Pure Epistemic Pluralism

Abstract

Section 1 offers some stage setting. Pluralist views have recently attracted considerable attention in different areas of philosophy. Truth and logic are cases in hand. According to the alethic pluralist, there are several ways of being true. According to the logical pluralist, there are several ways of being valid. Section 2 introduces epistemic pluralism through the work of Tyler Burge , Alvin Goldman , and William Alston . In the work of these authors, we find pluralism about, respectively, epistemic warrant (Burge), justification (Goldman), and desiderata (Alston). Section 3 investigates what rationale can be given for epistemic pluralism. Drawing on the literature on truth pluralism , I suggest that one rationale for adopting a pluralist view in epistemology is its wider scope. Pluralism puts one in a position to accommodate a wider range of cases of epistemic assessments. In Sect. 4, I do two things. First, I explain why the distinction between epistemic monism and epistemic pluralism is most interestingly drawn at the level of non-derivative epistemic goods . Second, I make the observation that, at a very fundamental level, the varieties of epistemic pluralism presented in Sect. 2 are not particularly pure in nature. This is because they are all combined with veritic unitarianism, i.e. the view that there are several epistemic goods but that truth is the only non-derivative one. What, other than truth, might qualify as goods of this kind? Section 5 offers some preliminary considerations on this question, drawing on the work of Michael DePaul and Jonathan Kvanvig . In Sect. 6, I present two kinds of collapse arguments, each meant to show that pluralism is inherently unstable. I first consider each argument in the case of truth and then transpose them to epistemology. In Sect. 7, I respond to both collapse arguments.

Earlier versions of the chapter were presented at the University of Connecticut, the European Epistemology Network Meeting (Autonomous University of Madrid, 2014), Kyoto University (2015), and the Pluralisms Research Seminar (Yonsei University, 29 April 2016). I would like to thank the following people for discussion: Michael Blome-Tillmann, Colin Caret, Annalisa Coliva , Aaron Cotnoir, Crawford Elder, Pascal Engel, Filippo Ferrari, Peter Graham, Mikkel Gerken, Jonathan Jenkins Ichikawa, Bjørn Jespersen, Masashi Kasaki, Michael Lynch , Sebastiano Moruzzi, Jennifer Nado, Sherri Roush, Nico Silins, and Jeremy Wyatt. Special thanks go to Luca Zanetti for especially illuminating discussions. While writing this paper, I was fortunate enough to be supported by grants no. 2013S1A2A2035514 and 2016S1A2A2911800 from the National Research Foundation of Korea. I gratefully acknowledge this support.

Appendix: G G

G ₁ , … , G _n are the non-derivative goods endorsed by the pure epistemic pluralist. F ₁ , … , F _m are the features that unify G ₁ , … , G _n according to the Unity Challenge. G _G is the property of being G ₁ or … or G _n . This appendix argues that G _G possesses the features F ₁ , … , F _m as a matter of necessity.

In giving my argument, I assume that the pluralist endorses two epistemic goods , G ₁ and G ₂, and that these epistemic goods are unified by (and only by) the feature of being a goal of enquiry:

(GE _G1):

G ₁ is a goal of enquiry.

(GE _G2):

G ₂ is a goal of enquiry.

I make the assumptions just stated for expository purposes only. Nothing in my argument hangs on them.

Now, (GE _G1) and (GE _G2) capture a feature of epistemic goods rather than ev-bearers. However, the ‘goal feature’ can be transposed to the level of ev-bearers as follows:

$\left( {\text{GE}_{G1}^{*} } \right)$ :

For all x, if G ₁(x), then x is a goal of enquiry.

$\left( {\text{GE}_{G2}^{*} } \right)$ :

For all x, if G ₂(x), then x is a goal of enquiry.

where x ranges over ev-bearers. ⁴⁶ Combining (G ^∗) (from Sect. 6) with the idea that the unifying feature of epistemic goods is to be a goal of enquiry yields:

(G ^∗GE):

G ^∗ is the property such that, necessarily, for all x: if G ^∗(x), then x is a goal of enquiry.

Here, G ^∗ is the supposed unique epistemic good. Above, I claimed that G _G —the property of being G ₁ or … or G _n —has $F_{1} , \ldots ,F_{n}$ necessarily. Assuming (G ^∗GE) supporting this claim amounts to showing the following:

(G ^∗GE−GG) Necessarily, for all x: if x is G ₁ or x is G ₂, then x is a goal of enquiry.

Let offer support for (G ^∗GE−GG). In order to do so, I rely on strengthened versions $\left( {\text{GE}_{G1}^{*} } \right)$and $\left( {\text{GE}_{G2}^{*} } \right)$:

$\left( {\text{GE}_{G1\square }^{*} } \right)$ :

Necessarily, for all x, if G ₁(x), then x is a goal of enquiry.

$\left( {\text{GE}_{G2\square }^{*} } \right)$ :

Necessarily, for all x, if G ₂(x), then x is a goal of enquiry.

(G ^∗GE-GG) is a consequence of $\left( {\text{GE}_{G1\square }^{*} } \right)$ and $\left( {\text{GE}_{G2\square }^{*} } \right)$.

$(GE_{G1}^{*} )$ and $(GE_{G2}^{*} )$ can be strengthened to $\left( {\text{GE}_{G1\square }^{*} } \right)$ and $\left( {\text{GE}_{G2\square }^{*} } \right)$ because the status of G ₁ and G ₂ as goals of enquiry is not a mere coincidence. It is modally robust. This modal robustness is explained by the modal robustness of the connection between the status of G ₁ and G ₂ as goods and their status as goals of enquiry together with the modal robustness of G ₁ and G ₂’s status as epistemic goods . Turn first to the modal robustness of the good–goal link. G ₁ and G ₂ are goals of enquiry because they are epistemic goods . This means that, in any possible case, if something is G ₁, then it is a goal of enquiry. Similarly for G ₂. So we have:

(G-GE_□):

Necessarily, for all x, if x is epistemically good, then x is a goal of enquiry.

Consider now the status of G ₁ and G ₂ as epistemic goods . They do not enjoy this status as a mere coincidence. The status of G ₁ and G ₂ as epistemic goods is modally robust—a matter of necessity—because they are non-derivative epistemic goods . The epistemic goodness of G ₁ and G ₂ is not due to their bearing some relation to some other good. The goodness of G ₁ and G ₂ resides solely in G ₁ and G ₂ themselves. Thus, whenever G ₁ is instantiated, this by itself—or on its own—is epistemically good. Similarly for G ₂. Here, I mean to use ‘whenever’ to indicate generality along two dimensions. The first dimension is intra-worldly: within a given possible world, the instantiation of G ₁ and G ₂ is always epistemically good. The other dimension is inter-worldly: the instantiation of G ₁ and G ₂ is epistemically good across all possible worlds. Taking on board these two dimensions of generality, we get:  

(G ₁-G^□):

Necessarily, for all x, if G ₁(x), then x is epistemically good.

(G ₂-G^□):

Necessarily, for all x, if G ₂(x), then x is epistemically good.

(G ₁-G^□) together with (G-GE_□) delivers $\left( {\text{GE}_{G1\square }^{*} } \right)$, while (G ₂-G^□) together with (G-GE_□) delivers $\left( {\text{GE}_{G2\square }^{*} } \right)$.

Claim: (G ₁-G^□) and (G-GE^□) entails $\left( {\text{GE}_{G1\square }^{*} } \right)$. Argument: assume (G ₁-G^□) and (G-GE)—that is, necessarily, for all x, if G ₁(x), then x is epistemically good and also, necessarily, for all x, if x is epistemically good, then x is a goal of enquiry. Now consider an arbitrary possible world w and an arbitrary ev-bearer x. Suppose G ₁(x) in w. Then, by (G ₁-G^□), x is epistemically good in w. By (G-GE^□), x is a goal of enquiry in w. Hence, in w, if G ₁(x), then x is a goal of enquiry. Since w and x were both arbitrary, we have: necessarily, for all x, if G ₁(x), then x is a goal of enquiry. A completely analogous argument shows that (G ₂-G^□) and (G-GE^□) entails $\left( {\text{GE}_{G2\square }^{*} } \right)$.

Having established $\left( {\text{GE}_{G1\square }^{*} } \right)$ and $\left( {\text{GE}_{G2\square }^{*} } \right)$, let us return to the task of showing that (G ^∗GE-GG) is a consequence of $\left( {\text{GE}_{G1\square }^{*} } \right)$ and $\left( {\text{GE}_{G2\square }^{*} } \right)$. Here is the argument: assume (GE _G1) and (GE _G2)—that is, necessarily for all x, if G ₁(x), then x is a goal of enquiry, and also, necessarily, for all x, if G ₂(x), then x is a goal of enquiry. Now, for an arbitrary world w and an arbitrary ev-bearer x, assume G ₁(x) or G ₂(x). Suppose that G ₁(x). By $\left( {\text{GE}_{G1\square }^{*} } \right)$, x is a goal of enquiry. Suppose that G ₂(x). By $\left( {\text{GE}_{G2\square }^{*} } \right)$, x is a goal of enquiry. Thus, whether G ₁(x) or G ₂(x), x is a goal of enquiry. Hence, x is a goal of enquiry. So, if G ₁(x) or G ₂(x), then x is a goal of enquiry. Since x was arbitrary, we can generalize on x: for all x, if G ₁(x) or G ₂(x), then x is a goal of enquiry. Since w was arbitrary as well, we can necessitate: necessarily, for all x, if G ₁(x) or G ₂(x), then x is a goal of enquiry. This shows that (G ^∗GE-GG) follows from $\left( {\text{GE}_{G1\square }^{*} } \right)$ and $\left( {\text{GE}_{G2\square }^{*} } \right)$, i.e. that if G ₁ and G ₂ are both goals of enquiry as a matter of necessity, then so, too, is the disjunctive property of being G ₁ or G ₂.

Let me restate what I take the significance of this result to be. Being a goal of enquiry is the feature that, in the argument just given, is assumed to be the unifying of the pluralist’s epistemic goods —and so the feature in terms of which the generic epistemic good G ^∗ is characterized. It has been shown that the property of being G ₁ and G ₂ has this feature—i.e. it is a goal of enquiry—as a matter of necessity. I thus take it that, by the lights of (G ^∗), being G ₁ or G ₂ is a generic epistemic good. Now, if G ^∗ is identified with being G ₁ and G ₂, we have a somewhat familiar good on our hands—namely, a specific case of G _G . As argued earlier, the pure epistemic pluralist does not have to worry about this specific epistemic good.

Now, the argument just given proceeded on the assumption that there are two epistemic goods and precisely one feature that unifies these two goods. Can the argument be generalized? Yes. As said, the assumptions of two goods and one unifying feature were made for the sake of exposition. Nothing in the argument seems to hang on either of them, and so it would seem that the argument can be generalized in at least two ways. First, the argument can be generalized to cover an arbitrary number of non-derivative epistemic goods , meaning in this particular context an arbitrary number of disjuncts. Second, the argument can be generalized to cover an arbitrary number of unifying features. Given non-derivative epistemic goods G ₁ , … , G _n what we have, then, is the conclusion that the corresponding disjunctive property will satisfy (G ^∗), i.e. it will possess the unifying features as a matter of necessity:

$\left( {G_{Fm - GG}^{*} } \right)$ :

Necessarily, for all x: if x is G ₁ or … or x is G _n , then F ₁(x), …, F _m (x).

The generalized argument supports the conclusion that the epistemic good characterized as the disjunction of the pluralist’s non-derivative goods, however many, is going to possess the unifying features as a matter of necessity.