Ideas, Evidence, and Method: Hume's Skepticism and Naturalism Concerning Knowledge and Causation

Abstract

This is a rich, ambitious, and original study. Graciela De Pierris aims to give a new account of the relationship between Hume's skepticism and his naturalism. Notoriously, Hume gives skeptical arguments targeting our best methods of inquiry, such as inductive reasoning. So, how can he—in good intellectual conscience—endorse and rely on those very methods, in his own naturalistic investigation of the human mind (his “science of man”)? Some scholars answer that, despite his skeptical arguments, Hume does not ultimately conclude that we lack justification for beliefs formed through induction or the other methods of inquiry that he employs as a scientist of man. Others answer that his skeptical arguments do not target his own preferred conception of these methods.De Pierris rejects both of these approaches. In her view, Hume argues for a form of skepticism that is “radical,” in that it targets his own preferred conception of our best methods of inquiry—the one that he endorses, as a practicing scientist of man. However, his “radical skepticism” and his “naturalism” can coexist because they constitute two different but mutually complementary “standpoints,” which must both “remain permanently available to all reflective thinkers” (22). In developing and defending this view, De Pierris offers an original interpretation not just of Hume but of early modern philosophy more broadly. She claims that Hume's views have been distorted by scholars who approach them from contemporary perspectives, such as cognitive psychology or naturalized epistemology. She therefore aims to understand Hume's skepticism and naturalism by placing them in the philosophical context of his own time. In doing so, she gives a novel account of the major intellectual fault lines in early modern philosophy.De Pierris divides early modern philosophers along two cross-cutting lines. First, she divides them based on their theories of mental content and “ultimate evidence.” Descartes and Locke accept what she calls “the presentational-phenomenological model of apprehension and ultimate evidence” (hereafter, PPM). According to PPM, we can be acquainted with an “ostensively given” particular content (for example, a particular sensory image) prior to grasping any general contents, such as formal rules of inference, and “ultimate evidence” derives from such acquaintance with particular contents. In opposition to these philosophers, Leibniz accepts the “logical-conceptual model,” according to which we must grasp a system of universal logical rules prior to any particular content, and “ultimate evidence” derives from this system of rules. (De Pierris's conception of “ultimate evidence” seems to have several components, including at least the following: ultimate evidence is the best possible evidence available to us; it confers certainty; and all other evidence derives from and depends on it.)With respect to this first division, De Pierris argues that Hume's theory of impressions and ideas is a version of PPM and is therefore a normative epistemological theory, not just a descriptive psychological one. Hume's version of PPM is the basis for his skeptical arguments, including his arguments for radical skepticism. Here, De Pierris assigns particular importance to Hume's distinction between natural and philosophical relations. As she interprets it, this distinction concerns two different kinds of mental activity: (i) unreflective, associative processes (natural relations) and (ii) reflective assessments of the evidence that particular contents are, or are not, related in certain ways (philosophical relations). Her view that Hume's notion of a philosophical relation is normative (insofar as it concerns the assessment of evidence) is original and informs her interpretation of his argument about induction in Treatise 1.3.6, discussed below.De Pierris's second division is between “mechanical philosophers,” who accept the “demonstrative ideal” of natural knowledge, and philosophers who accept Newton's rival, inductivist ideal. According to the former group, causal relationships are akin to geometric ones—an effect is a “necessary, quasi-geometric consequence” of its cause (164)—and science aims at a priori, demonstrative knowledge of them. Descartes and Leibniz accept this view. And, although Locke denies that we can attain a priori, demonstrative knowledge of nature, he accepts this as an unattainable ideal; this explains why he denies that we can have a genuine science of nature. In contrast, Newton rejects the demonstrative conception of natural knowledge, even as an unattainable ideal. According to his rival methodology, science ideally aims not at demonstration but at inductive “proof”—that is, inductive inference from a perfectly constant conjunction that we have experienced. De Pierris argues that Hume belongs in the Newtonian camp, based on several considerations, including a comparison of Hume's “rules by which to judge of causes and effects” with Newton's “rules for the study of natural philosophy” in the Principia. She sees this as “an important new argument against the skeptical realist interpretation of Hume” (187n258), which, she claims, wrongly assimilates his position to Locke's (14). However, she nowhere considers a striking passage from Hume's Dialogues Concerning Natural Religion, where Philo suggests that “the whole economy of the universe” may be conducted by a necessity like that of arithmetic, “though no human algebra can furnish a key, which solves the difficulty” (Hume 1947, 191). This passage presents a view like that which De Pierris attributes to Locke, and skeptical realist interpreters have made much of it, so it is disappointing that she neglects to discuss it.De Pierris interprets Hume's skepticism and naturalism in the context of these divisions among early modern philosophers. As she understands it, his naturalism consists chiefly in his endorsement of Newtonian inductive methodology. It is therefore a normative position, not just a descriptive one. Hume's naturalism also involves a distinctive conception of causal necessity, as a projection of “the normative force of our best inductive methods . . . onto nature itself” (176). If I understand this correctly, it means that, for Hume, our conception of causal necessity just is the conception of conformity to those regularities that our best inductive methods tell us that we should believe to obtain. This “epistemological notion” of causal necessity differs from both the mechanical philosophers’ quasi-geometrical notion and the regularity theory of causation that some scholars attribute to Hume, which are both “ontological” (176).Hume's skepticism, on the other hand, derives from his version of PPM. De Pierris finds two main lines of skeptical argument in Hume. One concerns mathematical knowledge; the other concerns natural knowledge and, relatedly, the nature of causation. Each argument has two stages. At the first stage, Hume targets his philosophical opponents’ conception of the knowledge in question. Numerous scholars think that his skeptical arguments end here. But De Pierris holds that these arguments have a second stage, at which Hume targets his own preferred conception of the knowledge in question—the conception that he endorses, as part of his naturalism. This is where his skepticism becomes “radical.”De Pierris's “skeptical inductivist” interpretation of Treatise 1.3.6 (and the corresponding passages of the “Abstract” and first Enquiry) is particularly interesting and original. She claims that this section gives a skeptical argument targeting “causation as a philosophical relation” (240)—that is, targeting the best evidence we can have, when reflectively comparing two phenomenologically given particulars, that they are causally related. For Hume, Newton's inductive methodology specifies what such evidence consists in: experience of a perfectly constant conjunction, which serves as the basis for an inductive “proof,” rather than a mere probability. According to Newton, an inductive proof relies, implicitly or explicitly, on the principle that nature is “ever consonant with itself” (as he puts it in the Principia). In Treatise 1.3.6, Hume argues that we have no good reason to accept this Newtonian uniformity principle. This is a normative, epistemological form of skepticism that targets Hume's own preferred conception of natural knowledge. Hence, De Pierris's interpretation differs from the “cognitive mechanisms” approach of Don Garrett and David Owen, who claim that Treatise 1.3.6 argues for a purely descriptive, nonnormative conclusion about the psychological processes that cause our beliefs about the unobserved. De Pierris's interpretation also differs from Peter Millican's (2002) skeptical interpretation of the corresponding section of the first Enquiry, according to which Hume's argument targets a conception of inductive reasoning that he himself does not endorse.For De Pierris, then, Hume gives skeptical arguments targeting what are, by his own lights, our best methods of scientific inquiry. So how can he continue to endorse and rely on those methods, as a practicing scientist of man? De Pierris claims that Hume's philosophy involves two different “standpoints,” each with its own normative standards. From the “standpoint of common life and science,” Hume endorses and employs Newtonian inductive methodology. From the “radical skeptical standpoint,” he argues that we lack justification for the uniformity principle that underwrites this very methodology. De Pierris claims that these two standpoints cannot be integrated but are nonetheless mutually complementary because those who occupy the standpoint of common life and science (including Newton and his followers) are frequently tempted into religion and superstition, and the radical skeptical standpoint serves to counter this temptation. She contrasts this interpretation with what she calls “the vehicle view,” on which Hume's skeptical arguments serve merely as a preliminary stage, clearing the way for his positive, naturalistic science of man. On her interpretation, by contrast, the radical skeptical standpoint must remain permanently available to the scientist of man, to serve as a check on our superstitious inclinations. She also contrasts her interpretation with Garrett's, which appeals to Hume's “Title Principle,” that “where reason is lively, and mixes itself with some propensity, it ought to be assented to. Where it does not, it can never have any title to operate on us” (Treatise 1.4.7.11/270).1 According to Garrett, this principle gives Hume epistemic permission to pursue his science of man, and to retain a significant degree of belief in its findings, despite the skeptical considerations that he raises. De Pierris gives some interesting objections to this proposal and claims that Hume recommends philosophy not because we have a “propensity” to it but because it is a safer guide than superstition.I am not convinced that De Pierris succeeds in removing the tension between Hume's skepticism and naturalism. For two reasons, it seems to me that the Humean scientist of man, as characterized by De Pierris, is irrational by his or her own lights.First, I can find only one argument in this book that supports Newtonian methodology. It proceeds by using Hume's version of PPM to eliminate the mechanical philosophers’ alternative methodology: Hume argues that a cause and its effect are always phenomenologically separable, and therefore lack an internal necessary connection that would allow us to demonstrate the existence of one from that of the other—contrary to the mechanical philosophers’ “demonstrative ideal” of natural knowledge. However, Hume's version of PPM is also the basis for his radical skeptical argument against Newtonian methodology (225, 237n297). So, someone who accepts Hume's version of PPM is rationally committed to rejecting both the mechanical philosophy and the Newtonian alternative. And so, it is irrational to accept Newtonian methodology on the strength of Hume's PPM-based argument against the mechanical philosophy.Second, according to De Pierris, Treatise 1.4.1 argues that Hume's own “rules by which to judge of causes and effects” epistemically require us to reduce our degree of belief in the conclusion of any piece of reasoning to zero. The Humean scientist of man endorses these rules. So, if he or she continues to believe any conclusions of reasoning, then this is irrational, by his or her own lights.It therefore seems to me that De Pierris's Hume cannot, in good intellectual conscience, pursue his science of man. In contrast, Garrett's Hume (who accepts the Title Principle) is epistemically permitted do so.There is much of value in this book beyond its main line of argument about Hume's skepticism and naturalism and their intellectual context. It contains interesting and original discussions of early modern theories of mathematical knowledge, Locke's relation to Newton, and Hume's view of space and geometry. Concerning this last view, De Pierris rejects, and develops an alternative to, two influential interpretations: the “discrete space” interpretation due to Donald Baxter, James Franklin, and others; and the “actual parts” interpretation due to Thomas Holden. De Pierris proposes that, contrary to the “discrete space” interpretation, Hume does not aim to replace Euclidean geometry with an alternative, finitist geometry. Instead, he accepts Euclidean geometry as a demonstrative science, albeit an inexact one that cannot achieve the same degree of certainty as algebra and arithmetic. And contrary to the “actual parts” interpretation, a given spatially extended whole is not composed of a determinate number of indivisible minima whose existence is metaphysically prior to that of the whole. Instead, extended wholes have a kind of priority over the minima that compose them: minima are defined in terms of processes of diminution or division performed upon such wholes.Ideas, Evidence, and Method is essential reading not just for Hume scholars, but for all philosophers interested in early modern epistemology, philosophy of mind, philosophy of mathematics, and philosophy of science.I thank Andrew Chignell and Don Garrett for helpful feedback on earlier drafts.