1. Introduction
According to Ernest Sosa, if one wants to attain better knowledge, then one must first seriously face the prospect of establishing warrant for one’s beliefs, for there is an important difference between having already warranted grounds for one’s knowledgeable beliefs — i.e., beliefs that amount to knowledge — and using these grounds as a starting point from which knowledge progresses to better knowledge (Sosa, Reference Sosa2009, p. 170). Without these grounds, epistemic progress is curtailed due to two factors: one, an inability to ascertain where in our progress to better knowledge we already are, and two, consequently, an incapacity to further such progress. Sosa’s epistemological project therefore encompasses accounts of not only the constitution of our epistemic capacities, but also the ways in which we can become epistemically better off in our beliefs.Footnote 1 His project involves multiple dimensions, from the distinction between animal and reflective knowledge, to the multiple levels of aptness about belief, to even the manner in which abductive/normative explanations for our epistemic competences capacitate the acquisition of safer and more secure beliefs.Footnote 2
However, little has been discussed regarding some of the implicit commitments Sosa makes that undergird his project. Two of the most integral are the a priori warrant for the reliability of our epistemic competences, established through the inevitability of our reliance on such competences to have beliefs at all, and the augmentation of our competences’ warrant that only the abductive explanation of their workings can provide (Sosa, Reference Sosa2009, p. 151, Reference Sosa2011b, pp. 46, 58, Reference Sosa2012b, p. 313). I interpret these to be implicit commitments because, although Sosa spends much space discussing the establishment and augmentation of warrant for our knowledge, he devotes relatively little space to spelling out that these are based on, respectively, an a priori grounding and a specific modal intuition concerning abductive explanation.Footnote 3
Although there is enough significance in each commitment to address them separately, in general they jointly inform the possibility of progressing up the epistemic aptness hierarchy, from mere aptness, to meta-aptness, to knowing full-well (the condition of being apt-because-meta-apt — called, the ABMA-condition), and finally to securely knowing full-well (called, the secure ABMA-condition).Footnote 4 Aptness, meta-aptness, and the ABMA-condition are each conditioned on different competences, while the secure ABMA-condition entails the manifestation of the ABMA-condition on a much more safely guaranteed competence than what is required just for the ABMA-condition. In other words, what meta-aptness is to aptness regarding safety of belief is what the secure ABMA-condition is to the ABMA-condition in terms of safety of a belief’s underlying competence. Now, a priori warrant of our epistemic competences’ reliability explains the grounds for our aptness, meta-aptness, and the ABMA-condition, while abduction-augmented warrant of such reliability explains how the secure ABMA-condition can be garnered.
Therefore, understanding how the highest rungs of knowledge can be attained relies on an understanding of Sosa’s employment of a priori warrant and how it conditions the rest of the aptness hierarchy. I therefore focus in this article on how Sosa comprehends the a priori warrant of our apt beliefs, meta-apt beliefs, and ABMA-beliefs. I leave for a future project the fleshing out of the modal intuition behind the secure ABMA-condition.Footnote 5 Specifically, I argue for two claims:
-
(1) Sosa’s establishment of warrant for our competences’ reliability, as a function of the inevitability of our reliance thereof, is best viewed in a priori terms. Specifically, apt beliefs come from competences whose presupposition of reliability is established a priori due to us needing to rely on them to have beliefs at all. Moreover, apt beliefs can become ABMA-beliefs not only because of this a priori establishment of warrant for apt beliefs (ABMA-beliefs are apt beliefs, after all) but also because of the need to rely on competences that deliver the very perspectives on our apt beliefs by which they acquire the ABMA-condition in the first place.
-
(2) A proper understanding of Sosa’s more mature view of the a priori — i.e., seeing it as a mode of warrant establishment that exists alongside other modes of warrant establishment, such as the brute fact of the reliability of our disposition to take some of our beliefs at face value — clarifies how coming to terms with Sosa’s way of constructing ABMA-beliefs naturally lends itself to an explanation of why they are the best candidates to represent how our beliefs can become epistemically better off. Specifically, their warrant is established in a manner that most closely concords with our natural constitution as epistemic agents. Our epistemic competences exhibit features and functions that span multiple epistemic orders — ranging from those that explain aptness for our beliefs to those that explain their meta-aptness — that are integrated in a fashion that is insufficiently accounted for by meta-aptness alone.
Section 2 introduces Sosa’s aptness hierarchy. Section 3 motivates the need for the a priori establishment of warrant. The bulk of this article is spent, in Section 4, fleshing out Sosa’s account of the a priori, the role it plays in the aptness hierarchy, and why Sosa looks towards other foundational modes of warrant establishment to secure warrant for our epistemic competences. Section 5 then concludes.
2. Aptness, Meta-Aptness, and the ABMA-Condition
Sosa’s general distinction between animal and reflective knowledge is meant to highlight how one can be worse and better off epistemically. Construing better knowledge as consisting of “safer, better justified, and more reliably acquired beliefs,” at least in an earlier formulation, Sosa contends that “reflective knowledge is a higher epistemic accomplishment if” a belief meets a two-part requirement of understanding, one, “why it is […] apt, or true because [competently derived],” and two, “how the way in which it is sustained is reliably truth-conducive” (Sosa, Reference Sosa2009, pp. 136, 138). In other words, better knowledge amounts to understanding not only that a belief is true, but also how epistemic agents acquire and maintain true beliefs such that their faculties of belief acquisition and maintenance may be regarded as truth-conducive at all — i.e., as indicating a belief’s being both aptly derived from those faculties and sustained in one’s understanding.Footnote 6 Attaining this bipartite understanding of why a belief is truth-conducive and how it is sustained truth-conducively is supposed to augment a belief’s safety status by its being more reliably acquired, and thus to help it avoid being luckily true in the way that characterises Gettier agents.Footnote 7
This is meant to highlight a modal safety condition for reflective knowledge, in that the safety of one’s animal beliefs ought to be assessed if its epistemic status is to be augmented at all. This is, of course, still in line with Sosa’s espousal of fallibilism, so it is not required for reflective knowledge that one rule out every possibility that may contravene the truth of one’s beliefs (Sosa, Reference Sosa2009, p. 198). Indeed, if one does not infallibly know that P, then one can still have reflective knowledge that P without having to rule out ¬P.Footnote 8 It is also in line with his fallibilism that what one does have to rule out does not have to be so precisely and accurately defined that ruling it out always unproblematically constitutes a necessary condition for reflective knowledge.Footnote 9
Now, animal (first-order) knowledge consists in aptly derived beliefs, whereas reflective (second-order) knowledge requires not only an understanding of this apt derivation, wherein the animal belief is sustained as an object of one’s reflection, but also that this understanding be apt as well. If one’s reflection, tantamount to one’s epistemic perspective on animal beliefs, is not itself aptly made, then it simply would not amount to knowledge in the way how animal knowledge amounts to knowledge by the fulfilment of its aptness condition. Thus, the types of beliefs constituting our reflective knowledge can be dually characterised in their aptness: As composing our epistemic perspective, they are the apt second-order meta-beliefs featuring in our reflective knowledge — this aptness being proper to reflective knowledge itself — but as beliefs whose objects are first-order animal beliefs, they entail the meta-aptness of these animal beliefs (Sosa, Reference Sosa2011a, p. 13n). Meta-aptness expresses an apt animal belief’s apt perspectival context, whereby we can distinguish an apt belief perspectivally situated to be meta-apt from a belief that is apt because it is meta-apt. The former requires, at minimum, an apt epistemic perspective on one’s apt beliefs, while the latter requires that the perspective do epistemic work on those beliefs. It is only once a belief attains this (apt-because-meta-apt) condition — call it, ABMA-condition — that animal knowledge rises to a higher rung of reflective knowledge, to knowing full-well (Sosa, Reference Sosa2011a, Chapter 1).
This ABMA-condition enables the acquisition of epistemically better knowledge in the first place — specifically, better first-order beliefs — for apt beliefs that are not ABMA-beliefs could have easily been inapt had the situation been different. This means that the ABMA-condition is one whereby the very apt formation of beliefs can itself be more truth-conducively safeguarded via reflection,Footnote 10 such that ABMA-beliefs become safer, better justified, and more reliably acquired than mere apt-but-not-ABMA-beliefs.
Stated differently, for Sosa, apt beliefs “track nature,” meta-beliefs require “also awareness of how one knows,” and ABMA-beliefs augment a belief’s aptness when one’s reflective awareness precludes “the unreliability of [their belief-forming] faculties” (Sosa, Reference Sosa2009, p. 199). Here, talk of reliability and unreliability characterises belief-forming faculties and competences (faculties and competences, for short) that are structured in particular ways but can be broadly categorised, ranging from our basic sensorial faculties to those competences more relevant for conscious deliberation. There is a relevant distinction between faculties and competences though. Competences are instances of a general genus of dispositions that includes dispositions of physical objects to behave in certain ways under certain situations.Footnote 11 We can shed light on the distinction between faculties and competences by looking at their relation to knowledge. For the case of animal knowledge, it entails belief derived from a reliable first-order faculty but not vice versa, while belief derived from a complete first-order competence entails animal knowledge and vice versa. This is because, while faculty refers specifically to one’s internal sources of belief formation, competence additionally incorporates one’s dispositions to have knowledgeable beliefs. Therefore, aptness of belief is a feature of first-order competences, meta-aptness of belief is a feature of first-order competences evaluated by second-order ones, and the ABMA-condition of belief is a feature of first-order competences guided, in a way, by second-order ones.Footnote 12
Consequently, the initial question with which we started this discussion — asking how one ought to establish warrant for our knowledgeable beliefs — can be further specified as one of asking how one ought to establish warrant for our competences such that the transition from apt beliefs to ABMA-beliefs becomes motivated. This article explains how we can construct from Sosa’s works a viable account of this transition as based on a priori warrant establishment conferred by the inevitability of relying on our competences, and how these grounds express the connections Sosa sees between, on one hand, aptness, meta-aptness, and the ABMA-condition of our beliefs, and, on the other hand, our natures as epistemic agents. By these connections, our beliefs can become epistemically better off as ABMA-beliefs only because the latter express a type of belief whose warrant best realises our natures as epistemic agents in terms of our animal epistemic endowment and broad competence genus (Sosa, Reference Sosa2009, pp. 237–238, Reference Sosa2021, p. 212n). This is significant for our purposes since both terms are made at least partially intelligible by the a priori grounds of inevitable competence reliance. We owe, in part, the success with which we can talk about our animal epistemic endowment, and the broader generic forms tying some of our various competences together, to such grounds. The other part we owe to our abductive theorisation and explication of what our epistemic natures consist of.Footnote 13
However, before delving into the analysis proper, it is intriguing to note that Sosa’s employment of the a priori is sparingly mentioned in the literature.Footnote 14 Relatedly, much of Sosa’s later works focus on fleshing out a positive account of telic normativity, moving away from his initial involvement with the inevitability of needing to rely on competences in order to have beliefs at all and towards a more fleshed out account of what constitutes such competences and how they exhibit a teleologically metaphysical ordering (Sosa, Reference Sosa2015, Reference Sosa2021). In other words, it is a movement from an a priori mode of establishing warrant for our competences to a more a posteriori one,Footnote 15 or a movement from the task of determining that we have competences to one of filling in the details of what these competences are.Footnote 16 Still, the project of figuring the a priori into the picture of establishing that we have competences, as Sosa began, and that they are reliable because they inevitably are relied upon, has yet to be fully appreciated. Before discussing the a priori properly, it would benefit us to first contextualise it within Sosa’s broader conception of the general purpose of establishing warrant in relation to knowledge acquisition, and how this normally turns problematic.
3. Setting up the Problem of Establishing Warrant
To briefly outline, Sosa seeks an establishment of such warrant that avoids the problems associated with circularity and infinite regress. For Sosa (Reference Sosa2012b, §4), how one avoids circularity is by showing in a relevant, non-problematic manner that knowledge acquisition obtains, while infinite regress is avoided by not being constrained by the need to have that acquisition be independently checked.Footnote 17 The issue here is captured in Crispin Wright’s observation, in rather internalist leanings, of a general problem “with the apparent implication […] that knowledge of anything requires knowledge of the satisfaction of enabling conditions of that knowledge,” that being, “those who cannot police their knowledge/warrant cannot get any, [and] those who can have to accomplish some kind of supertask before they can get any” (Wright, Reference Wright2011, p. 30). In terms of gaining epistemic warrant or justification for, say, the aptness of one’s epistemic perspective, one may ask if they can even non-circularly obtain such justification in a manner independent from how the belief on which you have that perspective is justified by that very perspective. More precisely, one can ask whether the presuppositions themselves about our knowledge being justified must be independently evaluated for that knowledge’s warrant to be established.
Wright concludes that even attempting an answer places us in a dilemma. Answering Yes amounts to “infinite regress or eventual vicious circularity” (Wright, Reference Wright2011, p. 31); answering No, though, either makes our knowledge claims purely ad hoc, or susceptible “to the idea that one can get warrant for the relevant presuppositions just by getting it […] for the parent propositions and reasoning to the presuppositions as very immediate conclusions” (Wright, Reference Wright2011, p. 32).Footnote 18 Sosa obviously desires neither answer, and one might contend that Wright’s more internalist framing of the problem of warrant establishment ought to not bother Sosa, but it is difficult to see how acquiring presuppositional warrant for our knowledge can come in any other variants than these:
-
(1) warrant from our knowledge-relevant first-/second-order faculties directly forming/sustaining belief in that proposition (circular),
-
(2) continuing warrant from other sources/faculties (regress),
-
(3) warrant from inference from the claimed proposition itself (invalid/trivial), or,
-
(4) non-inferential warrant (brute claim).
(1), (2), and (3) are obviously problematic due to the parenthetical comments, and (4) risks ad hoc presuppositions. However, Sosa’s approach, fleshed out more fully in Section 4, is a reinterpretation of (4) whereby warrant is established for our competences by way of a presupposed reliability of their constituent faculties, presupposed in a rational albeit non-knowledgeable way.
4. External Warrant and the A Priori
Section 4.1 introduces the sense in which the grounds of our epistemic warrant are a priori. Section 4.2 expounds on what Sosa means by a priori warrant and how it can accommodate aptness, meta-aptness, and the ABMA-condition in light of the argument that our competences are reliable because we cannot help but rely upon them. Finally, Section 4.3 further motivates this reading of Sosa’s use of the a priori by seeing how it can be naturally applied within a response to the criticism that competences can remain unreliable despite our need to rely on them.
4.1. Reliability as a Function of Inevitability of Reliance
Sosa’s appeal to the a priori is primarily an externalist one, in that it is meant to ground warrant for our competences that deliver apt beliefs without needing to appeal to the reasons we have for the warrant of any one of our beliefs. Sharon Mason’s understanding of the internalism/externalism distinction is instructive here:
[A] theory of knowledge is externalist if the ability to make an assessment of one’s own epistemic standing is not required for knowledge. A theory of knowledge is internalist if the ability to make some kind of assessment of one’s own epistemic standing is taken to be a necessary condition for knowledge. (Mason, Reference Mason2019, pp. 240–241)
This lines up nicely with Sosa’s own epistemological externalism, given his denial of the idea that what a belief must attain to become knowledgeable is it “having the backing of reasons or arguments” (Sosa, Reference Sosa2009, p. 168). Sosa’s externalism therefore makes any reason-based claim to knowledge ultimately need to rely “on already attained knowledge of the external” (Sosa, Reference Sosa2009, p. 170), whether animal or reflective.
We can specify this externalism more concretely through Sosa’s formalisation of a broad condition for knowledge attainment in terms of, T: “[a] belief X amounts to knowledge if and only if it satisfies conditions C” (Sosa, Reference Sosa2009, p. 174), wherein C constitutes the justificatory conditions that form belief as knowledgeable belief. One may ask, though, whether T is true, and whether knowledge truly does consist in a belief that satisfies C. To avoid arguing for T circularly or in a regress, one ought not to, respectively, attribute knowledge as part of C’s explanatory role for knowledge, or further argue that the knowledgeable belief, that knowledge consists in a C-satisfying belief, itself meets C (Sosa, Reference Sosa2009, pp. 174–176). The epistemological theory, T, is an externalist-friendly one because how one justifies knowledgeable belief as one that meets C does not have to itself appeal to independent justification via reasons — the justification is self-supporting in a way, albeit qualified such that it must be through meeting the specific conditions in C, and not any other, that belief can manifest as knowledge (Sosa, Reference Sosa2009, pp. 176–177).
How ought one theorise about C’s conditions? For Sosa, good theorisation derives from abductive reasoning about the constitution of our first- and second-order competences — i.e., their Skill, Shape, and Situation conditions.Footnote 19 What does such reasoning look like? Sosa asks, for instance,
[i]s it proposed that when ordinarily we accept the presence of a hand before us, we do know, and know on the basis of an abductive inference; or is it proposed rather that in such circumstances we have resources that would enable us to know if only we used those resources to make effective abductive arguments? (Sosa, Reference Sosa2009, p. 179)
Sosa denies this second proposal,
since it leaves our ordinary [first-order] beliefs in a position like that of a theorem accepted through a guess or a blunder, […] one that we have not come close to proving at the time when we are just guessing or blundering. (Sosa, Reference Sosa2009, p. 179)
He also doubts the viability of the first proposal if it is meant to infer abductively a theory about external reality from mere phenomenal qualities. This is because, if one does not simply stipulate for such qualities “the external reality to be inferred” from them, then non-phenomenal characteristics of external reality just would not be available “if we restrict ourselves to beliefs about the [qualitative] character of our experience” (Sosa, Reference Sosa2009, pp. 179–180). This grants the possibility of inferring from mere phenomenal characteristics the non-existence of external reality. So, if we cannot just infer knowledge of the external world from reasons about our phenomenal experiences, then without just stipulating such knowledge, what options are we left with?
Sosa’s (Reference Sosa2009, p. 151) answer is twofold: one, we must rely on abductive explanations about why our capacity to know is reliably tied to the world; two, most importantly, such explanations are epistemically substantiated because, in a sense, they are constructed through the appropriate use of those competences whose very reliability we are trying to explain. The second part of the answer is meant to appeal to the fact that, before we can judge and evaluate our knowledge claims, and before we can theorise about how we know, inevitably we must rely on and use those competences constituting our epistemic capacities. Consequently, for Sosa, not only can we not know our competences “to be reliable except by relying on them at some level,” but we do know that our competences are reliable and under what conditions they can be relied upon (Sosa, Reference Sosa2011b, p. 46). How this knowledge manifests is based primarily on an implicit apprehension of these reliability conditions before our understanding of our competences becomes subsequently fleshed out by their proper appropriation in abductive theorisation about these very conditions (Sosa, Reference Sosa2012b, p. 313).
This all may seem dubiously circular, for what is occurring is that one must rely on the reliability of one’s competences to theorise about that very reliability reliably! Surely epistemic warrant derived in this fashion is illegitimate. Sosa disagrees, as there is something about the inevitability of relying on our competences that legitimises the claim that there is implicit knowledge of their reliability.Footnote 20 This is a subtle claim, for Sosa is not merely assuming that the reliability of our competences is as such due to arbitrarily stipulating it as such; instead, they are reliable in their capacity to make us understand their own reliability, a capacity that is conditioned on the inevitability of our reliance on them — i.e., the inevitability of our reliance thereof capacitates their reliability as such and our understanding of said reliability.
There are two distinctions we need to make here. First, the fact that we understand something through the reliable doxastic deliverances of a competence is not the same as the fact of our capacity to understand something as a function of a competence’s capaciousness to deliver reliable beliefs, capacious in the epistemic potency of that very capacity. Second, and similarly, that we understand is not the same as the inevitability of our appropriation of the capacity to understand to understand in some limited fashion. These distinctions reflect the difference between the epistemic significance of our beliefs and the epistemic significance of our inability to fully escape having beliefs, or to escape an understanding in terms of those beliefs. In effect, our competences, at the most fundamental analysis, amount to whatever we must use to have knowledge at all. Indeed, for Sosa, the presupposition of reliability for our rational competences does “have a kind of a priori [epistemic] status” by way of our inevitability of reliance thereof (Sosa, Reference Sosa2011b, p. 58, italics added) — i.e., it is seemingly a priori true that we have such reliable competences by virtue of our inevitable reliance on them, however a posteriori constituted they may be in animal or reflective knowledge. This constitution is what we need to discover a posteriori, fleshed out via abductive explication and theorisation.
Another way of seeing how the a priori fits in with Sosa’s earlier epistemological project is by situating it within the context of his transcendental argument, at the end of Knowing Full Well, against scepticism about the reliability of our competences (Sosa, Reference Sosa2011a, pp. 152–157). Sosa argues that the claim that “our [competences] do not have disabling origins (e.g., ones that involve powerful and systemic deception)” is not only reasonable but is also “clearly [the] best rational option […] to take” (Sosa, Reference Sosa2011a, p. 157n16). This is to be opposed to either disbelief or even suspension of judgement on the matter on account of these two options being “epistemically self-defeating” (Sosa, Reference Sosa2011a, p. 155). This is because they implicitly appeal to, for their reliability, the very competences that these options are questioning, thereby defeating justification for either option. In other words, if I even doubt that my competences are reliable in a general sense, then I cannot know that I would thereby be reliable in asserting that my competences may be unreliable (Cormik, Reference Cormik2020; Neta, Reference Neta2016; Silva, Reference Silva2013).
It is easy to see how one’s epistemic position, by going through the above argument, on the reliability of their competences takes on an a priori character. Without having any a posteriori apprehension as to the particular constitution of any one of our competences, Sosa’s transcendental argument can only function to establish a priori warrant for the presupposition of reliability of our competences — i.e., there is a sense in which we cannot help but rely on our competences to even lodge doubts about them.
This is not to say that the function wholly succeeds. To say that our competences have gained a priori warrant because we cannot be rationally coherent in our doubts about their reliability is to say something that is still consistent with our competences being objectively unreliable — i.e., the a priori establishment of warrant is insensitive to the facts. Now, Sosa does seem to register this when discussing how the reasons provided by the transcendental argument are not “ultima facie” (Sosa, Reference Sosa2011a, p. 157n16) — i.e., warrant can be defeasibly established for what is ultimately false — but even if the argument can only provide ultima facie reasons to take one’s competences as reliable, if it provides reasons at all, there is still a way to undermine even this capacity for reasons provision. The basis for this is simply the possibility that lodging reasonable doubts need not appeal to the reliability of competences to doubt. Indeed, a reasonable doubt can be sourced from either a reliable or unreliable competence to doubt. After all, the evaluative label of reliability does not apply to doubts in the manner that it applies to beliefs, given that, between the two, only beliefs have content for which truth would be a relevant category — e.g., it is nonsensical to call true an instance of doubting that there is a red ball in front of me in the way we might for believing that there is one. As such, it should be reasonable to doubt the general reliability of our competences, for doing so could avoid considering, or being obliged to consider, any reason we might think we have that our competence to doubt is reliable, thereby incurring no charge of incoherence in our doxastic attitudes.
Notwithstanding, the above points show how the a priori fits rather neatly within Sosa’s conception of warrant establishment, whether this conception ultimately pans out in guaranteeing some extent of reliability for our competences, either generally as referring to the believer inevitably relying on competences to have beliefs at all, or to the sceptic inevitably relying on competences to have doubts at all. However, to truly appreciate how Sosa’s notion of ABMA-beliefs illustrates not only his more mature stance on how the a priori of warrant establishment relates to warrant establishment more generally, but also why ABMA-beliefs are the best candidates of belief to represent how our beliefs can become epistemically better off, we need to flesh out a number of technical details and distinctions.
4.2. Directions of Warrant Establishment for the Aptness Hierarchy
4.2.1. Set-Up
Ram Neta offers a way of clarifying what this a prioricity of warrant establishment may mean by involving it in a tension between establishing warrant for the presupposition of reliability about our reason-based knowledge sources — these, considering Sosa, are composed of our rational competences — and establishing warrant for the knowledge itself. On one hand, these rational competences are what we have been discussing in relation to abductive theorisation specifically and to beliefs more generally — i.e., those on which we must rely to reason about our competences and have beliefs at all, respectively. On the other hand, as I discuss in Section 4.3, we possess a-rational competences that are responsible for other kinds of doxastic deliverances besides belief. Supposedly, warrant for both our rational and a-rational competences can be established a priori simply given the inevitability of our reliance thereof. This distinction is helpful because focusing on our rational competences anticipates Sosa’s use of a-rational competences to address issues — e.g., circularity — that the a priori reading of warrant establishment has. Therefore, outlining the relation between Sosa’s a priori and our rational competences illustrates just how ostensibly inescapable the role of the inevitability of competence reliance is in the way in which Sosa conceptualises the aptness hierarchy, at least up until the ABMA-condition, as involving both rational and a-rational competences.
Neta asks,
what is the difference between
-
(a) the epistemic status of the presupposition that must be aptly made for the reason-based source to deliver knowledge on that occasion, and
-
(b) the epistemic status of the knowledge delivered by that source on that occasion?
The smaller the difference between (a) and (b), the closer Sosa comes to the view that, for our knowledge-sources to give us knowledge, we must have a priori knowledge of truths concerning the success and reliability of our reason-based knowledge sources. But the bigger the difference between (a) and (b), the closer Sosa comes to the view that the operation of a single reason-based knowledge source on a single occasion can create knowledge without the benefit of coherence: the knowledge-source, operating by itself on a particular occasion, can give us reflective knowledge of its own reliability, and its own safe success on that occasion. (Neta, Reference Neta2011, p. 16)
Call (a) the presupposition status of knowledge, and (b) the belief status of knowledge. Positive determination of (a) and (b) amount to presupposition warrant and belief warrant, respectively. What is being asked is how Sosa wants to establish the warrant of both the rational basis of our knowledge, its presupposition warrant, and the knowledge itself, its belief warrant. In other words, we can interpret presupposition warrant as involving warrant underlining the reliability of our rational competences and belief warrant as involving warrant for the knowledge they deliver — specifically, for the aptness of the delivered knowledgeable belief. Given that knowledgeable beliefs come from a complete competence, both presupposition and belief warrant involve kinds of warrant that are relevant to the warrant of rational competences, or, rational competence warrant for short.
With this in mind, we can interpret Neta’s two possibilities as follows:
-
(1) A rational competence’s warrant must be established by prior warranted presuppositions about its success and reliability before it can deliver apt beliefs if the warrant for both the presupposition and the apt belief come as a package deal.Footnote 21
-
(2) It is not the case that a rational competence’s warrant must be established by prior warranted presuppositions about its success and reliability before it can deliver apt beliefs, if the warrant for both the presupposition and the apt belief do not need to manifest concurrently.
Accordingly, presupposition warrant is necessary for rational competence warrant if both belief and presupposition warrants come as a package deal. However, if belief and presupposition warrants do not come as a package deal, then given the connection between knowledgeable belief and competence, the only relation we are justified in asserting is that belief warrant is equivalent to rational competence warrant, which is already the case if both belief and presupposition warrants did come as a package deal. It is just that, in the presupposition/belief package deal case, belief warrant is established by rational competence warrant, whereas in the non-package deal case, the reverse obtains. This gives us a reinterpretation of (1) and (2):
-
(1*) Presupposition warrant necessarily establishes rational competence warrant, itself being what necessarily establishes belief warrant, if belief and presupposition warrants come as a package deal.
-
(2*) Belief warrant can establish rational competence warrant if belief and presupposition warrants do not come as a package deal.
Now, since warrant for our rational competences is established a priori by the inevitability of our reliance thereof, then the above analysis gives us ample reason to conclude that all three warrant types — rational competence, belief, and presupposition — are also a priori. It is easy to see this in the package deal case, given that, if rational competence warrant establishes belief warrant, then it is reasonable to infer that this accomplishes the fact that the a prioricity of the former is inherited in the latter — i.e., the a prioricity of one explains that of the other. Furthermore, since presupposition and belief warrants come as a package deal, it is also reasonable to infer that their a prioricity does to. For the non-package deal case, since belief warrant can establish rational competence warrant, it would be strange if belief warrant is able to explain everything except for the a prioricity of rational competence warrant. Therefore, belief warrant is also a priori in character since, naturally, the best way for the explanatory relation to obtain is for the a prioricity of one to explain that of the other. This leaves us with presupposition warrant in the non-package deal case. Nonetheless, for Sosa (Reference Sosa2011b, p. 58), it too attains a priori status.
With that out of the way, we still have some unanswered queries. If presupposition and belief warrants do not come as a package deal, then how are they nonetheless related, if at all? Also, and perhaps most pressing for us, how would one establish the truth of (1*) and (2*)? Answering these will help in understanding how Sosa can be viewed as espousing both (1*) and (2*).
Now, what would it mean for (1*) and (2*) to be true? For (1*), one reason it could be true is if something about what it means for presupposition and belief warrants to come as a package deal is also something about what it means for presupposition warrant to ultimately establish belief warrant. For (2*), one reason it could be true is if something about what it means for presupposition and belief warrants to not come as a package deal can also be something about what it means for belief warrant to establish rational competence warrant.
The case for (2*) is relatively straightforward in some contexts. Presupposition and belief warrants coming apart would lead to belief warrant being needed to establish rational competence warrant, given that presupposition warrant may not be available. This would be especially straightforward in those cases wherein belief warrant involves warrant for knowledge about the reliability of the very rational competence that delivers the knowledge that attains belief warrant — i.e., reflective knowledge of that competence’s reliability, as Neta (Reference Neta2011, p. 16) mentions. Moreover, the fact that (2*) is only straightforward in some contexts does not hurt its chances, given that (2*) requires only that, by presupposition and belief warrants not coming as a package deal, belief warrant can establish rational competence warrant, not that there must be such an occurrence. Additionally, since a prioricity is involved in both warrant types, (2*) permits explanation of a priori rational competence warrant by something other than a priori belief warrant. This explanatory burden can certainly be taken up by a priori presupposition warrant, which further strengthens the case for (2*) since this permittance is opened up by presupposition and belief warrants not needing to be taken together.
However, what would it mean for the tight association involved in (1*) between presupposition and belief warrants to guarantee the need for the former to establish the latter, via the former needing to establish rational competence warrant? Generally, (1*) necessitates two things from presupposition warrant’s concurrence with belief warrant. One, its function of excluding any warrant instances not characterisable as presupposition warrant from establishing a priori rational competence warrant. Two, its function of excluding belief warrant from establishing a priori presupposition warrant — i.e., that rational competences cannot be used to derive a priori knowledge of their own reliability. These functions are indeed quite reasonable on the grounds that presupposition warrant is already given as a priori due to warrant for our rational competences, established a priori due to inevitability of reliance thereof, undergirding belief warrant, which comes together with presupposition warrant. As such, the first and second functions become more assured since other sources of a priori warrant need not be called upon to establish rational competence warrant and presupposition warrant, respectively.
The main take away is that both functions of (1*) serve to establish presupposition warrant as not only the only way to establish rational competence warrant, but as itself not establishable by either rational competence or belief warrants. (2*) gives belief warrant a similar role to what (1*) gives to presupposition warrant. Belief warrant is warrant for a belief delivered from a rational competence, which in (2*) is warranted by said belief. Presupposition warrant is warrant associated with what, for Neta, is “knowledge of truths concerning the success and reliability” of our rational competences (Neta, Reference Neta2011, p. 16). However, in keeping (1*) and (2*) separate, this presuppositional knowledge in (1*) should not be delivered from the same rational competence that delivers the beliefs figuring in belief warrant in (2*), or at least it should not be included among such beliefs. What then delivers the presuppositional knowledge required to establish a priori rational competence warrant if presuppositional and belief warrants must be taken together? It is here that we must now bring up Sosa’s notion of a-rational competences — a-rational due to presuppositional knowledge not being delivered from rational competences — and its implicit connection with a priori warrant establishment.Footnote 22
First, though, let us construct a model from which this connection becomes more obvious, one that can be gleamed from Sosa’s aptness hierarchy. Interestingly, we can read Sosa’s aptness hierarchy as incorporating every direction of warrant establishment outlined above as functioning within distinct contexts. For presuppositional warrant establishing rational competence warrant, this represents external establishment of the warrant we have for our apt beliefs, external in the sense of being derived from warranted rational competences and not beliefs acting as reasons for said establishment. For belief warrant establishing rational competence warrant, this represents internal establishment of warrant, precisely because belief warrant can justify reasons for rational competence warrant. Given that belief warrant can express warrant for either apt, meta-apt, or ABMA-beliefs, presuppositional warrant establishing rational competence warrant makes sense in the context of how our apt beliefs derive their warrant from underlying competences and belief warrant establishing rational competence warrant makes sense when considering how meta-apt beliefs are made — i.e., meta-beliefs (apt second-order reflective knowledge) transforming first-order apt beliefs into meta-apt beliefs, perhaps by augmenting the warrant of those apt beliefs’ underlying competences through their capacity as an epistemically potent perspective on such competences’ reliability.
I noted earlier in Section 4.2 that we have good reason to believe that belief, presupposition, and rational competence warrants are all a priori. The adequacy of the above model is thus further augmented by incorporating a prioricity into its structure. I argue that we can flesh out such a model utilising the different directions of warrant establishment expressed in (1*) and (2*) to show how the warrant of both apt and even ABMA-beliefs must, according to Sosa, be a priori established.
4.2.2. Technical Exposition
We start with noting that the particular character of belief warrant differs depending on whether it establishes rational competence warrant or whether it is being established by rational competence warrant by being established by presupposition warrant. The picture I have in mind is as follows. Let us first consider the warrant we have for, respectively, a perceptual belief that there is a red ball and the first-order rational perceptual competence of our eyes. Call this belief, B1, and its corresponding competence, C1. The belief warrant associated with B1 is B1 warrant, the rational competence warrant of C1 is called C1 warrant, and the presupposition warrant behind C1 warrant is called p-C1 warrant. Like the warrant direction in (1*), we can express B1 warrant as that which is established by p-C1 warrant, the warrant for our presupposition of reliability of C1, by having B1 warrant be grounded on the a priori basis of inevitably relying on C1.Footnote 23 Next, we can augment B1 warrant to the warrant for the barest levels of our reflective knowledge that there is a red ball (call the meta-apt first-order belief constituting the base of this knowledge, B1*), by fleshing out the content of p-C1 warrant such that it establishes warrant for a rational competence capable of delivering B1*. Let us call this (B1*)-delivering competence, C1*. This transformation happens due to a separate warrant for a reflective perspective whose object is this fleshed out presupposition warrant of C1*, or, p-C1* warrant. Call this reflective perspective, B2, which is distinct from B1* — they are of different epistemic orders — and plays a role in the warrant establishment direction expressed in (2*). B2 may also involve the claim that “we might have only the barest knowledge that somehow through the eyes (say) we can competently access relevant truths about shapes and colors” (Sosa, Reference Sosa2011b, p. 58). B2 and B1* are, respectively, the second-order meta-belief and first-order meta-apt belief that constitute our reflective knowledge that there is a red ball.
In other words, our rational competence capable of supporting reflective knowledge that there is a red ball (C1*) is internally warranted from a perspective (B2) about the initial rational competence warrant of the first-order perceptual competence (C1 warrant) capable of supporting first-order perceptual beliefs that there is a red ball (B1). Specifically, this is done by C1 being set in greater relief by the belief warrant involved in a reflective perspective (B2 warrant) on the presupposition warrant justifying the (reflective knowledge)-supporting competence (p-C1* warrant) — i.e., the warranted presupposition of reliability of C1*, p-C1*, which we might have only the barest knowledge of as true. This is to be contrasted with B1’s being externally warranted, ultimately, by the inevitability of reliance of C1.Footnote 24
To summarise, this process deals with three distinct instances of belief warrant: an initial one expressing warranted first-order perceptual beliefs (B1 warrant), an augmented variant thereof expressing warranted first-order perceptual beliefs (B1* warrant), and a third one expressing a warranted second-order perspective (B2 warrant) accomplishing this augmentation by indirectly establishing B1* warrant through directly establishing its underlying p-C1* warrant. B1 warrant is underlined by its corresponding C1 warrant, a priori established by reliance inevitability considerations. We can now give an analogous story explaining how B2 warrant is underlined by its own a priori established C2 warrant, given that it represents a second-order reflective competence that must have been relied upon to deliver the reflective perspective in question. All three belief warrants involve apt beliefs, but only B1 is not also involved in a meta-apt belief; the second belief is a meta-apt animal belief, while the third belief is a reflective meta-belief.
The model I have just described implies a description of how we can finally derive an ABMA-belief that there is a red ball. Note that presupposition warrant can be further augmented, in an a posteriori manner, via increasingly expansive coherent reasoning about the presuppositions themselves — i.e., having B2 warrant be augmented to B2* warrant, in an analogous manner to how B1 warrant was augmented to that of B1* warrant by B2. In other words, this would obtain by having a further reflective perspective, B3, whose object is p-C2*, the presupposition of reliability of the rational competence underlining the newly warrant-augmented reflective perspective, that sets in greater relief C2, the rational competence underlining the initial reflective perspective. Here, B3 warrant could express another warranted belief that, say, what we are able to say about what our eyes allow us to know regarding shapes and colours is generally true but relative to the normal conditions of light and eye health.
Thus, while B2 is warranted by C2, B1* is warranted by the fact that p-C1* is the object of B2, which involves a coherent explanatory picture on the presuppositions of reliability underlying the rational competence capable of delivering the barest reflective knowledge that there is a red ball. Therefore, C2 warrant establishing B2 warrant amounts to an external grounding while B2 warrant establishing p-C1* warrant is derivatively internal because it is ultimately grounded in the externally established warrant exhibited by C2 warrant establishing B2 warrant. Analogously, B2* is warranted by the fact that p-C2* is the object of B3, which involves a subsequent coherent explanatory picture on the presuppositions of reliability underlying the rational competence capable of delivering the initial coherent explanatory picture involved in B2. Resultantly, B3 warrant establishing p-C2* warrant is derivatively internal on account of the latter’s basis on the externally established warrant exhibited by the fact that B3 warrant is established by warrant of C3, the rational competence capable of delivering B3.
A significant consequence of the above outline is that beliefs of an n+1 epistemic order employ the belief of the nth order that is relevant to the former belief’s object. This allows the fact of B3 warrant establishing p-C2* warrant, along with warrant for p-C2 and C2 being a priori established due to reliance inevitability considerations, make at least partial sense of our interpretation of Sosa in Section 4.1, that the inevitability of our reliance on our capacities entails both their reliability as such and the reliability of our understanding of their reliability. This makes more sense if we regard competences of different orders (e.g., C1 and C2) as incorporated within a broader competence genus correlative with our capacity for reflective knowledge (Sosa, Reference Sosa2009, p. 237n), meaning that we cannot escape, in our perspectives about warrant for a capacity’s reliability, the deliverances of that very capacity. In other words, inevitably relying on a competence is a way to partially derive its augmented warrant due to the competence’s warranted deliverances being involved in higher-order deliverances that take a perspective on it — e.g., the claim that we can access perceptual truths via our eyes is already part of the higher-order belief that what we say about what our eyes can access is generally true. In short, first-order aptness already paves part of the way for first-order meta-aptness. Now, does this also pave the way for the ABMA-condition?
Given that B2’s object is p-C1*, and that B1* is meant to represent the barest reflective upgrade of B1 — i.e., representing a better epistemic position due to the addition of an apt epistemic perspective on the initial perceptual belief that there is a red ball (this makes B1* both apt and meta-apt) — B2* then represents the barest reflective upgrade of B2. This, ideally, should grant a modest upgrade to the epistemic position of p-C1*, and consequently of B1*. The question then becomes: Is this epistemically augmented B1* an instance of ABMA-belief? I argue below that the kind of warrant needed for an ABMA-belief is a priori, which is not a surprising result given that p-C1 and p-C2 warrants, establishing warrants for competences that deliver, respectively, apt beliefs and meta-beliefs, are themselves a priori established by the inevitability of relying on their associated rational competences. The somewhat surprising result, however, is that the a prioricity of ABMA-belief warrant makes it so that the ABMA-condition is not just of a higher level than meta-aptness, but of a different type of level altogether.
Remember, we want not just meta-apt but also apt-because-meta-apt beliefs — e.g., a belief that there is a red ball that is not just more meta-apt than its less justified version, but one whose aptness is grounded on and explained by its meta-aptness. As such, it would not do to simply keep improving the justification of the meta-apt belief, B1*, by increasingly grander epistemic perspectives, or even higher-order beliefs, like those of B3, B4, etc. What is required is a distinct kind of warrant grounding, hinging on the meta-apt status of the ABMA-belief. Recall that a belief is meta-apt due to an epistemic perspective on that belief, and a perspective is warranted because the underlying second-order competence is warranted — a belief that is apt is true because it is reliable, and reliability is a function of our competence being of working order. If B1** expresses the ABMA-belief that there is a red ball, then its warrant would still, based on the warrant direction of (1*), be externally established by p-C1** warrant, or, warrant for the presupposition of reliability of a rational competence, C1**, capable of delivering ABMA-beliefs. However, given that ABMA-beliefs require at least that second-order epistemic perspectives be somehow about first-order beliefs, perspectives that grant them the ABMA-condition, then p-C1** warrant would need to be based on a quite different kind of second-order competence compared to something like C2*, which, as indicated above, is already responsible for the epistemically augmented B1*.
Let us provisionally call this ABMA-responsible second-order competence, X. First, note that X can be neither C2 nor C2*, representing, respectively, a second-order competence delivering apt meta-beliefs and a better justified version of said competence now delivering meta-apt meta-beliefs. This is because B1** warrant cannot be established by either C2 warrant or C2* warrant: The former too easily conflates the warrant direction for meta-beliefs with that of ABMA-beliefs, while the latter employs the kinds of meta-beliefs that are not aptness-grounding in the manner needed for the ABMA-condition. In other words, nothing about the different warrant types to which C2 and C2* warrants translate necessitates for B1** a categorically distinct warrant compared to what are involved in B2 and B2* warrants.Footnote 25
What we need therefore is a differently justified second-order competence delivering beliefs whose warrant establishes ABMA first-order beliefs. Let this ABMA-establishing competence and its delivered belief be, respectively, C2M and B2M. Let us also see how B1** warrant being established by C2M warrant plays out. The first obvious direction of warrant establishment is from C2M to B2M. Recall that B2M warrant represents warrant for the second-order perspective that grants the type of meta-aptness that grounds ABMA-beliefs, and so B2M warrant should also establish B1** warrant perspectivally a posteriori through B2M’s object being p-C1**, or, the presupposition of reliability for the rational competence capable of delivering the ABMA-belief, B1**. We thus have derived B1** warrant being established by C2M warrant, justifying taking C2M as X. Now, how are we to understand C2M warrant? C1** delivers B1**, but does C2M deliver apt beliefs ala B2, or meta-apt beliefs ala B2*? The deliverances from C2M cannot themselves be ABMA-beliefs, for then we would need to inquire as to how these beliefs are warranted, and thus the warrant directions would regress infinitely up the epistemic order hierarchy.
Effectively, understanding the difference between meta-aptness and the ABMA-condition is, for Sosa, to understand the difference between C2 and C2M warrants, or, between meta-aptness-sufficient second-order aptness and ABMA-condition-sufficient second-order warrant. Now, progress in understanding this difference can be made by comparing how B2 warrant establishes p-C1* warrant with how B2M warrant establishes C1** warrant. Recall that the former direction of warrant establishment makes sense of the claim that the inevitability of our reliance on our capacities entails both their reliability as such and the reliability of our understanding of their reliability, in that inevitably relying on a competence is a way to partially derive its augmented warrant. p-C1* warrant, after all, starts out as p-C1 warrant before its associated competence delivers a belief that becomes involved with the reflective perspective, B2, which augments p-C1 warrant into p-C1* warrant. Stated differently, normal meta-aptness can start out as normal aptness.
However, C1** warrant does not start out as C1 warrant, because these are two categorically distinct types of aptness in how they are grounded, even if both warrant types are directed at the same proposition, i.e., that there is a red ball. Therefore, if B1** did not start out as normally apt, then how was it initially justified? Surely the belief did not start out as unwarranted, for that would contravene Sosa’s insistence that establishing warrant for our beliefs does not come about by a process of ex nihilo garnering of justification (Sosa, Reference Sosa2009, Chapter 10.5). It also could not have started out as meta-apt, as this would somehow require the transition from a rational competence capable of delivering meta-apt beliefs to one capable of delivering ABMA-beliefs, such as what would be exhibited in a transition from C1* to C1**, which requires a transition, of establishing warrants, from C2 warrant to C2M warrant. However, this represents two categorically distinct types of aptness in terms of how they ground different first-order warrants — i.e., B2 warrant establishes p-C1* warrant, and thus B1* warrant, a posteriori while B1** warrant, although established through B2M warrant, is also a priori established by way of C1** warrant in a fashion that is unavailable to B1* warrant by way of C1* warrant.
4.2.3. Exposition of A Priori Warrant Establishment of ABMA-Beliefs
To suggest a way forward, remember that the ABMA-condition implies that one’s perspective on a belief does epistemic work on that belief. Since a belief being apt means that it is true because competently derived, in this case by a complete competence, ABMA-beliefs are therefore those beliefs whose meta-aptness makes good on the notion that they are true because competently derived. In other words, the epistemic perspective on an ABMA-belief is how that belief was competently derived — more specifically, it is the second-order competence delivering said perspective that matters for how the first-order ABMA-belief is derived competently. This paints a scene of first-order beliefs coming into being with an already formed second-order perspective conditioning their aptness — e.g., my belief that there is a red ball in front of me is warranted because I have already formed a perspective on it claiming that we can access various perceptual truths with our eyes. This scene differs from both normally apt beliefs that can persist without such a perspective and normally meta-apt beliefs that were already apt prior to the perspective. As well on this picture is the important distinction in second-order beliefs between those whose aptness is irrelevant to first-order aptness and those whose aptness coincides with, mutually reinforces, or perhaps even equates to first-order aptness.
What all this means is that, while the inevitability of relying on a competence — functioning to partially derive, a priori, that competence’s augmented warrant — is usually taken to apply to the relation between apt and meta-apt beliefs, what ABMA-beliefs offer is a way to have this a priori derivation of augmented warrant be fully complete. This is because only through ABMA-beliefs can the inevitability of our reliance on our competences entail both their reliability as such and the reliability of our understanding of their reliability, which conforms to Sosa’s notion of correlative justification of warrant established at the same time for both belief and competence (Sosa, Reference Sosa2009, p. 232, Reference Sosa2011b, pp. 51–52). After all, even though our perspective on ABMA-beliefs is distinct from them, both are mutually integral in ways not found between apt second-order and meta-apt first-order beliefs. Said differently, while it is true that generally we cannot escape from the deliverances of our first-order perceptual competences when forming a perspective on them, we specifically cannot escape the perspectives themselves when forming ABMA-beliefs that are of a lower order than them.
However, we need to more precisely determine how this fully complete a priori derivation of augmented warrant comes into being. It is this needed reliance on the epistemic perspective that must be leveraged to explain why we can have ABMA-beliefs as opposed to just meta-apt beliefs. The issue is that we can redescribe our reliance on meta-aptness-relevant perspectives to be analogous to perspective reliance for ABMA-beliefs: We need to rely on epistemic perspectives to even have meta-apt beliefs at all. Indeed, we need to rely on meta-beliefs to have both meta-apt beliefs and ABMA-beliefs. How, then, can the inevitability of competence reliance be able to distinguish between establishment of the ABMA-condition and that of mere meta-aptness? Thankfully, the fact is that the package deal between ABMA-beliefs and their ABMA-condition-granting meta-beliefs is not identical to that between meta-apt beliefs and their associated meta-beliefs — i.e., meta-apt beliefs start out as apt yet non-meta-apt beliefs although ABMA-beliefs do not. This is because said fact opens up space for a viable explanation of the needed distinction in a priori derivation of warrant that differentiates competence reliance for meta-aptness establishment from that for ABMA-condition establishment.
Simply, while we must rely on some competences to have meta-apt beliefs, only other competences can fulfil the role of establishing warrant for our ABMA-beliefs. In other words, the competences we must rely upon to have perspectivally contextualised apt beliefs (meta-aptness) are not necessarily the competences we must rely upon to have perspectivally augmented apt beliefs (ABMA-condition). The a priori grounding of warrant comes from assuming that we have different kinds of warranted beliefs and consequently seeking an explanation for this difference in terms of the inevitability of relying on different competences underlying such beliefs.
We have now given substance to the claim that, according to Sosa, what grounds the reliability of our knowledge-conducive reasons — whether apt, meta-apt, or apt-because-meta-apt — is a matter-of-fact reliance on their underlying first- or second-order competences — e.g., C1 for apt beliefs, C1*/C2* for meta-apt beliefs, and C2M for ABMA-beliefs — and this constitutes some implicit grounding for said reasons. Additionally, and most revelatory, this implicit grounding expresses an a priori grasp, knowledge even, of those competences’ inherent reliability, structured around the inevitability of our reliance thereof. This implicit knowledge then gets substantiated in its warrant through explanatory abduction a posteriori, fleshing out presuppositions of reliability about this knowledge, which is primarily external.
There are important distinctions though. While p-C1 and p-C2 experience a priori warrant establishment via inevitability of reliance on their corresponding competences, and p-C1* and p-C2* experience a posteriori warrant establishment via abductive inferences sourced from the epistemic perspectives (B2 and B3, respectively) on them, only p-C1** experiences both a priori and a posteriori warrant establishment in the way that it does through the perspective, B2M, taken on it. This is not to conflate the two warrant types, but to locate them appropriately: p-C1** warrant only comes about through B2M warrant, indicating the kind of a posteriori warrant establishment that only epistemic perspectives on a competence’s presupposition of reliability can give, but C2M and p-C1** warrants come as a package deal, implying that the a priori mode of warrant establishment comes about due to the inevitability of relying on both C2M and C1** for the corresponding beliefs they deliver. This package-deal reliance grounds a posteriori warrant establishment, in that it is possible only because of the baseline a priori establishment of the reliability of competences like C2M. This is in part resultant from the fact that C2M warrant does not require a posteriori warrant establishment from some perspective taken on it, as well as because, without C2M warrant, the a posteriori warrant-augmenting functionality of (B2M)-involved abductive theorisation about C1**’s reliability would not be established as itself a reliable function.
Let us further make sense of how both warrant modes can co-exist. Note that, although we deploy reasons in our abductive inferences for a posteriori warrant establishment, once those inferences are soundly given, they cannot be defeated by reasons in the way that internally grounded knowledge can, precisely because what is defeated in the latter are claims dealing with what we know, not claims dealing with how we come to know at all. For instance, inferring p-C1** warrant from B2M warrant is a reasons-based process whose aim is the warranting of a single belief, B1**, while this inference already being sound (true) is instead an indication of how we come to know through our belief, B1**, due to the causal establishment of p-C1** warrant by B2M warrant. The former case is an instance of internal warrant establishment and so is susceptible to defeat via reasons. The latter case is an instance of external warrant establishment and so is not susceptible to such defeat since it implies a priori warrant as a function of inevitably relying on this very causal process to have the ABMA-belief, B1** — i.e., in acquiring ABMA-beliefs, we inevitably must rely on our capacities for abductive theorisation.Footnote 26
For example, reasoning about the reliability of second-order faculties can often take the form of an infinite regress, such as when additionally establishing aptness for second-order beliefs necessitates an additional round of risk assessment besides what was required for assessing modal risk of first-order beliefs (Buhr et al., Reference Buhr, Jansen, Kiesling, Bahr and Seidel2016, p. 24) — e.g., the modal risk in the belief that there is a red ball may differ from the modal risk in the belief that one knows that there is a red ball in ways that preclude any risk evaluation from ever being practically settled. These requirements for modal risk assessment can compound indefinitely if the chain of reasoning is not halted. Sosa desires to halt the regress without losing epistemic warrant by, first, pointing out that we inevitably must rely on second-order competences if we want to reflect on our knowledge at all, and second, arguing that whatever we abduce as a theory explaining competence reliability does not have to itself be established by way of additional risk assessment, since its epistemic warrant would have already been established externally at the ground level by a priori reliance inevitability considerations.
4.3. The A Priori and the Dispositional
We have just been discussing how the a priori can apply across the aptness hierarchy even in establishing warrant for our ABMA-beliefs. Now, there is obviously a worry to be had: It is one thing to have a belief’s warrant be derived from an epistemic perspective thereof, but another thing entirely to conclude that this warrant is so because we cannot but rely on the underlying competences to have such a belief at all. After all, one may in fact be relying on something unreliable, and so the claim that something like B2M warrant grounds C1** warrant is not guaranteed just because we cannot but rely on the corresponding competences.Footnote 27 Otherwise, this would amount to claiming that the belief that “we might have only the barest knowledge that somehow through the eyes (say) we can competently access relevant truths about shapes and colors” (Sosa, Reference Sosa2011b, p. 58) is warranted on account of us needing to rely on the perceptual competence to derive beliefs, such as the belief that there is a red ball in front of me. It is not enough for us to end with an adumbration of how the a priori matters to the aptness hierarchy, for if the a priori warrant itself fails to hold, then the entire hierarchy falls flat and loses all warrant. There is a line of argumentation Sosa deploys that can be read, through our direction of warrant establishment model in Section 4.2.2, as contributing to a fuller maturation of his stance concerning the a priori, which appeals to another mode of establishing the brute reliability of our competences that, especially in the case for ABMA-beliefs, is meant to go hand in hand with the a priori mode of inevitable competence reliance.
First, let us consider Sosa’s discussion of the justificatory standing we have to beliefs delivered by our first-order perceptual faculty since it gives possibly the most telling and illuminating description of his response:
Can’t we agree that there are dispositions to take visual experience at face value in play when we visually perceive, e.g., that a wall we see is red? Won’t there then be a state that is the basis for that disposition, and isn’t it plausible to describe the person as being disposed to believe that he sees something red upon having a visual experience as if he does? […] [T]hese dispositions require proper epistemic standing if they are to do their proper epistemic justificatory work, and this standing must not all be earned through reasoning based on antecedently justified deliverances of those very faculties. If we can agree on all that, I will argue further that such proper standing can be earned largely through the fact that the dispositions in question are just brutely reliable as part of a normal human’s natural endowment, even if that standing can then be boosted through rational means. Thus, it can be boosted through a perspective developed through the deliverances of such animal competences. Once it is thus boosted, it ascends to the level of reflective standing, which means that the belief is now defensible from certain doubts pressed against the alleged competence whence it derives, if any, and against the circumstances of the operation of that competence, and how propitious these are. (Sosa, Reference Sosa2009, p. 223n, italics added)
This rational boosting of our dispositions’ epistemic standing can come via abductive theorisation, but they are externally warranted to begin with due to them being brutely reliable. It is also important to note that these dispositions are not identical to the presuppositions of our competences being reliable, for such presuppositions are rational in nature — they are reasons, after all — and thus confer internal warrant.
The difference between presupposition and disposition can be further specified as follows: Our presupposition that perception is reliably veridical is a commitment that seems “to be empirical and not the sort of thing that could have a priori foundational status” (Sosa, Reference Sosa2009, p. 236n, italics added). This still coheres with our argument in Section 4.2 that, say, p-C1 warrant must be a priori established even though it involves a presupposition about competence reliability. This is because p-C1 warrant being a priori established does not entail that the presupposition, p-C1, itself has a priori foundational status, for what establishes p-C1 warrant — that being the argument that C1 is reliable because we inevitably rely on it and its deliverances — is instead what has such an a priori status. Perceptual competences are therefore presuppositional in their reliability and thus must have their warrant grounded elsewhere. However, dispositions are not presuppositional in their reliability, for they are a-rational, or, “not reason-based” (Sosa, Reference Sosa2009, p. 237n) — i.e., although they have as their objects the deliverances from the reason-based competence itself,Footnote 28 they are not grounded on presuppositions about that competence’s reliability. Dispositions to take, say, perceptual beliefs at face value are instead warranted because they are deliverances from a second-order competence that is epistemic in nature yet a-rational, thereby granting its deliverances external warrant. Consequently, these dispositions are reliable because their originating second-order competence is itself reliable (Sosa, Reference Sosa2009, p. 237n. cf., Sosa, Reference Sosa2015, p. 202). In short, (internal warrant)-granting presuppositions are to reason-based first-order competences what (external warrant)-granting second-order competences are to a-rational dispositions, at least in terms of what grounds whose reliability.Footnote 29
Sosa sees a-rational second-order competences as informing our reflective scientific knowledge: Such competences have their epistemic worth established simply from the fact of being part of one’s animal endowment about which scientific reflection aims at theorising abductively (Sosa, Reference Sosa2009, p. 238). It is important to note that Sosa is not being an infallibilist about scientific knowledge concerning our animal endowment, nor is he one to argue for the primary status of our perceptual knowledge’s reliability through post hoc abductive inference (Sosa, Reference Sosa2009, pp. 178–180). Rather, it is “[t]hrough our growing commonsense and scientific knowledge […] [that] we see our modes of rational basing [e.g., through presuppositions] and other belief acquisition [e.g., through dispositions] as sufficiently reliable” (Sosa, Reference Sosa2009, p. 240).Footnote 30 Said differently, a-rational second-order competences, among the other externally warranted competences we have, impart brutely reliable knowledge through our dispositions before abductive inference can establish a coherent reflective perspective on such knowledge.Footnote 31 As such, warrant for the reflective capabilities affording these epistemic perspectives is based in part on these a-rational second-order competences, or, more specifically, on the implicit commitments that we may intuitively abide by through our already warranted dispositions (Sosa, Reference Sosa2009, pp. 242–243).Footnote 32
Interestingly, for Sosa, our perceptual competences as well as the a-rational second-order ones justifying their reliability are both “constitutive of the faculty of vision” that he takes to be in a sense “fuller” than that of mere perception (Sosa, Reference Sosa2009, p. 237n). This would allow us to interpret Sosa’s more mature stance on the relation between first- and second-order competences as one where they constitute a more overarching competence that is epistemically justified via the brute reliability of its second-order a-rational competence. We thus have this picture of Sosa’s epistemological account: From the brute reliability of this latter group of a-rational second-order competences — reliable on the basis of constituting part of our baseline animal endowment — we derive epistemic warrant for the presuppositions of reliability for all reason-based competences, both first- and second-order, that are also — and this is important — a priori warranted because we cannot help but rely on them for our beliefs.
Well, what of externally warranted animal knowledge derived from a-rational first-order competences? Does it share in the justificatory bounty of, say, the fuller faculty of vision consisting of the relevant reason-based first-order and a-rational second-order competences? Now, it should not be a stretch to read Sosa as considering both a-rational first- and second-order competences as justified by being part of our animal endowment, given that a-rational competences in general are meant to be brutely reliable due to our natural constitution as epistemic agents. However, these two concepts of animal endowment and fuller competence genus are not necessarily equivalent since they select for different features. Nonetheless, given that a-rational second-order competences feature in both — e.g., our disposition to take perceptual deliverances at face value is part of both our natural epistemic endowment and fuller faculty of perception — and that it would be strange if our, say, baseline a-rational first-order reliability as it pertains to perceptual animal knowledge is not part of the fuller faculty of vision, then it should be permissible to consider a-rational first-order competences as warranted externally as a consequence of being part of a fuller faculty genus, and not just as a consequence of it constituting our natural animal endowment.
Let us summarise. For our reason-based competences, we can now name the relevant internal and external variants of warrant. Regarding our reason-based first-order competence, C1, we augment its warrant internally through a warranted epistemic perspective on it (B2), but C1 warrant is externally established in part on our already warranted disposition to regard its deliverances as reliable — call this disposition, D2 — and in other part a priori on the inevitability of relying on it for its deliverance, B1. D2, being a doxastic deliverance of sorts, is warranted generally by a reliable a-rational competence, C+2, whose warrant is established, as usual, a priori due to the inevitability of relying on it for D2, but not presuppositionally — i.e., there is no p-C+2 warrant. Externally warranted animal knowledge derived from a-rational first-order competences, furthermore, comprises a distinct, a-rational first-order epistemic dimension. We can represent this as C+1 and K+1 for the competence and its deliverance, respectively.
This interpretation of Sosa’s use of dispositional and a priori modes of warrant establishment is supported by the above-mentioned notions of animal endowment and fuller competence genus.Footnote 33 This is because the distinction between the a priori and the dispositional can be made to map onto the distinction between those notions rather comfortably. C+1 and C+2 are warranted a priori as part of the animal epistemic endowment we naturally have — i.e., animal because a-rational. C1 and C+2 jointly represent the broader genus of a perceptual competence we have, consisting of a rational competence delivering our first-order perceptual belief and an a-rational second-order competence delivering a disposition to take the rational competence’s deliverances at face value. This broader perceptual competence is warranted by D2, our brutely reliable disposition to take the deliverances of C1 at face value. This mapping further links dispositional warrant to a priori warrant across both the first and second epistemic orders.
The significance of this link will become readily apparent once we bring back the topic of ABMA-beliefs. However, for now, we can infer from the above discussion a nuanced picture of the relation between parts of our natural animal epistemic endowment and our broader perceptual competence: There are parts of this competence that can be explained as being part of such an endowment, but not all — e.g., the more reflective parts — and as such we can say that we are naturally endowed with both dispositional and a priori modes of warrant for our first-order beliefs.
It is intriguing that Sosa appeals to two modes — the a priori (inevitability of competence reliance) and dispositional (brute reliability of competences) — in establishing our competences’ warrants. One leverages our inability to rely on anything else besides our competences to have beliefs, while the other leverages the brute reliability of our disposition to take those beliefs at face value. The question now becomes: Which mode is more foundational? We have already considered the issues arising from letting the a priori be more foundational — i.e., we can inevitably rely on unreliable competences — but choosing the dispositional instead merely begs the question as to how one establishes the brute reliability of said dispositions. Alternatively, arguing for our dispositions’ warrant by citing that we cannot help but rely on our competences to have such dispositions is obviously problematic since it just reintroduces the a priori at the ground level.
A plausible solution comes, (un)surprisingly, from the above interpretation of ABMA-beliefs as first-order beliefs that come into being while already attended to by a second-order perspective conditioning their aptness. The rational competences delivering ABMA-beliefs need, for their warrant, epistemic perspectives that explanatorily flesh out the presupposition of reliability of such competences in a manner that does not make their deliverances merely meta-apt. This cannot be guaranteed solely just because we cannot but rely — for our ABMA-beliefs — on competences that deliver either first-order beliefs or the perspectives by which they become ABMA-beliefs. This guarantee may instead also come about by a brutely reliable disposition to take the deliverance of first-order competences at face value while also being relevant to both first and second epistemic orders. If so, then what turns up is a tight relation between the a priori grasp of competence reliability and the brute reliability of dispositions to take that competence’s deliverance at face value that is not found in either apt or meta-apt beliefs.
In effect, the fact that we cannot help but rely on our first-and second-order competences to have the beliefs we have perfectly coincides with our disposition to take those first-order beliefs at face value, at least when applied to ABMA-beliefs, whereby instead of having dispositions be just about first-order competences, they are now also indirectly about second-order competences. This is because ABMA-beliefs are characterised as first-order beliefs that are in some way responsive to or conditioned on second-order ones: For apt beliefs, the dispositional and a priori modes of warrant establishment only coincide within the same epistemic order, not across orders, while for meta-apt beliefs, the dispositional and a priori modes come together across different orders but not necessarily at the same time. In the meta-apt case, while the a priori mode is at work across two orders simultaneously, the dispositional is at best stuck at the mere aptness level, at the first order, given that what the disposition is about concerns only the first order. As such, only for ABMA-beliefs do we have a package-deal mode of establishing warrant, wherein one cannot have the dispositional without the a priori, and vice versa, across two epistemic orders simultaneously.Footnote 34
What this coinciding of the two warrant modes says is that warrant for our capacities and beliefs at different epistemic orders can be established all at once. This is significant because the warrant coming from aptness or meta-aptness exhibits worrying shortcomings: Aptness by itself cannot explain how our first-order beliefs can be better off as a function of the perspectives we have on them, and so it all too easily invokes in equal measure the problematic assertions that our competences are justified because we cannot help but rely on them to have beliefs at all and that our disposition to take their deliverances at face value is brutely reliable. Meta-aptness by itself — although contributing to such an explanation — does so while succumbing to concerns that surface by eschewing attempts at answering the question of whether the dispositional or a priori modes of establishing warrant for our competences is foundational; not answering this is to not answer how one ought to flesh out warrant because there is no resultant account of how first-order competences are responsive to the perspectives we have on their deliverances. This responsiveness is contingent on whether the a priori or dispositional is more foundational, for what is left for a first-order competence to respond to in order to be augmented in its warrant depends on what its baseline reliability already is.
The ABMA-condition, however, explains how our first-order beliefs can be better off in terms of both warrant modes being indispensable to one another, for this indispensability most closely concords with how warrant would naturally be established for our competences anyway, multivariably structured as they are to include multiple epistemic orders — e.g., the belief that there is a red ball whenever we see a red ball is epistemically better off as an ABMA-belief than a merely apt or meta-apt one since that belief being responsive to or even conditioned on a corresponding perspective on its aptness indicates two facts: one, the need to rely on both first- and second-order competences, and two, the disposition to take a first-order belief at face value as a function of our need to rely on not only a first-order competence but also a second-order competence. This makes the disposition more warranted by fleshing out its connection to two competences instead of just one.
Our fuller competence genus, including our first-order competences and our second-order dispositions to take their deliverances at face value, is part of our animal endowment, including our externally warranted first- and second-order competences, after all. As such, any competence to have beliefs that is structured a priori-ly and dispositionally in its warrant, in terms of these modes being mutually indispensable across different orders, will best respect this link between fuller competence genus and animal endowment. This is due to, as argued earlier in this section, how the link neatly maps over the distinction between a priori and dispositional modes of warrant establishment. Importantly, this is just to say that what are linked are the dispositional and a priori modes across different epistemic orders. Given that ABMA-beliefs are structured in this way — they allow for our dispositions to be about two different orders since we have to inevitably rely on competences of those two orders to have ABMA-beliefs at all — they are the best candidates to represent how our beliefs can become epistemically better off, because how their warrant is established most closely concords with our natural constitution as epistemic agents, whose competences constitute both a fuller genus and our natural animal endowment that span multiple epistemic orders.
For example, the meta-apt belief that there is a red ball in front of me is based in part on a perspective on the corresponding first-order competence, a perspective that may include the belief that our faculty of vision is generally reliable. We can also have an associated disposition to take our first-order perceptual beliefs at face value. However, even if such a disposition is brutely reliable, and the presupposition of reliability for the underlying first- and second-order rational competences of the meta-apt belief is a priori established in its warrant, how this disposition permits, or at least coincides with, the translation of such warrant to the kind ABMA-beliefs are endowed with is not a trivial occurrence. However this occurrence obtains, it would need to be differentiated from the kind of epistemic value that is afforded to meta-apt beliefs that are not ABMA-beliefs. Looking at the warrant structure of ABMA-beliefs gives us the way forward. The reflective belief that our vision is generally reliable becomes ABMA-conducive only if it indicates that the disposition to take our perceptual beliefs at face value figures in both our natural animal epistemic endowment and broader visual competence genus, where the second-order a-rational competence underlying this disposition is grouped together with the a-rational and rational first-order perceptual competences that deliver, respectively, our animal knowledge that there is a red ball in front of us and our more explicit, reasons-malleable belief that there is such a ball. Consequently, our disposition to take our first-order perceptual beliefs at face value is tied to the broader suite of our first-order perceptual competences. This means that our reflective perspective on our first-order belief that there is a red ball in front of us delivers an epistemic boon to both our animal epistemic endowment and broader visual competence genus. It is a holistic augmentation that is denied for reflective perspectives that only grant meta-aptness, not the ABMA-condition. This may also be how one explains the reflective perspective now being indirectly about said disposition because it directly pertains to the latter’s doxastic object, implying that the disposition itself is now relevant to both epistemic orders in a fuller way that is impossible for mere meta-aptness.
5. Conclusion
To reiterate, our purpose here is not to evaluate whether Sosa’s account of how the warrants for our knowledge are established succeeds. Indeed, even the above dual picture of the dispositional and a priori modes of warrant establishment may ultimately fail to secure such warrant: Our dispositions to take the deliverances of our competences at face value can themselves be unreliable, given that the very epistemic resources used to account for our animal endowment and the characteristics of, say, our fuller genus of perceptual competence originate from competences we have yet to guarantee warrant for. Our purpose here, recall, is to argue for two related claims:
-
(1) Sosa’s establishment of warrant for our competences’ reliability, as a function of the inevitability of our reliance thereof, is best viewed in a priori terms. Specifically, apt beliefs come from competences whose presupposition of reliability is established a priori due to us needing to rely on them to have beliefs at all. Moreover, ABMA-beliefs are established in their warrant due to a specific package-deal kind of reliance on both first- and second-order competences, both rational and a-rational, that is not needed to establish warrant for either apt beliefs or meta-beliefs.
-
(2) The association between the a priori mode of warrant establishment and that which comes from the brute fact of the reliability of our disposition to take some of our beliefs at face value is of significant interpretive value. In a more precise sense, coming to terms with this association clarifies how ABMA-beliefs naturally place themselves as the best candidates of apt belief to explain how our beliefs can become epistemically better off. Specifically, their warrant structure spans a range of epistemic orders in a particular arrangement that plausibly illustrates how we are constituted as epistemic agents through our animal epistemic endowment and broad competence genus.
Think of it this way. The establishment of warrant for our competences seeks to avoid the pitfalls of internalist justification for their deliverances, namely that positively assessing any one belief fails to necessarily secure positive outcomes for any other belief. Additionally, the establishment of warrant for our beliefs seeks to avoid the pitfalls of externalist justification for their underlying competences, namely that successfully thinking about competences in a way that neutralises the requirement of posing an assessment of our individual beliefs itself requires the warranted use of individual beliefs. Sosa’s justificatory approach being primarily externalist is susceptible to this latter pitfall, even with an appeal to the a priori mode of establishing warrant by noting the inevitability of competence reliance. The introduction of a dispositional mode of warrant establishment by itself also does not do much to aid in this regard because it too can be viewed in an externalist light.
Nevertheless, we can conceive of Sosa’s own brand of external warrant by illustrating more precisely how ABMA-beliefs are meant to secure for us a state of being epistemically better off in our beliefs. This is because these beliefs have as their foundation a tight connection between a priori and dispositional modes of warrant establishment: The a priori reliability of our first- and second-order competences and the brute reliability of our disposition to take first-order beliefs at face value come together to establish warrant for first-order beliefs in tandem with their corresponding second-order epistemic perspectives. This is supposed to afford epistemic betterment for our first-order beliefs because the a priori/dispositional connection entails that our disposition to take first-order beliefs at face value extends not just to first-order beliefs but to the second-order perspectives on which those beliefs are conditioned — we must inevitably rely on both first- and second-order competences to have ABMA-beliefs, after all — thus augmenting warrant for the first-order beliefs by bringing more competences into the already brutely reliable disposition’s justificatory purview. If it is possible to establish a priori warrant for our beliefs and the presupposition of reliability for the competences underlying them, then the warrant structure of our ABMA-beliefs and their corresponding rational and a-rational competences provides us with significant advantages in explaining how such warrant translates into the epistemic status of ABMA-beliefs being better off than either apt or meta-apt beliefs.
Data availability statement
I do not analyse or generate any datasets, because my work proceeds within a theoretical and mathematical approach.
Acknowledgements
I would like to thank Albert Atkin for his conceptual supervision on this work, as well as the journal’s anonymous referees for their guidance on extensive edits made to an earlier draft.
Funding statement
This work was supported by the International Macquarie University Research Excellence Scholarship (“iMQRES MRES”) (Road to Research). Allocation Number: 20213681.
Competing interests
The author has no conflict of interest to declare.
Appendices
A — List of Abbreviations
B1/B1*/B1**: Apt/meta-apt/ABMA first-order perceptual belief that there is a red ball
B2/B2 M : Apt second-order meta-belief whose object is p-C1*/p-C1**
B2*: Meta-apt second-order meta-belief whose object is p-C1*
B3: Apt third-order meta-belief whose object is p-C2*
C1/C1*/C1**: First-order rational perceptual competence delivering B1/B1*/B1**
C2/C2*/C2 M : Second-order rational reflective competence delivering B2/B2*/B2M
p-C1/p-C1*/p-C1**/P-C2*: Presupposition of reliability of C1/C1*/C1**/C2*
D2: Disposition to take B1 at face value
K + 1: A-rational animal knowledge that there is a red ball
C + 2: A-rational second-order competence delivering D2
C + 1: A-rational first-order competence delivering K+1
B — Warrant, Types, and Tokens
Sosa characterises internalist warrant for knowledge in terms of some belief B, “a complete token motivating rationale R, [and] the relevant rationale type R’ (of which R is a token)” (Sosa, Reference Sosa2009, p. 214n3). The relation between B, R, and R’ can factorise into causal and non-causal justificatory components. For the former, R motivates B “in virtue of some of [R’s] specific features. These will be features causally relevant to the believer’s hosting that particular token belief at that time, with its specific representational content” (Sosa, Reference Sosa2009, p. 213n). The constitutive causal relationality of the content grants B its tendency to be true, and, in situations wherein this tendency is actualised, the belief is apt, but only because said tendency originates from a general R’ that instantiates particular Rs reliably causing Bs being delivered as true (Sosa, Reference Sosa2009, pp. 213n, 214). In this case, B would belong to B’, a general doxastic type whose relevant content — “the sort that abstracts from [B tokens]” — is motivated from R’ (Sosa, Reference Sosa2009, p. 214n3).Footnote 35 For the non-causal justificatory component, it differs from the causal in terms of the source for B. According to Sosa (Reference Sosa2009), a causal source for B
would be the rational basis [R] of the constitutive belief [B], whereas in another sense the epistemic [i.e., non-causal] ‘source’ of one’s knowledge would be the source of the epistemic justification of [B], and this would be [B’s] having the [R] that is in the earlier sense its source. (Sosa, Reference Sosa2009, p. 216n7)
Having animal knowledge be warranted internally, then, amounts to attaining an R for a B without having R act as the epistemic justification of B, since B would not have that R — i.e., R is not knowingly deployed to justify B by the former’s membership as a token of some generally reliable R’. Externalist warrant for animal knowledge would constitute the reliable deliverance of B without some causal R-B relation — i.e., reliable causation of B would be because of some feature of the objectual fact represented by the propositional aspect of B, but not as some reason for B. Internalist warrant for reflective knowledge would be, non-causally, some B having some R, as in, apprehending R in its capacity as a token of some R’ that reliably produces true Bs in a general fashion as tokens of a generally true B’ type. Lastly, externalist warrant for reflective knowledge would be most like internalist warrant for animal knowledge in analogous terms of the latter’s reliance on an R-B causal relation without that relation being apprehended as epistemically justificatory. The salient difference, though, is in the perspective one takes regarding said apprehension: Just as R justifies B causally in internalist justification of animal knowledge, externalist warrant for reflective knowledge would entail the causal R-B relation that is within the perspectival jurisdiction of one’s reflective knowledge itself being justified causally by a broader R’-B’ relation, without this latter relation being apprehended as epistemically justificatory.
The relevant difference between internalist and externalist warrant for reflective knowledge, then, is that, while the latter admits of an already justificatory R’-B’ causal relation, the former further requires that relation to be apprehended, even known, to be justificatory. This knowledge requirement is the proper target for circularity/infinite regress critiques, as knowledge of R-B as justificatory either presupposes knowledge of a broader R’-B’ relation (circularity) or procrastinates the fulfilment of this knowledge requirement via an indefinitely broadening (R-B)-to-(R’-B’) perspective (infinite regress). Externalist warrant for reflective knowledge is therefore the most significant aspect of Sosa’s epistemology, because a sub-set thereof — externalist a priori warrant — plays a constitutive role in the R’-B’ relation’s justificatory function in a way where R’-B’ is quasi-known to be generally reliable (Sosa, Reference Sosa2009, p. 222). Quasi-knowledge avoids circularity/infinite regress critiques because it is not a proper knowledge requirement, but it also avoids, supposedly, ad hoc critiques because it is not tantamount to a random assumption.
C — A Priori Warrant and Inferentially and Hierarchically Basic Knowledge
To clarify why dispositions are associated with brutely reliable knowledge, let us discuss how similarly the disposition-presupposition distinction follows a distinction Sosa mentions between inferentially basic and hierarchically basic knowledge — IBK and HBK, respectively. To elucidate, IBK grounds inferential chains and so is not itself based on inference, in that “its sources are not inferential,” while HBK, even if inferentially based, can be “held unaccompanied by knowledge that its [inferential] basis […] is reliable” Sosa (Reference Sosa2009, p. 228).Footnote 36 Now, IBK is also HBK, since, by definition, one cannot know that sources for IBK are reliable lest the latter no longer remains inferentially basic. However, HBK is not necessarily IBK, since just because HBK is held epistemically unaccompanied in the above fashion does not mean that it cannot subsequently be so accompanied, in which case it would not amount to IBK. For Sosa, HBK affords “a basis for knowing that the sources of one’s knowledge are reliable” (Sosa, Reference Sosa2009, p. 228); however, this reasoning should also apply to IBK, given that it surely is possible for knowledge that the sources for our HBK are reliable to be inferentially basic. How this pans out can motivate an HBK-IBK distinction along a presupposition-disposition one.
Supposedly, the reliability of our perceptual competence can be presupposed. Do presuppositions constitute knowledge? Perhaps, but this would depend on whether the belief-forming processes underwriting them are reliably constituted. If yes, then how so? If dispositions are directed towards the presuppositions themselves, at least in the manner adumbrated in Note 29, then presuppositional reliability can be conferred by dispositional reliability. Dispositional reliability itself is a specific instance of a broader reliability genus constitutive of, as Sosa calls them, animal epistemic competences. These competences have their “proper epistemic standing,” one, derived “from being part of the animal endowment of an epistemically well-functioning human being,” and two, “attained without the aid of any [meta-belief] that endorses any such competence as reliable” (Sosa, Reference Sosa2009, pp. 238, 238n). This no-meta-belief condition on establishing our epistemic competences’ reliability may as well be an admission that dispositions — which come from such competences — attain IBK status.Footnote 37 These epistemic competences of our animal endowment are second-order and already externally warranted, since warrant for this type of competence grounds all our other instances of knowledge.
Consequently, do presuppositions attain HBK status? Surely they must if their reliability is derived (inferred) from that of dispositions, and we can have such presuppositions without having accompanying knowledge of this, or even that they are reliable at all. In other words, HBK can count as knowledge held unaccompanied by the inferential chain that had derived it, and where justification for the reliability of this inferential source of our HBK can come from the warrant of a disposition from which said inference originated — e.g., the presuppositions grounding our basic knowledge of the reliability of our perceptual capacity can be inferred from our disposition to take what we see as veridical. In a similar vein, dispositions count as IBK in the idea that our quasi-knowledge that said presuppositions are true is non-inferentially derived because it is not inferred from epistemic competences but is simply delivered from them. More specifically, presuppositions can be inferred, perhaps from the quasi-knowledge characteristic of our dispositions, while dispositions count as quasi-knowledge and are not inferred since they are a-rational — i.e., not reason-constituted.
Open access
