know, it seems are in the same position as BonJour, as these non-sensory beliefs call for a mode of non-sensory, a priori knowing.
It is a plausible project to rebuff such an argument by attempting to find some method by which normative facts are sensorily justified. This could be accomplished perhaps, by a Humean sense of approbation (moral-good-feeling), and disapprobation (moral-bad-feeling). This answer could be considered the Devittian rejoinder to the BonJourian question of normative facts, in that we would be arguing that ‘normative facts’ are in fact mere posits that aim to provide causal explanations for our ‘moral’ sensory experiences.
The problem with such an objection as I see it, is that it runs into a version of the Moorean open-question argument (16, 2012). The Moorean open-question argument is a metaethical argument that puts forth that for any attempt to explain normativity with a naturalistic occurrence, such as sense experience, it remains an open question: “Is normativity entirely accounted for by x?”. In this case, then, it is to insist on the BonJourian gap between any sort of sense experience, and any sort of non-sensory normative implication. All of this, is essentially to notice the Humean is-ought problem through a BonJourian lens. Without some non-sensory initial spark, it is easy to think of all of our sense data having as much import for our behaviour as a movie screening at a cinema; it would all be there, but it wouldn’t mean anything for us. To the extent, therefore, that we realise normative facts as things we can know, it seems in a BonJourian fashion we commit to some non-sensory form of a priori justification.
All of this, however, need not lead us to non-naturalism, nor to invent some form of ‘rational insight’. Rather, I think this is easily squared away with evolutionary naturalism. In the same way that we readily accept that birds ‘know’ to fly south in the winter, and spiders ‘know’ when and how to spin webs, it should strike us as relatively unsurprising that we come pre-loaded with certain normative implications for certain sensory experience: we ‘know’ we ought to eat when we are hungry, ‘know’ we ought to avoid that which causes pain. Furthermore, it need not be the case, that we come pre-loaded with all normative implications, but indeed, all we need is some sort of spark or initial inferential rule by which we might go on to further develop others.
Our story with justification, then, is akin to that of chicken and egg. It seems like we need justification to get justification, so how does it ever arrive? I claim, then, that we never make the leap from a situation of ‘no justification’, to ‘justification’, but rather we start, axiomatically pre-loaded as justified, with a beginning set of axioms of inference and normativity. Certainly we do not retain these initial rules, but equally certainly are never without them, entirely adrift. As such, it seems that to the extent that we take ourselves to be justified, and that an important part of justification is normative implication, and that normative implication cannot be entirely sensory, that we commit to some non-sensory, a priori justification. I have argued, however, that this need not lead us to BonJourian ‘rational insight’ but is easily square-awayable with our routine evolutionary naturalism.
8. Conclusion
Where have we got to? Have we succeeded in our goals of blending the BonJourian, Devittian, and entitlement solutions to the a priori/a posteriori divide? In the beginning, I put forth that all three solutions captured desirable intuitions, but came at a heavy price. BonJour’s solution captured the intuition of requiring justification for all claims, but seemed to invent a form of justification in order to deal with non-sensory beliefs. Devitt’s proposal, that all could be accounted for in a Quinean fashion, achieved an elegance of unified justification, but came the cost of circular justification. Finally, the entitlement solution maintained that we could take ourselves to be axiomatically justified, but did so ironically at the cost of dispensing with the requirement that all claims had to be justified. Furthermore, I argued that both BonJour and Devitt circled around the problem of epistemic regress, with both of them circling entire inferential systems and asking the regressive question “Yes, what justifies that?”. I proposed we might come to a better conception via three moves. The first move was to endorse the same foundationalism based on sensory experience that both BonJour and Devitt endorse, as a way of securing at least some secure justified grounds. Secondly, I argued that almost all non-sensory claims such as metaphysical claims, claims about the unobserved past, present and future, and claims about mechanisms of theoretical science could be accounted for in terms of Quinean ‘best guess’ posits. Thirdly, I argued that we in fact take ourselves to be justified in treating our inferences with normative implication as an axiomatic starting point, and that this normativity represents a special case of BonJourian non-sensory belief, which cannot be accounted for, in a Humean is-ought fashion, by sensory input. I put forth that this need not lead us to BonJourian non-naturalistic rational insight, but is easily squared away with our standard evolutionary naturalist picture, in which we come pre-loaded with the normative implications of certain sensory experiences. How ought we regard the a priori/a posteriori distinction, then? Intact, but changed. From Devitt, we are to interpret almost all daily justifications as a posteriori, referring either to foundational sense experience, or causal posits. From BonJour, we plug the regress with an a priori starting point spark. And from the entitlement solution, we take ourselves, from the very beginning, to be justified.
References
BonJour, Laurence. 2013a. “In Defense of the a Priori.” In Contemporary Debates in Epistemology, edited by Matthias Steup, 177–184. Newark: John Wiley & Sons, Incorporated.
BonJour, Laurence. 2013b. “Reply to Devitt.” In Contemporary Debates in Epistemology, edited by Matthias Steup, 195–197. Newark: John Wiley & Sons, Incorporated.
BonJour, Laurence. 2013c. “Last Rejoinder.” In Contemporary Debates in Epistemology, edited by Matthias Steup, 200–201. Newark: John Wiley & Sons, Incorporated.
Casullo, Albert. 2015. “Four Challenges to the a Priori–a Posteriori Distinction.” Synthese 192 (9): 2701–24. https://doi.org/10.1007/s11229-013-0341-x.
Devitt, Michael. 2013a. “Reply to BonJour.” In Contemporary Debates in Epistemology, edited by Matthias Steup, 197–199. Newark: John Wiley & Sons, Incorporated.
Devitt, Michael. 2013b. “There Is No a Priori.” In Contemporary Debates in Epistemology, edited by Matthias Steup, 185–194. Newark: John Wiley & Sons, Incorporated.
Jenkins, Carrie. 2012. “A Priori Knowledge: The Conceptual Approach.” In The Continuum Companion to Epistemology, edited by Andrew Cullison. Continuum.
Klein, Peter D. 2012. “Infinitism.” In The Continuum Companion to Epistemology, edited by Andrew Cullison, 73–91. New York: Continuum.
Kornblith, Hilary. 2000. “The Impurity of Reason.” Pacific Philosophical Quarterly 81 (1): 67–89. https://doi.org/10.1111/1468-0114.00095.
Moore, G. E. 2012. Principia Ethica. United States: Dover Publications.
Quine, W. V. 1951. “Main Trends in Recent Philosophy: Two Dogmas of Empiricism.” The Philosophical Review 60 (1): 20–43. https://doi.org/10.2307/2181906.
Teixeira, Célia. 2022. “Epistemic Analyticity Reconsidered.” Pacific Philosophical Quarterly 103 (2): 280–92. https://doi.org/10.1111/papq.12347.
Williamson, Timothy. 2013. “How Deep Is the Distinction Between a Priori and a Posteriori Knowledge?” In The a Priori in Philosophy, edited by Albert Casullo and Joshua C. Thurow, 291–312. Oxford: Oxford University Press.
Wright, Crispin. 2004a. “Intuition, Entitlement and the Epistemology of Logical Laws.” Dialectica 58 (1): 155–75.
Wright, Crispin, and Martin Davies. 2004b. “On Epistemic Entitlement.” Proceedings of the Aristotelian Society, Supplementary Volumes 78: 167–245.