Loeb -Psychology, Epistemology, And Skepticism in Hume_s Argument About Induction

7/27/2019 Loeb -Psychology, Epistemology, And Skepticism in Hume_s Argument About Induction

1/18

Synthese (2006) 152:321338DOI 10.1007/s11229-006-9008-1

O R I G I N A L PA P E R

Psychology, epistemology, and skepticism in Humes

argument about induction

Louis E. Loeb

Published online: 19 September 2006 Springer Science+Business Media B.V. 2006

Abstract Since the mid-1970s, scholars have recognized that the skeptical interpre-tation of Humes central argument about induction is problematic. The science ofhuman nature presupposes that inductive inference is justified and there are endorse-ments of induction throughout Treatise Book I. The recent suggestion that I.iii.6 isconfined to the psychology of inductive inference cannot account for the epistemicflavor of its claims that neither a genuine demonstration nor a non-question-begging

inductive argument can establish the uniformity principle. For Hume, that inductiveinference is justified is part of the data to be explained. Bad argument is thereforeexcluded as the cause of inductive inference; and there is no good argument to causeit. Does this reinstate the problem of induction, undermining Humes own assumptionthat induction is justified? It does so only if justification must derive from reason,from the availability of a cogent argument. Hume rejects this internalist thesis; induc-tions favorable epistemic status derives from features of custom, the mechanism thatgenerates inductive beliefs. Hume is attractedto this externalist posture because it pro-vides a direct explanation of the epistemic achievements of children and non-humananimalscreatures that must rely on custom unsupplemented by argument.

Keywords Hume Induction Reason Externalism Internalism Garrett

1 Introduction

Do we yet understand Humes project in his main argument about induction inTreatise I.iii.6? Is the argument skeptical? Is it even so much as epistemological,skeptical or otherwise? Or is it merely psychological? Although the literature hasmade much progress on these questions, we have not reached a full understanding of

Humes position. I hope to fill in some missing pieces of the interpretive puzzle.

L. E. Loeb (B)Department of Philosophy, University of Michigan,435 S. State Street, Ann Arbor, MI 48109-1003, USAe-mail: [email protected]


2/18

322 Synthese (2006) 152:321338

2 The skeptical interpretation of I.iii.6

According to the skeptical interpretation of I.iii.6, the section advances the problemof induction, in the fashion of Russell in the Problems of Philosophy.1 This interpre-

tation, which held sway for decades in the middle of the twentieth century, has itsattractions.2 It is philosophically interesting. Also, it coheres with the Beattie-Reidtradition of interpretation, on which Hume is utterly destructive, an arch-skeptic withrespect to causal necessity, the causal principle (that every new existence, or newmodification of existence, has a cause), the external world, the substantial self, andso forth.3 Finally, the skeptical interpretation, unsurprisingly, has some basis in thetext. Section I.iii.6 does supply key premises of the skeptical problem: first, inductiveinference presupposes that nature is uniform; second, there is no demonstrative argu-ment to show that nature is uniform; and third, no probable argument could showthat nature is uniform, without begging the question. Hume would also seem to draw

the arguments conclusion, that there is no justification whatsoever for belief in theuniformity principle and hence for inductive inference: he writes that there is noreason (T 92; cf. 91, 139) to draw an inference from the unobserved to the observed.

In order to assess the skeptical interpretation, it is necessary to insulate the ques-tion of whether Hume is a skeptic specifically with respect to induction from hisoverall skeptical tendencies. No one denies that Hume at least takes us near the brinkof deeply skeptical results.4 He argues that our faculties lead to contradictory beliefsabout the existence of matter (T231, EHU154155).5 More dramatically, Hume writesin the concluding section of Book I: I am ready to reject all belief and reasoning, and

can look upon no opinion even as more probable or likely than another (T268269).This seems to imply rejection of beliefs about the unobserved as a special case, andon the ground that all beliefs are equally probable, presumably because they are notjustified at all. At the same time, it is important to distinguish among different linesof argument to the latter conclusion. Incoherence in our faculties, for example, mightbe a ground for skepticism, but it is a ground distinct from the Russellian problem ofinduction.6 The issue before us is whether in I.iii.6 Hume advances skepticism on thebasis of this problem.

Arnold and Broughton independently produce a consideration that should givepause.7 In the I.iv.7 build-up to his readiness to reject all belief, Hume relies on a

1 Russell (1912, chap VI).2 For some prominent examples, see Prichards 1932 lectures on Hume, Russell (1945), Will (1947),and Popkin (1951), Ayers 1960 Oxford inaugural lecture, Flew (1961), Stove (1965), Bennett (1971),Popper (1972), Hacking (1975), and Stroud (1977). The Prichard lectures are in his (1950) and theAyer lecture in his (1963).3 This is the interpretation Kemp Smith attacked (1905, 1941).4 Although see Williams (2004).5 Unless otherwise noted, quotations from Hume are from the following editions: D. F. Norton andM. J. Norton, (Eds.), A Treatise of Human Nature (Oxford: Oxford University Press, 2000), for theTreatise and theAbstract; and T. L. Beauchamp, (Ed.),An Enquiry concerning Human Understanding(Oxford: Clarendon Press, 1999), for the first Enquiry. Page references, however, are to the followingeditions: EHUL. A. Selby Bigge, (Ed.), Enquiries Concerning the Human Understanding and Con-cerning the Principles of Morals, 3rd edn by P. H. Nidditch (Oxford: Oxford University Press, 1975);TL. A. Selby-Bigge, (Ed.), A Treatise of Human Nature, 2nd edn, with text revised and variantreadings, by P. H. Nidditch (Oxford: Oxford University Press, 1978), for the Treatise and the Abstract.6 Noonan (1999, pp 129131) is clear about this distinction.7 Arnold (1983) and Broughton (1983).


3/18

Synthese (2006) 152:321338 323

series of what he takes to be prior results: the contradiction about matter in I.iv.4(T265266); the discovery in I.iii.14 that causation involves no connexion or tieoutside the mind (T266267); and a dangerous dilemma arising from the claim inI.iv.1 that the understanding subverts itself (T267268). Whereas Humes own foot-

notes cite the three earlier sections, he neither provides a note to I.iii.6 nor relies onthe argument of that section. If Hume arrives at skepticism on the basis of consider-ations advanced in I.iii.6, surely he would appeal to them in the concluding section ofBook I.

Not only does Hume not refer to I.iii.6 in the course of his skeptical tailspin in I.iv.7,Humes project in the Treatise, his argumentative practices, and his explicit statementsabout induction, commit him to the thesis that inductive inference is justified. Theevidence against the skeptical interpretation has steadily accumulated since the 1970s,to the point that the interpretation has effectively been demolished.8 This is not tosay there are no hold-outs, but that holding out is now untenable.9 It would be well

to flesh this out a bit.First, Humes project. The Treatise carries the subtitle, BEING AN ATTEMPT

TO INTRODUCE THE EXPERIMENTAL METHOD OF REASONING INTOMORAL SUBJECTS (T xi). The result, in part, is the elaboration of a variety ofassociative mechanisms: for example, association by the relation of cause and effect,in Book I; the double association of impressions and ideas, in Book II; and the asso-ciative mechanism of sympathy, in Books II and III. Hume pursues the science ofhuman nature across all three books of the Treatise.

Second, Humes argumentative practices in support of this project. In the initial

section of the work, Hume relies on inductive evidence to establish his principle thatevery simple idea exactly resembles a preceding impression (T45). Just two sectionsfollowing the main argument about induction, Hume appeals to experience andexperiments to confirmindeed, to provehis associationist account of belief(T99, 100). Many subsequent sections are replete with experiments in support ofHumes psychological hypotheses. Book II includes a section Experiments to con-firm this system, that is, Humes system of the indirect passions.

Third, Humes explicit statements about induction. Hume tends to assimilate induc-tive inference to causal inference and thus often uses the latter as a stand-in for theformer.10 Hume writes of causal inference to belief in the unobserved as leading to

epistemic success. In I.iii.2, causation is the only relation that enables the mind togo beyond what is immediately present to the senses, either to discover the realexistence or the relations of objects (T73, emphasis added); the relation of causationinforms us of existences and objects, which we do not see or feel (T74, empha-sis added). There are similar passages subsequent to I.iii.6. In I.iii.8, a person whostops his journey at a river foresees the consequences of his proceeding forward; andhis knowledge of these consequences is conveyd to him by past experience (T103,emphasis addedcf. 104, 148); in I.iii.9, the relation of cause and effect brings usacquainted with such existences, as . . . lie beyond the reach of the senses and memory

8 For some early inventories of the evidence, see Beauchamp and Mappes (1975), Connons 1976conference paper on Humes naturalism, Winters (1979), and Beauchamp and Rosenberg (1981), aswell as Arnold (1983) and Broughton (1983). Connon was published in 1979.9 For attempts to resist the case against the skeptical interpretation, see Penelhum (1992) and Winkler(1999).10 Price (1940, p 25; 1969, pp 176179), Passmore (1952, pp 2934), and Pears (1990, pp 7172).


4/18

324 Synthese (2006) 152:321338

(T108, emphasis added). In theAbstract, No matter of fact can beproved but from itscause or its effect. Nothing can be known to be the cause of another but by experience(T654, emphasis added).11 Similarly, Hume writes: cause and effect . . . tis the only[connexion or relation of objects], on which we can found a just inference from one

object to another (T89, emphasis added). This endorsement occurs within Humesmain argument about induction. At I.iii.13, causal inference is just and conclusive(T144). The striking continuity in Humes positive evaluations of inductive inferencethroughout Part iii is difficult to explain on the hypothesis that I.iii.6 constitutes askeptical turning-point.

What is more, even after I.iii.6 Hume recognizes gradations in inductive evidence.At I.iii.13, Of unphilosophical probability, he provides an inventory of degree[s]of evidence (T153154) that includes proofs and also probability, beliefs based onobservation of conjunctions that are not constant or on infrequent observation ofconjunctions (T130131, 142). Within probability, there are degrees of force (T130)

and evidence (T131, 154). Hume devotes I.iii.13 to unphilosophical probabil-ity (T143), in contrast to kinds of probability [that] are receivd by philosophers(T143).12 All this, even though the problem of induction levels all inductive inference.

There is extensive evidence that Hume endorses inductive inference.13 How mightproponents of the skeptical interpretation of I.iii.6 respond? Russell thought Humesimply inconsistent.14 Stroud attempts to transmute this inconsistency into the verymessage of Humes philosophy. In Strouds hands, Hume is purposefully calling atten-tion to the inevitability of the inconsistency, for himself and everyone else, given theuniversality and irresistibility of inductive inference. Stroud writes: Humes theory

taken all together shows that most of our beliefs mustbe wrong or unreasonable; giventhe way we are, we could not have those beliefs unless that were true.15 Hume ismaking a point about an epistemic feature of the human condition. Strouds interpre-tation is much more interesting than Russells charge of inconsistency, but it equallypresupposes that Hume is a skeptic about induction. We have seen that this cannotbe correct.

3 The descriptivist interpretation

My critical focus will be on a different and more influential line of interpretation,which I dub descriptivism.16 The descriptivist reading grants that Hume does not

11 Much as causal inference constitutes epistemic success in the Treatise, we have in the Enquiry:Had not the presence of an object instantly excited the idea of those objects, commonly conjoinedwith it, all our knowledge must have been limited to the narrow sphere of our memory and senses(EHU55, emphasis added); and this: The existence . . . of any being can only beproved by argumentsfrom its cause or its effect; and these arguments are founded entirely on experience (EHU164,emphasis added).12 Similarly, in the Enquiry, some events are more probable (EHU58) than others; and, in a pas-

sage emphasized by Meeker (1998, p 38), A wise man. . .

proportions his belief to the evidence(EHU110cf. 117).13 The section Evidence that Hume Considers Inductive Inference Justified in Loeb (forthcoming)provides detail and, especially at item (g), additional kinds of evidence.14 Russell (1945, p 672). See also Flew (1986, pp 5657).15 Stroud (1991, p 276).16 Noonan calls this view literalism (1999, pp 119128).


5/18

Synthese (2006) 152:321338 325

advance skepticism about induction in I.iii.6, but takes this to be a symptom of theabsence of normative epistemology in the section. Hume does not there engage ineither negative or positive epistemological evaluation of induction. Rather, I.iii.6 iswholly given over to the causes of inductive inference; it is an exercise in cogni-

tive psychology. Of course, Humes investigation may result in fairly high-level ortheoretical conclusions. The descriptivist label does not signal a contrast betweenlow-level observation and higher-level theory, but rather one between empirical andnormative inquiry.

Any viable version of descriptivism will have to concede that Hume in some sensepresupposes the justification of inductive inference. After all, he relies on inductiveinference throughout Part iii. According to the descriptivist, Humes psychologicalinvestigation does not undermine the commitment to the legitimacy of inductiveinference. This is because I.iii.6 does not so much as address the epistemic status ofinductive inference. The psychological study of that section is neutral with respect to

normative questions.In order to get descriptivism off the ground, its proponents must neutralize the

passages in which Hume says there is no reason for inductive inference to theunobserved. One common descriptivist strategy proceeds in two steps. The first is toconstrue the no reason passages as somewhat awkward formulations of the claimthat inductive inference is not caused or produced by reason. The second step is totake this claim to be directed ad hominem at a conception of reason Hume himselfrejects. Descriptivists can disagree on whether Humes point is directed at Cartesian,or Lockian, or perhaps any non-associative conception of reason.17 The important

point is that the claim that inductive inference is not due to reason in a specializedsense need not commit Hume to the claim that inductive inference is unjustified; thenegative psychological claim about the causation of inductive inference leaves openthe question of justification.

A compelling and relatively pure version of the descriptivist interpretation maybe found in the paper by Broughton mentioned earlier. More recent literature gen-eralizes the interpretation beyond I.iii.6. The idea is that the whole of Part iii ofBook I is purely descriptive. It is not until Part iv, perhaps quite late in Part iv, thatHume undertakes any epistemologicalevaluation of inductive inference. On the mostpromising version of this position, due to Don Garrett, Part iii is descriptive of the

psychology of epistemic evaluation and not itself normative.18 This generalization ofa descriptivist reading beyond I.iii.6 has an important advantage. It provides a foot-hold for explaining away the implicit and explicit commitments to inductive inference,both before and after the main argument about induction; they are merely part of thedata for psychological study. Much as psychology can investigate inductive inference,it can investigate why we approve of inductive inference (where we do approve ofit). At least in Part iii, however, these commitments do not issue from Hume quaepistemologistso the descriptivist tells us.

Even a generalized descriptivist interpretation faces serious difficulties. Aware that

descriptivism constitutes a large-scale interpretation with considerable resources to

17 For some variants, see Garrett (1997, ch. 4), Millican (2002, Sections 1 and 10.3), and Disarmingthe Evidence for the Traditional Interpretation, in Loeb (forthcoming). Although he is a descriptivist,Garrett does not himself accept this approach.18 Garrett (1997, esp. chap. 4, 7, 10). For Owen (1999, chap. 6, 8, 9), I.iii is descriptive of epistemicdistinctions internal to a practice.


6/18

326 Synthese (2006) 152:321338

ward off criticism, I sketch a few of the problems. In the first place, whether or notHume operates with a specialized conception of reason within I.iii.6, Hume appliesreason and its cognates to causal inference in I.iii.6 subsequent to the main argu-ment about induction (T93), in I.iii.7 (T94, 95, 97n), and also in I.iii.8, after attributing

causal inference to custom (T103, 104, 105). This terminology would seem to carrya positive epistemic connotation. Descriptivists can contend that, following I.iii.6,reason simply refers to whatever faculty causes inductive inference.19 Let me offerone consideration that tells against this. In I.iii.9, Hume distinguishes within the fac-ulty of association between reason, which includes all probable reasonings, andthe imagination, which includes whimsies and prejudices, which are rejected underthe opprobrious character of being the offspring of the imagination (T117118n).20

Earlier in the section, within his discussion of two systems of realtiesone based onperception and memory, the other on causal inferenceHume contrasts beliefs aris-ing from custom and the relation of cause and effect with beliefs that are merely

the offspring of the imagination (T108). Similarly, in I.iii.13 the kinds of probabilityreceived by philosophers are reasonable foundations of belief and opinion (T143,emphasis added). Further, the I.iii.9 distinctions anticipate that in I.iv.4, between thepermanent, irresistible, and universal principlessuch as the customary transitionfrom causes to effects, and from effects to causes, which Hume there accepts,and the principles, which are changeable, weak, and irregular, which are observdonly to take place in weak minds (T225). I think it difficult to resist reading theoccurrences of reason and its cognates after I.iii.6 as honorifics.21

In the second place, the descriptivist interpretation must strain to accommodate

a number of features of Humes positive evaluations of inductive inference. Thereis the sheer mass and coherence of these positive assessmentsinductive inferenceas constituting an epistemic success, as just, and as generating degrees of evidence.22

Garrett suggests that, were Hume not to offer such evaluations, he would provide acounterexample to his own psychological theory, which arguably . . . entails that wewill, on the whole, continue to approve epistemically of our engaging in [such infer-ence], so long as it continues to succeed.23 This is not persuasive. Even if approvalof causal inference is unavoidable, putting such commitments in writing is not. Fur-ther, some of Humes evaluations are second-order. Hume writes at I.iii.13: We shallafterwards take notice of some general rules, by which we ought to regulate our

judgment concerning causes and effects (T149). Humes footnote references I.iii.15,Rules by which to judge of causes and effects, where he offers eight rules, all theLOGIC I think proper to employ in my reasoning (T175). It is a stretch to suggestthat Hume is merely reporting that we are psychologically determined to engage inand approve of higher-order evaluations.24 In addition, Hume does not say things heought to be expected to say on the descriptivist interpretation. Hume does not pause

19 Cf. Garrett (1997, chap. 4; 1998, pp 17880; 2002, Appendix).20 The term probability has a narrower meaning at T124.21 Loeb (2002, Section II.3), Millican (2002, Section 12).22 For skirmishes with descriptivist treatments of some of the texts, see Loeb (2002, pp 4344nn.1213, 47n.19, 102nn.23, 131n.43).23 Garrett (1997, pp. 7980; cf. 157158). Similarly, Fogelin appeals to Humes thesis that we arepsychologically compelled to undertake epistemic assessments in order to explain why the positiveevaluations of inductive inference do not undermine the skeptical interpretation of I.iii.6 (1985,pp 148149).24 For these points, see Loeb (2002, p 45n.15).


7/18

Synthese (2006) 152:321338 327

to note that his own positive evaluations are offered as nothing but part of the datafor psychological study. Similarly, when Hume utilizes inductive inferenceappealingto experiments and the results they prove, he does so unflinchingly, withoutcautionary comment, in each Book of the Treatise.

In the third place, we have observed that Hume writes that inductive inferenceis just not only in I.iii.6 itself (T89), but also later in I.iii.13, Of unphilosophicalprobability (T144). Beyond that, Hume applies just and its cognates to inductiveinference well into Part iv: in I.iv.2, Of skepticism with regard to the senses (T 216)and in I.iv.4, Of the modern philosophy (T225). The latter passage, where Humeis giving an example of the operation of the permanent, irresistible, and universalprinciples, is well-known: One who concludes somebody to be near him, when hehears an articulate voice in the dark, reasons justly and naturally; tho that conclusionbe derived from nothing but custom. . . (T225).25 What is the descriptivist reading tomake of the continuity in Humes seemingly positive assessments of causal inference

across Parts iii and iv of Book I?26 These assessments occur as late as I.iv.4. If Humeis not yet engaged in normative epistemology there, when does this enterprise set in?I.iv.5 and I.iv.6, Of the immateriality of the soul and Of personal identity, are byand large given over to metaphysics. Descriptivism seems to require that cognitivepsychology does not give way to normative epistemology until the final section ofBook I.27 This seems an odd role for a section entitled Conclusion of this book. Thealternative is to concede that the Treatise contains normative epistemology all theway back, that is, at least as early as Part iii. 28

4 The epistemic character of I.iii.6

Let us put to the side the difficulties I have canvassed for the generalized descriptivistinterpretation. Descriptivism cannot survive considerations internal to I.iii.6 itself.All descriptivists are committed to the thesis that Hume is not engaged in normativeepistemology in that section. According to the interpretation, Humes point is merelypsychological, that reason does not cause or determine us to make inductive infer-ences; rather, inductive inferences are caused by some other faculty. This is a negativeconclusion within cognitive psychology. On the descriptivist interpretation, I.iii.6 has

25 There is also continuity in Humes positive assessment of causal inference across the Treatise andthe Enquiry. In Sections 10 and 11, Of Miracles and Of a Particular Providence and of a FutureState, there are a half dozen references to just reasoning or reasoners, or to reasoning justly(EHU 110, 113, 136, 139, 142, 145)these in a context where Hume is relying on his distinctionbetween legitimate and illegitimate forms of causal inference to criticize arguments for God based onmiracles and design. See also notes 11 and 12.26 For this question, and its bearing on the descriptivist contention that Humes use of just and itscognates in Part iii is not normative at all, see Loeb (2002, pp 43n.12, 4344n.13).27

Hence Garrett (1997, pp 95, 230, 232). Cf. Owen (1999, esp. pp 205206).28 Why would Hume defer normative epistemology until after Part iii? Garrett sometimes portraysHume as engaged in a Quinean enterprise, with the cognitive faculties investigating themselves (1998,p 186). Garrett has suggested (in conversation) that Hume withholds his epistemological evaluationuntil all the empirical data are in. It would be surprising, however, if Hume thought that he couldever possess all the relevant psychological data. In any event, why would he not provide provisionalepistemic assessments of inductive inference and other belief-forming mechanisms as the empiricalinvestigation proceeds? Cf. Loeb (2002, pp 1415, 8892; 2004, Section II.2).


8/18

328 Synthese (2006) 152:321338

no epistemic force at all. Against this, I will argue that Hume does make a negativeepistemic point, albeit one that falls short of radical skepticism.

I begin by noting an ambiguity that tends to beset the descriptivist reading. Con-sider a formulation of Garretts: [Hume] is arguing that we do not adopt induction

on the basis of recognizing an argument for its reliability.29

Are advocates of thedescriptivist reading supposing that Hume restricts the arguments he is canvassing togood arguments, or must he also be on the look-out for bad arguments that seem tobe good ones and thus manage to cause inductive inference?30 The ambiguity here isHumes own. In both I.iii.6 (T89) and the corresponding section of the first Enquiry(EHU32, 37, 38), he uses the terminology of founded and foundation withoutexplicitly saying whether he is looking for a psychological or an epistemic founda-tion.31 Hume does not pausenot even at critical juncturesto distinguish merelypsychological and epistemic notions.

The interpretive issue depends upon two cases. First, is Humes point that no

argument that purports to be demonstrative causes inductive inference, or that noargument that is in fact sound causes such inference? Similarly, is his point thatno probable argument that purports to be cogent causes inductive inference, or thatno genuinely cogent probable argument does so? On these questions, I am in agree-ment with Millican: Hume focuses only on legitimate forms of argument, which onGarretts principles he has no right to do. Millican was the first to advance a numberof the points that follow.32

Humes intentions are clear in the case of demonstrative argument. He claimsthat there is no demonstrative reasoning to prove that nature is uniform (T89, 92;

EHU35). This looks like a success verb; a proof is presumably a successful dem-onstration. Hume shows no interest in the question of whether inductive inference iscaused by reasoning, however seductive, that falls short of a genuine proof of the uni-formity principle. The descriptivist reading cannot explain Humes silence in regardto this possibility. Garrett replies that such arguments, even if convincing, wouldnot be instances of genuine demonstrations operating alone.33 Granted, but wereHume merely pursuing cognitive psychology, he would need to exclude the possi-bility that inductive inference is caused by such impure reasoning. Bad argumentssway our thinking and practices all the time. Hume was certainly alert to purport-edly demonstrative arguments that are flawed. After providing a general argument to

show that there is no demonstration of the causal principle, he devotes five additionalparagraphs to showing that its purported demonstrationsdue to Hobbes, Clarke,Locke, and othersare fallacious and sophistical (T80). All this in I.iii.3. If the only

29 Garrett (1997, p 92). Garrett now puts no weight on an argument specifically for the reliability ofinductive inference (2002, p 333n.30).30 The same question could be asked of formulations of Loeb (2002, Section II.2) and (forthcoming).31 This ambiguity, perhaps even vacillation, is but one instance of Humes tendency to use terminol-ogy that straddles purely psychological and epistemic notions. For example, he writes that causationis the only relation that produces assurance (T73, 124; cf. 106, EHU27) in matters of fact that havenot been observed. Does assurance mean mere belief, a psychological state, or does it imply a statethat involves knowledge, and hence carry epistemic implications? See Loeb (2002, pp 61, 73, 77).32 Millican (1998, Section VII; 2002, pp 157158). The quotation is from his (1998, p 150).33 Garrett (1998, p 187). Garrett also finds rather implausible the hypothesis that inductive infer-ence depends upon fallacious attempted demonstrations. But Hume is alert to a wide range ofpsychological distortions both to belief, as in unphilosophical probability, and sentiment. SeeLoeb (2002, chap. IV).


9/18

Synthese (2006) 152:321338 329

issue in play in I.iii.6 is whether a demonstrative argument causes inductive inference,it is difficult to see why demonstrations that are subtly flawed, but have an air ofplausibility, could not do the trick.

I turn to the case of probable reasoning. A preliminary point. Why does Hume fail

to consider inductive arguments that are not good ones? Garretts explanation is thateven bad probable arguments rely on inductive extrapolation from the observed to theunobserved; for this reason, they can no more be the underlying cause of inductiveinference than good or philosophical ones can.34 This line of defense dependsupon Garretts view that I.iii.6 seeks to explain the entire practice of inductive infer-ence, rather than individual inductive inferences.35 This is dubious. It is custom or pastexperience that causes individual inductive inferences. Hume is clear in I.iii.16 thatwe cannot give the ultimate reason, why past experience and observation producessuch an effect (T179, cf. EHU5455). The ultimate reason for inductive inference,and hence the cause of the operation of custom itself, lie beyond Humes explanatory

ambitions.36

Even bracketing this matter, the textual situation with regard to probable reasoningis more complex than in the case of demonstrative argument. Hume writes in theTreatise: The same principle cannot be both the cause and effect of another (T90).This causal formulation perhaps opens the door to taking the issue to be confinedto the causation of belief in the uniformity principle.37 This way of putting things,however, disappears in the Abstract:

All probable arguments are built on the supposition, that there is this conformitybetwixt the future and the past, and therefore can never prove it . . . . [O]ur expe-

rience in the past can be a proof of nothing for the future, but upon a supposition,that there is a resemblance betwixt them. (T651652)

This reproduces language in the Treatise: probability is founded on the presumptionof a resemblance betwixt those objects, of which we have had experience, and those,of which we have had none; and therefore tis impossible this presumption can arisefrom probability (T90). The implicit point is that a probable argument for the unifor-mity of nature, the conformity of unobserved to observed, would be question-begging,presupposing the resemblance it seeks to establish. This is explicit in the Enquiry: Toendeavor . . . the proof [that the future will be conformable to the past] by probable

arguments. . .

must be evidently going in a circle, and taking that for granted, whichis the very point in question (EHU3536). Humes uptake on the causal formulationin the Treatise is significant: The same principle cannot be both the cause and effectof another; and this is, perhaps, the only proposition concerning that relation [of cau-sation], which is either intuitively or demonstratively certain (T90). Hume cannotresist the causal formulation, which affords the opportunity to inject a clever, butgratuitous, slap at reason.38 Taken literally, however, the formulation runs askew tothe underlying point about circularity, and thus can appear to support the descriptivistreading. This is ultimately noise in the text.

34 Garrett (1998, p 187).35 See Garrett (1997, esp. pp 9192; 2002, Appendix). Broughton (2004) contrasts her version of theposition with that of Garrett and also begins to part company with the pure descriptivist position.36 For development of the point, see Loeb (2004, Part III).37 As at Garrett (1997, pp 82, 9192).38 Bennett writes that the causal formulation embodies a joke (2001, p 261).


10/18

330 Synthese (2006) 152:321338

Perhaps one could both brush off the Abstractand try to insist that Humes claimsabout probable arguments in the Treatise are merely psychological, whereas in theEnquiry they are epistemic. It is difficult to see any advantage in such maneuvers,since it is clear that Humes claims in the case of demonstrative arguments for the

uniformity principle carry epistemic weight. Humes negative argument about reasonmanifests an epistemic dimension in the cases of both demonstration and probability.

5 Justified inductive inference as Humes explanandum in I.iii.6

What does explain Humes failure to consider the possibility that reason supplies con-siderations that, though defective, are sufficiently appealing to cause us to makeinductive inference? The answer must be that Hume imposes an epistemic con-straint on any causal explanation of inductive inference: the explanation of our

making inductive inferences must be compatible with their being justified.39 It seemsplausibleplausible enough to attribute the thought to Humethat if inductive infer-ences are caused by bad or defective reasoning, then they would not be justified, inviolation of the constraint. This explains why Hume confines his attention to good orcogent considerations of reasonto genuine demonstrations and to probable reason-ing that is not question-begging. Given that inductive inference is justified, it couldnot be caused by reasonbad reasoning is excluded, and there is no good reasoningto cause it.40 Attributing the epistemic constraint to Hume also accounts for his will-ingness to say, in the course of his main argument about induction itself, that cause

and effect. . .

tis the only [connexion or relation of objects], on which we can founda just inference from one object to another (T89). This is a statement of the data tobe explained: the making of justified inductive inference. (The assumption, of course,is that some inductive inferences are justified, not that they all are.) The descriptivistreading obscures the epistemic structure of Humes argument.

My suggestion does not reduce to the earlier point that Humes project in thescience of human nature and his argumentative practices presuppose or assume thatinductive inference is justified. This is a commonplace. As we have seen, even de-scriptivists must allow that Hume presupposes the justification of inductive inference.As Broughton puts things, there is a natural presupposition, one Hume shares,

that inductive inference is reasonable.41 The linchpin in my interpretation is not thatHumes engaging in inductive inference presupposes that it is justified, but that inI.iii.6 he incorporates this positive epistemic status into his description of the subjectmatter under investigation. Let me clarify the distinction here. Suppose Hume utilizesinductive inference in the course of the I.iii.6 investigation into the causes of inductive

39 This differs from Connons view that Hume takes inductive inference to be justified and seeksa description of what is involved in making inductive inference (1976, p. 129). Not any descriptionwould be compatible with justification. Criteria for justification are operating in the background.40 We cannot regard the constraint as merely ad hominem against philosophers who maintain that

reason does explain the belief in the uniformity principle. These friends of reason doubtless think thatthe arguments that figure in such explanations are good ones. Hume could defeat their position byshowing that it is not the case that good reasons cause inductive inference. This would not, however,be sufficient to establish the position descriptivists attribute to Hume: that inductive inference is notdue to reason. A showing that good reasons do not cause inductive inference leaves out of account thepossibility that inductive inference is due to a defective employment of reason, rather than to custom.We must therefore suppose that the assumption that inductive inference is justified is Humes own.41 Broughton (1983, pp 5, 12). See also Garrett (1998, p 186).


11/18

Synthese (2006) 152:321338 331

inference. This would not itself explain the role of epistemic considerations in the sec-tion. In order to explain this, we would have to take the assumption that inductiveinference is justified to be operative at two levels: in the method of investigation, butalso in Humes understanding of the phenomenon he seeks to explain.

Another way to see the distinction is as follows. It is clear that Hume relies on induc-tive inference in I.iii.8, where he argues from experience and observation (T101),from experiments (T99, 100, 102), that causal inference is due to custom or habit.By contrast, the I.iii.6 discussion of the uniformity principle does not utilize inductiveinference. Rather, Hume is relying on philosophical reflectionconsiderations aboutthe conceivability of the falsehood of the uniformity principle, considerations aboutthe presuppositions of any probable argumentto reveal limitations in demonstra-tive and probable arguments. In the core of I.iii.6 (paragraphs four through eleven)the assumption that inductive inference is justified is part and parcel of the phenom-enon under investigation, but not integral to the method of investigation itself. The

main argument in I.iii.6 is shot through with epistemological claims and can hardly beregarded merely as an exercise in descriptive psychology.

6 Humes opposition to epistemic parasitism

A closely related textual consideration arises in connection with I.iii.8. Recall that aperson who stops his journey at a river foresees the consequences of his proceedingforward. Hume adds some important comments: custom operates before we have

time for reflection, without a moments delay; we draw the inference from pastexperience without reflecting on it (T104, emphases added). The riverbank passageposes a challenge to the descriptivist reading: if in I.iii.6 Hume is merely making thenegative point that inductive inference is not caused by reason, why does he not restcontent with the empirical observation that reflection and reason play no role in typ-ical cases of basic inductive inference? Indeed, Hume does make this point in I.iii.6:the imagination of itself supplies the place of . . . reflection, and does so withouta moments delay (T93). This simple observation would seem to dispose of reasonas the causal explanation of inductive inference.42 If so, how is the descriptivist toexplain the central position of the considerations about demonstrative and probable

arguments in the section?An argument in Of skepticism with regard to the senses heightens the puzzle.

Hume asks whether it be the senses, reason, or the imagination, that produces(T188) the belief in continued and independent objects. This looks a lot like asking,as Hume does in I.iii.6 (T8889, 92), whether it is reason or the imagination thatproduces inductive inference. Here is Humes case against reason in I.iv.2: whateverconvincing arguments philosophers may fancy they can produce to establish the beliefof objects independent of the mind, tis obvious these arguments are known but tovery few, and that tis not by them, that children, peasants, and the greatest part of

mankind are inducd (T193) to believe in body. Hume relies on the empirical obser-vation that most of us believe in body without interposing any reflection or reason.If that is a good argument against reasons causing the belief in body, why do notanalogous considerationsthe person at the riverbanksuffice to show that reasondoes not cause inductive inference?

42 See Millican (1998, p 152) and Broughton (forthcoming).


12/18

332 Synthese (2006) 152:321338

Why cannot Hume rest content to observe that basic inductive inferences proceedwithout a moments delay; that they are not caused by reflection or reason, and afortiori they are not caused by any good or cogent deployment of reason? The answeris that he wants to foreclose a broad strategy for explaining why inductive inference

is justified. Suppose that reason, fully deployed, could produce cogent reflections infavor of the uniformity principle. In that event, friends of reasonthose who thinkthat the justification of inductive inference somehow resides in reason might pro-pose a position along the following lines. Even if the person at the riverbank does nothimself invoke the cogent reasoning, even if he is not so much as aware of it, his induc-tive inferences might nevertheless possess justification in virtue of the availability ofthe cogent argument; such justification would be indirect, parasitic upon the existenceof reasoning available to a more reflective person. Let us call this strategy epistemicparasitism.

How might indirect justification be thought to arise? Consider Descartes. We attain

unshakable certainty, scientific knowledge, that the material world exists only if wefirst prove the existence of a non-deceiving God. Where does that leave those of uswho have not proved this, and hence do not possess certain or scientificlets sayloftyknowledge? Evidently, such knowledge of the material world as you and Ipossess is second-rate. One way to flesh this out is to say that, even though we havenot proved a non-deceiving God, we do possess a fragment of the system of beliefsrequired for lofty knowledge. Perhaps our sketchy belief system confers on us second-rate knowledge. Alternatively, perhaps the fact that we could obtain lofty knowledge,were we fully to exercise our faculty of reason, confers second-rate knowledge. Either

way, our lowly knowledge derives from the availability of a cogent argument for anon-deceiving God.43

Epistemic parasitism cannot be defeated by the observation that ordinarycognizersas at the riverbankdo not reason or reflect; the whole point of parasitismis that justification for perfectly ordinary persons is derivative from the availability oflofty knowledge that they do not themselves possess. Hume wants to preempt para-sitism by showing that putative lofty knowledge is unattainable in the first place, sothat no other knowledge can be second-rate. In connection with inductive inference,Hume does take himself to show that lofty knowledge is unattainable; there is no(demonstrative or probable) argument available to reason that justifies belief in the

uniformity principle. We thus cannot regard the argument about induction in I.iii.6 asa descriptive exercise in cognitive psychology. Hume wants to quash overly reflectivistaccounts of the way the mind ordinarily works, but he also needs to take on overlyintellectualist accounts of the way the mind could work even in the case of an ideal,fully reflective, cognizer.44 To this end, the riverbank argument can at best play a rolethat is supplementary to the main argument about induction.45

What about Treatise I.iv.2, where Hume does employ an empirical argument: chil-dren, peasants, and the greatest part of mankind have no knowledge of any con-siderations of reason that support the belief in body? In this context, there is no

43 I say more about these themes in Some Context and Humes Epistemic Options in Loeb(forthcoming) and (2004, Sections II.1 and IV.4).44 Broughton (forthcoming) emphasizes the first of the two ambitions.45 Cf. Millican (1998, pp 151152). Hume concludes the riverbank passage: This removes all pretext,if there yet remains any, for asserting that the mind is convincd by reasoning (T104, emphasis added)of the uniformity principle.


13/18

Synthese (2006) 152:321338 333

harm in employing the short argument. In Part iii, Humes claim that reason couldnot cause us to make causal inferences, given that such inferences are justified, isaccompanied by the thesis that all assurance in unobserved matters of fact is basedon causal inference (T7374, 87, 89, 107108) and the identification of reason with

causal inference (T117118n.1). Both the terminological stipulation (T193; cf. 231)and the substantive thesis about the scope of causal inference (T193, 212; cf. 198, 216)are imported into I.iv.2. Part iii has destroyed parasitism as an account of knowledgeof any unobserved matters of fact. In Part iv, the earlier care in addressing parasitismis no longer required.

We have seen that a generalized descriptivist interpretation faces numerous diffi-culties. Hume persistently provides positive assessments of inductive inference, anddoes so without comment or qualification. These assessments include evaluations ofsecond-order inductive principles. There is the continuity in his posture toward induc-tive inferences across Part iii and much of Part iv of Book I, and also a recurring

normative distinction between reason and the imagination. In addition, the river-bank passage in I.iii.8, together with the character of Humes main argument aboutinduction, pose severe difficulties for the descriptivist account of I.iii.6. Even the mostcasual reader will be struck by the pronounced epistemic flavor of the section.

7 Humes epistemological externalism

Does recognition of the epistemic character of I.iii.6 saddle us with the skeptical inter-

pretation of the section after all? It would, if Hume thought that a belief is justifiedonly if it is supported or supportable by good argument. Roughly speaking, this is aninternalist assumption. According to epistemological internalism, the justification ofa belief depends exclusively upon the beliefs one holds orin versions that tolerateparasitismupon beliefs that are available, beliefs one could hold. The Russellianproblem of induction relies on internalist presuppositions in moving from the claimthat there is no cogent (demonstrative or probable) argument for the uniformity ofnature, to the conclusion that no inductive belief has any justification. Even supposingthat demonstrative and probable arguments exhaust the field, the Russellian skep-tic requires the premise that an inductive belief can be justified only by argument,

argument taking as premises other beliefs one holds or could hold.46In light of the massive evidence that Hume is not a skeptic about induction, he must

reject this internalist way of thinking. Hume does contend that there are no cogentarguments for the uniformity of nature. Yet, he labels inductive inference just inthe course of his main argument about induction and maintains his commitment toinductive inference in the remainder of Part iii and at least much of Part iv. This isunaccountable, on the supposition that Hume is an internalist.47 Hume contributeskey premises for the skeptical argument about induction, but he does not share the in-ternalist framework that is also necessary to generate its conclusion. In order to avoid

attributing skepticism to Hume, we need not suppose that his argument in I.iii.6 isonly directed against a specialized conception of reason (Section The descriptivist

46 Garrett notes that the skeptical reading requires attributing this premise to Hume and observesthat he nowhere advances it (1997, pp 8283). As we shall see, Hume rejects the premise. Indeed, itis rejected implicitly as early as I.iii.6.47 If he is an internalist, as characterized above. There is room to attribute to Hume a negativecoherence theory of justification. See Loeb (2001).


14/18

334 Synthese (2006) 152:321338

interpretation). Skepticism can be avoided, even if there is no good argument of anysortsupporting the uniformity principle.

Let us see more concretely how Hume can stand by his positive epistemic assess-ment of inductive inference, though such inference is not justified in virtue of

argument. Humes negative claim in I.iii.6, that inductive inference is not due toreason, leads to the positive claim in I.iii.8, that it is due to custom. In I.iii.12, Of theprobability of causes, Hume extends the role of custom to the explanation of formsof probabilistic or statistical inference that are receivd by philosophers (T143).Once reason or argument has been ruled out, custom is the only resource Humehas available to account for inductions positive epistemic status. I do not mean thatinductive inference is justified simply because it is due to custom or repetition. Humestresses in I.iii.9 that education or indoctrination is also due to custom, but is dis-claimd by philosophy, as a fallacious ground of assent (T118).48 At the same time, ifinductive inference is justified, but not by argument, custom is the only place to look

for its justification. There must be some feature of custom, as it operates in inductiveinference, that generates justification.

In pioneering work, Kemp Smith in effect identified the irresistibility and unavoid-ability of custom as the crucial feature.49 A number of other features of custom, how-ever, suggest themselves as candidates for the central role in Humes epistemology.Perhaps the operation of custom confers justification insofar as it is an operation of ahealthy organism, or because it is adaptive, or because it is reliable. These options giverise to proper function, adaptivist, and reliabilist interpretations of Hume.50 My ownview is that what matters for Hume is customs tendency to infix ideas, to produce

steady beliefs.51

Adjudicating among these alternativesbetter yet, understandinghow the texts and Humes historical position constrain the choice among them is thecentral challenge in coming to terms with Humes epistemological position.52 Whatmatters here is that the various options appealing to irresistibility, proper function-ing, adaptiveness, reliability, stabilityare externalist theories. According to thesetheories, the epistemic status of a belief depends, at least in part, upon naturalisticfacts about the mechanism that produces it. Hume is offering an externalist accountof why inductive inference is justified. In a recent book on the problem of induction,reliabilism is the first solution considered.53 Russell, an internalist and foundationalist,had no such solution remotely in view.

Attributing externalism to Hume explains how he could reject internalism andthus avoid drawing a skeptical conclusion at the close of I.iii.6. It also explains why hecannot rely on the riverbank considerations, at least insofar as Hume wants to argueagainst internalism; they do not defeat parasitism as an internalist strategy. An exter-nalist interpretation has considerable explanatory power and coheres with Humesfocus on custom as the mechanism that underpins inductive inference.

48 Whether Hume can sustain this epistemic distinction between two kinds of custom is a difficult

matter central to his epistemological project. See Loeb (2002, Section VII.5).49 Kemp Smith (1905, 1941). See Loeb (2002, pp 2025) for difficulties.50 For a proper function interpretation, see Craig (1987, p 81) and Wolterstorff (1996, p 166n.6); foradaptivism, Schmitt (1992, pp 6872); for reliabilism, Schmitt (1992) and (2004).51 I advance this view in Loeb (2002). I amend my position in (2004).52 For a start on this project, see Loeb (2004, Part III).53 Howson (2000).


15/18

Synthese (2006) 152:321338 335

Of course, we would want to locate direct textual evidence for Humes external-ism. I cannot construct the case here, but I can show that there is such evidence. Theopening paragraph of I.iv.4 is one central place to look. A person who infers from thearticulate voice that someone is near reasons justly and naturally. By contrast:

[O]ne, who is tormented he knows not why, with the apprehension of spectres inthe dark, may, perhaps, be said to reason, and to reason naturally too: But thenit must be in the same sense, that a malady is said to be natural; as arising fromnatural causes, tho it be contrary to health . . . (T225226)

This resonates with adaptivist theories. So too does I.iii.10, where Nature has . . .chosen a medium between perceptions, on the one hand, and the unsteadiness ofmere ideas, on the other, so that we might avoid future goods and evils (T119). Themedium is belief, and steadiness in the service of guiding action is its natural function.

This passage also resonates with proper function and stability theories. Section I.iii.16is germane as well to externalism in the Treatise.In discussing the riverbank passage, I suggested that Hume seeks to thwart episte-

mic parasitism. His commitment to externalism lies at the root of this objective. Thepressure toward parasitism arises within an internalist framework. From an internalistperspective, the arguments required to justify fundamental beliefsthe uniformity ofnature, the existence of the external world, and so forthmust meet some thresholdof articulation and sophistication. Suppose the internalist sets the threshold low. Evenso, it is implausible that honest gentlemen (T272) or landholders, peasants (T193),and the vulgar at large, as well as children (T193), reach the required threshold. The

internalist must resort to providing, in the first instance, an account of the justifica-tion of suitably idealized cognizers.54 Justification and knowledge for the vulgar andchildren is derivative from that of idealized cognizers.

Within externalism, this pressure toward parasitism vanishes. Custom operates inthe vulgar and in children; the favorable epistemic status of their inductive beliefscan be explained in terms of the relevant features of custom, without any referenceto belief systems available to more reflective persons. Indeed, it is part of Humesenterprise to secure the epistemic standing of many beliefs in non-human animals.This is explicit in the Enquiry: the most ignorant and stupid peasants, nay infants, nayeven brute beasts, improve by experience, and learn the qualities of natural objects,by observing the effects, which result from them (EHU39, emphasis added; cf.105).The externalist theories provide a direct explanation of the epistemic achievementsof non-human animals. The Treatise, as well as the Enquiry, devotes a section to thecognitive achievements, the reason, of non-human animals. Custom operates allthe way down, from the most reflective philosophers to non-human animals. Humesappeal to custom is thus central to his constructive project of explaining how beliefs ofall these organisms can constitute knowledge in their own right. Epistemic parasitismis antithetical to Humes competing picture, in which the most ordinary creaturespossess full-fledged knowledge.

We might summarize some of our results as follows. Section I.iii.6 can be neitherpurely descriptive nor purely skeptical in intent. Not purely descriptivethat induc-tive inference is justified is built into the characterization of the phenomenon to be

54 This formulation applies both for foundationalist and coherentist versions of internalism. I leaveopen the extent to which the latter are in play in Hume. See Humes Epistemic Options in Loeb(forthcoming).


16/18

336 Synthese (2006) 152:321338

explained and constrains the admissible explanations. And not purely skepticalHumes endorsement of induction survives both the negative claim that inductiveinference is not caused by reason and the constructive claim that it is caused bycustom. This does not leave Hume and his interpreters in the position of having to

walk a fine line. To the contrary, externalism represents a broad swath of middleground, allowing Hume to be engaged in normative epistemology without embracingthe Russellian problem of induction. The difficulty has been to get this territory intocommentators field of vision. As Garrett observes, Hume relies on inductive infer-ence before, during, and after I.iii.6.55 We can suppose that Hume just could nothelp himself; or we can recognize that his commitment to induction is sustained byan externalist orientation in epistemology, and hence is impervious to internalist orRussellian attack.

At this juncture, we are positioned to entertain a way of thinking about the moti-vation for descriptivism. Consider the old saw that Hume confuses psychology with

philosophy. On its face, this is uncharitable. We might think of proponents of descrip-tivism as opting for a more charitable version of this line: Hume does not confusepsychology with philosophy; he self-consciouslyseparates them. As we have seen, thisis not Hume. To the contrary, it requires us to jettison one of his most distinctivephilosophical contributions: that the psychology of belief-formation, the characteris-tics of the psychological mechanisms that produce belief, are critical determinants ofthe epistemic status of belief.

Who would have thought Humes view a confusion? The answer, painting witha broad brush, is interpreters under the sway of logical empiricism, where we find

the claim that philosophy, and hence epistemology, is a priori. Psychology is at bestrelevant to the context of discovery, not to the context of justification. Theoriesof evidence are essentially theories of a priori confirmation relations. For Hume tothink that evidence depends upon genetic facts about belief then appears to be aconfusion. This would have bothered positivistically inclined interpreters. After all,logical positivism claimed Hume as one of its own insofar as he stressed the dividebetween relations of ideas and matters of fact, in the service of an onslaught onmetaphysics. Interpreters in the positivist tradition thus strived to save Hume fromhimself.

The positivist tradition, however, predates more naturalistic strands in 20th century

epistemology. Positivists were either foundationalists (as with Schlick) or coherentists(as with Neurath), but internalists either way. They did not contemplate, and cer-tainly did not accept, externalist theories. We can think of descriptivism as resultingfrom damage control within an interpretive tradition that does not see externalismas an option. I am unsure that in aligning himself with externalism Hume was aheadof his time, but he was certainly ahead of his positivistically-inclined 20th centuryinterpreters.

Acknowledgements I am grateful to Garrett and the other participants in the April, 2005 WakeForest University Hester Conference on Humes Naturalism for lively and useful discussion. I am

indebted especially to Broughton and Millican for helping me advance my thinking on the topic ofthis paper.

55 Garrett (1997, p 78).


17/18

Synthese (2006) 152:321338 337

References

Arnold, N. S. (1983). Humes skepticism about inductive inference. Journal of the History of Philoso-phy, 21, 3155.

Ayer, A. J. (1963). A concept of a person and other essays. London: Macmillan.

Beauchamp, T. L., & Mappes, T. A. (1975). Is Hume really a sceptic about induction? AmericanPhilosophical Quarterly, 12, 119129.

Beauchamp, T. L., & Rosenberg, A. (1981). Hume and the problem of causation, New York: OxfordUniversity Press.

Bennett, J. (1971). Locke, Berkeley, Hume: central themes. Oxford: Clarendon Press.Bennett, J. (2001). Learning from six philosophers (Vol. 2). Oxford: Clarendon Press.Broughton, J. (1983). Humes skepticism about causal inferences. Pacific Philosophical Quarterly,

64, 318.Broughton, J. (2004). Causal inference and the uniformity principle. American Philosophical Associa-

tion Central Division.Broughton, J. (forthcoming). Humes explanation of causal inference. In P. Hoffman, D. Owen, &

G. Yaffe (Eds.), Early modern metaphysics: essays in honor of Vere Chappell. Toronto:

Broadview Press.Connon, R. W. (1979). The naturalism of Hume revisited. In D. F. Norton, N. Capaldi, & W. L. Robison

(Eds.), McGill Hume studies, studies in Hume and Scottish philosophy (vol. 1, pp. 121145). SanDiego: Austin Hill Press.

Craig, E. (1987). The mind of god and the works of man. Oxford: Clarendon Press.Flew, A. (1961). Humes philosophy of belief, a study of his first inquiry. London: Routledge & Kegan

Paul.Flew, A. (1986). David Hume: philosopher of moral science. Oxford: Basil Blackwell.Fogelin, R. J. (1985). Humes skepticism in the treatise of human nature. London: Routledge.Garrett, D. (1997). Cognition and commitment in Humes philosophy. Oxford: Clarendon Press.Garrett, D. (1998). Ideas, reason, and skepticism: response to my critics. Hume Studies, 24, 171194Garrett, D. (2002). Hume on testimony concerning miracles. In P. Millican (Ed.), Reading Hume on

human understanding (pp. 301334). Oxford: Clarendon Press, Oxford University Press.Hacking, I. (1975). The emergence of probability. Cambridge: Cambridge University Press.Howson, C. (2000). Humes problem: induction and the justification of belief. Oxford: Clarendon Press.Kemp Smith, N. (1905). The naturalism of Hume (I). Mind, 14, 149173.Kemp Smith, N. (1941). The philosophy of David Hume: a critical study of its origins and central

doctrines. Macmillan, London (Reprinted in 1966, by New York: St. Martins Press).Loeb, L. E. (2001). Integrating Humes accounts of belief and justification. Philosophy and Phenom-

enological Research, 63, 279303.Loeb, L. E. (2002). Stability and justification in Humes Treatise. New York: Oxford University Press.Loeb, L. E. (2004). Stability and justification in Humes Treatise, another looka response to Erin

Kelly, Frederick Schmitt, and Michael Williams. Hume Studies, 30, 339404.Loeb, L. E. (forthcoming). Inductive inference in Humes philosophy. In E. Radcliffe (Ed.), Blackwell

Companion to Hume. Oxford: Blackwell.Meeker, K. (1998). Hume: radical skeptic or naturalized epistemologist? Hume Studies, 24, 3152.Millican, P. (1998). Hume on reason and induction: epistemology or cognitive science? Hume Studies,

24, 141159.Millican, P. (2002). Humes sceptical doubts concerning induction. In P. Millican (Ed.), Reading Hume

on human understanding (pp. 107173). Oxford: Clarendon Press.Noonan, H. W. (1999). Hume on knowledge. London: Routledge.Owen, D. (1999). Humes reason, Oxford: Oxford University Press.Passmore, J. (1952/1968). Humes intentions. London: Gerald Duckworth.Pears, D. (1990). Humes system: an examination of the first book of his treatise. Oxford: Oxford

University Press.

Penelhum, T. (1992). David Hume: an introduction to his philosophical system. West Lafayette, IN:Purdue University Press.Popkin, R. H. (1951). David Hume: his Pyrrhonism and his critique of Pyrrhonism. Philosophical

Quarterly, 1, 385407 (Reprinted in V. Chappell (Ed.) (1965). Hume: a collection of critical essays(pp. 5398). Garden City, NY: Doubleday, Anchor Books.

Popper, K. R. (1972). Objective knowledge: an evolutionary approach. Oxford: Clarendon Press.


18/18

338 Synthese (2006) 152:321338

Price, H. H. (1940). The permanent significance of Humes philosophy. Philosophy, 15, 737(Reprinted in A. Sesonske & N. Fleming (Eds.), Human understanding: studies in the philos-ophy of David Hume (pp. 533). California: Wadsworth, Belmont).

Price, H. H. (1969). Belief. London: George Allen & Unwin.Prichard, H. A. (1950). Knowledge and perception. Oxford: Clarendon Press.

Russell, B. (1912). The problems of philosophy. London: Butterworth.Russell, B. (1945). A history of western philosophy. New York: Simon and Schuster.Schmitt, F. F. (1992). Knowledge and belief. London: Routledge.Schmitt, F. F. (2004). Loeb on stability and justification in Humes Treatise. Hume Studies, 30, 297327.Stove, D. C. (1965). Hume, probability, and induction. Philosophical Review, 74, 160177 (Reprinted

in V. Chappell (Ed.), (1965). Hume: a collection of critical essays (pp. 187212). Garden City, NY:Doubleday, Anchor Books).

Stroud, B. (1977). Hume. London: Routledge & Kegan Paul.Stroud, B. (1991). Humes scepticism: natural instincts and philosophical reflection. Philosophical

Topics, 19, 271291 (Reprinted in M. Atherton (Ed.), (1999). The empiricists: critical essays onLocke, Berkeley, and Hume (pp. 229252). Lanham, MD: Rowman & Littlefield).

Will, F. (1947). Will the future be like the past? Mind, 56, 332347.

Williams, M. (2004). The unity of Humes philosophical project, Hume Studies, 30, 265296.Winkler, K. (1999). Humes inductive skepticism. In M. Atherton (Ed.), The empiricists: critical essays

on Locke, Berkeley, and Hume (pp. 183212), Lanham, MD: Rowman & Littlefield.Winters, B. (1979). Hume on reason. Hume Studies, 5, 2035.Wolterstorff, N. (1996). John Locke and the ethics of belief. Cambridge: Cambridge University Press.

Documents

Loeb -Psychology, Epistemology, And Skepticism in Hume_s Argument About Induction