Epistemic humility and the principle of sufficient reason

ABSTRACT

According to the unrestricted version of the Principle of Sufficient Reason (PSR), every truth has an explanation. I argue that there is defeasible methodological justification for belief in an unrestricted PSR. The argument is based on considerations about our cognitive limitations. It is possible that our cognitive limitations prevent us from even recognizing the explanatorily open character of some propositions we can now represent: the fact that these propositions are explicable in the first place. If this is the case, then a proper recognition of our limitations and a standing goal of reasoning to attain complete explanatory knowledge justify the adoption of an unrestricted PSR.

KEYWORDS:

• Humility
• explanation
• principle of sufficient reason
• goals of reasoning
• inexpressible facts

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The role of the PSR in Spinoza’s thought and its overall significance are a matter of current debate. See, for example, (Garber 2015; Della Rocca 2015; Lin 2018).

2 One possible exception is Kant’s discussion of the PSR as a regulative ideal of human reasoning. See Kant ([1787] 1998). He accords a central role to the idea of completely unified knowledge as a necessary goal of human reasoning. For recent discussion of Kant’s views on the PSR, see Longuenesse (2005); Lu-Adler (2021). The present paper can be seen as one possible expression of the underlying Kantian idea of the PSR as a regulative idea of Reason, but I do not take myself to be contributing to an interpretative debate on Kant’s position.

3 Della Rocca has recently questioned the coherence of this distinction. See Della Rocca (2021). According to Della Rocca, the distinction between propositions apt for explanation and those not apt for explanation really amounts to a restriction of the PSR. I do not think introducing this distinction amounts to a restriction of the PSR. I agree that it is often difficult to determine objectively whether one is giving reasons to think some proposition in L is not apt for explanation as opposed to giving reasons to think that it has no explanation, even though it is apt for explanation. Nonetheless, the difficulty is dialectical and does not stem from lack of intelligibility in the distinction itself. One can get a grip on the distinction by reflecting on uncontroversial cases of propositions which are not apt for explanation, and I think some of the examples that I give in the text are uncontroversial.

4 A referee has pointed out that assuming explanation to be irreflexive may be controversial in this context. For example, the assumption has been questioned in the early modern period by Descartes (1984–91; AT 10:108–111), Spinoza (1985; E1Def1), and Leibniz (1989; AG 210) through the idea of self-causation. If grounding is taken to be metaphysical explanation, then the reflexivity of grounding could be one way to cast doubt on the irreflexivity of explanation. For recent arguments that grounding is reflexive, (see Jenkins 2011; Bliss 2018; Wilson 2014; Rodriguez-Pereyra 2015). I think that if allowing explanation to be reflexive implies that all truths are apt for explanation, then the resulting position is a reductio of the assumption of reflexivity. We ought to be able to distinguish between an explanatory relation which is a mere artefact of our representations and an explanatory relation which is objective. The target of this paper is explanation understood as an objective dependence relation. One of the ways such a distinction is secured is through the methodological rejection of views that would immediately trivialize the search for objective dependence relations which could play the explanation role. The view that all truths in a language are apt for explanation is one such view. After all, how can we be sure that a relation which allows for self-explanation is an objective dependence relation of the requisite kind? We take causal or nomological relations to be explanatory in the objective sense, for example, because they play a significant role in the discovery of new truths. An objective explanatory relation would make sense of and justify inferences such as inference to the best explanation. Relations which are mere artefacts of our representations could not presumably play such a role. It is hard to see how self-explanation can figure in the pursuit of new truths. This is not to say that positive argument cannot be given for this position but to say that there is a strong methodological presumption against acceptance of this position. There is then a strong methodological presumption against taking explanation to be reflexive.

5 Discussion on explanation of necessary truths and specifically mathematical ones can also be found in Pruss (2006, Ch. 1).

6 Dasgupta (2016) argues that essentialist facts are not apt for explanation. For an opposing view, see Raven (2021).

7 I remain neutral on the question whether there are any contingent truths which are not apt for explanation.

8 The difference between simply having no explanation and failing to be explanation apt then amounts to the following. In the former case, we can at least specify how the world would have to be for a given truth to have an explanation, even if in fact it doesn’t have one in the actual world. In the latter case, we are not in a position to specify what objective dependence relation could be invoked to explain the relevant truth. For example, some truths in the actual world may only be explicable if there were deterministic causation. We can at least specify what kind of connection among facts would yield an explanation in this case. When a truth fails to be explanation apt, no such connection can be in principle invoked.

9 What is an example of theoretical confusion that could underlie mistaken judgments about explanation aptness? Any example here is bound to be controversial but consider possible explanations of the existence of God. If any adequate explanation must invoke an ontological argument for the existence of God, and if the success of an ontological argument requires there to be an objective dependence relation between facts about existence and facts about meaning and there can be no such relation, then explanations of God’s existence would involve the relevant kind of theoretical confusion.

10 L is taken here to be a natural language.

11 For a dissenting view, according to which conservation laws are not really explained by symmetry principles but are simply correlated with them, see Brown and Holland (2004). Comprehensive discussion on the conditions under which symmetry principles can be said to play an explanatory role can be found in Lange (2016).

12 Notice that E-N only requires that for each member of the series of languages there are some explanatorily open propositions in that language which have explanations in a future language. It does not require that all explanatorily open propositions in a given language have explanations in a future language. It also doesn’t require that the set of explanatory propositions (the set of accepted explanatory truths) be held constant. The restriction to some explanatorily open propositions can cover cases in which propositions cease to be explanatorily open in future languages because discarded theoretical vocabulary means they cannot be formulated and are no longer taken to be true. A referee suggests the example of the transition from classical aether theory to special relativity whereby an older and no longer adequate language is discarded together with the explanatory connections it made possible. I thank the referee for making me clarify this point.

13 According to alternative views, humility does not really involve evaluation of one’s epistemic states but rather a tendency to underestimate one’s self-worth. See Driver (2001, ch.2). Alternatively, it involves a tendency to minimize one’s intellectual self-importance including intellectual status and entitlements. See Roberts and Wood (2007, ch.9). Even if epistemic humility includes these other features, they are not relevant to the epistemic context in the way the disposition to recognize one’s limitations is. I set aside these additional components to humility for the purposes of this paper.

14 One might wonder whether the ultimate source of justification for U-PSR is really circular. If inductive inference from past experience justifies the assumption that there are cognitively inaccessible explanations, a PSR sceptic can certainly block this inference. As others have observed, reliance on induction seems to presuppose the validity of the PSR (Della Rocca 2010; Pruss 2006). If there was a break from past experience, this would be inexplicable and PSR prohibits inexplicable discontinuities in experience. Moreover, if epistemic humility itself has a rational basis in past experience, i.e., in species-level observations on the types of limitations cognitive agents like us face in general, then appeal to epistemic humility does not seem to help with breaking the circle of justification. This is a general problem about the relationship between induction and the validity of the PSR which I cannot resolve here. I doubt that PSR ultimately justifies inductive inference. The latter is too basic to admit of justification. If any justification is possible, it is likely to appeal to constitutive considerations about the role of inductive inference in constituting what can count as a reason for belief.

15 Criticisms of various versions of these arguments can be found in Kleinschmidt (2013). Della Rocca (2010) does not discuss restricted versions of the PSR. For extensive discussion of various versions of these arguments, see Pruss (2006, Ch. 1 and Ch. 17).