Bonjour Knowledge and Skepticism

Laurence BonJourAnn Baker

Philosophical Problems

Longman Publishers51 Madison Avenue

New York, NY 10010www.ablongman.com

ISBN-10: 0205539378 / ISBN-13: 9780205539376(Please use above number to order your exam copy.)

© 2008

s a m p l e c h a p t e rThe pages of this Sample Chapter may have

slight variations in final published form.

Visit www.ablongman.com/replocator to contact your local Allyn & Bacon/Longman representative.

An Annotated Anthology, 2/e

2Knowledgeand Skepticism

42

*It has become clear as a result of relatively recent discussion that the traditional definition of knowledge is inadequate in thatthere are cases in which the three conditions are satisfied but it is intuitively clear that genuine knowledge is not present. Intu-itively, these are cases in which the belief is true in an accidental way that does not connect appropriately with the way inwhich it is justified. The problem has become known as “the Gettier problem,” named after the philosopher who firstdiscovered it, and the interested student can easily find discussions of it in philosophical dictionaries and encyclopedias.

The selections in this chapter are concerned with two main issues in the general areaof philosophy known as the theory of knowledge or epistemology, both having to dowith whether and how knowledge of certain specified kinds is possible, and with the

possibility and plausibility of skepticism with respect to the kinds of knowledge in question.The first and larger section of the chapter focuses on one of the most central issues of mod-ern philosophy: the problem of the external world, which asks whether and how beliefs aboutmaterial objects outside the mind are justified on the basis of sense perception. A secondsection deals with the problem of induction, which asks whether and why it is justifiable toinfer from apparent regularities found in observation to more general conclusions that extendbeyond those observations.

Before we discuss these more specific problems, it will be helpful to say somethingabout the concept of knowledge itself. According to a definition often characterized astraditional or standard, knowledge is (1) belief that is (2) true and (3) adequately justified.Problems can be raised about each of these three elements, but it will be adequate for thepurposes of this chapter to understand belief as the mental acceptance or affirmation of apropositional claim and truth as the correspondence or agreement of that claim withwhatever it is about. The nature of justification in the sense relevant to knowledge hasbeen the subject of widespread debate in recent philosophy, but the authors of all the pres-ent selections would have accepted the view that justification is having a good reason orwarrant for thinking that the claim in question is true.*

One other preliminary remark is perhaps worth making: the issues discussed in thischapter depend heavily on concepts and terminology that are relatively technical and dif-ficult. You will need to work at mastering these, and you should only expect gradual suc-cess, especially at first. This also applies to the material in this introduction: we have keptit short so that you can reread it easily, and if you do so, you will find concepts and issuesthat were initially puzzling gradually becoming clear.

C H A P T E R

DO WE HAVE KNOWLEDGE OF THE EXTERNAL WORLD? 43

Do We have Knowledge of the External World?It seems entirely obvious from the standpoint of both common sense and science, first, thatwe have a great deal of knowledge of material objects, and of the material world generally;and, second, that the basis of this knowledge is sense experience. The basic issue is how thisworks, and in particular how sense experience provides us with justification of the sort justexplained for the beliefs about material objects that we arrive at on the basis of it. One ini-tial question concerns the nature of sense experience itself: when we have the sense expe-rience of, for example, a table, what is it that we are aware of in the most direct or immedi-ate way—what is it that is “directly before the mind”? Contrary to the apparent view ofcommon sense, most of the philosophers in this section would reply that the most direct orimmediate object of awareness in sense experience is not the external physical object (thetable) whose presence the experience seems in some way to reflect, but rather a subjectivemental entity: a perceptual idea, in the terminology of John Locke and George Berkeley, ora sense-datum (plural: sense-data), in the terminology of more recent philosophers. BothRené Descartes and Locke take such a view more or less for granted, as does LaurenceBonJour; Berkeley argues for it at considerable length; and Thomas Reid rejects it. SextusEmpiricus also seems to have something like this view of sense experience in mind.

If this view of the nature of sense experience is accepted, then the question becomes howclaims about material objects can be justified on the basis of awareness of sensory ideas orsense-data. One alternative here is representative realism: the view that perceptual beliefsabout material objects can be justified by an argument or inference starting from an awarenessof the ideas or sense-data that lead us to have those beliefs and seem in some way to representthe corresponding objects. Descartes, Locke, and BonJour all defend versions of such a view,while Berkeley and Reid reject the possibility of any cogent inference of this sort.

A second, initially less obvious alternative is to say that material objects, rather thanexisting outside of and independently of the mind, are in fact themselves nothing more thansystematic collections or patterns of sensory ideas or sense-data, including both those thatare actually experienced and those that would be experienced under various conditions. Allsuch views are versions of idealism. Berkeley defends one specific version of idealism,according to which the ideas that make up “real” material objects are systematically put intoour human minds by God (who is thus the ultimate source of their reality). A more recentversion of idealism that dispenses with God and takes the availability of orderly patterns ofsense-data to be itself the most fundamental, not further explainable fact about reality iscalled phenomenalism; this view is explained and criticized by BonJour. (Views of thissort are likely to seem quite puzzling and hard to take seriously from a common-sensestandpoint, and you will have to work to get them clearly in mind.)

A third alternative, seemingly much closer to common sense than either of the others, isto reject the common denominator of these other two views, the idea that the immediate objectof awareness in perceptual experience is a sensory idea or sense-datum, and hold instead thatwe are directly or immediately aware of material objects themselves, with no need for the sortof justifying inference invoked by the representative realists. Once referred to as naïve real-ism because it was thought to be philosophically untenable in an obvious way, such a view hasmore recently been referred to simply as direct realism. A version of direct realism isdefended by Reid, who argues that the acceptability of our sensory beliefs about the materialworld is a “first principle” of human thought, one that requires no further defense.

A final alternative is skepticism: in relation to the problem of knowledge of the externalworld, the view that we have no such knowledge and that the impression that we do is simplyan illusion. One way to argue for skepticism is by appeal to the problems that face the variouspositive accounts of how such knowledge works, problems that turn out to be very serious.But it is also possible to argue more directly for a skeptical conclusion. The final selection inthis part of the chapter is a brief excerpt from perhaps the most famous representative of theancient skeptical school of philosophy, Sextus Empiricus, who argues that we can have noknowledge of the true nature of external objects.

Is Induction Justified?The problem of induction concerns the justification for inferring from regularities observedin experience to more general claims. Thus suppose that a large number of cases of someobservable property or category A have been observed (by various observers and underwidely varied collateral conditions)—for example, a large number of eggs laid by robinshave been observed in many different locations with as much variation of other conditions(temperature, time of year, elevation, etc.) as the habits of robins permit. Suppose alsothat all the observed cases of A have also been cases of some further observable propertyor category B—all of the observed robin’s eggs have been blue-green and speckled.(A more general version of the problem would include the possibility that some definite andstable fraction m/n of A’s, rather than all of them, have been B’s.) Given an observationalpremise of this sort, inductive reasoning of the most standard sort leads to the conclusionthat all cases of A (observed or not, past, present, or future) are also cases of B—that allrobin’s eggs are blue-green and speckled. (Or in the more general version, the conclusionwould be that approximately m/n of all A’s are B’s.)

It is obvious that reasoning of this sort is generally taken for granted by common senseand very extensively relied upon by science. (Here you should stop to reflect on how verymany of the things that we think that we know about the world rely on reasoning of thissort: claims about which foods are safe and nourishing, about the behavior of various sortsof animals, about weather conditions, about human behavior in various sorts of circum-stances, about the properties of various sorts of minerals and other substances, etc., etc.)The question is whether such reasoning is rationally justified, and if so, why: that is, whysuch a conclusion is genuinely likely to be true whenever the corresponding premise istrue. Hume considers this issue in the more specific context of a causal claim (with A thesupposed cause and B the supposed effect, and observed A’s all being followed by B’s).Having posed the problem, he argues for the skeptical view that there is no rational justi-fication at all for an inductive conclusion because there is no cogent reasoning from theobservational premise to the inductive conclusion. (He also offers a psychological accountof why we reason in this way.) Wesley Salmon recapitulates Hume’s account and thensurveys a number of different responses, rejecting most of them but arguing that oneresponse is at least promising. A. C. Ewing (in the course of a more general discussion ofa priori knowledge) defends one of the responses that Salmon rejects: a rationalist solu-tion that holds that inductive reasoning is justified a priori.

Appendix: A Priori Justification and KnowledgeA third epistemological issue that is, on some views at least, highly relevant to these othertwo is the issue of a priori knowledge, knowledge whose justification does not depend onsense experience. Salmon and Ewing differ fundamentally on the nature of such knowledgeand, as a result, on whether an a priori justification of induction is possible. And while thisis not discussed very explicitly in the readings, the sort of representative realist solution tothe problem of the external world that is defended by Descartes, Locke, and BonJour wouldapparently also have to rely on the sort of a priori knowledge that Ewing accepts andSalmon rejects: a priori knowledge that is not merely definitional (or “tautological”) incharacter. For only knowledge of this sort could apparently justify the inference from claimsabout sensory experience to claims about external material objects—as BonJour explains(following Hume), such an inference could not be justified by experience since there is noway from a representative realist standpoint to experience a connection or correlationbetween experience and such objects. What follows is a brief introduction to a priori knowl-edge and the questions it raises.

The issue of a priori knowledge is structured around three main distinctions, all of themcomplicated and subtle—and easily confused with one another. First, there is a metaphysicaldistinction between two ways in which a proposition or claim can be true or false. Some

44 CHAPTER 2 KNOWLEDGE AND SKEPTICISM

APPENDIX: A PRIORI JUSTIFICATION AND KNOWLEDGE 45

propositions, most philosophers agree, are logically or metaphysically necessary: true in anyworld or situation that is logically or metaphysically possible; while others are logically ormetaphysically contingent: true in some logically or metaphysically possible worlds and notin others. Thus, for example, propositions of logic and mathematics are usually held to benecessary in this sense, while most propositions about things and happenings in the materialworld are contingent—contingent truths if they are in fact true in the actual world. (Someunfortunate propositions are necessarily false: false in every logically or metaphysicallypossible world; the proposition that 2 + 2 = 5 is an example.)

Second, there is the epistemological distinction between two different ways in which aproposition can be justified: two different sorts of reasons for thinking that it is true (andderivatively, assuming that knowledge requires justification, two different kinds of knowl-edge). While it seems obvious that much of our knowledge depends for its justification onsense experience and perhaps also other, similar sorts of experience, such as introspectiveexperience (and so is justified empirically or a posteriori), it has seemed equally obvious tomost philosophers that there is knowledge that does not depend on sense experience for itsjustification, but is instead a priori: justified via pure reason or thought alone. Here the mostobvious examples are again the claims of logic and mathematics, but there are many othersorts of claims that have been claimed (rightly or wrongly) to be justified a priori: metaphys-ical claims (for example, the claim that there must be an explanation for everything that hap-pens or that space has only three dimensions); various claims about common-sense proper-ties and relations (that nothing can be red all over and green all over at the same time or thatall cubes have 12 edges); and also some ethical claims (such as that causing unnecessary suf-fering is morally wrong or that racial discrimination is unjust). In cases like these, it isclaimed, someone who adequately understands the claim in question can simply “see” on anintuitive basis that it is true, indeed that it must be true. The central issues here are: First, howis such a priori knowledge possible, given that it is not justified by sense experience? Whereexactly does the justification for such claims come from? If the appeal is to a priori intuition,what does such intuition amount to and how exactly does it work? And, second, what spe-cific sorts of things are knowable on this sort of basis? In particular, does a priori knowledgeinclude only essentially trivial (even though perhaps sometimes complicated) matters ofdefinition (such as the claim that all bachelors are unmarried)—claims often referred to astautologies or, more technically, as “analytic”?

Third, there is a logical or structural distinction between two kinds of propositions,analytic and synthetic. The explicit formulation of this distinction derives from the greatGerman philosopher Immanuel Kant. As Kant defines the notion, an analytic propositionis one of subject-predicate form whose predicate is included in its subject, either explicitly(for example, the claim that all tall men are tall) or implicitly (for example, the claim thatall bachelors are unmarried). The point of the distinction for Kant is that if a proposition isanalytic, then it is seemingly quite easy to see how it can be justified and known a priori:simply by understanding the concepts involved and noticing the relation of containment.What is problematic is how any synthetic proposition—one that does not have this sort ofform or logical structure, that is, in which the predicate is not contained in the subject—could still be justified and known a priori, though this is something that Kant nonethelessbelieves to occur. (What makes this distinction especially tricky is that some philosophers,such as Salmon in some places, have retained these terms and the basic idea that a certainsort of logical structure can be used to explain a priori justification, while altering Kant’saccount of what this structural feature is. This raises the issue, often not addressed veryexplicitly, of whether the new account of analyticity has the same capacity to explain apriori justification that the old one did, something that cannot be assumed just because thesame term “analytic” is used. You should keep this question in mind as you read Salmon.)

The main opposed positions on the issue of the a priori are versions of empiricism andrationalism. Empiricism is a general view about human cognition that involves two ratherdifferent main theses, each of them having to do with the relation between cognition andsense experience, and each of them reflected in Locke’s thought, though one much more


clearly and unambiguously than the other. The first thesis, most standardly referred to asconcept empiricism (but it could also, in relation to the terminology used by Locke andothers, be called idea empiricism), is a claim about where and how the human mindacquires the ideas or concepts that it uses to think about the world or indeed about anything.According to the concept empiricist, all concepts are derived from sense experience (con-strued broadly so as to also include introspective experience). Here the main opposing viewis the claim, held by Descartes and his rationalist successors, that at least some ideas or con-cepts are innate, “programmed” into the mind at birth (with this “programming” usuallyattributed to God).

The second main strand of empiricism (the one that is more relevant to the main issueabout a priori knowledge) lacks a completely standard label, but will be referred to here asjustificatory empiricism. This is a view about the reasons or warrant for thinking that beliefsor propositional claims are true. According to one version of justificatory empiricism, all of thejustification for claims that are not mere logical or definitional tautologies (not analytic inKant’s sense or perhaps in some other sense of that term) must derive from sense experience(again construed broadly so as to include introspection). A version of this moderate empiricistview is defended by Salmon, in the course of his discussion of induction.* The main alterna-tive to moderate empiricism is the moderate rationalist view that a priori justification and apriori knowledge include more than mere definitional tautologies or analytic propositions.Instead, it is claimed, the human mind has the capacity for direct a priori insight into certainnecessary features of reality. Here the main examples would be the kinds of claims briefly listedabove, none of which are, according to the rationalist, mere matters of definition or analytic inany epistemologically helpful sense. (The moderate rationalist does not claim that all justifica-tion and knowledge is a priori—a view that very few philosophers since perhaps Plato haveever held.) Such a view is defended by Ewing.

Do We Have Knowledge of the External World?

René DescartesRené Descartes (1596–1650), a French philosopher and mathematician, wasone of the most important and influential philosophers of all time. Descartesis called “the father of modern philosophy” because several of the centralproblems and themes of modern (post-Renaissance) philosophy first appearin his work: most fundamentally, the insistence on beginning with questionsabout knowledge (epistemological questions) rather than questions aboutreality (metaphysical questions). Among the more specific philosophicalproblems that first appear clearly in Descartes are the problem of the exter-nal world (concerning how claims about material objects can be justified onthe basis of sense experience) and the mind-body problem (see Chapter 4).The Meditations is his most important and influential work.

Descartes’s fundamental motivation for asking questions about knowl-edge is reflected in the very first sentence of the Meditations: “Several yearshave now passed since I first realized how numerous were the false opinionsthat in my youth I had taken to be true, and thus how doubtful were all thosethat I had subsequently built upon them.” Descartes lived in a time of greatintellectual ferment, when medieval views were still being replaced by moremodern ones and the scientific revolution was just under way. It was clear to

*There is also a more radical version of empiricism that denies the existence of a priori justification or knowledge of any sort;this view is not represented in the present anthology. The main proponent of this more radical version of empiricism is theAmerican philosopher and logician W. V. O. Quine.

RENÉ DESCARTES: FROM MEDITATIONS ON FIRST PHILOSOPHY 47

From Meditations on First Philosophy

him on the basis of the many conflicting opinions and arguments with whichhe was confronted, that many of the opinions he had previously accepted fromvarious sources were very likely to be false. The central issue of theMeditations is how to correct this situation. In writing the Meditations,Descartes is attempting to lay bare his process of thought so that the readercan follow along with him, think with him, and thereby come to the same con-clusions he does.

Descartes’s solution to the problem of how to eliminate erroneous beliefsis to accept only beliefs that are indubitable, in the sense of being incapableof being mistaken and thereby certain. His method for achieving such cer-tainty is to systematically doubt whole categories of belief, withholding assentfrom them on the basis of the mere possibility that they are false. He is thussuggesting that genuine knowledge requires justification that is conclusive:reasons that are strong enough to guarantee the truth of the claim in question.(Thus understood, Descartes agrees with the traditional conception of knowl-edge with one addition: knowledge is not merely justified true belief, but ratherconclusively justified true belief.)

By the end of Meditation One, Descartes has suspended belief in everyopinion that he has considered, first on the basis of the possibility that he mightbe dreaming and later on the basis of the much more radical possibility thathe might be deceived by an all-powerful “evil genius.” But in Meditation Two,he finds a belief that cannot be doubted for even the latter of these reasons: thebelief that he himself exists as a thinking thing. Later he identifies other beliefsabout which he thinks he can be certain on essentially the same basis:beliefs about his various conscious states of mind. And so the project inMeditations Three and Six is to reclaim as knowledge the previously suspendedbeliefs (or at least as many of them as possible) on the basis of the indubitablebeliefs identified in Meditation Two. In the end, you will have to try to judgewhether or not Descartes is successful at securing a foundation on the basis ofwhich his previously held beliefs can be justified: that is, on the basis of whichconclusive reasons can be given for thinking that those beliefs are true.

1 Stop and ThinkCan you think of things thatyou believed when you were

younger but no longer believe aretrue? Might some (or even many)of the things that you now believeturn out to be false in the sameway? (Suppose someone asked youfor your reasons for being confi-dent about the truth of variousbeliefs of which you are confident(think of specific examples): whatwould your answer be?)

2 DefinitionThis is Descartes’s famous

method of doubt: doubt anythingthat can possibly be false, therebyarriving (if anything is left) at cer-tainty. Notice that he is not seekingmerely psychological certainty.Many people are convinced ofvarious things (perhaps that Godexists), even though they can giveno reasons for thinking that theyare true, but this is not the kind ofcertainty that will help to avoiderror (since something that ispsychologically certain can still befalse). Instead he is looking for akind of certainty that guaranteestruth: beliefs for which there is aconclusive reason.

MEDITATION ONE: ConcerningThose Things That Can Be Called

into DoubtSeveral years have now passed since I first real-ized how numerous were the false opinions thatin my youth I had taken to be true, and thus howdoubtful were all those that I had subsequentlybuilt upon them. And thus I realized thatonce in my life I had to raze everything to theground and begin again from the originalfoundations, if I wanted to establish anythingfirm and lasting in the sciences. But the taskseemed enormous, and I was waiting until Ireached a point in my life that was so timely thatno more suitable time for undertaking theseplans of action would come to pass. For this rea-son, I procrastinated for so long that I would

1

henceforth be at fault, were I to waste the timethat remains for carrying out the project bybrooding over it. Accordingly, I have today suit-ably freed my mind of all cares, secured formyself a period of leisurely tranquility, and amwithdrawing into solitude. At last I will applymyself earnestly and unreservedly to this gen-eral demolition of my opinions.

Yet to bring this about I will not need to showthat all my opinions are false, which is perhapssomething I could never accomplish. But reasonnow persuades me that I should withhold myassent no less carefully from opinions that arenot completely certain and indubitable thanI would from those that are patently false. Forthis reason, it will suffice for the rejection of allof these opinions, if I find in each of them somereason for doubt. Nor therefore need I sur-vey each opinion individually, a task that wouldbe endless. Rather, because undermining thefoundations will cause whatever has been built

2

From Meditations on First Philosophy, 3rd ed., trans-lated by Donald A. Cress (Indianapolis: Hackett, 1993).

3

Rather than scrutinizingbeliefs individually, Descartesproposes to examine thefoundations upon which whole cate-gories of beliefs rest; the first suchcategory, discussed in the followingparagraphs, is beliefs justified bysense experience. (The idea of afoundation for belief or knowledgeappeals to an architecturalmetaphor—think about the analogybetween building construction andcognition that it suggests.)


upon them to crumble of its own accord, I willattack straightaway those principles whichsupported everything I once believed.

Surely whatever I had admitted until now asmost true I received either from the senses orthrough the senses. However, I have noticed thatthe senses are sometimes deceptive; and it is amark of prudence never to place our completetrust in those who have deceived us even once.

But perhaps, even though the senses dosometimes deceive us when it is a question ofvery small and distant things, still there aremany other matters concerning which one sim-ply cannot doubt, even though they are derivedfrom the very same senses: for example, that Iam sitting here next to the fire, wearing mywinter dressing gown, that I am holding thissheet of paper in my hands, and the like. But onwhat grounds could one deny that these handsand this entire body are mine? Unless perhaps Iwere to liken myself to the insane, whose brainsare impaired by such an unrelenting vapor ofblack bile that they steadfastly insist that theyare kings when they are utter paupers, or thatthey are arrayed in purple robes when they arenaked, or that they have heads made of clay, orthat they are gourds, or that they are made ofglass. But such people are mad, and I wouldappear no less mad, were I to take their behav-ior as an example for myself.

This would all be well and good, were I not aman who is accustomed to sleeping at night, andto experiencing in my dreams the very samethings, or now and then even less plausible ones,as these insane people do when they are awake.How often does my evening slumber persuademe of such ordinary things as these: that I amhere, clothed in my dressing gown, seated next tothe fireplace—when in fact I am lying undressedin bed! But right now my eyes are certainly wideawake when I gaze upon this sheet of paper. Thishead which I am shaking is not heavy with sleep.I extend this hand consciously and deliberately,and I feel it. Such things would not be so distinctfor someone who is asleep. As if I did not recallhaving been deceived on other occasions even bysimilar thoughts in my dreams! As I considerthese matters more carefully, I see so plainly thatthere are no definitive signs by which to distin-guish being awake from being asleep. As aresult, I am becoming quite dizzy, and this dizzi-ness nearly convinces me that I am asleep.

4

3

Let us assume then, for the sake of argument,that we are dreaming and that such particularsas these are not true: that we are opening oureyes, moving our head, and extending ourhands. Perhaps we do not even have such hands,or any such body at all. Nevertheless, it surelymust be admitted that the things seen duringslumber are, as it were, like painted images,which could only have been produced in thelikeness of true things, and that therefore atleast these general things—eyes, head, hands,and the whole body—are not imaginary things,but are true and exist. For indeed when paintersthemselves wish to represent sirens and satyrsby means of especially bizarre forms, theysurely cannot assign to them utterly newnatures. Rather, they simply fuse together themembers of various animals. Or if perhaps theyconcoct something so utterly novel that nothinglike it has ever been seen before (and thus issomething utterly fictitious and false), yet cer-tainly at the very least the colors from whichthey fashion it ought to be true. And by thesame token, although even these generalthings—eyes, head, hands, and the like—couldbe imaginary, still one has to admit that at leastcertain other things that are even more simpleand universal are true. It is from these compo-nents, as if from true colors, that all thoseimages of things that are in our thought arefashioned, be they true or false.

This class of things appears to include cor-poreal nature in general, together with itsextension; the shape of extended things; theirquantity, that is, their size and number; as wellas the place where they exist; the time throughwhich they endure, and the like.

Thus it is not improper to conclude from thisthat physics, astronomy, medicine, and all theother disciplines that are dependent upon theconsideration of composite things are doubtful,and that, on the other hand, arithmetic, geome-try, and other such disciplines, which treat ofnothing but the simplest and most generalthings and which are indifferent as to whetherthese things do or do not in fact exist, containsomething certain and indubitable. For whetherI am awake or asleep, two plus three make five,and a square does not have more than foursides. It does not seem possible that such obvi-ous truths should be subject to the suspicion ofbeing false. 6

5

4Restatement/ Summary This paragraph containsDescartes’s reason for

doubting this first category ofbeliefs.

RR

5

The suggestion is that thecreativity involved in dreams

is limited to the recombination ofelements derived in some otherway.This means that the dreaminghypothesis does not provide areason for thinking that everythinginvolved in our experience is mis-taken: the basic elements may stillbe true, even if the ways in whichthey are combined are mistaken.

RR

6

Thus while sciences that dealwith complex things are called

into question by the dreaminghypothesis, those that deal withthese simpler elements are not.Theexamples Descartes gives at theend of this paragraph suggest thathe has in mind at least primarilybeliefs justified on a completely dif-ferent basis than sense experience:a priori beliefs justified by reasonor rational thought, rather thansense experience (which could bejust a dream).

RR


Be that as it may, there is fixed in my mind acertain opinion of long standing, namely thatthere exists a God who is able to do anything andby whom I, such as I am, have been created. Howdo I know that he did not bring it about that thereis no earth at all, no heavens, no extended thing,no shape, no size, no place, and yet bringing itabout that all these things appear to me to existprecisely as they do now? Moreover, since I judgethat others sometimes make mistakes in mattersthat they believe they know most perfectly, mayI not, in like fashion, be deceived every time I addtwo and three or count the sides of a square, orperform an even simpler operation, if that can beimagined? But perhaps God has not willed that Ibe deceived in this way, for he is said to besupremely good. Nonetheless, if it were repug-nant to his goodness to have created me such thatI be deceived all the time, it would also seem for-eign to that same goodness to permit me to bedeceived even occasionally. But we cannot makethis last assertion.

…

Accordingly, I will suppose not a supremelygood God, the source of truth, but rather an evilgenius, supremely powerful and clever, who hasdirected his entire effort at deceiving me. I willregard the heavens, the air, the earth, colors,shapes, sounds, and all external things as noth-ing but the bedeviling hoaxes of my dreams,with which he lays snares for my credulity. I willregard myself as not having hands, or eyes, orflesh, or blood, or any senses, but as neverthe-less falsely believing that I possess all thesethings. I will remain resolute and steadfast inthis meditation, and even if it is not within mypower to know anything true, it certainly iswithin my power to take care resolutely to with-hold my assent to what is false, lest thisdeceiver, however powerful, however clever hemay be, have any effect on me. But thisundertaking is arduous, and a certain lazinessbrings me back to my customary way of living.I am not unlike a prisoner who enjoyed an imag-inary freedom during his sleep, but, when helater begins to suspect that he is dreaming, fearsbeing awakened and nonchalantly conspireswith these pleasant illusions. In just the sameway, I fall back of my own accord into my oldopinions, and dread being awakened, lest thetoilsome wakefulness which follows upon a

8

7

peaceful rest must be spent thenceforward not inthe light but among the inextricable shadows ofthe difficulties now brought forward.

MEDITATION TWO: Concerning theNature of the Human Mind: ThatIt Is Better Known Than the BodyYesterday’s meditation has thrown me intosuch doubts that I can no longer ignore them,yet I fail to see how they are to be resolved. Itis as if I had suddenly fallen into a deepwhirlpool; I am so tossed about that I can nei-ther touch bottom with my foot, nor swim up tothe top. Nevertheless I will work my way upand will once again attempt the same path Ientered upon yesterday. I will accomplish thisby putting aside everything that admits of theleast doubt, as if I had discovered it to be com-pletely false. I will stay on this course until Iknow something certain, or, if nothing else,until I at least know for certain that nothing iscertain. Archimedes sought but one firm andimmovable point in order to move the entireearth from one place to another. Just so, greatthings are also to be hoped for if I succeed infinding just one thing, however slight, that iscertain and unshaken.

Therefore I suppose that everything I see isfalse. I believe that none of what my deceitfulmemory represents ever existed. I have nosenses whatever. Body, shape, extension, move-ment, and place are all chimeras. What then willbe true? Perhaps just the single fact that nothingis certain.

But how do I know there is not somethingelse, over and above all those things that I havejust reviewed, concerning which there is noteven the slightest occasion for doubt? Is therenot some God, or by whatever name I might callhim, who instills these very thoughts in me? Butwhy would I think that, since I myself could per-haps be the author of these thoughts? Am I notthen at least something? But I have alreadydenied that I have any senses and any body. StillI hesitate; for what follows from this? Am I sotied to a body and to the senses that I cannotexist without them? But I have persuaded myselfthat there is absolutely nothing in the world: nosky, no earth, no minds, no bodies. Is it then thecase that I too do not exist? But doubtless I didexist, if I persuaded myself of something. But

7 Editor Comment

Here is an issue that willbecome important later on

in Meditation Six: God, beingperfectly good, would seeminglynot have made me so that I fall intoerror all the time.The problem isthat God’s goodness seemsincompatible with my making anyerrors at all, which I obviously do.(No resolution of the issue isoffered at this point.)

8

Here is Descartes’s secondand much more important

reason for doubting whole cate-gories of belief, one that is muchmore powerful than the one thatappeals to the possibility of dream-ing because it calls many morebeliefs into question, includingapparently those (like 2 + 3 = 5)that result from a priori reason.

Do any beliefs at all escapethis second reason for doubt?

RR


9

The insight here is the onethat Descartes expresses

more succinctly in another work(his Discourse on Method) as (inLatin) “Cogito ergo sum,” “I think,therefore I am”—often referred tojust as the Cogito.What exactly isthe claim that is supposed to beindubitable? Is Descartes right thateven the evil genius is unable todeceive him about the truth of thatclaim: to make him believe that it istrue when it is really false?

there is some deceiver or other who is supremelypowerful and supremely sly and who is alwaysdeliberately deceiving me. Then too there is nodoubt that I exist, if he is deceiving me. And lethim do his best at deception, he will never bringit about that I am nothing so long as I shall thinkthat I am something. Thus, after everything hasbeen most carefully weighed, it must finally beestablished that this pronouncement “I am, Iexist” is necessarily true every time I utter it orconceive it in my mind.

But I do not yet understand sufficiently whatI am—I, who now necessarily exist. And sofrom this point on, I must be careful lest I unwit-tingly mistake something else for myself, andthus err in that very item of knowledge that Iclaim to be the most certain and evident of all.Thus, I will meditate once more on what Ionce believed myself to be, prior to embarkingupon these thoughts. For this reason, then, I willset aside whatever can be weakened even to theslightest degree by the arguments broughtforward, so that eventually all that remains isprecisely nothing but what is certain andunshaken.

What then did I [formerly] think I was? Aman, of course. But what is a man? Might I notsay a “rational animal”? No, because then Iwould have to inquire what “animal” and“rational” mean. And thus from one question Iwould slide into many more difficult ones. Nordo I now have enough free time that I want towaste it on subtleties of this sort. Instead, permitme to focus here on what came spontaneouslyand naturally into my thinking whenever I pon-dered what I was. Now it occurred to me firstthat I had a face, hands, arms, and this entiremechanism of bodily members: the very sameas are discerned in a corpse, and which Ireferred to by the name “body.” It next occurredto me that I took in food, that I walked about,and that I sensed and thought various things;these actions I used to attribute to the soul. Butas to what this soul might be, I either did notthink about it or else I imagined it a rarified I-know-not-what, like a wind, or a fire, or ether,which had been infused into my coarser parts.But as to the body I was not in any doubt. On thecontrary, I was under the impression that I knewits nature distinctly. Were I perhaps tempted todescribe this nature such as I conceived it in mymind, I would have described it thus: by “body,”

9

I understand all that is capable of being boundedby some shape, of being enclosed in a place, andof filling up a space in such a way as to excludeany other body from it; of being perceived bytouch, sight, hearing, taste, or smell; of beingmoved in several ways, not, of course, by itself,but by whatever else impinges upon it. For it wasmy view that the power of self-motion, and like-wise of sensing or of thinking, in no waybelonged to the nature of the body. Indeed I usedrather to marvel that such faculties were to befound in certain bodies.

But now what am I, when I suppose that thereis some supremely powerful and, if I may bepermitted to say so, malicious deceiver whodeliberately tries to fool me in any way he can?Can I not affirm that I possess at least a smallmeasure of all those things which I have alreadysaid belong to the nature of the body? I focus myattention on them, I think about them, I reviewthem again, but nothing comes to mind. I amtired of repeating this to no purpose. But whatabout those things I ascribed to the soul? Whatabout being nourished or moving about? Since Inow do not have a body, these are surely nothingbut fictions. What about sensing? Surely this toodoes not take place without a body; and Iseemed to have sensed in my dreams manythings that I later realized I did not sense. Whatabout thinking? Here I make my discovery:thought exists; it alone cannot be separated fromme. I am; I exist—this is certain. But for howlong? For as long as I am thinking; for perhapsit could also come to pass that if I were to ceaseall thinking I would then utterly cease to exist.At this time I admit nothing that is not necessar-ily true. I am therefore precisely nothing but athinking thing; that is, a mind, or intellect, orunderstanding, or reason—words of whosemeanings I was previously ignorant. Yet I am atrue thing and am truly existing; but what kindof thing? I have said it already: a thinking thing.

What else am I? I will set my imagination inmotion. I am not that concatenation of memberswe call the human body. Neither am I even somesubtle air infused into these members, nor awind, nor a fire, nor a vapor, nor a breath,nor anything I devise for myself. For I havesupposed these things to be nothing. Theassumption still stands; yet nevertheless I amsomething. But is it perhaps the case that thesevery things which I take to be nothing, because


they are unknown to me, nevertheless are in factno different from that “me” that I know? This Ido not know, and I will not quarrel about it now.I can make a judgment only about things that areknown to me. I know that I exist; I ask now whois this “I” whom I know? Most certainly, in thestrict sense the knowledge of this “I” does notdepend upon things of whose existence I do notyet have knowledge. Therefore it is not depen-dent upon any of those things that I simulate inmy imagination. . . .

But what then am I? A thing that thinks. Whatis that? A thing that doubts, understands,affirms, denies, wills, refuses, and that alsoimagines and senses.

Indeed it is no small matter if all of these thingsbelong to me. But why should they not belong tome? Is it not the very same “I” who now doubtsalmost everything, who nevertheless understandssomething, who affirms that this one thing is true,who denies other things, who desires to knowmore, who wishes not to be deceived, who imag-ines many things even against my will, who alsonotices many things which appear to come fromthe senses? What is there in all of this that is notevery bit as true as the fact that I exist—even if Iam always asleep or even if my creator makesevery effort to mislead me? Which of these thingsis distinct from my thought? Which of them canbe said to be separate from myself? For it is soobvious that it is I who doubt, I who understand,and I who will, that there is nothing by which itcould be explained more clearly. But indeed it isalso the same “I” who imagines; for although per-haps, as I supposed before, absolutely nothingthat I imagined is true, still the very power ofimagining really does exist, and constitutes a partof my thought. Finally, it is this same “I” whosenses or who is cognizant of bodily things as ifthrough the senses. For example, I now see alight, I hear a noise, I feel heat. These things arefalse, since I am asleep. Yet I certainly do seem tosee, hear, and feel warmth. This cannot be false.Properly speaking, this is what in me is called“sensing.” But this, precisely so taken, is nothingother than thinking.

From these considerations I am beginning toknow a little better what I am. But it still seems(and I cannot resist believing) that corporealthings—whose images are formed by thought,and which the senses themselves examine—aremuch more distinctly known than this mysterious

10

“I” which does not fall within the imagination.And yet it would be strange indeed were I to graspthe very things I consider to be doubtful,unknown, and foreign to me more distinctly thanwhat is true, what is known—than, in short,myself. But I see what is happening: my mindloves to wander and does not yet permit itself tobe restricted within the confines of truth. So be itthen; let us just this once allow it completely freerein, so that, a little while later, when the time hascome to pull in the reins, the mind may morereadily permit itself to be controlled.

Let us consider those things which arecommonly believed to be the most distinctlygrasped of all: namely the bodies we touch andsee. Not bodies in general, mind you, for thesegeneral perceptions are apt to be somewhatmore confused, but one body in particular. Letus take, for instance, this piece of wax. It hasbeen taken quite recently from the honeycomb;it has not yet lost all the honey flavor. It retainssome of the scent of the flowers from which itwas collected. Its color, shape, and size are man-ifest. It is hard and cold; it is easy to touch. Ifyou rap on it with your knuckle it will emit asound. In short, everything is present in it thatappears needed to enable a body to be known asdistinctly as possible. But notice that, as I amspeaking, I am bringing it close to the fire. Theremaining traces of the honey flavor are disap-pearing; the scent is vanishing; the color ischanging; the original shape is disappearing.Its size is increasing; it is becoming liquid andhot; you can hardly touch it. And now, when yourap on it, it no longer emits any sound. Does thesame wax still remain? I must confess that itdoes; no one denies it; no one thinks otherwise.So what was there in the wax that was sodistinctly grasped? Certainly none of the aspectsthat I reached by means of the senses. For what-ever came under the senses of taste, smell, sight,touch or hearing has now changed; and yet thewax remains.

Perhaps the wax was what I now think it is:namely that the wax itself never really wasthe sweetness of the honey, nor the fragranceof the flowers, nor the whiteness, nor the shape,nor the sound, but instead was a body that a shorttime ago manifested itself to me in these ways,and now does so in other ways. But just what pre-cisely is this thing that I thus imagine? Let usfocus our attention on this and see what remains

10

Here Descartes is claimingthat there are many more

things that are beyond doubt in thesame way that his own existenceis—thus greatly expanding thefoundation of (alleged) certaintyupon which he will eventually tryto rebuild his knowledge.What arethese further things, and is he rightthat they are indubitable in the sameway, even given the evil genius possi-bility? (See Discussion Question 2.)


after we have removed everything that does notbelong to the wax: only that it is somethingextended, flexible, and mutable. But what is it tobe flexible and mutable? Is it what my imagina-tion shows it to be: namely, that this piece of waxcan change from a round to a square shape, orfrom the latter to a triangular shape? Not at all;for I grasp that the wax is capable of innumerablechanges of this sort, even though I am incapableof running through these innumerable changesby using my imagination. Therefore this insightis not achieved by the faculty of imagination.What is it to be extended? Is this thing’s exten-sion also unknown? For it becomes greater inwax that is beginning to melt, greater in boilingwax, and greater still as the heat is increased.And I would not judge correctly what the wax isif I did not believe that it takes on an even greatervariety of dimensions than I could ever graspwith the imagination. It remains then for me toconcede that I do not grasp what this wax isthrough the imagination; rather, I perceive itthrough the mind alone. The point I am makingrefers to this particular piece of wax, for the caseof wax in general is clearer still. But what is thispiece of wax which is perceived only by themind? Surely it is the same piece of wax that Isee, touch, and imagine; in short it is the samepiece of wax I took it to be from the very begin-ning. But I need to realize that the perception ofthe wax is neither a seeing, nor a touching, nor animagining. Nor has it ever been, even though itpreviously seemed so; rather it is an inspectionon the part of the mind alone. This inspection canbe imperfect and confused, as it was before, orclear and distinct, as it is now, depending on howclosely I pay attention to the things in which thepiece of wax consists.

…

But what am I to say about this mind, that is,about myself? For as yet I admit nothing else tobe in me over and above the mind. What, I ask,am I who seem to perceive this wax so dis-tinctly? Do I not know myself not only muchmore truly and with greater certainty, but alsomuch more distinctly and evidently? For if Ijudge that the wax exists from the fact that I seeit, certainly from this same fact that I see thewax it follows much more evidently that Imyself exist. For it could happen that what I seeis not truly wax. It could happen that I have no

eyes with which to see anything. But it is utterlyimpossible that, while I see or think I see (I donot now distinguish these two), I who think amnot something. Likewise, if I judge that the waxexists from the fact that I touch it, the sameoutcome will again obtain, namely that I exist. IfI judge that the wax exists from the fact that Iimagine it, or for any other reason, plainly thesame thing follows. But what I note regardingthe wax applies to everything else that is exter-nal to me. Furthermore, if my perception of thewax seemed more distinct after it becameknown to me not only on account of sight ortouch, but on account of many reasons, one hasto admit how much more distinctly I am nowknown to myself. For there is not a single con-sideration that can aid in my perception of thewax or of any other body that fails to make evenmore manifest the nature of my mind. But thereare still so many other things in the mind itselfon the basis of which my knowledge of it can berendered more distinct that it hardly seemsworth enumerating those things which emanateto it from the body.

But lo and behold, I have returned on my ownto where I wanted to be. For since I now know thateven bodies are not, properly speaking, perceivedby the senses or by the faculty of imagination, butby the intellect alone, and that they are notperceived through their being touched or seen,but only through their being understood, I mani-festly know that nothing can be perceived moreeasily and more evidently than my own mind. Butsince the tendency to hang on to long-held beliefscannot be put aside so quickly, I want to stop here,so that by the length of my meditation this newknowledge may be more deeply impressed uponmy memory.

MEDITATION THREE: ConcerningGod, That He Exists

I will now shut my eyes, stop up my ears, andwithdraw all my senses. I will also blot out frommy thoughts all images of corporeal things, orrather, since the latter is hardly possible, I willregard these images as empty, false and worth-less. And as I converse with myself alone andlook more deeply into myself, I will attempt torender myself gradually better known and morefamiliar to myself. I am a thing that thinks, thatis to say, a thing that doubts, affirms, denies,

1111

Descartes uses the waxexample to argue for this

claim (that bodies themselves arenot properly perceived by thesenses nor by the faculty of imagi-nation, but by the intellect alone).

RR


understands a few things, is ignorant of manythings, wills, refrains from willing, and alsoimagines and senses. For as I observed earlier,even though these things that I sense or imaginemay perhaps be nothing at all outside me,nevertheless I am certain that these modes ofthinking, which are cases of what I call sensingand imagining, insofar as they are merely modesof thinking, do exist within me.

In these few words, I have reviewed everythingI truly know, or at least what so far I have noticedthat I know. Now I will ponder morecarefully to see whether perhaps there may beother things belonging to me that up until now Ihave failed to notice. I am certain that I am athinking thing. But do I not therefore also knowwhat is required for me to be certain of anything?Surely in this first instance of knowledge, thereis nothing but a certain clear and distinct percep-tion of what I affirm. Yet this would hardly beenough to render me certain of the truth of athing, if it could ever happen that something thatI perceived so clearly and distinctly were false.And thus I now seem able to posit as a generalrule that everything I very clearly and distinctlyperceive is true.

Be that as it may, I have previously admittedmany things as wholly certain and evident thatnevertheless I later discovered to be doubtful.What sort of things were these? Why, the earth,the sky, the stars, and all the other things Iperceived by means of the senses. But what wasit about these things that I clearly perceived?Surely the fact that the ideas or thoughts of thesethings were hovering before my mind. But evennow I do not deny that these ideas are in me. Yetthere was something else I used to affirm,which, owing to my habitual tendency to believeit, I used to think was something I clearly per-ceived, even though I actually did not perceive itat all: namely, that certain things existed outsideme, things from which those ideas proceededand which those ideas completely resembled.But on this point I was mistaken; or rather, if myjudgment was a true one, it was not the result ofthe force of my perception.

But what about when I considered somethingvery simple and easy in the areas of arithmeticor geometry, for example that two plus threemake five, and the like? Did I not intuit them atleast clearly enough so as to affirm them astrue? To be sure, I did decide later on that I must

13

12

doubt these things, but that was only because itoccurred to me that some God could perhapshave given me a nature such that I might bedeceived even about matters that seemed mostevident. But whenever this preconceived opin-ion about the supreme power of God occurs tome, I cannot help admitting that, were he to wishit, it would be easy for him to cause me to erreven in those matters that I think I intuit asclearly as possible with the eyes of the mind.On the other hand, whenever I turn my attentionto those very things that I think I perceive withsuch great clarity, I am so completely persuadedby them that I spontaneously blurt out thesewords: “let anyone who can do so deceive me;so long as I think that I am something, he willnever bring it about that I am nothing. Nor willhe one day make it true that I never existed, forit is true now that I do exist. Nor will he evenbring it about that perhaps two plus three mightequal more or less than five, or similar items inwhich I recognize an obvious contradiction.”And certainly, because I have no reason forthinking that there is a God who is a deceiver(and of course I do not yet sufficiently knowwhether there even is a God), the basis fordoubting, depending as it does merely on theabove hypothesis, is very tenuous and, so tospeak, metaphysical. But in order to removeeven this basis for doubt, I should at the firstopportunity inquire whether there is a God, and,if there is, whether or not he can be a deceiver.For if I am ignorant of this, it appears I am nevercapable of being completely certain aboutanything else.

However, at this stage good order seems todemand that I first group all my thoughts intocertain classes, and ask in which of them truth orfalsity properly resides. Some of these thoughtsare like images of things; to these alone does theword “idea” properly apply, as when I think of aman, or a chimera, or the sky, or an angel, orGod. Again there are other thoughts that takedifferent forms: for example, when I will, orfear, or affirm, or deny, there is always something that I grasp as the subject of my thought,yet I embrace in my thought something morethan the likeness of that thing. Some of thesethoughts are called volitions or affects, whileothers are called judgments.

Now as far as ideas are concerned, if they areconsidered alone and in their own right, without

15

14

12

Here Descartes summarizesthe main result of Meditation

Two: the tiny amount of knowledgethat has (allegedly) escaped thedoubt resulting from the evil geniuspossibility. Stated in the first person(as is obviously appropriate), I canknow that I am thinking and alsothat I am thinking (experiencing,doubting, wondering) about variousthings, but not that anythingbeyond myself and my ownconscious states exists.

RR

13

This is Descartes’s suggestionfor a general rule or principle

that captures the way in which hisbelief about his own existenceescapes the doubt.

But is he right about this? Is“clarity and distinctness” the

reason that I can’t doubt my ownexistence? In fact, Descarteshimself proceeds to show that itis not, since he goes on to questionwhether things that are clear anddistinct are always true, eventhough he never questions hisown existence.

RR

14

Since Descartes has alreadysaid that God wouldn’t

deceive us in this way, it is betterto take him to be talking about theevil genius here.

RR

15

Here Descartes might seemto be questioning even his

own existence and states of mind,but the subsequent discussionshows he is not really doing so. (Ifhe did, he would have nothing tostart from in attempting to proveGod’s existence.) Thus the issueseems to be whether he can safelyaccept things other than his ownexistence and states of mind thatseem “clear and distinct” (thingslike 2 + 3 = 5).The evil geniuscould deceive him about suchthings, so the only way to be sure isto prove the existence of God (andthereby the nonexistence of theevil genius).


16

So Descartes is worriedabout whether his thoughts

about the world are accurate, notabout whether his thoughts abouthis thoughts are accurate, for (heclaims) he could never be in errorabout those.

being referred to something else, they cannot,properly speaking, be false. For whether it is ashe-goat or a chimera that I am imagining, it is noless true that I imagine the one than the other.Moreover, we need not fear that there is falsity inthe will itself or in the affects, for although I canchoose evil things or even things that are utterlynon-existent, I cannot conclude from this that itis untrue that I do choose these things. Thus thereremain only judgments in which I must take carenot to be mistaken. Now the principal and mostfrequent error to be found in judgments consistsin the fact that I judge that the ideas which are inme are similar to or in conformity with certainthings outside me. Obviously, if I were toconsider these ideas merely as certain modes ofmy thought, and were not to refer them to any-thing else, they could hardly give me any subjectmatter for error.

Among these ideas, some appear to me to beinnate, some adventitious, and some produced byme. For I understand what a thing is, what truthis, what thought is, and I appear to have derivedthis exclusively from my very own nature. Butsay I am now hearing a noise, or looking at thesun, or feeling the fire; up until now I judged thatthese things proceeded from certain things out-side me, and finally, that sirens, hippogriffs, andthe like are made by me. Or perhaps I can eventhink of all these ideas as being adventitious, oras being innate, or as fabrications, for I have notyet clearly ascertained their true origin.

But here I must inquire particularly into thoseideas that I believe to be derived from thingsexisting outside me. Just what reason do I havefor believing that these ideas resemble thosethings? Well, I do seem to have been so taughtby nature. Moreover, I do know from experiencethat these ideas do not depend upon my will, norconsequently upon myself, for I often noticethem even against my will. Now, for example,whether or not I will it, I feel heat. It is for thisreason that I believe this feeling or idea of heatcomes to me from something other than myself,namely from the heat of the fire by which I amsitting. Nothing is more obvious than the judg-ment that this thing is sending its likeness ratherthan something else into me.

I will now see whether these reasons are pow-erful enough. When I say here “I have been sotaught by nature,” all I have in mind is that I amdriven by a spontaneous impulse to believe this,

18

17

16

and not that some light of nature is showing methat it is true. These are two very different things.For whatever is shown me by this light of nature,for example, that from the fact that I doubt, itfollows that I am, and the like, cannot in any waybe doubtful. This is owing to the fact that therecan be no other faculty that I can trust as much asthis light and which could teach that these thingsare not true. But as far as natural impulses areconcerned, in the past I have often judged myselfto have been driven by them to make the poorerchoice when it was a question of choosing agood; and I fail to see why I should place anygreater faith in them than in other matters.

Again, although these ideas do not dependupon my will, it does not follow that they nec-essarily proceed from things existing outsideme. For just as these impulses about which Ispoke just now seem to be different from mywill, even though they are in me, so too perhapsthere is also in me some other faculty, one notyet sufficiently known to me, which producesthese ideas, just as it has always seemed up tonow that ideas are formed in me without anyhelp from external things when I am asleep.

And finally, even if these ideas did proceedfrom things other than myself, it does not there-fore follow that they must resemble those things.Indeed it seems I have frequently noticed a vastdifference in many respects. For example, I findwithin myself two distinct ideas of the sun. Oneidea is drawn, as it were, from the senses. Now itis this idea which, of all those that I take to bederived from outside me, is most in need ofexamination. By means of this idea the sunappears to me to be quite small. But there isanother idea, one derived from astronomical rea-soning, that is, it is elicited from certain notionsthat are innate in me, or else is fashioned by mein some other way. Through this idea the sun isshown to be several times larger than the earth.Both ideas surely cannot resemble the same sunexisting outside me; and reason convinces methat the idea that seems to have emanated fromthe sun itself from so close is the very one thatleast resembles the sun.

All these points demonstrate sufficiently thatup to this point it was not a well-founded judg-ment but only a blind impulse that formed thebasis of my belief that things existing outside mesend ideas or images of themselves to me throughthe sense organs or by some other means. 20

19

17

An innate idea is one thatis programmed into a person atbirth (by God, in Descartes’s view).Ideas produced by me are onesthat I make up or that are “ficti-tious.” What then is an “adventi-tious” idea? What is the third alter-native besides being innate or being“produced by me”?

19

By “the natural light,”Descartes seems to mean some-thing like a priori insight into why aclaim must be true—the sort ofinsight that is reflected in the claimseeming “clear and distinct.” Hispoint is that while I naturally andspontaneously believe what thesenses seem to tell me, there isnothing in that case that showsme that the claims in questionmust be true.

But, as we have seen, the evilgenius can apparently deceive

me even about things (other thanmy own existence and states ofmind) that seem to be shown bythe natural light or that seem clearand distinct (such as 2 + 3 = 5).

20

Descartes concludes that thetwo initial reasons (see

Annotation 18) have no real force.RR

18

Here are two initial reasonsfor the view that sense expe-

rience accurately reflects thenature of external things.

RR


But still another way occurs to me for inquiringwhether some of the things of which there areideas in me do exist outside me: insofar as theseideas are merely modes of thought, I see noinequality among them; they all seem to proceedfrom me in the same manner. But insofar as oneidea represents one thing and another ideaanother thing, it is obvious that they do differ verygreatly from one another. Unquestionably, thoseideas that display substances to me are somethingmore and, if I may say so, contain within them-selves more objective reality than those whichrepresent only modes or accidents. Again, theidea that enables me to understand a supremedeity, eternal, infinite, omniscient, omnipotent,and creator of all things other than himself,clearly has more objective reality within it thando those ideas through which finite substancesare displayed.

Now it is indeed evident by the light of naturethat there must be at least as much [reality] in theefficient and total cause as there is in the effect ofthat same cause. For whence, I ask, could aneffect get its reality, if not from its cause? Andhow could the cause give that reality to the effect,unless it also possessed that reality? Hence it fol-lows that something cannot come into being outof nothing, and also that what is more perfect(that is, what contains in itself more reality) can-not come into being from what is less perfect. Butthis is manifestly true not merely for those effectswhose reality is actual or formal, but also forideas in which only objective reality is consid-ered. For example, not only can a stone whichdid not exist previously not now begin to existunless it is produced by something in which thereis, either formally or eminently, everything that isin the stone; nor heat be introduced into a subjectwhich was not already hot unless it is done bysomething that is of at least as perfect an order asheat—and the same for the rest—but it is alsotrue that there can be in me no idea of heat, or ofa stone, unless it is placed in me by some causethat has at least as much reality as I conceive tobe in the heat or in the stone. For although thiscause conveys none of its actual or formal realityto my idea, it should not be thought for that rea-son that it must be less real. Rather, the verynature of an idea is such that of itself it needs noformal reality other than what it borrows from mythought, of which it is a mode. But that a particu-lar idea contains this as opposed to that objective

22

21

reality is surely owing to some cause in whichthere is at least as much formal reality as there isobjective reality contained in the idea. For if weassume that something is found in the idea thatwas not in its cause, then the idea gets that some-thing from nothing. Yet as imperfect a mode ofbeing as this is by which a thing exists in the intel-lect objectively through an idea, nevertheless it isplainly not nothing; hence it cannot get its beingfrom nothing.

Moreover, even though the reality that I amconsidering in my ideas is merely objectivereality, I ought not on that account to suspectthat there is no need for the same reality to beformally in the causes of these ideas, but that itsuffices for it to be in them objectively. For justas the objective mode of being belongs to ideasby their very nature, so the formal mode ofbeing belongs to the causes of ideas, at least tothe first and preeminent ones, by their verynature. And although one idea can perhapsissue from another, nevertheless no infiniteregress is permitted here; eventually some firstidea must be reached whose cause is a sort ofarchetype that contains formally all the realitythat is in the idea merely objectively. Thusit is clear to me by the light of nature that theideas that are in me are like images that can eas-ily fail to match the perfection of the thingsfrom which they have been drawn, but whichcan contain nothing greater or more perfect.

And the longer and more attentively I examineall these points, the more clearly and distinctly Iknow they are true. But what am I ultimately toconclude? If the objective reality of any of myideas is found to be so great that I am certain thatthe same reality was not in me, either formally oreminently, and that therefore I myself cannot bethe cause of the idea, then it necessarily followsthat I am not alone in the world, but that some-thing else, which is the cause of this idea, alsoexists. But if no such idea is found in me, I willhave no argument whatsoever to make mecertain of the existence of anything other thanmyself, for I have conscientiously reviewed allthese arguments, and so far I have been unable tofind any other.

Among my ideas, in addition to the one thatdisplays me to myself (about which there can beno difficulty at this point), are others that repre-sent God, corporeal and inanimate things, angels,animals, and finally other men like myself.

24

23

22

Here is the first of two impor-tant metaphysical principles

involved in Descartes’ argument: theprinciple that any cause must have atleast as much reality (must be at leastas perfect as) as its effect.He appliesthis to both formal or actual reality(reality in the ordinary sense) andobjective reality. (The labels “formal”and “objective” strike our contempo-rary ears as terminologically back-wards. What we would mean by“objective reality” is what Descartesmeans by “formal reality”.

21

Another approach is to seewhether the claim that some

of my ideas correspond to thingsexisting outside my mind can bederived from their content: whatthey seem to represent or depict.

Descartes puts this interms of the obscure (to us,anyway) medieval conception ofobjective reality: reality as an objectof thought. (So, for example, SantaClaus would have “objective reality”when anyone has a thought of SantaClaus.) In his view, different ideascontain different degrees of objec-tive reality, corresponding to thedegree of perfection of the object ofthe idea. It follows, he thinks, thatthe idea of God contains moreobjective reality than any other idea.

RR

23

Here is the second metaphys-ical principle involved in

Descartes’s argument: the principlethat any cause must have at least asmuch reality (must be at least asperfect as) as its effect. He appliesthis to both formal or actual reality(reality in the ordinary sense) andobjective reality. (The labels “formal”and “objective” strike our contem-porary ears as terminologicallybackwards.What we would mean by“objective reality” is what Descartesmeans by “formal reality”.

24

Thus the issue becomeswhether I have any idea that

I myself could not, according to theforegoing principle, be the cause of.If so, I can know on this basis thatsomething outside myself exists.

RR


As to the ideas that display other men, or ani-mals, or angels, I easily understand that theycould be fashioned from the ideas that I have ofmyself, of corporeal things, and of God—evenif no men (except myself), no animals, and noangels existed in the world.

As to the ideas of corporeal things, there isnothing in them that is so great that it seems inca-pable of having originated from me. For if I inves-tigate them thoroughly and examine each oneindividually in the way I examined the idea ofwax yesterday, I notice that there are only a veryfew things in them that I perceive clearly and dis-tinctly: namely, size, or extension in length,breadth, and depth; shape, which arises from thelimits of this extension; position, which variousthings possessing shape have in relation to oneanother; and motion, or alteration in position. Tothese can be added substance, duration, and num-ber. But as for the remaining items, such as lightand colors, sounds, odors, tastes, heat and coldand other tactile qualities, I think of these only ina very confused and obscure manner, to theextent that I do not even know whether they aretrue or false, that is, whether the ideas I have ofthem are ideas of things or ideas of non-things.

For although a short time ago I noted thatfalsity properly so called (or “formal” falsity) isto be found only in judgments, nevertheless thereis another kind of falsity (called “material”falsity) which is found in ideas whenever theyrepresent a non-thing as if it were a thing. Forexample, the ideas I have of heat and cold fall sofar short of being clear and distinct that I cannottell from them whether cold is merely the priva-tion of heat or whether heat is the privation ofcold, or whether both are real qualities, orwhether neither is. And because ideas can onlybe, as it were, of things, if it is true that cold ismerely the absence of heat, then an idea that rep-resents cold to me as something real and positivewill not inappropriately be called false. The sameholds for other similar ideas.

Assuredly I need not assign to these ideas anauthor distinct from myself. For if they werefalse, that is, if they were to represent non-things,I know by the light of nature that they proceedfrom nothing; that is, they are in me for no otherreason than that something is lacking in mynature, and that my nature is not entirely perfect.If, on the other hand, these ideas are true, thenbecause they exhibit so little reality to me that I

25

cannot distinguish it from a non-thing, I see noreason why they cannot get their being from me.

As for what is clear and distinct in the ideas ofcorporeal things, it appears I could have bor-rowed some of these from the idea of myself:namely, substance, duration, number, and what-ever else there may be of this type. For instance,I think that a stone is a substance, that is to say,a thing that is suitable for existing in itself; andlikewise I think that I too am a substance.Despite the fact that I conceive myself to be athinking thing and not an extended thing,whereas I conceive of a stone as an extendedthing and not a thinking thing, and hence thereis the greatest diversity between these two con-cepts, nevertheless they seem to agree with oneanother when considered under the rubric ofsubstance. Furthermore, I perceive that I nowexist and recall that I have previously existed forsome time. And I have various thoughts andknow how many of them there are. It is in doingthese things that I acquire the ideas of durationand number, which I can then apply to otherthings. However, none of the other componentsout of which the ideas of corporeal things arefashioned (namely extension, shape, position,and motion) are contained in me formally, sinceI am merely a thinking thing. But since these areonly certain modes of a substance, whereas I ama substance, it seems possible that they are con-tained in me eminently.

Thus there remains only the idea of God. Imust consider whether there is anything in thisidea that could not have originated from me.I understand by the name “God” a certain sub-stance that is infinite, independent, supremelyintelligent and supremely powerful, and thatcreated me along with everything else thatexists—if anything else exists. Indeed all theseare such that, the more carefully I focus myattention on them, the less possible it seems theycould have arisen from myself alone. Thus, fromwhat has been said, I must conclude that Godnecessarily exists.

For although the idea of substance is in me byvirtue of the fact that I am a substance, that factis not sufficient to explain my having the idea ofan infinite substance, since I am finite, unlessthis idea proceeded from some substance whichreally was infinite.

Nor should I think that I do not perceive theinfinite by means of a true idea, but only

26

25

Here Descartes has inmind at least approximately thedistinction between primary andsecondary qualities (about whichLocke will have much more to say).

26

The crucial idea is, of course,the idea of God.RRCan this idea be causedonly by me, given the two

principles?


through a negation of the finite, just as I per-ceive rest and darkness by means of a negationof motion and light. On the contrary, I clearlyunderstand that there is more reality in an infi-nite substance than there is in a finite one. Thusthe perception of the infinite is somehow priorin me to the perception of the finite, that is, myperception of God is prior to my perception ofmyself. For how would I understand that I doubtand that I desire, that is, that I lack somethingand that I am not wholly perfect, unless therewere some idea in me of a more perfect being,by comparison with which I might recognize mydefects?

…

But perhaps I am something greater than Imyself understand. Perhaps all these perfectionsthat I am attributing to God are somehow in mepotentially, although they do not yet assert them-selves and are not yet actualized. For I nowobserve that my knowledge is gradually beingincreased, and I see nothing standing in the wayof its being increased more and more to infinity.Moreover, I see no reason why, with my knowl-edge thus increased, I could not acquire all theremaining perfections of God.And, finally, if thepotential for these perfections is in me already, Isee no reason why this potential would not suf-fice to produce the idea of these perfections.

Yet none of these things can be the case. First,while it is true that my knowledge is graduallybeing increased and that there are many thingsin me potentially that are not yet actual, never-theless, none of these pertains to the idea ofGod, in which there is nothing whatever that ispotential. Indeed this gradual increase is itself amost certain proof of imperfection. Moreover,although my knowledge may always increasemore and more, nevertheless I understand thatthis knowledge will never by this means be actu-ally infinite, because it will never reach a pointwhere it is incapable of greater increase. On thecontrary, I judge God to be actually infinite, sothat nothing can be added to his perfection. . . .

Nor can one fancy that perhaps several partialcauses have concurred in bringing me intobeing, and that I have taken the ideas of the var-ious perfections I attribute to God from a varietyof causes, so that all of these perfections arefound somewhere in the universe, but not alljoined together in a single being—God. On the

27

contrary, the unity, the simplicity, that is, theinseparability of all those features that are inGod is one of the chief perfections that I under-stand to be in him. Certainly the idea of the unityof all his perfections could not have been placedin me by any cause from which I did not also getthe ideas of the other perfections; for neithercould some cause have made me understandthem joined together and inseparable from oneanother, unless it also caused me to recognizewhat they were.

. . . Indeed I have no choice but to concludethat the mere fact of my existing and of therebeing in me an idea of a most perfect being, thatis, God, demonstrates most evidently that Godtoo exists.

All that remains for me is to ask how I receivedthis idea of God. For I did not draw it from thesenses; it never came upon me unexpectedly, asis usually the case with the ideas of sensiblethings when these things present themselves (orseem to present themselves) to the external senseorgans. Nor was it made by me, for I plainly canneither subtract anything from it nor addanything to it. Thus the only option remaining isthat this idea is innate in me, just as the idea ofmyself is innate in me.

To be sure, it is not astonishing that increating me, God should have endowed me withthis idea, so that it would be like the mark of thecraftsman impressed upon his work, althoughthis mark need not be something distinct fromthe work itself. But the mere fact that Godcreated me makes it highly plausible that I havesomehow been made in his image and likeness,and that I perceive this likeness, in which theidea of God is contained, by means of the samefaculty by which I perceive myself. That is,when I turn the mind’s eye toward myself, Iunderstand not only that I am something incom-plete and dependent upon another, somethingaspiring indefinitely for greater and greater orbetter things, but also that the being on whomI depend has in himself all those greaterthings—not merely indefinitely and potentially,but infinitely and actually, and thus that he isGod. The whole force of the argument rests onthe fact that I recognize that it would be impos-sible for me to exist, being of such a nature asI am (namely, having in me the idea of God),unless God did in fact exist. God, I say, thatsame being the idea of whom is in me: a being

29

28

27

One reason for thinking thatI cannot be the cause of my

idea of God is that it involves theproperty of infinity and I am onlyfinite. Descartes considers theresponse that the idea of infinity isonly the negation of the idea of thefinite and claims that, on the con-trary, the idea of infinity or of aninfinite being comes first.

RR

29

Here Descartes refers to thethree options mentioned at

the point in the text whereAnnotation 17 occurs.

RR

28

The result that is supposedto follow from the ingredi-

ents already discussed finallyemerges here: only a genuinelyexisting God could be the cause ofDescartes’s idea of God. And onceDescartes knows that God exists,he can trust that his clear anddistinct ideas are true.The restof Meditation Three raises andattempts to answer variousobjections to his overall argument.

But there is a more seriousobjection that was raised by

Descartes’s contemporaries: isn’tDescartes arguing in a circle byassuming that he can trust the twometaphysical principles beforeestablishing that God exists? (SeeDiscussion Question 3.)

RR


having all those perfections that I cannotcomprehend, but can somehow touch with mythought, and a being subject to no defectswhatever. From these considerations it is quiteobvious that he cannot be a deceiver, for it ismanifest by the light of nature that all fraud anddeception depend on some defect.

But before examining this idea more closelyand at the same time inquiring into other truthsthat can be gathered from it, at this point I wantto spend some time contemplating this God, toponder his attributes and, so far as the eye of mydarkened mind can take me, to gaze upon, toadmire, and to adore the beauty of this immenselight. For just as we believe by faith that thegreatest felicity of the next life consists solely inthis contemplation of the divine majesty, so toowe now experience that from the same contem-plation, although it is much less perfect, thegreatest pleasure of which we are capable in thislife can be perceived.

…

MEDITATION SIX: Concerning theExistence of Material Things, and

the Real Distinction BetweenMind and Body

It remains for me to examine whether materialthings exist. Indeed I now know that they canexist, at least insofar as they are the object ofpure mathematics, since I clearly and distinctlyperceive them. For no doubt God is capable ofbringing about everything that I am capable ofperceiving in this way. And I have never judgedthat God was incapable of something, exceptwhen it was incompatible with my perceiving itdistinctly. . . .

But I am in the habit of imagining many otherthings, over and above that corporeal nature whichis the object of pure mathematics, such as colors,sounds, tastes, pain, and the like, though not sodistinctly. And I perceive these things better bymeans of the senses, from which, with the aid ofthe memory, they seem to have arrived at the imag-ination. Thus I . . . must see whether I can obtainany reliable argument for the existence of corpo-real things from those things that are perceived bythe mode of thinking that I call “sense.”

First of all, to be sure, I will review here all thethings I previously believed to be true because I

31

30

had perceived them by means of the senses andthe causes I had for thinking this. Next I willassess the causes why I later called them intodoubt. Finally, I will consider what I must nowbelieve about these things.

So first, I sensed that I had a head, hands, feet,and other members that comprised this bodywhich I viewed as part of me, or perhaps even asthe whole of me. I sensed that this body wasfound among many other bodies, by which mybody can be affected in various beneficial orharmful ways. I gauged what was opportune bymeans of a certain sensation of pleasure, andwhat was inopportune by a sensation of pain. Inaddition to pain and pleasure, I also sensedwithin me hunger, thirst, and other suchappetites, as well as certain bodily tendenciestoward mirth, sadness, anger, and other suchaffects. And externally, besides the extension,shapes, and motions of bodies, I also sensedtheir hardness, heat, and other tactile qualities. Ialso sensed light, colors, odors, tastes, andsounds, on the basis of whose variety I distin-guished the sky, the earth, the seas, and the otherbodies, one from the other. Now given the ideasof all these qualities that presented themselvesto my thought, and which were all that I properlyand immediately sensed, still it was surely notwithout reason that I thought I sensed things thatwere manifestly different from my thought,namely, the bodies from which these ideasproceeded. For I knew by experience that theseideas came upon me utterly without my consent,to the extent that, wish as I may, I could notsense any object unless it was present to a senseorgan. Nor could I fail to sense it when it waspresent. And since the ideas perceived by sensewere much more vivid and explicit and even, intheir own way, more distinct than any of thosethat I deliberately and knowingly formedthrough meditation or that I found impressed onmy memory, it seemed impossible that theycame from myself. Thus the remaining alterna-tive was that they came from other things. SinceI had no knowledge of such things except fromthose same ideas themselves, I could not helpentertaining the thought that they were similarto those ideas. Moreover, I also recalled thatthe use of the senses antedated the use of reason.And since I saw that the ideas that I myselffashioned were not as explicit as those that Iperceived through the faculty of sense, and were

32

30

This amounts to the claimthat clear and distinct per-

ception is at least a good criterionfor what is possible.

RR

31

We now come to the mainissue: does sense experience

yield any good reason for thinkingthat external material bodies exist?

RR

32

This is one of the reasonsmentioned earlier at the

point in the text where Annotation20 occurs. It is also one of thosethat Locke appeals to (in thefollowing selection) in his attemptto argue, without any appeal toGod, for the existence of materialobjects corresponding to senseexperience.

RR


for the most part composed of parts of the latter,I easily convinced myself that I had absolutelyno idea in the intellect that I did not have before-hand in the sense faculty. Not without reason didI judge that this body, which by a certain specialright I called “mine,” belongs more to me thandid any other. For I could never be separatedfrom it in the same way I could be from otherbodies. I sensed all appetites and feelings in andon behalf of it. Finally, I noticed pain and plea-surable excitement in its parts, but not in otherbodies external to it. But why should a certainsadness of spirit arise from some sensation orother of pain, and why should a certain elationarise from a sensation of excitement, or whyshould that peculiar twitching in the stomach,which I call hunger, warn me to have somethingto eat, or why should dryness in the throat warnme to take something to drink, and so on? Iplainly had no explanation other than that I hadbeen taught this way by nature. For there is noaffinity whatsoever, at least none I am aware of,between this twitching in the stomach and thewill to have something to eat, or between thesensation of something causing pain and thethought of sadness arising from this sensation.But nature also seems to have taught me every-thing else as well that I judged concerningthe objects of the senses, for I had alreadyconvinced myself that this was how things were,prior to my assessing any of the arguments thatmight prove it.

Afterwards, however, many experiencesgradually weakened any faith that I had in thesenses. Towers that had seemed round from afaroccasionally appeared square at close quarters.Very large statues mounted on their pedestalsdid not seem large to someone looking at themfrom ground level. And in countless other suchinstances I determined that judgments inmatters of the external senses were in error. . . .To these causes for doubt I recently added twoquite general ones. The first was that everythingI ever thought I sensed while awake I couldbelieve I also sometimes sensed while asleep,and since I do not believe that what I seem tosense in my dreams comes to me from thingsexternal to me, I saw no reason why I shouldhold this belief about those things I seem to besensing while awake. The second was that, sinceI was still ignorant of the author of my origin (orat least pretended to be ignorant of it), I saw

33

nothing to prevent my having been so consti-tuted by nature that I should be mistaken evenabout what seemed to me most true. As tothe arguments that used to convince me of thetruth of sensible things, I found no difficultyresponding to them. For since I seemed drivenby nature toward many things about which rea-son tried to dissuade me, I did not think thatwhat I was taught by nature deserved much cre-dence. And even though the perceptions of thesenses did not depend on my will, I did not thinkthat we must therefore conclude that they camefrom things distinct from me, since perhapsthere is some faculty in me, as yet unknown tome, that produces these perceptions.

But now, having begun to have a betterknowledge of myself and the author of my ori-gin, I am of the opinion that I must not rashlyadmit everything that I seem to derive from thesenses; but neither, for that matter, should I calleverything into doubt.

First, I know that all the things that I clearly anddistinctly understand can be made by God such asI understand them. For this reason, my abilityclearly and distinctly to understand one thingwithout another suffices to make me certain thatthe one thing is different from the other, sincethey can be separated from each other, at least byGod. The question as to the sort of power thatmight effect such a separation is not relevant totheir being thought to be different. For this rea-son, from the fact that I know that I exist, and thatat the same time I judge that obviously nothingelse belongs to my nature or essence except that Iam a thinking thing, I rightly conclude that myessence consists entirely in my being a thinkingthing. And although perhaps (or rather, as I shallsoon say, assuredly) I have a body that is veryclosely joined to me, nevertheless, because on theone hand I have a clear and distinct idea ofmyself, insofar as I am merely a thinking thingand not an extended thing, and because on theother hand I have a distinct idea of a body, inso-far as it is merely an extended thing and not athinking thing, it is certain that I am really distinctfrom my body, and can exist without it.

Moreover, I find in myself faculties for certainspecial modes of thinking, namely the faculties ofimagining and sensing. I can clearly and distinctlyunderstand myself in my entirety without thesefaculties, but not vice versa: I cannot understandthem clearly and distinctly without me, that is,

35

34

33

This paragraph describesDescartes’s previous faith in

his senses. He made many assump-tions about the existence of bod-ies and the accuracy of the beliefsbased on his sense experiences,but he recognizes that he had noreasons or arguments for thecorrectness of these assumptions.

Think about this issue: whatreason do you have for

thinking that the belief you haveright now that there is a physicalbook in front of you (a beliefbased on your visual sense experi-ence) is true?

RR

34

This is an allusion to the evilgenius.RR

35

Here is an argument for theconclusion that a person (his

or her mind) is distinct from his orher body. (See DiscussionQuestion 4.)

RR


without a substance endowed with understandingin which they inhere, for they include an act ofunderstanding in their formal concept. Thus I per-ceive them to be distinguished from me as modesfrom a thing. . . . Now there clearly is in me a pas-sive faculty of sensing, that is, a faculty for receiv-ing and knowing the ideas of sensible things; butI could not use it unless there also existed, eitherin me or in something else, a certain active facultyof producing or bringing about these ideas. Butthis faculty surely cannot be in me, since it clearlypresupposes no act of understanding, and theseideas are produced without my cooperation andoften even against my will. Therefore the onlyalternative is that it is in some substance differentfrom me, containing either formally or eminentlyall the reality that exists objectively in the ideasproduced by that faculty, as I have just notedabove. Hence this substance is either a body, thatis, a corporeal nature, which contains formally allthat is contained objectively in the ideas, or else itis God, or some other creature more noble than abody, which contains eminently all that is con-tained objectively in the ideas. But since God isnot a deceiver, it is patently obvious that he doesnot send me these ideas either immediately byhimself, or even through the mediation of somecreature that contains the objective reality of theseideas not formally but only eminently. For sinceGod has given me no faculty whatsoever for mak-ing this determination, but instead has given me agreat inclination to believe that these ideas issuefrom corporeal things, I fail to see how God couldbe understood not to be a deceiver, if these ideaswere to issue from a source other than corporealthings. And consequently corporeal things exist.Nevertheless, perhaps not all bodies exist exactlyas I grasp them by sense, since this sensory graspis in many cases very obscure and confused. Butat least they do contain everything I clearly anddistinctly understand—that is, everything, consid-ered in a general sense, that is encompassed in theobject of pure mathematics.

As far as the remaining matters are concerned,which are either merely particular (for example,that the sun is of such and such a size or shape,and so on) or less clearly understood (for exam-ple, light, sound, pain, and the like), even thoughthese matters are very doubtful and uncertain,nevertheless the fact that God is no deceiver(and thus no falsity can be found in my opinions,unless there is also in me a faculty given me by

36

God for the purpose of rectifying this falsity)offers me a definite hope of reaching the trutheven in these matters. And surely there is nodoubt that all that I am taught by nature hassome truth to it; for by “nature,” taken generally,I understand nothing other than God himself orthe ordered network of created things which wasinstituted by God. By my own particular natureI understand nothing other than the combinationof all the things bestowed upon me by God.

There is nothing that this nature teaches memore explicitly than that I have a body that isill-disposed when I feel pain, that needs foodand drink when I suffer hunger or thirst, and thelike. Therefore, I should not doubt that there issome truth in this.

By means of these sensations of pain, hunger,thirst and so on, nature also teaches not merelythat I am present to my body in the way a sailoris present in a ship, but that I am most tightlyjoined and, so to speak, commingled with it, somuch so that I and the body constitute one sin-gle thing. For if this were not the case, then I,who am only a thinking thing, would not sensepain when the body is injured; rather, I wouldperceive the wound by means of the pure intel-lect, just as a sailor perceives by sight whetheranything in his ship is broken. And when thebody is in need of food or drink, I should under-stand this explicitly, instead of having confusedsensations of hunger and thirst. For clearly thesesensations of thirst, hunger, pain, and so on arenothing but certain confused modes of thinkingarising from the union and, as it were, the com-mingling of the mind with the body.

Moreover, I am also taught by nature thatvarious other bodies exist around my body, someof which are to be pursued, while others are to beavoided. And to be sure, from the fact that I sensea wide variety of colors, sounds, odors, tastes,levels of heat, and grades of roughness, and thelike, I rightly conclude that in the bodies fromwhich these different perceptions of the sensesproceed there are differences corresponding tothe different perceptions—though perhaps thelatter do not resemble the former. And from thefact that some of these perceptions are pleasantwhile others are unpleasant, it is plainly certainthat my body, or rather my whole self, insofar asI am comprised of a body and a mind, can beaffected by various beneficial and harmfulbodies in the vicinity.

37

36

Here finally is Descartes’smain reason for thinking that

there are external things: My sen-sory ideas of external things can-not be produced by me, since theyare involuntary and the faculty thatproduces them is not an essentialpart of my nature as a thinkingthing.Therefore they must beproduced by something else, andGod would be a deceiver if he soarranged things that these ideaswere produced by something otherthan things. But God is not adeceiver (since deception wouldbe an imperfection).Therefore,external things must exist—thoughnot necessarily with all the specificfeatures that I experience. (SeeDiscussion Question 5.)

RR

37

Here is Descartes’s accountof the relation between his

mind and his body: they are twodistinct things (substances), butstill intricately connected with eachother.

RR


Granted, there are many other things that Iseem to have been taught by nature; neverthe-less it was not really nature that taught them tome but a certain habit of making reckless judg-ments. And thus it could easily happen that thesejudgments are false: for example, that any spacewhere there is absolutely nothing happening tomove my senses is empty; or that there is some-thing in a hot body that bears an exact likenessto the idea of heat that is in me; or that in a whiteor green body there is the same whiteness orgreenness that I sense; or that in a bitter or sweetbody there is the same taste, and so on; or thatstars and towers and any other distant bodieshave the same size and shape that they present tomy senses, and other things of this sort. . . .it is this nature that teaches me to avoid thingsthat produce a sensation of pain and to pursuethings that produce a sensation of pleasure, andthe like. But it does not appear that natureteaches us to conclude anything, besides thesethings, from these sense perceptions unless theintellect has first conducted its own inquiryregarding things external to us. For it seems tobelong exclusively to the mind, and not to thecomposite of mind and body, to know the truthin these matters. Thus, although a star affects myeye no more than does the flame from a smalltorch, still there is no real or positive tendencyin my eye toward believing that the star is nolarger than the flame. Yet, ever since my youth, Ihave made this judgment without any reason fordoing so. And although I feel heat as I drawcloser to the fire, and I also feel pain upon draw-ing too close to it, there is not a single argumentthat persuades me that there is something in thefire similar to that heat, any more than to thatpain. On the contrary, I am convinced only thatthere is something in the fire that, regardless ofwhat it finally turns out to be, causes in us thosesensations of heat or pain. And although theremay be nothing in a given space that moves thesenses, it does not therefore follow that there isno body in it. But I see that in these and manyother instances I have been in the habit of sub-verting the order of nature. For admittedly I usethe perceptions of the senses (which are prop-erly given by nature only for signifying to themind what things are useful or harmful to thecomposite of which it is a part, and to that extentthey are clear and distinct enough) as reliablerules for immediately discerning what is the

38

essence of bodies located outside us. Yet theysignify nothing about that except quiteobscurely and confusedly.

I have already examined in sufficient detailhow it could happen that my judgments arefalse, despite the goodness of God. But a newdifficulty now arises regarding those verythings that nature shows me are either to besought out or avoided, as well as the internalsensations where I seem to have detected errors,as for example, when someone is deluded by afood’s pleasant taste to eat the poison hiddeninside it. In this case, however, he is driven bynature only toward desiring the thing in whichthe pleasurable taste is found, but not toward thepoison, of which he obviously is unaware. I canonly conclude that this nature is not omniscient.This is not remarkable, since man is a limitedthing, and thus only what is of limited perfectionbefits him.

But we not infrequently err even in thosethings to which nature impels us. Take, forexample, the case of those who are ill and whodesire food or drink that will soon afterwards beinjurious to them. . . . It therefore remains toinquire here how the goodness of God does notprevent “nature,” thus considered, from beingdeceptive.

Now my first observation here is that there isa great difference between a mind and a body inthat a body, by its very nature, is always divisi-ble. On the other hand, the mind is utterly indi-visible. For when I consider the mind, that is,myself insofar as I am only a thinking thing, Icannot distinguish any parts within me; rather, Iunderstand myself to be manifestly one com-plete thing. Although the entire mind seems tobe united to the entire body, nevertheless, werea foot or an arm or any other bodily part to beamputated, I know that nothing has been takenaway from the mind on that account. Nor can thefaculties of willing, sensing, understanding, andso on be called “parts” of the mind, since it isone and the same mind that wills, senses, andunderstands. On the other hand, there is no cor-poreal or extended thing I can think of that Imay not in my thought easily divide into parts;and in this way I understand that it is divisible.This consideration alone would suffice to teachme that the mind is wholly diverse from thebody, had I not yet known it well enough in anyother way. 41

40

39

39

Descartes’s view is that thespecific perceptions of the

senses should not be taken toreveal the actual nature of externalobjects, but only as a guide to whatthings are useful or harmful.Thesensation of heat does notdescribe any property that is actu-ally in the fire, but nonethelesswarns us not to get too close to it.Thus the fact that people who donot think carefully about such mat-ters are sometimes misled doesnot show that God is a deceiver.

RR

41

As a part of his explanationof the fact that our nature is

liable to errors of this last sort,Descartes argues again that hismind is distinct from his body, thistime appealing to the properties ofdivisibility and indivisibility. (SeeDiscussion Question 4.)

RR

38

Here Descartes lists varioussorts of mistaken judgments

that his sensory experiences tempthim to make. Locke and Berkeleywill have more to say about thesematters in the following selections.

RR

40

There is still a problem, how-ever, since our sense experi-

ence sometimes misleads us evenconcerning which things arehelpful or harmful.

RR


My second observation is that my mind is notimmediately affected by all the parts of thebody, but only by the brain, or perhaps even byjust one small part of the brain, namely, by thatpart where the “common” sense is said to reside.Whenever this part of the brain is disposed inthe same manner, it presents the same thing tothe mind, even if the other parts of the body areable meanwhile to be related in diverse ways.Countless experiments show this, none of whichneed be reviewed here.

My next observation is that the nature of thebody is such that whenever any of its parts canbe moved by another part some distance away, itcan also be moved in the same manner by any ofthe parts that lie between them, even if this moredistant part is doing nothing. For example, in thecord ABCD, if the final part D is pulled, the firstpart A would be moved in exactly the same man-ner as it could be, if one of the intermediate partsB or C were pulled, while the end part Dremained immobile. Likewise, when I feel apain in my foot, physics teaches me that this sen-sation took place by means of nerves distributedthroughout the foot, like stretched cords extend-ing from the foot all the way to the brain. Whenthese nerves are pulled in the foot, they also pullon the inner parts of the brain to which theyextend, and produce a certain motion in them.This motion has been constituted by nature so asto affect the mind with a sensation of pain, as ifit occurred in the foot. But because these nervesneed to pass through the shin, thigh, loins, back,and neck to get from the foot to the brain, it canhappen that even if it is not the part in the footbut merely one of the intermediate parts that isbeing struck, the very same movement willoccur in the brain that would occur were the footbadly injured. The inevitable result will be thatthe mind feels the same pain. The same opinionshould hold for any other sensation.

My final observation is that, since any givenmotion occurring in that part of the brainimmediately affecting the mind produces butone sensation in it, I can think of no betterarrangement than that it produces the one sen-sation that, of all the ones it is able to produce,is most especially and most often conducive tothe maintenance of a healthy man. Moreover,experience shows that all the sensationsbestowed on us by nature are like this. Hencethere is absolutely nothing to be found in them

that does not bear witness to God’s power andgoodness. Thus, for example, when the nervesin the foot are agitated in a violent and unusualmanner, this motion of theirs extends throughthe marrow of the spine to the inner reaches ofthe brain, where it gives the mind the sign tosense something, namely, the pain as if it isoccurring in the foot. This provokes the mindto do its utmost to move away from the cause ofthe pain, since it is seen as harmful to the foot.But the nature of man could have been so con-stituted by God that this same motion in thebrain might have indicated something else tothe mind: for example, either the motion itselfas it occurs in the brain, or in the foot, or insome place in between, or something elseentirely different. But nothing else would haveserved so well the maintenance of the body.Similarly, when we need something to drink, acertain dryness arises in the throat that movesthe nerves in the throat, and, by means of them,the inner parts of the brain. And this motionaffects the mind with a sensation of thirst,because in this entire affair nothing is moreuseful for us to know than that we need some-thing to drink in order to maintain our health;the same holds in the other cases.

From these considerations it is utterly appar-ent that, notwithstanding the immense good-ness of God, the nature of man, insofar as it iscomposed of mind and body, cannot helpbeing sometimes mistaken. For if some cause,not in the foot but in some other part throughwhich the nerves extend from the foot to thebrain, or perhaps even in the brain itself, wereto produce the same motion that would nor-mally be produced by a badly injured foot, thepain will be felt as if it were in the foot, and thesenses will naturally be deceived. For since anidentical motion in the brain can only bringabout an identical sensation in the mind, and itis more frequently the case that this motion iswont to arise on account of a cause that harmsthe foot than on account of some other thingexisting elsewhere, it is reasonable that themotion should always show pain to the mind assomething belonging to the foot rather than tosome other part. And if dryness in the throatdoes not arise, as is normal, because takingsomething to drink contributes to bodilyhealth, but from a contrary cause, as happensin the case of someone with dropsy, then it is


far better that it should deceive on that occa-sion than that it should always be deceptivewhen the body is in good health. The sameholds for the other cases.

This consideration is most helpful, not onlyfor my noticing all the errors to which my natureis liable, but also for enabling me to correct oravoid them without difficulty. To be sure, I knowthat all the senses set forth what is true morefrequently than what is false regarding whatconcerns the welfare of the body. Moreover, Ican nearly always make use of several of them inorder to examine the same thing. Furthermore, Ican use my memory, which connects currenthappenings with past ones, and my intellect,which now has examined all the causes of error.Hence I should no longer fear that those thingsthat are daily shown me by the senses are false.On the contrary, the hyperbolic doubts of the lastfew days ought to be rejected as ludicrous. Thisgoes especially for the chief reason for doubt-ing, which dealt with my failure to distinguishbeing asleep from being awake. For I now noticethat there is a considerable difference betweenthese two; dreams are never joined by the mem-ory with all the other actions of life, as is thecase with those actions that occur when one is

42

awake. For surely, if, while I am awake, someonewere suddenly to appear to me and then imme-diately disappear, as occurs in dreams, so that Isee neither where he came from nor where hewent, it is not without reason that I would judgehim to be a ghost or a phantom conjured up inmy brain, rather than a true man. But when thesethings happen, and I notice distinctly where theycome from, where they are now, and when theycome to me, and when I connect my perceptionof them without interruption with the whole restof my life, I am clearly certain that these percep-tions have happened to me not while I wasdreaming but while I was awake. Nor oughtI have even the least doubt regarding the truth ofthese things, if, having mustered all the senses,in addition to my memory and my intellect, inorder to examine them, nothing is passed on tome by one of these sources that conflicts withthe others. For from the fact that God is nodeceiver, it follows that I am in no way mistakenin these matters. But because the need to getthings done does not always permit us theleisure for such a careful inquiry, we must con-fess that the life of man is apt to commit errorsregarding particular things, and we mustacknowledge the infirmity of our nature.

43

Discussion Questions

1. Descartes finally finds something indubitable inMeditation Two: his own existence. One questionabout this is whether Descartes’s insight reallytakes the form of an argument, as the formulation“I think, therefore I am” suggests.On this view, thepremise would be the claim (in the first person) “Ithink,” from which the conclusion “I exist” isinferred.There is little doubt about the cogency ofthis inference, but the obvious question is why thepremise is itself justified. Could the evil geniusdeceive me about the truth of this premise? Whyor why not? Or is it perhaps, as some have sug-gested,a mistake to think of the Cogito as an argu-ment at all, rather than as just one unitary insight?And what exactly would that insight be?

2. In addition to the claim that he cannot be deceivedabout his own existence, Descartes also claimsthat he cannot be deceived about the existenceand content of his specific conscious states ofmind. (See Annotation 10.) This is a crucial step,since very little could be inferred from his exis-tence alone. Is he right that the evil genius couldnot deceive him about such things? Why or whynot? (In thinking about this, it may be helpful toconsider carefully just what form such deception

would take: what would the evil genius do if hewere trying to deceive me into thinking that I hada state of mind—perhaps a certain sensory expe-rience—that I didn’t really have?)

3. In Meditation Three, Descartes attempts to provethe existence of God in order to establish that hecan safely accept claims that seem to him to beclear and distinct—which seems to amount to thesame thing as claims revealed by the “light ofnature.” (See Annotation 19.) But to do this,he hasto appeal to two general metaphysical principles(see Annotations 22 and 23).How does Descartesknow that these principles are true? (Look care-fully at what he says.) Can he safely accept theseprinciples on this basis before he has proved theexistence of God? What does this show aboutDescartes’s main argument in Meditation Three:that God exists and therefore that he can trust hisclear and distinct ideas?

4. Descartes offers two arguments for the conclu-sion that his mind and his body are two distinctentities, one in the passage at Annotation 35 andthe second in the passage at Annotation 41. For-mulate each of these arguments as carefully as youcan and then assess their cogency.

43

And here, finally, is aresponse to the concern

about dreaming.How does he think thatdreams and waking experi-

ence can be distinguished? Is heright about this?

RR

42

The preceding paragraphscontain Descartes’s explana-

tion of why the sorts of mistakesas to what is desirable or harmfulthat result from various sorts ofabnormal conditions do not countagainst the goodness of God orshow that he is a deceiver.

RR


5. How compelling is Descartes’s main argument forthe existence of external things (seeAnnotation 36)?One issue is whether this argument really yields thesort of conclusive reason for such beliefs thatDescartes is seeking. And a second issue is just what

specific conclusions about external things can be jus-tified in this way: do we come to know merely thatthere are things of some sort out there, or can weknow more specific facts about the nature and prop-erties of particular things?

Like Descartes, the English philosopher John Locke (1632–1704) was one ofthe most influential philosophers of all time, making major contributions toepistemology, metaphysics, and political philosophy. Locke was the founderof the philosophical movement that has come to be called British Empiricism(which also includes George Berkeley, David Hume, and John Stuart Mill,among others). The Essay Concerning Human Understanding is his majorepistemological and metaphysical work and (like the Meditations) one of themost widely read of all philosophical books of all time, having attained thestatus of both a classic work and a widely used textbook for students almostas soon as it was published. (For more about Locke’s rather colorful life, seethe introduction to the selection from his Second Treatise of Government, inChapter 6.)

Perhaps the key term of Locke’s epistemology is “idea.” Ideas are saidto be the objects of the understanding when a person thinks, but this mustnot be taken to mean that we think only about ideas and not in someway about various other sorts of things, most of them external to the mind(including material objects). Locke’s view is that ideas are the immediateobjects of consciousness, that which is directly before the mind, butthat these ideas represent other things and thus allow us to think aboutthem. Thus it might be clearer to describe ideas as the vehicles of thinking,the means by which we think about other things. On this view, an idea isessentially the same thing or at least plays the same role as whatmore recent philosophers refer to with the term “concept”: it is somethingthat exists in or is a feature of the mind, and that allows it to thinkabout some specific sort of thing. (But Locke also has at least a tendencyto think of ideas as something like mental pictures or images, and it isdoubtful that images can do the job of accounting for thought in general.See Discussion Question 1.)

With regard to how the mind acquires ideas, Locke is a thoroughgoingproponent of concept empiricism (see the introduction to this chapter).He argues in Book I of the Essay (not reproduced here) that there is noreason to hold that any ideas are innate, and offers in Book II a detailedaccount of the derivation of various specific ideas from experience. (Thebest way to evaluate his general view is to think carefully about thesespecific examples.)

With regard to justificatory empiricism, Locke’s stance is more equivo-cal. Though the main goal of Book IV of the Essay is to give an account of thejustification of knowledge that relies heavily on sense experience, Locke alsoholds (though not in the parts of the book excerpted here) that there are claimsthat are justified a priori rather than empirically and that not all of these aremerely trivial matters of definition—“trifling truths,” as he calls them. Lockethus holds a view rather similar to that of Descartes, holding much of ourknowledge is justified by appeal to experience, but that some substantive,non-trivial claims are justified a priori. (His being standardly labeled an empiri-cist reflects the fact that the issue about the acquisition of ideas or conceptswas formulated both earlier and initially more clearly than the issue about thenature of justification.)

John Locke

JOHN LOCKE: FROM AN ESSAY CONCERNING HUMAN UNDERSTANDING 65

From An Essay Concerning Human Understanding

From An Essay Concerning Human Understanding,1689.

The main focus of Book IV is on the specific issue of the justification ofclaims about the external material world. It is fair to say that Locke strugglesmightily with this question, first giving a definition of knowledge that seems toleave no room for such knowledge and then gradually altering his position ashe proceeds. This part of his discussion culminates in Chapter XI of Book IV,where Locke attempts to argue from certain features of our experience to theprobable truth of the claims about material objects that we hold on the basisof that experience. This is a representative realist view similar to Descartes’sattempt in Meditation Six, but unlike Descartes, Locke does not invoke God asa basis for his solution.

. . . Were it fit to trouble thee with the historyof this Essay, I should tell thee that five or sixfriends, meeting at my chamber and discours-ing on a subject very remote from this, foundthemselves quickly at a stand, by the difficul-ties that rose on every side. After we hadawhile puzzled ourselves, without coming anynearer a resolution of those doubts whichperplexed us, it came into my thoughts that wetook a wrong course; and that before we setourselves upon inquiries of that nature, it wasnecessary to examine our own abilities andsee what objects our understandings were, orwere not, fitted to deal with. . . .

The commonwealth of learning is not at thistime without master-builders, whose mightydesigns, in advancing the sciences, will leavelasting monuments to the admiration of poster-ity; but everyone must not hope to be a Boyle ora Sydenham; and in an age that produces suchmasters as the great Huygenius and the incom-parable Mr. Newton, with some others of thatstrain, it is ambition enough to be employed asan under-labourer in clearing ground a little, andremoving some of the rubbish that lies in theway to knowledge. . . . .

Book II: Of IdeasChapter I: Of Ideas in General, and

Their Original

1. Every man being conscious to himself that hethinks, and that which his mind is applied aboutwhilst thinking being the ideas that are there, itis past doubt that men have in their minds sev-eral ideas such as are those expressed by thewords whiteness, hardness, sweetness, thinking,

1

motion, man, elephant, army, drunkenness andothers: it is in the first place then to be inquired,how he comes by them? . . .

2. Let us then suppose the mind to be, as wesay, white paper void of all characters, withoutany ideas. How comes it to be furnished?Whence comes it by that vast store which thebusy and boundless fancy of man has paintedon it with an almost endless variety? Whencehas it all the materials of reason and knowl-edge? To this I answer, in one word, fromexperience; in that all our knowledge isfounded, and from that it ultimately derivesitself. Our observation, employed either aboutexternal sensible objects, or about the internaloperations of our minds perceived and reflectedon by ourselves, is that which supplies ourunderstandings with all the materials of think-ing. These two are the fountains of knowledge,from whence all the ideas we have, or can nat-urally have, do spring.

3. First, our senses, conversant about particularsensible objects, do convey into the mind severaldistinct perceptions of things, according to thosevarious ways wherein those objects do affectthem. And thus we come by those ideas we haveof yellow, white, heat, cold, soft, hard, bitter,sweet, and all those which we call sensible quali-ties. . . . This great source of most of the ideaswe have . . . , I call SENSATION.

4. Secondly, the other fountain from whichexperience furnisheth the understanding withideas is the perception of the operations of ourown minds within us, as it is employed about theideas it has got; which operations, when the soulcomes to reflect on and consider, do furnish theunderstanding with another set of ideas, whichcould not be had from things without. And suchare perception, thinking, doubting, believing,

2

1

In these two passages, wehave a fairly standard con-

ception of epistemology and itsrole: to investigate the nature ofhuman understanding (humancognitive capacities) itself and toremove various sorts of confusionsand other mistakes about cognitionthat impede the quest for moresubstantive knowledge.

2

Here is Locke’s initial state-ment of concept empiricism,

which he will elaborate in the restof Book II.The mind is initially“white paper,” in Latin a tabula rasa,and it is only through experiencethat it comes to have the capacityto think about any specific thing.

RR


reasoning, knowing, willing, and all the differentactings of our own mind . . . as I call the othersensation, so I call this REFLECTION, the ideasit affords being such only as the mind gets byreflecting on its own operations within itself. . . .

5. The understanding seems to me not tohave the least glimmering of any ideas which itdoth not receive from one of these two.External objects furnish the mind with theideas of sensible qualities, which are all thosedifferent perceptions they produce in us; andthe mind furnishes the understanding withideas of its own operations.

These, when we have taken a full survey ofthem and their several modes, combinations, andrelations, we shall find to contain all our wholestock of ideas, and that we have nothing in ourminds which did not come in one of these twoways. Let anyone examine his own thoughts andthoroughly search into his understanding andthen let him tell me whether all the original ideashe has there are any other than of the objects ofhis senses, or of the operations of his mind, con-sidered as objects of his reflection. . . .

Chapter II: Of Simple Ideas

1. The better to understand the nature, manner,and extent of our knowledge, one thing is care-fully to be observed concerning the ideas wehave, and that is that some of them are simpleand some complex.

. . . The coldness and hardness which a manfeels in a piece of ice being as distinct ideas inthe mind as the smell and whiteness of a lily, oras the taste of sugar, and smell of a rose; andthere is nothing can be plainer to a man than theclear and distinct perception he has of thosesimple ideas; which, being each in itself uncom-pounded, contains in it nothing but one uniformappearance or conception in the mind, and isnot distinguishable into different ideas.

2. These simple ideas, the materials of all ourknowledge, are suggested and furnished to themind only by those two ways above mentioned,viz. sensation and reflection. When the under-standing is once stored with these simple ideas,it has the power to repeat, compare, and unitethem, even to an almost infinite variety, and socan make at pleasure new complex ideas. But itis not in the power of the most exalted wit orenlarged understanding, by any quickness orvariety of thought, to invent or frame one new

4

3

simple idea in the mind, not taken in by the waysbefore mentioned . . . .

3

Try to think of examples ofideas that derive from each

of the two main sources Lockementions, specifying in each casethe experience from which Lockewould claim that the idea isderived.

RR

4

Here is the basic conceptempiricist argument: a chal-

lenge to the opponent to identifyany idea that cannot be accountedfor in the empiricist way.

(You should attempt to dothis, in order to see how

plausible Locke’s claim is.)

RR

Chapter III: Of Ideas of One Sense

1. The better to conceive the ideas we receivefrom sensation, it may not be amiss for us toconsider them in reference to the different wayswhereby they make their approaches to ourminds and make themselves perceivable by us.

First, then, There are some which come intoour minds by one sense only.

Secondly, There are others that convey them-selves into the mind by more senses than one.

Thirdly, Others that are had from reflectiononly.

Fourthly, There are some that make themselvesway and are suggested to the mind by all the waysof sensation and reflection.

We shall consider them apart under theseseveral heads.

First, There are some ideas which haveadmittance only through one sense, which ispeculiarly adapted to receive them. Thus lightand colours, as white, red, yellow, blue, withtheir several degrees or shades and mixtures,as green, scarlet, purple, sea-green, and therest, come in only by the eyes. All kinds ofnoises, sounds, and tones, only by the ears.The several tastes and smells, by the nose andpalate. . . .

Chapter V: Of Simple Ideas of DiversSenses

The ideas we get by more than one sense are ofspace or extension, figure, rest, and motion. Forthese make perceivable impressions, both on theeyes and touch; and we can receive and conveyinto our minds the ideas of the extension, figure,motion, and rest of bodies, both by seeing andfeeling. . . .

Chapter VI: Of Simple Ideas ofReflection

1. The mind, receiving the ideas mentioned inthe foregoing chapters from without, when itturns its view inward upon itself and observesits own actions about those ideas it has, takesfrom thence other ideas, which are as capable tobe the objects of its contemplation as any ofthose it received from foreign things.

2. The two great and principal actions of themind, which are most frequently considered, and


which are so frequent that everyone that pleasesmay take notice of them in himself, are these two:

Perception, or Thinking; andVolition, or Willing.

…

Chapter VII: Of Simple Ideas of BothSensation and Reflection

1. There be other simple ideas which conveythemselves into the mind by all the ways of sen-sation and reflection, viz.:

Pleasure or Delight, and its opposite.Pain, or Uneasiness.Power.Existence.Unity.

…

7. Existence and unity are two other ideas thatare suggested to the understanding by everyobject without, and every idea within. Whenideas are in our minds, we consider them asbeing actually there, as well as we considerthings to be actually without us; which is, thatthey exist or have existence. And whatever wecan consider as one thing, whether a real beingor idea, suggests to the understanding the ideaof unity.

8. Power also is another of those simple ideaswhich we receive from sensation and reflection.For, observing in ourselves that we can at plea-sure move several parts of our bodies whichwere at rest; the effects, also, that natural bodiesare able to produce in one another occurringevery moment to our senses, we both these waysget the idea of power.

…

Chapter VIII: Some FurtherConsiderations Concerning Our Simple

Ideas

…

7. To discover the nature of our ideas the bet-ter, and to discourse of them intelligibly, it willbe convenient to distinguish them as they areideas or perceptions in our minds, and as theyare modifications of matter in the bodies thatcause such perceptions in us: that so we may notthink (as perhaps usually is done) that they are

5

exactly the images and resemblances of some-thing inherent in the subject: most of those ofsensation being in the mind no more the likenessof something existing without us, than thenames that stand for them are the likeness of ourideas, which yet upon hearing they are apt toexcite in us.

8. Whatsoever the mind perceives in itself, oris the immediate object of perception, thought,or understanding, that I call idea; and the powerto produce any idea in our mind, I call qualityof the subject wherein that power is. Thus asnowball having the power to produce in us theideas of white, cold, and round, the powers toproduce those ideas in us as they are in thesnowball I call qualities; and as they are sensa-tions or perceptions in our understandings, Icall them ideas; which ideas, if I speak ofsometimes as in the things themselves, I wouldbe understood to mean those qualities in theobjects which produce them in us.

9. Qualities thus considered in bodies are:First, such as are utterly inseparable from the

body, in what state soever it be; such as in all thealterations and changes it suffers, all the forcecan be used upon it, it constantly keeps; and suchas sense constantly finds in every particle ofmatter which has bulk enough to be perceived;and the mind finds inseparable from every parti-cle of matter, though less than to make itselfsingly be perceived by our senses. V.g., take agrain of wheat, divide it into two parts, each parthas still solidity, extension, figure, and mobility;divide it again, and it retains still the same qual-ities; and so divide it on, till the parts becomeinsensible: they must retain still each of them allthose qualities. For division (which is all that amill or pestle or any other body does uponanother in reducing it to insensible parts) cannever take away either solidity, extension, figure,or mobility from any body, but only makes twoor more distinct separate masses of matter, ofthat which was but one before; all which distinctmasses, reckoned as so many distinct bodies,after division make a certain number. These I calloriginal or primary qualities of body; which Ithink we may observe to produce simple ideas inus, viz. solidity, extension, figure, motion or rest,and number.

10. Secondly, such qualities which in truth arenothing in the objects themselves but powers toproduce various sensations in us by their

6

6

An idea in the mind is

one thing; the correspondingquality in an object is somethingquite different. (At the beginningof section 7, and elsewhere,Locke seems to say that ideashave two kinds of existence: inminds and in objects. But as heclarifies, by speaking of ideas “asthey are modifications of matterin the bodies that cause suchperceptions,” Locke means torefer to whatever in the bodycauses that idea—without assum-ing that it fits the content of theidea very closely.)

5

The idea of power is thegeneral idea of a causal capacity:the capacity of things of varioussorts to causally affect other thingsor be causally affected by them.


primary qualities, i.e. by the bulk, figure,texture, and motion of their insensible parts, ascolours, sounds, tastes, etc. These I callsecondary qualities. To these might beadded a third sort, which are allowed to be barelypowers, though they are as much real qualities inthe subject as those which I, to comply with thecommon way of speaking, call qualities, but fordistinction, secondary qualities. For the powerin fire to produce a new colour, or consistency inwax or clay, by its primary qualities, is as mucha quality in fire as the power it has to produce inme a new idea or sensation of warmth or burn-ing, which I felt not before, by the same primaryqualities, viz. the bulk, texture, and motion of itsinsensible parts.

…

15. . . . the ideas of primary qualities of bod-ies are resemblances of them, and their patternsdo really exist in the bodies themselves; but theideas produced in us by these secondary quali-ties have no resemblance of them at all. There isnothing like our ideas existing in the bodiesthemselves. They are, in the bodies we denomi-nate from them, only a power to produce thosesensations in us; and what is sweet, blue, orwarm in idea is but the certain bulk, figure, andmotion of the insensible parts in the bodiesthemselves, which we call so.

…

21. Ideas being thus distinguished and under-stood, we may be able to give an account howthe same water, at the same time, may producethe idea of cold by one hand and of heat by theother, whereas it is impossible that the samewater, if those ideas were really in it, should atthe same time be both hot and cold. For if weimagine warmth as it is in our hands to benothing but a certain sort and degree of motionin the minute particles of our nerves, or animalspirits, we may understand how it is possiblethat the same water may at the same time pro-duce the sensation of heat in one hand and coldin the other; which yet figure never does, thatnever producing the idea of a square by onehand which has produced the idea of a globe byanother. But if the sensation of heat and cold benothing but the increase or diminution of themotion of the minute parts of our bodies,caused by the corpuscles of any other body, it is

9

8

7

easy to be understood that, if that motion begreater in one hand than in the other, if a bodybe applied to the two hands, which has in itsminute particles a greater motion than in thoseof one of the hands, and a less than in those ofthe other, it will increase the motion of the onehand and lessen it in the other, and so cause thedifferent sensations of heat and cold thatdepend thereon.

…

23. The qualities, then, that are in bodies,rightly considered, are of three sorts:

First, The bulk, figure, number, situation, andmotion or rest of their solid parts. Those are inthem, whether we perceive them or no; andwhen they are of that size that we can discoverthem, we have by these an idea of the thing as itis in itself, as is plain in artificial things. These Icall primary qualities.

Secondly, The power that is in any body, byreason of its insensible primary qualities, tooperate after a peculiar manner on any of oursenses, and thereby produce in us the differentideas of several colours, sounds, smells, tastes,etc. These are usually called sensible qualities.

Thirdly, The power that is in any body, byreason of the particular constitution of itsprimary qualities, to make such a change in thebulk, figure, texture, and motion of anotherbody, as to make it operate on our senses differ-ently from what it did before. Thus the sun has apower to make wax white, and fire to make leadfluid. These are usually called powers.

The first of these, as has been said, I thinkmay be properly called real, original, orprimary qualities, because they are in the thingsthemselves, whether they are perceived or no;and upon their different modifications it is thatthe secondary qualities depend.

The other two are only powers to act differentlyupon other things, which powers result from thedifferent modifications of those primary qualities.

…

Chapter XI: Of Discerning, and OtherOperations of the Mind

…

9. The use of words then being to stand asoutward marks of our internal ideas, and those

10

7

Primary qualities aregenuinely in bodies as the ideas inquestion represent them, with themain argument being that bodiesare inconceivable without suchqualities. (Note here that solidity isthe only such quality that is notessentially spatial or geometrical incharacter.) Secondary qualitiesare powers (causal capacities) inbodies to produce ideas in us; thesepowers result from the “bulk, figure,texture, and motion” of the“insensible parts” of the bodies—that is, from their primary qualities(compare the list of these here withthe original one)—but the qualityrepresented by an idea of asecondary quality does not exist, asthus represented, in the object.

8

A third sort of quality is acausal capacity in the object

that is represented as such. Suchqualities are again a result of theprimary qualities of the insensibleparts of the body.

(Thus what is odd and misleading about secondary

qualities is that while they are, inLocke’s view, really no more thanpowers, they are represented asthough they were qualities on apar with primary qualities.)

RR

9

But a possible source of trou-ble here is the talk of ideas

resembling qualities in objects, some-thing Berkeley will strongly objectto.Think here of your idea of acertain shape (say, squareness) andan object whose surface has thequality in question. Is resemblancethe right way to think about therelation between these two?


ideas being taken from particular things, ifevery particular idea that we take in shouldhave a distinct name, names must be endless.To prevent this, the mind makes the particularideas received from particular objects tobecome general; which is done by consideringthem as they are in the mind such appear-ances, separate from all other existences andthe circumstances of real existence, as time,place, or any other concomitant ideas. This iscalled ABSTRACTION, whereby ideas takenfrom particular beings become generalrepresentatives of all of the same kind; andtheir names, general names, applicable towhatever exists conformable to such abstractideas. Such precise, naked appearances in themind, without considering how, whence, orwith what others they came there, the under-standing lays up (with names commonlyannexed to them) as the standards to rank realexistences into sorts, as they agree with thesepatterns, and to denominate them accordingly.Thus the same colour being observed to-dayin chalk or snow, which the mind yesterdayreceived from milk, it considers that appear-ance alone, makes it a representative of all ofthat kind; and having given it the namewhiteness, it by that sound signifies the samequality wheresoever to be imagined or metwith; and thus universals, whether ideas orterms, are made.

. . . the ideas first in the mind, it is evident,are those of particular things; from whence, byslow degrees, the understanding proceeds tosome few general ones; which, being takenfrom the ordinary and familiar objects ofsense, are settled in the minds with generalnames to them. Thus particular ideas are firstreceived and distinguished, and so knowledgegot about them; and next to them, the less gen-eral or specific, which are next to particular.For abstract ideas are not so obvious or easy,to children or the yet unexercised mind, as par-ticular ones. If they seem so to grown men, itis only because by constant and familiar usethey are made so. For when we nicely reflectupon them, we shall find that general ideas arefictions and contrivances of the mind thatcarry difficulty with them, and do not so eas-ily offer themselves as we are apt to imagine.For example, does it not require some painsand skill to form the general idea of a triangle

11

(which is yet none of the most abstract, com-prehensive, and difficult), for it must be nei-ther oblique nor rectangle, neither equilateral,equicrural, nor scalenon; but all and none ofthese at once. In effect, it is something imper-fect that cannot exist, an idea wherein someparts of several different and inconsistentideas are put together. It is true, the mind inthis imperfect state has need of such ideas, andmakes all the haste to them it can, for the con-veniency of communication and enlargementof knowledge, to both which it is naturallyvery much inclined. But yet one has reason tosuspect such ideas are marks of our imperfec-tion: at least, this is enough to show that themost abstract and general ideas are not thosethat the mind is first and most easilyacquainted with, nor such as its earliest knowl-edge is conversant about.

…

Chapter XII: Of Complex Ideas

…

6. . . . the ideas of substances are suchcombinations of simple ideas as are taken torepresent distinct particular things subsisting bythemselves, in which the supposed or confusedidea of substance, such as it is, is always the firstand chief. Thus if to substance be joined thesimple idea of a certain dull whitish colour, withcertain degrees of weight, hardness, ductility,and fusibility, we have the idea of lead; and acombination of the ideas of a certain sort offigure, with the powers of motion, thought, andreasoning, joined to substance, make theordinary idea of a man. . . .

Book IV: Of Knowledge andOpinion

Chapter I: Of Knowledge in General

1. Since the mind, in all its thoughts and reason-ings, hath no other immediate object but its ownideas, which it alone does or can contemplate, itis evident that our knowledge is only conversantabout them.

2. Knowledge then seems to me to be noth-ing but the perception of the connexion andagreement, or disagreement and repugnancy,of any of our ideas. In this alone it consists.

13

12

10

This example, also used byBerkeley, leads to the

argument from illusion.To seethe point, imagine that one of yourhands has previously been in icewater and the other in water as hotas you can stand,while the “samewater” in which they are now placedis lukewarm. If the experiences inquestion were direct apprehensionsof actual qualities of the water, thediscrepancy between them would beinexplicable.This is taken to showthat the directly experienced quali-ties are only ideas in the mind.

11

It is the alleged capacity ofabstraction that allows the

mind to think in general orabstract terms. Locke’s accountworks reasonably well for theexample discussed here. In whatfollows, we insert a later passage(from Book IV, Chapter VII,section 9) in which Locke discussesa different, more difficult, andunfortunately probably moretypical example: the abstract ideaof a triangle.

12

This is one of the placeswhere Locke seems to be

thinking of ideas as images or mentalpictures.

What would an abstractimage of a triangle (one that

represents not one particular sortof triangle but rather triangles in general) be like?

13

One particularly importantclass of complex ideas is ideas ofsubstances: concrete kinds of stuffor objects, such as lead or a man.By “the supposed or confused ideaof substance,” Locke means theunderlying idea of a substratumin which all the various qualitiespertaining to a certain kind ofthing exist, but which has noqualities in itself.This is apuzzling notion, which Berkeleywas quick to attack.


Where this perception is, there is knowl-edge; and where it is not, there, though we mayfancy, guess, or believe, yet we always comeshort of knowledge. . . .

3. But to understand a little more distinctlywherein this agreement or disagreement consists,I think we may reduce it all to these four sorts:

1. Identity, or diversity.2. Relation.3. Co-existence, or necessary connexion.4. Real existence.

4. First, As to the first sort of agreement ordisagreement, viz. identity or diversity. It is thefirst act of the mind, when it has any sentimentsor ideas at all, to perceive its ideas, and so far asit perceives them, to know each what it is, andthereby also to perceive their difference and thatone is not another. . . .

5. Secondly, The next sort of agreement ordisagreement the mind perceives in any of itsideas may, I think, be called relative, and isnothing but the perception of the relationbetween any two ideas, of what kind soever,whether substances, modes, or any other. . . .

6. Thirdly, The third sort of agreement ordisagreement to be found in our ideas, whichthe perception of the mind is employed about,is co-existence or non-co-existence in thesame subject; and this belongs particularly tosubstances . . . .

7. Fourthly, The fourth and last sort is that ofactual real existence agreeing to any idea.Within these four sorts of agreement or disagree-ment is, I suppose, contained all the knowledgewe have or are capable of. . . .

Chapter II: Of the Degrees of OurKnowledge

1. . . . if we will reflect on our own ways ofthinking, we shall find that sometimes the mindperceives the agreement or disagreement of twoideas immediately by themselves, without theintervention of any other; and this I think wemay call intuitive knowledge. . . . Thus the mindperceives that white is not black, that a circle isnot a triangle, that three are more than two andequal to one and two. Such kind of truths themind perceives at the first sight of the ideastogether, by bare intuition, without the inter-vention of any other idea; and this kind of

15

14knowledge is the clearest and most certain thathuman frailty is capable of. . . .

2. The next degree of knowledge is where themind perceives the agreement or disagreementof any ideas, but not immediately. . . . In thiscase then, when the mind cannot so bring itsideas together as by their immediate comparisonand as it were juxtaposition or applicationone to another, to perceive their agreement ordisagreement, it is fain, by the intervention ofother ideas (one or more, as it happens) to dis-cover the agreement or disagreement which itsearches; and this is that which we callreasoning. Thus the mind, being willing to knowthe agreement or disagreement in bignessbetween the three angles of a triangle and tworight ones, cannot by an immediate view andcomparing them do it, because the three anglesof a triangle cannot be brought at once and becompared with any one or two angles; and so ofthis the mind has no immediate, no intuitiveknowledge. In this case the mind is fain to findout some other angles to which the three anglesof a triangle have an equality, and finding thoseequal to two right ones, comes to know theirequality to two right ones.

3. Those intervening ideas which serve toshow the agreement of any two others arecalled proofs; and where the agreement ordisagreement is by this means plainly andclearly perceived, it is called demonstration: itbeing shown to the understanding, and the mindmade see that it is so. . . .

14. These two, viz. intuition and demonstration,are the degrees of our knowledge; whatever comesshort of one of these, with what assurance soeverembraced, is but faith or opinion, but not knowl-edge, at least in all general truths. There is, indeed,another perception of the mind, employed aboutthe particular existence of finite beings withoutus, which, going beyond bare probability and yetnot reaching perfectly to either of the foregoingdegrees of certainty, passes under the name ofknowledge. There can be nothing more certainthan that the idea we receive from an externalobject is in our minds: this is intuitive knowledge.But whether there be anything more than barelythat idea in our minds, whether we can thence cer-tainly infer the existence of anything without uswhich corresponds to that idea is that whereofsome men think there may be a question made:because men may have such ideas in their minds,

17

1614

Here is Locke’s initial defini-tion of knowledge, which

follows very naturally from theaccount of mental operations inthe previous paragraph. If all that isimmediately before the mind isideas, then what can knowledge bebut the perception of the ways inwhich they agree or disagree?

15

The immediate problem,however, is how this allows

for knowledge of things outsidethe mind, such as material objectsof various sorts. Locke’s firstattempt to solve this problem is toconstrue such knowledge as per-ception of a special sort of agree-ment of ideas, one of which is theidea of “actual real existence.”

But what is it for this idea toagree or disagree with oth-

ers, and why couldn’t it do so evenif the thing in question didn’t reallyexist?

RR

16

Intuitive knowledge is thebasis, in Locke’s view, for knowledgeof simple, self-evident truths likethe ones enumerated. Laterphilosophers will describe these asa priori.

17

A second kind of knowl-edge is demonstrative knowledge:knowledge that results from moreextended arguments or proofs, inwhich the individual steps areknown intuitively.


when no such thing exists, no such object affectstheir senses. But yet here I think we are providedwith an evidence that puts us past doubting: for Iask anyone whether he be not invincibly consciousto himself of a different perception, when he lookson the sun by day and thinks on it by night, whenhe actually . . . smells a rose or only thinks on that. . . odour? We as plainly find the difference thereis between any idea revived in our minds by ourown memory and actually coming into our mindsby our senses, as we do between any two distinctideas. If anyone say a dream may do the samething, and all these ideas may be produced in uswithout any external objects, he may please todream that I make him this answer: (1)That it is nogreat matter whether I remove his scruple or no:where all is but dream, reasoning and argumentsare of no use, truth and knowledge nothing.(2) That I believe he will allow a very manifestdifference between dreaming of being in the fireand being actually in it. But yet if he be resolvedto appear so sceptical as to maintain that what Icall being actually in the fire is nothing but adream, and that we cannot thereby certainly knowthat any such thing as fire actually exists withoutus, I answer: that we certainly finding that plea-sure or pain follows upon the application ofcertain objects to us whose existence we perceiveor dream that we perceive by our senses, this cer-tainty is as great as our happiness or misery,beyond which we have no concernment to know orto be. So that, I think, we may add to the two for-mer sorts of knowledge this also of the existenceof particular external objects, by that perceptionand consciousness we have of the actual entranceof ideas from them, and allow these three degreesof knowledge, viz. intuitive, demonstrative, andsensitive, in each of which there are differentdegrees and ways of evidence and certainty.

…

Chapter IV: Of the Reality ofKnowledge

1. I doubt not but my reader, by this time, may beapt to think that I have been all this while onlybuilding a castle in the air and be ready to say tome: To what purpose all this stir? Knowledge,say you, is only the perception of the agreementor disagreement of our own ideas; but whoknows what those ideas may be? Is thereanything so extravagant as the imaginations of

18

men’s brains? Where is the head that has nochimeras in it? Or if there be a sober and a wiseman, what difference will there be by your rulesbetween his knowledge and that of the mostextravagant fancy in the world? . . . If it be truethat all knowledge lies only in the perception ofthe agreement or disagreement of our own ideas,the visions of an enthusiast and the reasonings ofa sober man will be equally certain. . . .

3. It is evident the mind knows not thingsimmediately, but only by the intervention of theideas it has of them. Our knowledge, therefore,is real only so far as there is a conformitybetween our ideas and the reality of things. Butwhat shall be here the criterion? How shall themind, when it perceives nothing but its ownideas, know that they agree with thingsthemselves? . . .

Chapter XI: Of Our Knowledge of theExistence of Other Things

1. The knowledge of our own being we have byintuition. The existence of a GOD, reasonclearly makes known to us. . . .

The knowledge of the existence of any otherthing we can have only by sensation: for, therebeing no necessary connexion of real existencewith any idea a man hath in his memory, . . . noparticular man can know the existence of anyother being but only when, by actual operatingupon him, it makes itself perceived by him. Forthe having the idea of anything in our mind nomore proves the existence of that thing, than thepicture of a man evidences his being in theworld, or the visions of a dream make thereby atrue history.

2. It is therefore the actual receiving of ideasfrom without that gives us notice of the existenceof other things and makes us know that some-thing doth exist at that time without us whichcauses that idea in us, though perhaps we neitherknow nor consider how it does it. . . .

3. The notice we have by our senses of theexisting of things without us, though it be notaltogether so certain as our intuitive knowledgeor the deductions of our reason employed aboutthe clear abstract ideas of our own minds, yet itis an assurance that deserves the name of knowl-edge. If we persuade ourselves that our facultiesact and inform us right concerning the existenceof those objects that affect them, it cannot passfor an ill-grounded confidence: for I think

20

19

18

Here is Locke’s second, andrather different, attempt to

account for knowledge of externalobjects.Rather than an appeal toagreement of an idea of the thing inquestion with the idea of “actual realexistence,” Locke now seems to beappealing to the distinctive charac-ter of the ideas involved in senseperception, as contrasted with otherideas of the same object that areinvolved in memory or thought.

But it is far from clear justwhat this difference amounts

to or why it should be taken toindicate actual existence outside themind.Locke’s nervousness about thisissue is shown by the way heattempts to ridicule the skeptic—andalso by his saying that cognition ofthe sort in question“passes underthe name of knowledge,” whichsurely raises the issue of whether itreally is knowledge.

20

Here is another take on theissue of how external

existence is known: not byagreement with the idea of “actualreal existence” and not by thecharacter of the ideas, taken inthemselves, but by “the actualreceiving” of them “from without.”

But how, we may ask, is thisknown? What in our ideas

reveals whether or not theyhave been thus “received fromwithout”—that is, caused byexternal things?

RR

19

Here the problem is realizedand stated more clearly: how

can we know, via perception of ourideas, that they accurately depictsomething existing outside ourminds in the world?

RR


nobody can, in earnest, be so sceptical as to beuncertain of the existence of those things whichhe sees and feels. At least, he that can doubt sofar (whatever he may have with his ownthoughts) will never have any controversy withme, since he can never be sure I say anythingcontrary to his opinion. As to myself, I thinkGOD has given me assurance enough of theexistence of things without me, since, by theirdifferent application, I can produce in myselfboth pleasure and pain, which is one greatconcernment of my present state. This is certain:the confidence that our faculties do not hereindeceive us is the greatest assurance we are capa-ble of concerning the existence of materialbeings. . . . But besides the assurance we havefrom our senses themselves, that they do not errin the information they give us of the existenceof things without us when they are affected bythem, we are further confirmed in this assuranceby other concurrent reasons.

4. First, It is plain those perceptions areproduced in us by exterior causes affecting oursenses, because those that want the organs ofany sense never can have the ideas belonging tothat sense produced in their minds. This is tooevident to be doubted, and therefore we cannotbut be assured that they come in by the organsof that sense, and no other way. The organsthemselves, it is plain, do not produce them: forthen the eyes of a man in the dark wouldproduce colours, and his nose smell roses in thewinter; but we see nobody gets the relish of apineapple till he goes to the Indies, where it is,and tastes it.

5. Secondly, Because sometimes I find that Icannot avoid the having those ideas produced inmy mind. For though, when my eyes are shut, orwindows fast, I can at pleasure recall to mymind the ideas of light, or the sun, which formersensations had lodged in my memory: so I can atpleasure lay by that idea, and take into my viewthat of the smell of a rose, or taste of sugar. But,if I turn my eyes at noon towards the sun, Icannot avoid the ideas which the light or sunthen produces in me. So that there is a manifestdifference between the ideas laid up in mymemory (over which, if they were there only, Ishould have constantly the same power todispose of them and lay them by at pleasure) andthose which force themselves upon me and Icannot avoid having. And therefore it must

21

needs be some exterior cause and the briskacting of some objects without me, whoseefficacy I cannot resist, that produces thoseideas in my mind, whether I will or no. Besides,there is nobody who doth not perceive thedifference in himself between contemplating thesun as he hath the idea of it in his memory, andactually looking upon it: of which two, hisperception is so distinct that few of his ideas aremore distinguishable one from another, andtherefore he hath certain knowledge that theyare not both memory or the actions of his mindand fancies only within him, but that actualseeing hath a cause without.

6. Thirdly, Add to this, that many of thoseideas are produced in us with pain, which after-wards we remember without the least offence.Thus, the pain of heat or cold, when the idea ofit is revived in our minds, gives us no distur-bance, which when felt was very troublesome,and is again when actually repeated: which isoccasioned by the disorder the external objectcauses in our bodies when applied to it; and weremember the pain of hunger, thirst, or theheadache without any pain at all: which wouldeither never disturb us, or else constantly do it asoften as we thought of it, were there nothingmore but ideas floating in our minds andappearances entertaining our fancies, withoutthe real existence of things affecting us fromabroad. The same may be said of pleasureaccompanying several actual sensations. . . .

7. Fourthly, Our senses in many cases bearwitness to the truth of each other’s reportconcerning the existence of sensible thingswithout us. He that sees a fire may, if he doubtwhether it be anything more than a bare fancy,feel it too and be convinced by putting his handin it. Which certainly could never be put intosuch exquisite pain by a bare idea or phantom,unless that the pain be a fancy too: which yet hecannot, when the burn is well, by raising theidea of it, bring upon himself again.

…

8. But yet, if after all this anyone will be sosceptical as to distrust his senses and to affirmthat all we see and hear, feel and taste, think anddo during our whole being is but the series anddeluding appearances of a long dream, whereofthere is no reality, and therefore will questionthe existence of all things or our knowledge of

22

21

Again Locke shows hisuncertainty about the

question (the assurance in question“deserves the name of knowledge”)and attempts to refute the skepticthrough mockery.

Think about the reasons fortrusting our senses that are

offered in this paragraph. Do any ofthem have any serious weight?

22

These four “concurrent”reasons for thinking that our

senses “do not err” are the closestthat Locke comes to an argumentfor the truth of perceptual claimsabout the external world.Thinkabout them carefully.They are ofvery unequal weight, and each ofthem can be considered alone orin combination with some or allof the others. Is there a goodargument, or at least the beginningsof one, for the thesis that some ofour ideas accurately representexternal objects that really exist?(See Discussion Question 5.)


anything: I must desire him to consider that, ifall be a dream, then he doth but dream that hemakes the question, and so it is not much matterthat a waking man should answer him. But yet ifhe pleases he may dream that I make him thisanswer, that the certainty of things existing inrerum natura, when we have the testimony ofour senses for it, is not only as great as ourframe can attain to, but as our condition needs.For our faculties being suited not to the fullextent of being, nor to a perfect, clear, compre-hensive knowledge of things free from all doubtand scruple, but to the preservation of us inwhom they are, and accommodated to the use oflife: they serve to our purpose well enough ifthey will but give us certain notice of thosethings which are convenient or inconvenient tous. . . . And if our dreamer pleases to try whetherthe glowing heat of a glass furnace be barely awandering imagination in a drowsy man’s fancy,by putting his hand into it, he may perhaps bewakened into a certainty greater than he couldwish that it is something more than bare imagi-nation. So that this evidence is as great aswe can desire, being as certain to us as ourpleasure or pain, i.e. happiness or misery,beyond which we have no concernment, eitherof knowing or being. Such an assurance of theexistence of things without us is sufficient todirect us in the attaining the good and avoidingthe evil which is caused by them, which is theimportant concernment we have of being madeacquainted with them.

9. In fine then, when our senses do actuallyconvey into our understandings any idea, wecannot but be satisfied that there doth somethingat that time really exist without us which dothaffect our senses, and by them give notice of itselfto our apprehensive faculties, and actuallyproduce that idea which we then perceive; andwe cannot so far distrust their testimony as to

23

doubt that such collections of simple ideas, as wehave observed by our senses to be united together,do really exist together. But this knowledgeextends as far as the present testimony of oursenses, employed about particular objects that dothen affect them, and no further. For if I saw sucha collection of simple ideas as is wont to be calledman, existing together one minute since, and amnow alone, I cannot be certain that the same manexists now, since there is no necessary connexionof his existence a minute since with his existencenow: by a thousand ways he may cease to be sinceI had the testimony of my senses for his existence.And if I cannot be certain that the man I saw lasttoday is now in being, I can less be certain that heis so who hath been longer removed from mysenses and I have not seen since yesterday orsince the last year; and much less can I be certainof the existence of men that I never saw. And,therefore, though it be highly probable thatmillions of men do now exist, yet, whilst I amalone, writing this, I have not that certainty of itwhich we strictly call knowledge: thoughthe great likelihood of it puts me past doubt, andit be reasonable for me to do several things uponthe confidence that there are men (and men alsoof my acquaintance, with whom I have to do)now in the world; but this is but probability, notknowledge.

10. Whereby yet we may observe how foolishand vain a thing it is for a man of a narrow knowl-edge who, having reason given him to judge ofthe different evidence and probability of thingsand to be swayed accordingly, how vain, I say, itis to expect demonstration and certainty inthings not capable of it, and refuse assent to veryrational propositions and act contrary to veryplain and clear truths because they cannot bemade out so evident as to surmount everythe least (I will not say reason, but) pretence ofdoubting. . . .

23

Again he resorts to mockeryof the skeptic.RR


1. Locke often seems to be thinking of ideas as men-tal images or pictures.This would in effect meanthat to think of things is to have a sequence ofsuch images pass through our minds. Is this an ade-quate view of the general nature of thought? Onespecific problem is whether images as such, with-out anything accompanying them in the way that acaption accompanies a picture in the newspaper,really make claims that could be true or false.

2. How plausible is Locke’s account in Book II of howwe acquire our various concepts? Think of a rangeof examples, ask what Locke would say aboutthem, and then evaluate whether his view seemscorrect. Remember that the issue is not whetheryou can find an experience to which the ideaapplies in some way or that provides an instanceof the idea, but rather whether all the content ofthat idea can be found in explicit form in that


George Berkeley (1685–1753) was an Irish philosopher who was also an Angli-can bishop. Berkeley is another of the British Empiricists, and was greatly influ-enced by Locke. He is, however, most famous for his criticisms of Locke and hisradically different view of both the objects and nature of human knowledge, espe-cially perceptual knowledge. While both Locke and Berkeley believe that thedirect or immediate objects of our experience, including perceptual experience,are ideas in our minds, Locke nonetheless believes in a commonsensical worldof mind-independent material objects and thinks that we can have knowledge ofsuch objects through inference from our sense experience, thus holding a realistview of the material world. Berkeley, in contrast, thinks that no such inference canbe justified, and for this and other reasons holds the idealist view that ordinaryobjects (trees, tables, buildings, and so on) are nothing more than collections ofideas that are caused by God to occur in our minds. (Paradoxical though this mayseem, Berkeley thought that in advocating this view, he was defending commonsense against the threat of skepticism.)

Berkeley first defended idealism in his Principles of Human Knowledge(1710). When this book was poorly received, he attempted to restate its argu-ments in a more popularly accessible form in the Three Dialogues BetweenHylas and Philonous, including also an extended defense of the thesis that thedirect or immediate object of perceptual experience is always an idea in themind—something that is not defended very explicitly in either Locke’s Essayor Berkeley’s own Principles. The two characters in the dialogue are Hylas (thename comes from the Greek word for matter) and Philonous (again from theGreek, meaning lover of mind), with the latter being the spokesman for Berke-ley’s own views. Hylas begins by defending a commonsensical view, getsdriven into a roughly Lockean view, and eventually is forced to accept ideal-ism. (The Three Dialogues are usually regarded as one of the most successfulphilosophical uses of dialogue form, ranking with Plato’s dialogues andHume’s Dialogues Concerning Natural Religion—see Chapter 7.)

To help you discern the structure of the argument, we have insertedRoman numerals in brackets to mark the main divisions.

Essay—not included in the selections here), thiswould require that there are correspondingabstract ideas that provide the meanings of suchterms. But do such ideas actually exist? Think firstof the construal of them as images, consideringvarious cases.Does this work? If not, can you thinkof any alternative to the image view that wouldstill account for abstract thought?

5. Reconsider Locke’s four reasons for thinking thatour senses “do not err” in Chapter XI of Book IV.One of these is thoroughly question-begging (thatis, it assumes the very sort of knowledge thatLocke is trying to account for). Another, whileperhaps correct, doesn’t seem to point to thedesired conclusion in any very clear way.The othertwo, while more suggestive, don’t seem to workwhen taken individually, but may do better whentaken together.With these hints, try to sort outthese reasons and figure out which two of them,when taken together, do yield an argument forsomething like the conclusion Locke is aiming at.How strong is this argument? Can you think of anyanswer to it?

experience,whether the experience in itself is richenough to supply all of that content. (Here aresome cases to think about: the idea of the taste oflemon, the idea of a dog (not some particular dogor breed of dog, just dog in general), the idea ofreflection (to which Locke appeals), the idea oftrump (as it figures in various card games), the ideaof a politician, and the idea of an electron.)

3. The distinction between primary and secondaryqualities (in Book II,Chapter VIII) is one that Lockeinherits from earlier thinkers, including Descartes,and he thinks it is obvious enough not to needargument.Does Locke’s thesis here seem plausibleto you: for example,do shapes really exist in mate-rial objects, while colors as we experience themexist only in our minds, with only the power toproduce such an experience being in the object?Can you see any way to argue for this thesis?(Hint: think of what would be required to explainour having the experiences of these two qualities.)

4. Do we have abstract ideas? There are certainlyabstract terms in our language, and on Locke’s the-ory of meaning (expounded in Book III of the

George Berkeley

GEORGE BERKELEY: FROM THREE DIALOGUES BETWEEN HYLAS AND PHILONOUS 75

From Three Dialogues Between Hylas and Philonous

From Three Dialogues Between Hylas and Philo-nous, 1713.

The First Dialogue[I]

Philonous: Good morrow, Hylas: I did notexpect to find you abroad so early.

Hylas: It is indeed something unusual; but mythoughts were so taken up with a subject Iwas discoursing of last night, that finding Icould not sleep, I resolved to rise and take aturn in the garden.

Phil.: It happened well, to let you see whatinnocent and agreeable pleasures you loseevery morning. Can there be a pleasantertime of the day, or a more delightful sea-son of the year? That purple sky, thesewild but sweet notes of birds, the fragrantbloom upon the trees and flowers, thegentle influence of the rising sun, theseand a thousand nameless beauties ofnature inspire the soul with secret trans-ports; its faculties too being at this timefresh and lively, are fit for those medita-tions, which the solitude of a garden andtranquility of the morning naturally dis-pose us to. But I am afraid I interruptyour thoughts: for you seemed very intenton something.

…

Hyl.: I was considering the odd fate of those menwho have in all ages, through an affectationof being distinguished from the vulgar, orsome unaccountable turn of thought, pre-tended either to believe nothing at all, or tobelieve the most extravagant things in theworld. This however might be borne, if theirparadoxes and scepticism did not draw afterthem some consequences of general disad-vantage to mankind. But the mischief lieshere; that when men of less leisure see themwho are supposed to have spent their wholetime in the pursuits of knowledge, profess-ing an entire ignorance of all things, oradvancing such notions as are repugnant toplain and commonly received principles,they will be tempted to entertain suspicionsconcerning the most important truths,

1

which they had hitherto held sacred andunquestionable.

Phil.: I entirely agree with you, as to the illtendency of the affected doubts of somePhilosophers, and fantastical conceits ofothers . . . .

Hyl.: I am glad to find there was nothing in theaccounts I heard of you.

Phil.: Pray, what were those?Hyl.: You were represented in last night’s

conversation, as one who maintained themost extravagant opinion that ever enteredinto the mind of man, to wit, that there is nosuch thing as material substance in theworld.

Phil.: That there is no such thing as whatPhilosophers call material substance, I amseriously persuaded: but if I were made tosee anything absurd or sceptical in this, Ishould then have the same reason torenounce this, that I imagine I have now toreject the contrary opinion.

Hyl.: What! Can anything be more fantastical,more repugnant to common sense, or amore manifest piece of scepticism, than tobelieve there is no such thing as matter?

Phil.: Softly, good Hylas. What if it should prove,that you, who hold there is, are by virtue ofthat opinion a greater sceptic, and maintainmore paradoxes and repugnancies to com-mon sense, than I who believe no such thing?. . . Pray, Hylas, what do you mean by asceptic?

Hyl.: I mean what all men mean, one that doubtsof everything.

Phil.: He then who entertains no doubt concern-ing some particular point, with regard tothat point cannot be thought a sceptic.

Hyl.: I agree with you.Phil.: . . . does doubting consist in embracing the

affirmative or negative side of a question?Hyl.: In neither; for whoever understands

English, cannot but know that doubtingsignifies a suspense between both.

…

Phil.: How comes it then, Hylas, that youpronounce me a sceptic, because I deny whatyou affirm, to wit, the existence of matter?

2 2

The doctrine of materialsubstance is Berkeley’s inter-

pretation of Locke’s view ofmaterial objects, according towhich an object consists of asubstance (an independentlyexisting thing) that has variousqualities (for Locke, only primaryqualities).Thus material substanceis the sort of substance that existsoutside the mind.

1

This opening speech isintended to make clear at

the outset that Philonous is notdenying the existence of ordinaryobjects, such as those found innature. Instead, as we will see fur-ther, he is offering an interpretationof what that existence amounts to,one that Berkeley at least believesto be in accord with commonsense.


Since, for aught you can tell, I am as peremp-tory in my denial, as you in your affirmation.

Hyl.: Hold, Philonous, I have been a littleout in my definition; but every false step aman makes in discourse is not to beinsisted on. I said indeed, that a sceptic wasone who doubted of everything; but Ishould have added, or who denies the real-ity and truth of things.. . . What think you of distrusting the senses,of denying the real existence of sensiblethings, or pretending to know nothing ofthem. Is not this sufficient to denominate aman a sceptic?

Phil.: Shall we therefore examine which of us itis that denies the reality of sensible things,or professes the greatest ignorance of themsince, if I take you rightly, he is to beesteemed the greatest sceptic?

Hyl.: That is what I desire.

[II]

Phil.: What mean you by sensible things?Hyl.: Those things which are perceived by the

senses. Can you imagine that I mean any-thing else?

Phil.: Pardon me, Hylas, if I am desirous clearlyto apprehend your notions, since this maymuch shorten our inquiry. Suffer me then toask you this farther question. Are thosethings only perceived by the senses whichare perceived immediately? Or, may thosethings properly be said to be sensible,which are perceived mediately, or not with-out the intervention of others?

Hyl.: I do not sufficiently understand you.Phil.: In reading a book, what I immediately

perceive are the letters, but mediately, or bymeans of these, are suggested to my mindthe notions of God, virtue, truth, &c. Now,that the letters are truly sensible things, orperceived by sense, there is no doubt: but Iwould know whether you take the thingssuggested by them to be so too.

Hyl.: No certainly, it were absurd to think Godor virtue sensible things, though they maybe signified and suggested to the mind bysensible marks, with which they have anarbitrary connection.

Phil.: It seems then, that by sensible things youmean those only which can be perceivedimmediately by sense.

…

Hyl.: . . . I tell you once for all, that by sensiblethings I mean those only which areperceived by sense, and that in truth thesenses perceive nothing which they do notperceive immediately: for they make noinferences. The deducing therefore ofcauses or occasions from effects andappearances, which alone are perceived bysense, entirely relates to reason.

Phil.: This point then is agreed between us, thatsensible things are those only which areimmediately perceived by sense. You willfarther inform me, whether we immediatelyperceive by sight anything beside light, andcolors, and figures: or by hearing anythingbut sounds: by the palate, anything besidetastes: by the smell, beside odors: or by thetouch, more than tangible qualities.

Hyl.: We do not.Phil.: It seems therefore, that if you take away

all sensible qualities, there remains noth-ing sensible.

Hyl.: I grant it.Phil.: Sensible things therefore are nothing else

but so many sensible qualities, or combina-tions of sensible qualities.

Hyl.: Nothing else.

[III]

Phil.: Heat then is a sensible thing.Hyl.: Certainly.Phil.: Does the reality of sensible things consist

in being perceived? Or, is it somethingdistinct from their being perceived, and thatbears no relation to the mind?

Hyl.: To exist is one thing, and to be perceivedis another.

Phil.: I speak with regard to sensible thingsonly: and of these I ask, whether by theirreal existence you mean a subsistence exte-rior to the mind, and distinct from theirbeing perceived?

Hyl.: I mean a real absolute being, distinctfrom, and without any relation to, theirbeing perceived.

Phil.: Heat therefore, if it be allowed a realbeing, must exist without the mind.

Hyl.: It must.Phil.: Tell me, Hylas, is this real existence

equally compatible to all degrees of heat,

5

4

3

4

The initial suggestion is that itis sensible qualities, qualities

like color or shape or sound ortaste or odor, that are immediatelyperceived, and that sensible things(things immediately perceived) con-sist only of these qualities.The nextissue is then the nature or status ofsuch qualities, in particular whetherthey are inside or outside of themind, and the argument proceedsby considering separately variousspecific kinds of sensible qualities.

RR

5

Berkeley’s eventual thesis isthat such qualities are only

ideas in the mind and exist only bybeing perceived. Hylas attempts toattribute an independent existenceto them, starting with the firstexample of heat (and cold).

3

Most of the First Dialogue isconcerned with the question

of just what it is that is directly orimmediately perceived, directlybefore the mind, in senseperception—rather than being insome way inferred or perhapsarrived at via some other sort oftransition. Before reading further,ask yourself what the answer is tothis question.


which we perceive: or is there any reasonwhy we should attribute it to some, anddeny it to others? And if there be, pray letme know that reason.

Hyl.: Whatever degree of heat we perceive bysense, we may be sure the same exists in theobject that occasions it.

Phil.: What, the greatest as well as the least?Hyl.: I tell you, the reason is plainly the same in

respect of both: they are both perceived bysense; nay, the greater degree of heat ismore sensibly perceived; and consequently,if there is any difference, we are more cer-tain of its real existence than we can be ofthe reality of a lesser degree.

Phil.: But is not the most vehement and intensedegree of heat a very great pain?

Hyl.: No one can deny it.Phil.: And is any unperceiving thing capable of

pain or pleasure?Hyl.: No, certainly.Phil.: Is your material substance a senseless

being, or a being endowed with sense andperception?

Hyl.: It is senseless, without doubt.Phil.: It cannot therefore be the subject of pain.Hyl.: By no means.Phil.: Nor consequently of the greatest heat per-

ceived by sense, since you acknowledgethis to be no small pain.

Hyl.: I grant it.Phil.: What shall we say then of your external

object; is it a material substance, or no?Hyl.: It is a material substance with the sensible

qualities inhering in it.Phil.: How then can a great heat exist in it, since

you own it cannot in a material substance?I desire you would [clarify] this point.

Hyl.: Hold, Philonous, I fear I was out in yield-ing intense heat to be a pain. It should seemrather, that pain is something distinct fromheat, and the consequence or effect of it.

Phil.: Upon putting your hand near the fire, doyou perceive one simple uniform sensation,or two distinct sensations?

Hyl.: But one simple sensation.Phil.: Is not the heat immediately perceived?Hyl.: It is.Phil.: And the pain?Hyl.: True.Phil.: Seeing therefore they are both immedi-

ately perceived at the same time, and the

6

fire affects you only with one simple, oruncompounded idea, it follows that thissame simple idea is both the intense heatimmediately perceived, and the pain; andconsequently, that the intense heat immedi-ately perceived, is nothing distinct from aparticular sort of pain.

Hyl.: It seems so.Phil.: Again, try in your thoughts, Hylas, if you

can conceive a vehement sensation to bewithout pain, or pleasure.

Hyl.: I cannot.Phil.: Or can you frame to yourself an idea of

sensible pain or pleasure in general,abstracted from every particular idea ofheat, cold, tastes, smells? &c.

Hyl.: —I do not find that I can.Phil.: Does it not therefore follow, that sensible

pain is nothing distinct from those sensa-tions or ideas, in an intense degree?

Hyl.: It is undeniable; and to speak the truth, Ibegin to suspect a very great heat cannotexist but in a mind perceiving it.

Phil.: What! Are you then in that scepticalstate of suspense, between affirming anddenying?

Hyl.: I think I may be positive in the point.A very violent and painful heat cannot existwithout the mind.

Phil.: It has not therefore, according to you, anyreal being.

Hyl.: I own it.Phil.: Is it therefore certain, that there is no

body in nature really hot?Hyl.: I have not denied there is any real heat in

bodies. I only say, there is no such thing asan intense real heat.

Phil.: But did you not say before, that alldegrees of heat were equally real: or if therewas any difference, that the greater weremore undoubtedly real than the lesser?

Hyl.: True: but it was, because I did not then con-sider the ground there is for distinguishingbetween them, which I now plainly see. Andit is this: because intense heat is nothing elsebut a particular kind of painful sensation;and pain cannot exist but in a perceivingbeing; it follows that no intense heat canreally exist in an unperceiving corporealsubstance. But this is no reason why weshould deny heat in an inferior degree toexist in such a substance.

8

7

6

If the quality of heat existsoutside the mind, it must

exist in a material substance.Being a quality or property, itcouldn’t simply exist on its own.This is a rather deep and subtlemetaphysical point, which you willhave to think carefully about.

(“Inhere” is a metaphysicalterm for the relation between qual-ities and the substance they belongto—for example, the qualities ofredness, approximate sphericality,juiciness, etc., inhere in a ripe apple.)

7

Is this claim correct? Is anintense sensation of heat

the very same quality as the corre-sponding sensation of pain, so thatthere is only one idea and not two?

8

Having said “yes” to thequestion in the previous

annotation, Hylas is forced toadmit that heat of this degreecannot exist outside the mind.

RR


Phil.: But how shall we be able to discern thosedegrees of heat which exist only in themind, from those which exist without it?

Hyl.: That is no difficult matter. You know, theleast pain cannot exist unperceived; what-ever, therefore, degree of heat is a pain,exists only in the mind. But as for all otherdegrees of heat, nothing obliges us to thinkthe same of them.

Phil.: I think you granted before, that no unper-ceiving being was capable of pleasure, anymore than of pain.

Hyl.: I did.Phil.: And is not warmth, or a more gentle

degree of heat than what causes uneasiness,a pleasure?

Hyl.: What then?Phil.: Consequently it cannot exist without the

mind in any unperceiving substance, orbody.

Hyl.: So it seems.Phil.: Since therefore, as well those degrees of

heat that are not painful, as those that are,can exist only in a thinking substance; maywe not conclude that external bodies areabsolutely incapable of any degree of heatwhatsoever?

Hyl.: On second thoughts, I do not think it soevident that warmth is a pleasure, as that agreat degree of heat is a pain.

Phil.: I do not pretend that warmth is as great apleasure as heat is a pain. But if you grant itto be even a small pleasure, it serves tomake good my conclusion.

Hyl.: I could rather call it an indolence. It seemsto be nothing more than a privation of bothpain and pleasure. And that such a qualityor state as this may agree to an unthinkingsubstance, I hope you will not deny.

Phil.: If you are resolved to maintain thatwarmth, or a gentle degree of heat, is nopleasure, I know not how to convince youotherwise than by appealing to your ownsense. But what think you of cold?

Hyl.: The same that I do of heat. An intensedegree of cold is a pain; for to feel a verygreat cold, is to perceive a great uneasiness:it cannot therefore exist without the mind;but a lesser degree of cold may, as well as alesser degree of heat.

Phil.: Those bodies, therefore, upon whoseapplication to our own, we perceive a

moderate degree of heat, must be con-cluded to have a moderate degree of heat orwarmth in them: and those, upon whoseapplication we feel a like degree of cold,must be thought to have cold in them.

Hyl.: They must.Phil.: Can any doctrine be true that necessarily

leads a man into an absurdity?Hyl.: Without doubt it cannot.Phil.: Is it not an absurdity to think that the

same thing should be at the same time bothcold and warm?

Hyl.: It is.Phil.: Suppose now one of your hands hot, and

the other cold and that they are both at onceput into the same vessel of water, in anintermediate state; will not the water seemcold to one hand, and warm to the other?

Hyl.: It will.Phil.: Ought we not therefore by your princi-

ples to conclude, it is really both cold andwarm at the same time, that is, accordingto your own concession, to believe anabsurdity?

Hyl.: I confess it seems so.Phil.: Consequently, the principles themselves

are false, since you have granted that notrue principle leads to an absurdity.

Hyl.: But after all, can anything be more absurdthan to say, there is no heat in the fire?

Phil.: To make the point still clearer; tell me,whether in two cases exactly alike, weought not to make the same judgment?

Hyl.: We ought.Phil.: When a pin pricks your finger, does it not

rend and divide the fibres of your flesh?Hyl.: It does.Phil.: And when a coal burns your finger, does

it any more?Hyl.: It does not.Phil.: Since therefore you neither judge the

sensation itself occasioned by the pin, noranything like it to be in the pin; you shouldnot, conformably to what you have nowgranted, judge the sensation occasioned bythe fire, or anything like it, to be in the fire.

Hyl.: Well, since it must be so, I am content toyield this point, and acknowledge, that heatand cold are only sensations existing in ourminds: but there still remain qualitiesenough to secure the reality of externalthings.

10

99

Hylas resists saying that alldegrees of heat and cold exist

only in the mind, conceding thisonly for degrees of heat and coldextreme enough to involve pain.

(How plausible is the resulting view?)

RR

10

Here is quite a differentargument, one usually

regarded as substantially morecompelling: the argument from illu-sion (or from perceptual relativity).

Try to formulate this argumentmore explicitly. Philonous

argues that it is absurd to say thatboth of the experienced qualities arein the water, but does it follow (as iseventually concluded) that neither ofthem is? How might Philonous arguemore explicitly for this further claim?


[IV]

Phil.: But what will you say, Hylas, if it shallappear that the case is the same with regardto all other sensible qualities, and that theycan no more be supposed to exist withoutthe mind, than heat and cold?

Hyl.: Then indeed you will have done some-thing to the purpose; but that is what Idespair of seeing proved.

Phil.: Let us examine them in order. What thinkyou of tastes, do they exist without themind, or no?

Hyl.: Can any man in his senses doubt whethersugar is sweet, or wormwood bitter?

Phil.: Inform me, Hylas. Is a sweet taste a par-ticular kind of pleasure or pleasant sensa-tion, or is it not?

Hyl.: It is.Phil.: And is not bitterness some kind of uneasi-

ness or pain?Hyl.: I grant it.Phil.: If therefore sugar and wormwood are

unthinking corporeal substances existingwithout the mind, how can sweetness andbitterness, that is, pleasure and pain, agreeto them?

Hyl.: Hold, Philonous, I now see what it wasdeluded me all this time. You askedwhether heat and cold, sweetness and bit-terness, were not particular sorts of plea-sure and pain; to which I answered simply,that they were. Whereas I should have thusdistinguished: those qualities, as perceivedby us, are pleasures or pains, but not asexisting in the external objects. We mustnot therefore conclude absolutely, thatthere is no heat in the fire, or sweetness inthe sugar, but only that heat or sweetness,as perceived by us, are not in the fire orsugar. What say you to this?

Phil.: I say it is nothing to the purpose. Our dis-course proceeded altogether concerningsensible things, which you defined to be thethings we immediately perceived by oursenses. Whatever other qualities thereforeyou speak of, as distinct from these, I knownothing of them, neither do they at allbelong to the point in dispute. You mayindeed pretend to have discovered certainqualities which you do not perceive, andassert those insensible qualities exist in fireand sugar. But what use can be made of this

11

to your present purpose, I am at a loss toconceive. Tell me then once more, do youacknowledge that heat and cold, sweetnessand bitterness, (meaning those qualitieswhich are perceived by the senses) do notexist without the mind?

Hyl.: I see it is to no purpose to hold out, so Igive up the cause as to those mentionedqualities. Though I profess it sound oddly,to say that sugar is not sweet.

Phil.: But for your farther satisfaction, take thisalong with you: that which at other timesseems sweet, shall to a distempered palateappear bitter. And nothing can be plainer,than that divers persons perceive differenttastes in the same food, since that which oneman delights in, another abhors. And howcould this be, if the taste was somethingreally inherent in the food?

Hyl.: I acknowledge I know not how.Phil.: In the next place, odors are to be consid-

ered. And, with regard to these, I would fainknow whether what has been said of tastesdoes not exactly agree to them?Are they notso many pleasing or displeasing sensations?

Hyl.: They are.Phil.: Can you then conceive it possible that

they should exist in an unperceiving thing?Hyl.: I cannot.Phil.: Or can you imagine, that filth and ordure

affect those brute animals that feed on themout of choice, with the same smells whichwe perceive in them?

Hyl.: By no means.Phil.: May we not therefore conclude of smells,

as of the other forementioned qualities, thatthey cannot exist in any but a perceivingsubstance or mind?

Hyl.: I think so.Phil.: Then as to sounds, what must we think of

them: are they accidents really inherent inexternal bodies, or not?

Hyl.: That they inhere not in the sonorous bod-ies, is plain from hence; because a bellstruck in the exhausted receiver of an air-pump, sends forth no sound. The air there-fore must be thought the subject of sound.

Phil.: What reason is there for that, Hylas?Hyl.: Because when any motion is raised in the

air, we perceive a sound greater or lesser, tothe air’s motion; but without some motionin the air, we never hear any sound at all.

13

12

11

Can the conclusion regardingheat be extended to all other

sensible qualities as well? This is theissue in the next several sections.

RR

12

Here is a brief reprisal of theargument from illusion, in

application to taste.Think again of how theargument could be spelled

out more explicitly.

RR

13

The pain/pleasure argumentis applied to odor, raising the

same issue as before. No versionof the argument from illusion isoffered for qualities of smell,but you should be able toconstruct one.


Phil.: And granting that we never hear a soundbut when some motion is produced in theair, yet I do not see how you can infer fromthence, that the sound itself is in the air.

Hyl.: It is this very motion in the external air,that produces in the mind the sensation ofsound. For, striking on the drum of the ear,it causes a vibration, which by the auditorynerves being communicated to the brain,the soul is thereupon affected with the sen-sation called sound.

Phil.: What! Is sound then a sensation?Hyl.: I tell you, as perceived by us, it is a partic-

ular sensation in the mind.Phil.: And can any sensation exist without the

mind?Hyl.: No certainly.Phil.: How then can sound, being a sensation,

exist in the air, if by the air you mean a sense-less substance existing without the mind?

Hyl.: You must distinguish, Philonous, betweensound as it is perceived by us, and as it is initself; or (which is the same thing) betweenthe sound we immediately perceive and thatwhich exists without us. The former indeedis a particular kind of sensation, but the lat-ter is merely a vibrative or undulatorymotion in the air.

Phil. I thought I had already obviated that dis-tinction by the answer I gave when youwere applying it in a like case before. But tosay no more of that; are you sure then thatsound is really nothing but motion?

Hyl.: I am.Phil.: Whatever therefore agrees to real sound,

may with truth be attributed to motion.Hyl.: It may.Phil.: It is then good sense to speak of motion, as

of a thing that is loud, sweet, acute, or grave.Hyl.: I see you are resolved not to understand me.

Is it not evident, those accidents or modesbelong only to sensible sound, or sound inthe common acceptation of the word, but notto sound in the real and Philosophic sense,which, as I just now told you, is nothing buta certain motion of the air?

Phil.: It seems then there are two sorts of sound,the one vulgar, or that which is heard, theother Philosophical and real.

Hyl.: Even so.Phil.: And the latter consists in motion.Hyl.: I told you so before.

14

Phil.: Tell me, Hylas, to which of the senses,think you, the idea of motion belongs: tothe hearing?

Hyl.: No certainly, but to the sight and touch.Phil.: It should follow then, that according to

you, real sounds may possibly be seen orfelt, but never heard.

Hyl.: Look you, Philonous, you may if youplease make a jest of my opinion, but thatwill not alter the truth of things. I ownindeed, the inferences you draw me into,sound something oddly; but common lan-guage, you know, is framed by, and for theuse of, the vulgar: we must not thereforewonder, if expressions adapted to exactPhilosophic notions, seem uncouth and outof the way.

Phil.: Is it come to that? I assure, you, I imaginemyself to have gained no small point, sinceyou make so light of departing from com-mon phrases and opinions; it being a mainpart of our inquiry, to examine whosenotions are widest of the common road, andmost repugnant to the general sense of theworld. But can you think it no more than aPhilosophical paradox, to say that realsounds are never heard, and that the idea ofthem is obtained by some other sense. Andis there nothing in this contrary to natureand the truth of things?

Hyl.: To deal ingenuously, I do not like it. Andafter the concessions already made, I hadas well grant that sounds too have no realbeing without the mind.

[V]

Phil.: And I hope you will make no difficulty toacknowledge the same of colors.

Hyl.: Pardon me: the case of colors is very dif-ferent. Can anything be plainer, than thatwe see them on the objects?

Phil.: The objects you speak of are, I suppose,corporeal substances existing without themind.

Hyl.: They are.Phil.: And have true and real colors inhering in

them?Hyl.: Each visible object has that color which

we see in it.Phil.: How! Is there anything visible but what

we perceive by sight.Hyl.: There is not.

16

15

14

Hylas does not questionthat sound as heard is a sen-

sation in the mind, but attemptsto argue that sound in a differentsense exists as a vibration in theair. In the immediately followingpassage, Philonous insists againthat it is only immediatelyperceived qualities with whichthe argument is concerned.

RR

15

The further argument isoffered that sound as a

motion is not the sort of thing thatcould be heard. How compelling isthis? (Note that the issue doesn’treally matter very much, since thesensible qualities that are immedi-ately perceived have already beenconceded to exist only in the mind.

Think how Berkeley mighthave argued further for such

a claim about sound as heard byusing arguments of the same sortsas were applied to heat and tastes.)

RR

16

The claim that corporealobjects have the color we

see in them is at least roughly thecommon-sense view of color.


Phil.: And do we perceive anything by sense,which we do not perceive immediately?

Hyl.: How often must I be obliged to repeat thesame thing? I tell you, we do not.

Phil.: Have patience, good Hylas; and tell meonce more, whether there is anythingimmediately perceived by the senses,except sensible qualities. I know youasserted there was not: but I would now beinformed, whether you still persist in thesame opinion.

Hyl.: I do.Phil.: Pray, is your corporeal substance either a

sensible quality, or made up of sensiblequalities?

Hyl.: What a question that is! Who ever thoughtit was?

Phil.: My reason for asking was, because insaying, Each visible object has that colorwhich we see in it, you make visible objectsto be corporeal substances; which implieseither that corporeal substances are sensi-ble qualities, or else that there is somethingbeside sensible qualities perceived by sight:but as this point was formerly agreedbetween us, and is still maintained by you,it is a clear consequence, that your corpo-real substance is nothing distinct fromsensible qualities.

Hyl.: You may draw as many absurd conse-quences as you please, and endeavor toperplex the plainest things; but you shallnever persuade me out of my senses. Iclearly understand my own meaning.

Phil.: I wish you would make me understand ittoo. But since you are unwilling to haveyour notion of corporeal substance exam-ined, I shall urge that point no farther. Onlybe pleased to let me know, whether thesame colors which we see, exist in externalbodies, or some other.

Hyl.: The very same.Phil.: What! Are then the beautiful red and

purple we see on yonder clouds, really inthem? Or do you imagine they have inthemselves any other form, than that of adark mist or vapor?

Hyl.: I must own, Philonous, those colors arenot really in the clouds as they seem to be atthis distance. They are only apparent colors.

Phil.: Apparent call you them? How shall wedistinguish these apparent colors from real?

17

Hyl.: Very easily. Those are to be thought appar-ent, which appearing only at a distance,vanish upon a nearer approach.

Phil.: And those I suppose are to be thoughtreal, which are discovered by the most nearand exact survey.

Hyl.: Right.Phil.: Is the nearest and exactest survey made by

the help of a microscope, or by the naked eye?Hyl.: By a microscope, doubtless.Phil.: But a microscope often discovers colors

in an object different from those per-ceived by the unassisted sight. And in casewe had microscopes magnifying to anyassigned degree; it is certain, that noobject whatsoever, viewed through them,would appear in the same color which itexhibits to the naked eye.

Hyl.: And what will you conclude from all this?You cannot argue that there are really andnaturally no colors on objects: because byartificial managements they may be altered,or made to vanish.

Phil.: I think it may evidently be concluded fromyour own concessions, that all the colors wesee with our naked eyes, are only apparent asthose on the clouds, since they vanish upon amore close and accurate inspection, which isafforded us by a microscope. Then as to whatyou say by way of prevention: I ask you,whether the real and natural state of an objectis better discovered by a very sharp and pierc-ing sight, or by one which is less sharp?

Hyl.: By the former without doubt.Phil.: Is it not plain . . . that microscopes make

the sight more penetrating, and representobjects as they would appear to the eye incase it were naturally endowed with a mostexquisite sharpness?

Hyl.: It is.Phil.: Consequently the microscopical represen-

tation is to be thought that which best setsforth the real nature of the thing, or what itis in itself. The colors, therefore, by it per-ceived, are more genuine and real, thanthose perceived otherwise.

Hyl.: I confess there is something in what you say.

…

Phil.: The point will be past all doubt, if youconsider, that in case colors were real prop-erties or affections inherent in external

17

This is an argument againstLocke’s conception of

substance (see Annotation 13 tothe Locke selection): a material orcorporeal substance is supposedto be something that has sensiblequalities like color, but is itselfdistinct from them; but all weperceive is sensible qualities, sosuch a substance must be nothingmore than sensible qualities—orelse something that we do notperceive at all and so arguablyhave no idea at all of.

RR


bodies, they could admit of no alteration,without some change wrought in the verybodies themselves: but is it not evidentfrom what has been said, that upon the useof microscopes, upon a change happeningin the humors of the eye, or a variation ofdistance, without any manner of realalteration in the thing itself, the colors ofany object are either changed, or totallydisappear? Nay, all other circumstancesremaining the same, change but thesituation of some objects, and they shallpresent different colors to the eye. Thesame thing happens upon viewing anobject in various degrees of light. Andwhat is more known, than that the same-bodies appear differently colored bycandle-light from what they do in the openday? Add to these the experiment of aprism, which separating the heterogeneousrays of light, alters the color of any object;and will cause the whitest to appear of adeep blue or red to the naked eye. And nowtell me, whether you are still of opinionthat every body has its true real color inher-ing in it: and if you think it has, I wouldfain know farther from you, what certaindistance and position of the object, whatpeculiar texture and formation of the eye,what degree or kind of light is necessaryfor ascertaining that true color, and distin-guishing it from apparent ones.

Hyl.: I own myself entirely satisfied, that theyare all equally apparent; and that there is nosuch thing as color really inhering in exter-nal bodies, but that it is altogether in thelight. And what confirms me in this opinionis, that in proportion to the light, colors arestill more or less vivid; and if there be nolight, then are there no colors perceived.Besides, allowing there are colors on exter-nal objects, yet how is it possible for us toperceive them? For no external body affectsthe mind, unless it act first on our organs ofsense. But the only action of bodies ismotion; and motion cannot be communi-cated otherwise than by impulse. A distantobject therefore cannot act on the eye, norconsequently make itself or its propertiesperceivable to the soul. Whence it plainlyfollows, that it is immediately somecontiguous substance, which operating on

18

the eye occasions a perception of colors:and such is light.

Phil.: How! Is light then a substance?Hyl.: I tell you, Philonous, external light is

nothing but a thin fluid substance, whoseminute particles being agitated with a briskmotion, and in various manners reflectedfrom the different surfaces of outwardobjects to the eyes, communicate differentmotions to the optic nerves; which, beingpropagated to the brain, cause therein vari-ous impressions: and these are attendedwith the sensations of red, blue, yellow, &c.

Phil.: It seems then, the light does no more thanshake the optic nerves.

Hyl.: Nothing else.Phil.: And consequent to each particular motion

of the nerves, the mind is affected with asensation, which is some particular color.

Hyl.: Right.Phil.: And these sensations have no existence

without the mind.Hyl.: They have not.Phil.: How then do you affirm that colors are in

the light, since by light you understand acorporeal substance external to the mind?

Hyl.: Light and colors, as immediately perceivedby us, I grant cannot exist without the mind.But in themselves they are only the motionsand configurations of certain insensibleparticles of matter.

Phil.: Colors then, in the vulgar sense, or takenfor the immediate objects of sight, cannotagree to any but a perceiving substance.

Hyl.: That is what I say.Phil.: Well then, since you give up the point as

to those sensible qualities, which are alonethought colors by all mankind beside, youmay hold what you please with regard tothose invisible ones of the Philosophers.

It is not my business to dispute aboutthem; only I would advise you to bethinkyourself, whether considering the inquirywe are upon, it be prudent for you toaffirm, The red and blue which we see arenot real colors, but certain unknownmotions and figures which no man ever didor can see are truly so. Are not theseshocking notions and are not they subjectto as many ridiculous inferences, as thoseyou were obliged to renounce before in thecase of sounds?

19

18

Here is another version ofthe argument from illusion,

this time applied to colors.This ver-sion is more complicated, becausethe different color qualities are notexperienced at the same moment.The additional premise needed isthat the changes that yield differentcolor experiences involve no changein the object itself. (Philonous’sremark at the end of this last para-graph may suggest a way of dealingwith the issue, raised in Annotation10, of why the conclusion should bethat none of the experienced quali-ties are external to the mind, ratherthan merely that not all ofthem are.)

19

Hylas accepts the result thatcolor is not in bodies, but

suggests that it is in light instead(which would still be somethingexternal to the mind). But in tryingto explain how color is in light, heis forced to admit that the colorsensations we directly experienceare in the mind, even if theircause is external.

RR


[VI]

Hyl.: I frankly own, Philonous, that it is in vain tostand out any longer. Colors, sounds, tastes,in a word, all those termed secondary quali-ties, have certainly no existence without themind. But by this acknowledgment I mustnot be supposed to derogate anything fromthe reality of matter or external objects, see-ing it is no more than several Philosophersmaintain, who nevertheless are the farthestimaginable from denying matter. For theclearer understanding of this, you must knowsensible qualities are by Philosophersdivided into primary and secondary. The for-mer are extension, figure, solidity, gravity,motion, and rest. And these they hold existreally in bodies. The latter are those aboveenumerated; or briefly, all sensible qualitiesbeside the primary, which they assert areonly so many sensations or ideas existingnowhere but in the mind. But all this, I doubtnot, you are already apprised of. For mypart, I have been a long time sensiblethere was such an opinion current amongPhilosophers, but was never thoroughlyconvinced of its truth till now.

Phil.:You are still then of opinion that extensionand figures are inherent in externalunthinking substances.

Hyl.: I am.Phil.: But what if the same arguments which are

brought against secondary qualities, willhold good against these also?

Hyl.: Why then I shall be obliged to think, theytoo exist only in the mind.

Phil.: Is it your opinion, the very figure andextension which you perceive by sense, existin the outward object or material substance?

Hyl.: It is.

…

Phil.: . . . have you not acknowledged that noreal inherent property of any object can bechanged, without some change in the thingitself?

Hyl.: I have.Phil.: But as we approach to or recede from an

object, the visible extension varies, being atone distance ten or a hundred times greaterthan at another. Does it not therefore followfrom hence likewise, that it is not reallyinherent in the object?

20

Hyl.: I own I am at a loss what to think.Phil.: Your judgment will soon be determined,

if you will venture to think as freelyconcerning this quality, as you have doneconcerning the rest. Was it not admitted asa good argument, that neither heat nor coldwas in the water, because it seemed warmto one hand, and cold to the other?

Hyl.: It was.Phil.: Is it not the very same reasoning to

conclude, there is no extension or figure inan object, because to one eye it shall seemlittle, smooth, and round, when at the sametime it appears to the other, great, uneven,and angular?

Hyl.: The very same. But does this latter factever happen?

Phil.:You may at any time make the experiment,by looking with one eye bare, and with theother through a microscope.

Hyl.: I know not how to maintain it, and yet I amloath to give up extension, I see so manyodd consequences following upon such aconcession.

Phil.: Odd, say you? After the concessionsalready made, I hope you will stick atnothing for its oddness. . . .

Hyl.: I give up the point for the present, reserv-ing still a right to retract my opinion, incase I shall hereafter discover any falsestep in my progress to it.

[VII]Phil.: That is a right you cannot be denied.

Figures and extension being dispatched, weproceed next to motion. Can a real motionin any external body be at the same timeboth very swift and very slow?

Hyl.: It cannot.Phil.: Is not the motion of a body swift in a

reciprocal proportion to the time it takes upin describing any given space? Thus a bodythat describes a mile in an hour, movesthree times faster than it would in case itdescribed only a mile in three hours.

Hyl.: I agree with you.Phil.: And is not time measured by the succes-

sion of ideas in our minds?Hyl.: It is.Phil.: And is it not possible ideas should suc-

ceed one another twice as fast in your mind,

21

20

Hylas retreats to the Lockean distinction between

primary and secondary qualities,agreeing that secondary qualitiesexist only in the mind, but claim-ing that primary qualities exist inexternal objects. (See Annotation7 to the Locke selection and theassociated text.)

RR

21

Philonous proceeds to applythe argument from illusion to

figure (shape) and extension (size).RR


as they do in mine, or in that of some spiritof another kind.

Hyl.: I own it.Phil.: Consequently the same body may to

another seem to perform its motion overany space in half the time that it does toyou. And the same reasoning will hold as toany other proportion: that is to say, accord-ing to your principles (since the motionsperceived are both really in the object) it ispossible one and the same body shall bereally moved the same way at once, bothvery swift and very slow. How is this con-sistent either with common sense, or withwhat you just now granted?

Hyl.: I have nothing to say to it.Phil.: Then as for solidity; either you do not

mean any sensible quality by that word, andso it is beside our inquiry: or if you do, itmust be either hardness or resistance. Butboth the one and the other are plainlyrelative to our senses: it being evident, thatwhat seems hard to one animal, may appearsoft to another, who hath greater force andfirmness of limbs. Nor is it less plain, thatthe resistance I feel is not in the body.

Hyl.: I own the very sensation of resistance,which is all you immediately perceive, is notin the body; but the cause of that sensation is.

Phil.: But the causes of our sensations are notthings immediately perceived, and there-fore not sensible. This point I thought hadbeen already determined.

Hyl.: I own it was; but you will pardon me if Iseem a little embarrassed: I know not howto quit my old notions.

Phil.: To help you out, do but consider that ifextension be once acknowledged to haveno existence without the mind, the samemust necessarily be granted of motion,solidity, and gravity, since they all evi-dently suppose extension. It is thereforesuperfluous to inquire particularly con-cerning each of them. In denying exten-sion, you have denied them all to have anyreal existence.

…

Phil.: Can you even separate the ideas ofextension and motion, from the ideas of allthose qualities which they who make thedistinction, term secondary.

23

22

Hyl.: What! Is it not an easy matter, to considerextension and motion by themselves,abstracted from all other sensible qualities?Pray how do the mathematicians treat ofthem?

Phil.: I acknowledge, Hylas, it is not difficult toform general propositions and reasoningsabout those qualities, without mentioningany other; and in this sense to consider ortreat of them abstractedly. But how does itfollow that because I can pronounce theword motion by itself, I can form the idea ofit in my mind exclusive of body? Orbecause theorems may be made of exten-sion and figures, without any mention ofgreat or small, or any other sensible modeor quality; that therefore it is possible suchan abstract idea of extension, without anyparticular size or figure, or sensible quality,should be distinctly formed, and appre-hended by the mind? . . .

Hyl.: But what say you to pure intellect? May notabstracted ideas be framed by that faculty?

Phil.: Since I cannot frame abstract ideas at all, itis plain, I cannot frame them by the help ofpure intellect, whatsoever faculty you under-stand by those words. Besides, . . . thus muchseems manifest, that sensible things are onlyto be perceived by sense, or represented bythe imagination. Figures, therefore, andextension, being originally perceived bysense, do not belong to pure intellect. But foryour farther satisfaction, try if you can framethe idea of any figure, abstracted from allparticularities of size, or even from othersensible qualities.

Hyl.: Let me think a little—I do not find thatI can.

Phil.: And can you think it possible, that shouldreally exist in nature, which implies arepugnancy in its conception?

Hyl.: By no means.Phil.: Since therefore it is impossible even for

the mind to disunite the ideas of extensionand motion from all other sensible qualities,does it not follow, that where the one exist,there necessarily the other exist likewise?

Hyl.: It should seem so.Phil.: Consequently the very same arguments

which you admitted, as conclusive againstthe secondary qualities, are, without anyfarther application of force, against the

22

Here it is important to real-ize that what the argument is

concerned with is motion asdirectly experienced, not motionas measured (for example, by aspeedometer).The claim that timeis measured by the speed withwhich ideas succeed each other inthe mind is not particularlyplausible. Can you think of a betterargument here? (See DiscussionQuestion 4.)

23

Here is a quick gesture atthe argument from illusion

in application to solidity—youshould try to spell it out morefully. In the following paragraph,Philonous argues further thatother primary qualities presup-pose extension and so cannot beoutside the mind if it is not.


primary too. Besides, if you will trust yoursenses, is it not plain, all sensible qualitiescoexist, or, to them, appear as being in thesame place? Do they ever represent amotion, or figure, as being divested of allother visible and tangible qualities?

Hyl.: You need say no more on this head. I amfree to own, if there be no secret error oroversight in our proceedings hitherto, thatall sensible qualities are alike to be deniedexistence without the mind. But my fearis, that I have been too liberal in my for-mer concessions, or overlooked some fal-lacy or other. In short, I did not take timeto think.

…

[VIII]

Hyl.: I acknowledge, Philonous, that upon afair observation of what passes in mymind, I can discover nothing else, but thatI am a thinking being, affected with vari-ety of sensations; neither is it possible toconceive how a sensation should exist inan unperceiving substance. But then onthe other hand, when I look on sensiblethings in a different view, consideringthem as so many modes and qualities, If ind it necessary to suppose a materialsubstratum, without which they cannot beconceived to exist.

Phil.: Material substratum call you it? Pray, bywhich of your senses came you acquaintedwith that being?

Hyl.: It is not itself sensible; its modes and qual-ities only being perceived by the senses.

Phil.: I presume then it was by reflection andreason you obtained the idea of it.

Hyl.: I do not pretend to any proper positiveidea of it. However I conclude it exists,because qualities cannot be conceived toexist without a support.

Phil.: It seems then you have only a relativenotion of it, or that you conceive it not oth-erwise than by conceiving the relation itbears to sensible qualities.

Hyl.: Right.Phil.: Be pleased therefore to let me know

wherein that relation consists.Hyl.: Is it not sufficiently expressed in the term

substratum, or substance?

25

24

Phil.: If so, the word substratum should import,that it is spread under the sensible qualitiesor accidents.

Hyl.: True.Phil.: And consequently under extension.Hyl.: I own it.Phil.: It is therefore somewhat in its own nature

entirely distinct from extension.Hyl.: I tell you, extension is only a mode, and

matter is something that supports modes.And is it not evident the thing supported isdifferent from the thing supporting?

Phil.: So that something distinct from, andexclusive of, extension, is supposed to bethe substratum of extension.

Hyl.: Just so.Phil.: Answer me, Hylas. Can a thing be spread

without extension? Or is not the idea ofextension necessarily included inspreading?

Hyl.: It is.Phil.: Whatsoever therefore you suppose spread

under anything, must have in itself anextension distinct from the extension of thatthing under which it is spread.

Hyl.: It must.Phil.: Consequently every corporeal substance,

being the substratum of extension, must havein itself another extension by which it is qual-ified to be a substratum: and so on to infinity.And I ask whether this be not absurd in itself,and repugnant to what you granted just now,to wit, that the substratum was somethingdistinct from, and exclusive of, extension.

Hyl.: Aye but, Philonous, you take me wrong. Ido not mean that matter is spread in a grossliteral sense under extension. The wordsubstratum is used only to express in generalthe same thing with substance.

Phil.: Well then, let us examine the relationimplied in the term substance. Is it not thatit stands under accidents?

Hyl.: The very same.Phil.: But that one thing may stand under or

support another, must it not be extended?Hyl.: It must.Phil.: Is not therefore this supposition liable to

the same absurdity with the former?Hyl.:You still take things in a strict literal sense:

that is not fair, Philonous.Phil.: I am not for imposing any sense on your

words: you are at liberty to explain them as

25

Here is another attackon the Lockean idea of

substance, spelling out the argu-ment that was only hinted at in thetext corresponding to Annotation17. Hylas uses the variant termsubstratum for substance, with theidea being that material substanceunderlies or supports the qualitiesthat inhere in it. But since all ourideas are of qualities, he is forcedto admit that he has no real ideaof material substance itself.

24

A further argument againstthe separate existence of

primary qualities is that we haveno ideas of them in isolation fromsecondary qualities. At this point,Hylas surrenders and concedesthat all sensible qualities existonly in the mind.

Here, as elsewhere, ideasseem to be regarded as

images, with the point beingthat any image of, for example, aparticular shape must also involveother qualities like color thatdelineate the shape.

RR


you please. Only I beseech you, make meunderstand something by them. You tell me,matter supports or stands under accidents.How! Is it as your legs support your body?

Hyl.: No; that is the literal sense.Phil.: Pray let me know any sense, literal or not

literal, that you understand it in.—Howlong must I wait for an answer, Hylas?

Hyl.: I declare I know not what to say. I oncethought I understood well enough what wasmeant by matter’s supporting accidents. Butnow the more I think on it, the less can Icomprehend it; in short, I find that I knownothing of it.

Phil.: It seems then you have no idea at all, nei-ther relative nor positive of matter; youknow neither what it is in itself, nor whatrelation it bears to accidents.

Hyl.: I acknowledge it.Phil.: And yet you asserted, that you could not

conceive how qualities or accidents shouldreally exist, without conceiving at the sametime a material support of them.

Hyl.: I did.Phil.: That is to say, when you conceive the real

existence of qualities, you do withal con-ceive something which you cannot conceive.

[IX]

Hyl.: It was wrong I own. But still I fear there issome fallacy or other. Pray what think youof this? It is just come into my head, thatthe ground of all our mistake lies in yourtreating of each quality by itself. Now, Igrant that each quality cannot singly sub-sist without the mind. Color cannot withoutextension, neither can figure without someother sensible quality. But as the severalqualities united or blended together formentire sensible things, nothing hinders whysuch things may not be supposed to existwithout the mind.

Phil.: Either, Hylas, you are jesting, or have avery bad memory. Though indeed we wentthrough all the qualities by name one afteranother; yet my arguments, or rather yourconcessions, nowhere tended to prove, thatthe secondary qualities did not subsist eachalone by itself; but that they were not at allwithout the mind. Indeed in treating of fig-ure and motion, we concluded they couldnot exist without the mind, because it was

26

impossible even in thought to separatethem from all secondary qualities, so as toconceive them existing by themselves. . . .But (to pass by all that hath been hithertosaid, and reckon it for nothing, if you willhave it so) I am content to put the wholeupon this issue. If you can conceive it pos-sible for any mixture or combination ofqualities, or any sensible object whatever,to exist without the mind, then I will grantit actually to be so.

Hyl.: If it comes to that, the point will soon bedecided. What more easy than to conceive atree or house existing by itself, independentof, and unperceived by, any mind whatso-ever? I do at this present time conceivethem existing after that manner.

Phil.: How say you, Hylas, can you see a thingwhich is at the same time unseen?

Hyl.: No, that were a contradiction.Phil.: Is it not as great a contradiction to talk of

conceiving a thing which is unconceived?Hyl.: It is.Phil.: The tree or house therefore which you

think of, is conceived by you.Hyl.: How should it be otherwise?Phil.: And what is conceived is surely in the

mind.Hyl.: Without question, that which is conceived

is in the mind.Phil.: How then came you to say, you conceived

a house or tree existing independent and outof all minds whatsoever?

Hyl.: That was I own an oversight; but stay, letme consider what led me into it.—It is apleasant mistake enough. As I was think-ing of a tree in a solitary place, where noone was present to see it, methought thatwas to conceive a tree as existing unper-ceived or unthought of, not consideringthat I myself conceived it all the while. Butnow I plainly see, that all I can do is toframe ideas in my own mind. I may indeedconceive in my own thoughts the idea of atree, or a house, or a mountain, but that isall. And this is far from proving, that I canconceive them existing out of the minds ofall spirits.

Phil.: You acknowledge then that you cannotpossibly conceive, how any one corporealsensible thing should exist otherwise thanin a mind.

26

Philonous ruthlessly driveshome the point that Hylas

has no intelligible idea of eithersubstance itself or its relation toqualities.

RR


Hyl.: I do.Phil.: And yet you will earnestly contend for the

truth of that which you cannot so much asconceive.

Hyl.: I profess I know not what to think. . . .

[X]

Hyl.: . . . but inform me, Philonous, can we per-ceive or know nothing beside our ideas?

Phil.: As for the rational deducing of causesfrom effects, that is beside our inquiry.And by the senses you can best tell,whether you perceive anything which isnot immediately perceived. And I ask youwhether the things immediately perceived,are other than your own sensations orideas? Your have indeed more than once,in the course of this conversation, declaredyourself on those points; but you seem bythis last question to have departed fromwhat you then thought.

Hyl.: To speak the truth, Philonous, I think thereare two kinds of objects, the one perceivedimmediately, which are likewise calledideas; the other are real things or externalobjects perceived by the mediation of ideas,which are their images and representations.Now I own, ideas do not exist without themind; but the latter sort of objects do. I amsorry I did not think of this distinctionsooner; it would probably have cut shortyour discourse.

Phil.: Are those external objects perceived bysense, or by some other faculty?

Hyl.: They are perceived by sense.Phil.: How! Is there anything perceived by

sense, which is not immediately perceived?Hyl.: Yes, Philonous, in some sort there is. For

example, when I look on a picture or statueof Julius Caesar, I may be said after a man-ner to perceive him (though not immedi-ately) by my senses.

Phil.: It seems then, you will have our ideas,which alone are immediately perceived, tobe pictures of external things; and that thesealso are perceived by sense, inasmuch asthey have a conformity or resemblance toour ideas.

Hyl.: That is my meaning.Phil.: And in the same way that Julius Caesar, in

himself invisible, is nevertheless perceived

28

27by sight; real things, in themselves imper-ceptible, are perceived by sense.

Hyl.: In the very same.

…

Phil.: . . . I grant we may in one acceptation besaid to perceive sensible things mediatelyby sense: that is, when from a frequentlyperceived connection, the immediate per-ception of ideas by one sense suggests tothe mind others, perhaps belonging toanother sense, which are wont to be con-nected with them. For instance, when I heara coach drive along the streets, immediatelyI perceive only the sound; but from theexperience I have had that such a sound isconnected with a coach, I am said to hearthe coach. It is nevertheless evident, that intruth and strictness, nothing can be heardbut sound: and the coach is not then prop-erly perceived by sense, but suggested fromexperience. So likewise when we are said tosee a red-hot bar of iron; the solidity andheat of the iron are not the objects of sight,but suggested to the imagination by thecolor and figure, which are properly per-ceived by that sense. In short, those thingsalone are actually and strictly perceived byany sense, which would have been per-ceived, in case that same sense had thenbeen first conferred on us. As for otherthings, it in plain they are only suggested tothe mind by experience grounded on for-mer perceptions. But to return to your com-parison of Caesar’s picture, it is plain, if youkeep to that, you must hold the real things,or archetypes of our ideas, are not per-ceived by sense, but by some internal fac-ulty of the soul, as reason or memory. Iwould therefore fair know, what argumentsyou can draw from reason for the existenceof what you call real things or materialobjects. Or whether you remember to haveseen them formerly as they are in them-selves? Or if you have heard or read of anyone that did.

Hyl.: I see, Philonous, you are disposed toraillery; but that will never convince me.

Phil.: My aim is only to learn from you, the wayto come at the knowledge of material beings.Whatever we perceive, is perceived immedi-ately or mediately: by sense, or by reason

28

At this point, Hylas has finallyarrived at Locke’s main view:

we are immediately or directlyaware of ideas in the mind, butthese represent objects outside themind in virtue of the “conformityor resemblance” between the two.

27

to the realist viewthat objects exist outside

the mind, it would be possible forsuch an object to exist eventhough conceived by no one. Butif one attempts to conceive ofsuch a situation, it seems to followthat the object is both allegedlyunconceived but also conceived,which is seemingly acontradiction.Thus both thepossibility in question and theview that leads to it appear tobe inconceivable, and soapparently impossible.

This is one ofBerkeley’s favorite

arguments, one that is unique tohim and also very tricky. Can yousee any way to reply to it?

RR


and reflection. But as you have excludedsense, pray show me what reason you have tobelieve their existence; or what medium youcan possibly make use of, to prove it either tomine or your own understanding.

Hyl.: To deal ingenuously, Philonous, now Iconsider the point, I do not find I can giveyou any good reason for it. . . .

Phil.: . . . if I understand you rightly, you say ourideas do not exist without the mind; but thatthey are copies, images, or representationsof certain originals that do.

Hyl.: You take me right.Phil.: They are then like external things.Hyl.: They are.

…

Phil.: Which are material objects in themselves,perceptible or imperceptible?

Hyl.: Properly and immediately nothing can beperceived but ideas. All material thingstherefore are in themselves insensible, andto be perceived only by their ideas.

Phil.: Ideas then are sensible, and their arche-types or originals insensible.

Hyl.: Right.Phil.: But how can that which is sensible be like

that which is insensible? Can a real thing initself invisible be like a color; or a real thingwhich is not audible, be like a sound? In aword, can anything be like a sensation oridea, but another sensation or idea?

…

Hyl.: Upon inquiry, I find it is impossible forme to conceive or understand how anythingbut an idea can be like an idea. And it ismost evident, that no idea can exist withoutthe mind.

Phil.: You are therefore by your principlesforced to deny the reality of sensible things,since you made it to consist in an absoluteexistence exterior to the mind. That is tosay, you are a downright sceptic. So I havegained my point, which was to show yourprinciples led to scepticism.

Hyl.: For the present I am, if not entirely con-vinced, at least silenced.

Phil.: I would fain know what more you wouldrequire in order to a perfect conviction.Have you not had the liberty of explainingyourself all manner of ways? Were any little

30

29

slips in discourse laid hold and insisted on?Or were you not allowed to retract orreinforce anything you had offered, as bestserved your purpose? Has not everythingyou could say been heard and examinedwith all the fairness imaginable? In a word,have you not in every point been convincedout of your own mouth? And if you can atpresent discover any flaw in any of yourformer concessions, or think of any remain-ing subterfuge, any new distinction, color,or comment whatsoever, why do you notproduce it?

Hyl.: A little patience, Philonous. I am at pres-ent so amazed to see myself ensnared, andas it were imprisoned in the labyrinths youhave drawn me into, that on the sudden itcannot be expected I should find my wayout. You must give me time to look aboutme, and recollect myself.

Phil.: Hark; is not this the college bell?Hyl.: It rings for prayers.Phil.: We will go in then if you please, and meet

here again tomorrow morning. In the mean-time you may employ your thoughts on thismorning’s discourse, and try if you can findany fallacy in it, or invent any new means toextricate yourself.

Hyl.: Agreed.

The Second Dialogue[XI]

Hyl.: I beg your pardon, Philonous, for notmeeting you sooner. All this morning myhead was so filled with our late conversa-tion, that I had not leisure to think of thetime of the day, or indeed of anything else.

Phil.: I am glad you were so intent upon it, inhopes if there were any mistakes in your con-cessions, or fallacies in my reasonings fromthem, you will now discover them to me.

Hyl.: I assure you, I have done nothing ever sinceI saw you, but search after mistakes and fal-lacies, and with that view have minutelyexamined the whole series of yesterday’s dis-course: but all in vain, for the notions it ledme into, upon review, appear still more clearand evident; and the more I consider them,the more irresistibly do they force my assent.

Phil.: And is not this, think you, a sign that theyare genuine, that they proceed from nature,

29

Philonous raises the sameissue that Locke was grap-

pling with in Book IV, chapter XIof the Essay (pp. 71–73): what rea-son is there to believe in the exis-tence of external objects that arenever directly perceived or experi-enced? In the case of the coach andthe bar of iron, the claim is thatsome immediately perceived ideassuggest others previously perceivedalong with them—but this accountwill not work for something that isnever immediately perceived. Hylashas no reply.

30

Philonous now challenges theclaim that ideas resemble

external objects. It is hard to seehow an insensible, unperceivingthing can literally resemble a bit ofexperience like an idea or sensa-tion—how could two things asdifferent as that be literally alike?Hylas concedes that he can makeno sense of this.

RR


and are conformable to right reason? Truthand beauty are in this alike, that thestrictest survey sets them both off to advan-tage. While the false lustre of error anddisguise cannot endure being reviewed, ortoo nearly inspected.

Hyl.: I own there is a great deal in what you say.Nor can any one be more entirely satisfiedof the truth of those odd consequences, solong as I have in view the reasonings thatlead to them. . . .

Phil.: Well then, are you at length satisfied thatno sensible things have a real existence; andthat you are in truth an arrant sceptic?

Hyl.: It is too plain to be denied.Phil.: Look! Are not the fields covered with a

delightful verdure? Is there not somethingin the woods and groves, in the rivers andclear springs, that soothes, that delights,that transports the soul? At the prospect ofthe wide and deep ocean, or some hugemountain whose top is lost in the clouds, orof an old gloomy forest, are not our mindsfilled with a pleasing horror? Even in rocksand deserts, is there not an agreeable wild-ness? How sincere a pleasure is it to beholdthe natural beauties of the earth! . . . Is notthe whole system immense, beautiful,glorious beyond expression and beyondthought! What treatment then do thosePhilosophers deserve, who would deprivethese noble and delightful scenes of allreality? How should those principles beentertained, that lead us to think all the vis-ible beauty of the creation a false imaginaryglare? To be plain, can you expect this scep-ticism of yours will not be thought extrava-gantly absurd by all men of sense?

Hyl.: Other men may think as they please: butfor your part you have nothing to reproachme with. My comfort is, you are as much asceptic as I am.

Phil.: There, Hylas, I must beg leave to differfrom you.

Hyl.: What! Have you all along agreed to thepremises, and do you now deny the conclu-sion, and leave me to maintain those para-doxes by myself which you led me into?This surely is not fair.

Phil.: I deny that I agreed with you in thosenotions that led to scepticism. You indeedsaid, the reality of sensible things consisted

31

in an absolute existence out of the minds ofspirits, or distinct from their being per-ceived. And pursuant to this notion of real-ity, you are obliged to deny sensible thingsany real existence: that is, according to yourown definition, you profess yourself asceptic. But I neither said nor thought thereality of sensible things was to be definedafter that manner. To me it is evident, for thereasons you allow of, that sensible thingscannot exist otherwise than in a mind orspirit. Whence I conclude, not that theyhave no real existence, but that seeing theydepend not on my thought, and have anexistence distinct from being perceived byme, there must be some other mind whereinthey exist. As sure therefore as the sensibleworld really exists, so sure is there anyinfinite omnipresent spirit who containsand supports it.

. . . Sensible things do really exist: and ifthey really exist, they are necessarily per-ceived by an infinite mind: therefore thereis an infinite mind, or God. This furnishesyou with a direct and immediate demon-stration, from a most evident principle, ofthe being of a God. . . .

Take here in brief my meaning. It isevident that the things I perceive are myown ideas, and that no idea can exist unlessit be in a mind. Nor is it less plain that theseideas or things by me perceived, eitherthemselves or their archetypes, existindependently of my mind, since I knowmyself not to be their author, it being out ofmy power to determine at pleasure,what particular ideas I shall be affectedwith upon opening my eyes or ears.They must therefore exist in some othermind, whose will it is they should be exhib-ited to me. The things, I say, immediatelyperceived, are ideas or sensations, call themwhich you will. But how can any idea orsensation exist in, or be produced by,anything but a mind or spirit? This indeed isinconceivable; and to assert that which isinconceivable, is to talk nonsense: is it not?

Hyl.: Without doubt.Phil.: But on the other hand, it is very con-

ceivable that they should exist in, and beproduced by, a spirit; since this is nomore than I daily experience in myself,

34

33

32

32

Philonous denies that he is askeptic. Instead, he views

himself as merely offering a differ-ent and more correct account ofwhat the reality of objects likethose in nature (whose existencehe does not deny) amounts to:such things exist only as ideas inthe mind, and if they have an exis-tence that is independent of hismind (and those of others likehim), this must be because thereis some other mind (God’s mind)in which they exist.

RR

33

Here is an outline of one ofBerkeley’s arguments for the

existence of God, the details ofwhich are elaborated in thefollowing passages.

RR

34

Here is a reason for thinkingthat sensible things exist

independently of Philonous’s mind(or of any human mind): the ideasthat make them up are not withinhis (or any other human’s) volun-tary control. How compelling areason is this? (Compare thiswith the second of Locke’s “fourconcurrent reasons” in chapter XIof Book IV of the Essay.)

31

Having driven Hylas into whatHylas regards as skepticism

about the world of ordinaryobjects, Philonous launches intoanother oration on the glories ofnature, again suggesting that he isnot denying the existence of suchthings.

RR


inasmuch as I perceive numberless ideas;and by an act of my will can form a greatvariety of them, and raise them up in myimagination: though it must be con-fessed, these creatures of the fancy arenot altogether so distinct, so strong, vivid,and permanent, as those perceived by mysenses, which latter are called real things.From all which I conclude, there is amind which affects me every moment withall the sensible impressions I perceive.And from the variety, order, and mannerof these, I conclude the author of them tobe wise, powerful, and good, beyond com-prehension. . . . the things by me per-ceived are known by the understanding,and produced by the will, of an infinitespirit. And is not all this most plain andevident? Is there any more in it, than whata little observation of our own minds, andthat which passes in them not onlyenables us to conceive, but also obliges usto acknowledge?

[XII]

Hyl.: I think I understand you very clearly; andown the proof you give of a deity seems noless evident, than it is surprising. Butallowing that God is the supreme and uni-versal cause of all things, yet may not therebe still a third nature besides spirits andideas? May we not admit a subordinate andlimited cause of our ideas? In a word, maythere not for all that be matter?

. . . I would by no means be thought todeny that God or an infinite spirit is thesupreme cause of all things. All I contendfor, is, that subordinate to the supremeagent there is a cause of a limited and infe-rior nature, which concurs in the productionof our ideas, not by any act of will or spiri-tual efficiency, but by that kind of actionwhich belongs to matter, viz. motion.

Phil.: . . . you are loath to part with your oldprejudice. But to make you quit it moreeasily, I desire that, beside what has beenhitherto suggested, you will fartherconsider whether, upon supposition thatmatter exists, you can possibly conceivehow you should be affected by it? Orsupposing it did not exist, whether it be notevident you might for all that be affected

36

35

with the same ideas you now are, andconsequently have the very same reasons tobelieve its existence that you now can have.

Hyl.: I acknowledge it is possible we might per-ceive all things just as we do now, thoughthere was no matter in the world; neither canI conceive, if there be matter, how it shouldproduce any idea in our minds. . . .

I acknowledge you have proved that mat-ter is impossible; nor do I see what more canbe said in defense of it. But at the same timethat I give up this, I suspect all my othernotions. For surely none could be moreseemingly evident than this once was: andyet it now seems as false and absurd as everit did true before. But I think we havediscussed the point sufficiently for the pres-ent. The remaining part of the day I wouldwillingly spend, in running over in mythoughts the several heads of this morning’sconversation, and tomorrow shall be glad tomeet you here again about the same time.

Phil.: I will not fail to attend you.

The Third Dialogue[XIII]

Philonous: Tell me, Hylas, what are the fruitsof yesterday’s meditation? Has it con-firmed you in the same mind you were in atparting? Or have you since seen cause tochange your opinion?

Hylas:: Truly my opinion is, that all our opin-ions are alike vain and uncertain. What weapprove to-day, we condemn tomorrow. Wekeep a stir about knowledge, and spend ourlives in the pursuit of it, when, alas! weknow nothing all the while: nor do I think itpossible for us ever to know anything in thislife. Our faculties are too narrow and toofew. Nature certainly never intended us forspeculation.

Phil.: What! Say you we can know nothing,Hylas?

Hyl.: There is not that single thing in the world,whereof we can know the real nature, orwhat it is in itself.

Phil.: Will you tell me I do not really know whatfire or water is?

Hyl.: You may indeed know that fire appearshot, and water fluid: but this is no more thanknowing what sensations are produced in

37

35

Berkeley’s view is that theideas corresponding to real

objects (as opposed, for example,to merely imaginary ones) areproduced in us by God. He sug-gests that this can be understoodby analogy with our own weakerpower to deliberately imagine orthink of various things.

RR

36

This is an appeal to some-thing like the argument from

design (see Chapter 7).

37

Hylas can give no account ofhow matter would cause

ideas, and hence has to admit thathe could just as well have the sameideas if there were no matter, inwhich case there is no reason tobelieve in it. (The underlying con-cern here is how minds and bodiescan causally interact—see thediscussion in Chapter 3.)

RR


your own mind, upon the application of fireand water to your organs of sense. Theirinternal constitution, their true and realnature, you are utterly in the dark as to that.

Phil.: Do I not know this to be a real stone thatI stand on, and that which I see before myeyes to be a real tree?

Hyl.: Know? No, it is impossible you or any manalive should know it. All you know, is, thatyou have such a certain idea or appearance inyour own mind. But what is this to the realtree or stone? I tell you, that color, figure,and hardness, which you perceive, are notthe real natures of those things, or in the leastlike them. The same may be said of all otherreal things or corporeal substances whichcompose the world. They have none of themanything in themselves, like those sensiblequalities by us perceived. We should nottherefore pretend to affirm or know anythingof them, as they are in their own nature.

…

Phil.: . . . And so you are plunged into the deep-est and most deplorable scepticism that everman was. Tell me, Hylas, is it not as I say?

[XIV]

Hyl.: I agree with you. Material substance wasno more than an hypothesis, and a false andgroundless one too. I will no longer spendmy breath in defence of it. But whateverhypothesis you advance, or whatsoeverscheme of things you introduce in its stead,I doubt not it will appear every whit asfalse: let me but be allowed to question youupon it. That is, suffer me to serve you inyour own kind, and I warrant it shall con-duct you through as many perplexities andcontradictions, to the very same state ofscepticism that I myself am in at present.

Phil.: I assure you, Hylas, I do not pretend toframe any hypothesis at all. I am of a vulgarcast, simple enough to believe my senses, andleave things as I find them. To be plain, it ismy opinion, that the real things are those verythings I see and feel, and perceive by mysenses. These I know, and finding theyanswer all the necessities and purposes oflife, have no reason to be solicitous about anyother unknown beings. A piece of sensible

bread, for instance, would stay my stomachbetter than ten thousand times as much of thatinsensible, unintelligible, real bread youspeak of. It is likewise my opinion, that col-ors and other sensible qualities are on theobjects. I cannot for my life help thinking thatsnow is white, and fire hot. You indeed, whoby snow and fire mean certain external,unperceived, unperceiving substances, are inthe right to deny whiteness or heat to beaffections inherent in them. But I, who under-stand by those words the things I see and feel,am obliged to think like other folks. And as Iam no sceptic with regard to the nature ofthings, so neither am I as to their existence.That a thing should be really perceived by mysenses, and at the same time not really exist,is to me a plain contradiction; since I cannotprescind or abstract, even in thought, theexistence of a sensible thing from its beingperceived. Wood, stones, fire, water, flesh,iron, and the like things, which I nameand discourse of, are things that I know; . . .And I should not have known them, butthat I perceived them by my senses; andthings perceived by the senses are immedi-ately perceived; and things immediatelyperceived are ideas; and ideas cannot existwithout the mind; their existence thereforeconsists in being perceived; when thereforethey are actually perceived, there can be nodoubt of their existence. Away then with allthat scepticism, all those ridiculousPhilosophical doubts. What a jest is it fora Philosopher to question the existence ofsensible things, till he has it proved to himfrom the veracity of God: or to pretend ourknowledge in this point falls short of intuitionor demonstration? I might as well doubt ofmy own being, as of the being of those thingsI actually see and feel.

Hyl.: Not so fast, Philonous: you say you cannotconceive how sensible things should existwithout the mind. Do you not?

Phil.: I do.Hyl.: Supposing you were annihilated, cannot

you conceive it possible, that things per-ceivable by sense may still exist?

Phil.: I can; but then it must be in another mind.When I deny sensible things an existenceout of the mind, I do not mean my mind inparticular, but all minds. Now it is plain

38 38

Philonous’s contrary view isthat for a thing to exist is just

for the relevant ideas to exist, andthat to know those ideas is toknow all there is to know aboutthe nature of such a thing.

(The immediately followingpassage is aimed at

Descartes and Locke.)

RR


they have an existence exterior to my mind,since I find them by experience to be inde-pendent of it. There is therefore some othermind wherein they exist, during the inter-vals between the times of my perceivingthem: as likewise they did before my birth,and would do after my supposed annihila-tion. And as the same is true, with regard toall other finite created spirits; it necessarilyfollows, there is an omnipresent eternalmind, which knows and comprehends allthings, and exhibits them to our view insuch a manner, and according to such rulesas he himself has ordained, and are by ustermed the laws of nature.

…

[XV]

Hyl.: But according to your notions, what dif-ference is there between real things, andchimeras formed by the imagination, or thevisions of a dream, since they are allequally in the mind?

Phil.: The ideas formed by the imagination arefaint and indistinct; they have, besides, anentire dependence on the will. But the ideasperceived by sense, that is, real things, aremore vivid and clear, and being imprintedon the mind by a spirit distinct from us, havenot a like dependence on our will. There istherefore no danger of confounding thesewith the foregoing: and there is as little ofconfounding them with the visions of adream, which are dim, irregular, and con-fused. And though they should happen to benever so lively and natural, yet by their notbeing connected, and of a piece with the pre-ceding and subsequent transactions of ourlives, they might easily be distinguishedfrom realities. In short, by whatever methodyou distinguish things from chimeras onyour own scheme, the same, it is evident,will hold also upon mine. For it must be, Ipresume, by some perceived difference, andI am not for depriving you of any one thingthat you perceive.

…

Hyl.: What say you to this? Since, according toyou, men judge of the reality of things by

39

their senses, how can a man be mistaken inthinking the moon a plain lucid surface,about a foot in diameter; or a square tower,seen at a distance, round; or an oar, withone end in the water, crooked?

Phil.: He is not mistaken with regard to theideas he actually perceives; but in the infer-ences he makes from his present percep-tions. Thus in the case of the oar, what heimmediately perceives by sight is certainlycrooked; and so far he is in the right. But ifhe thence conclude, that upon taking theoar out of the water he shall perceive thesame crookedness; or that it would affecthis touch, as crooked things are wont to do:in that he is mistaken. In like manner, if heshall conclude from what he perceives inone station, that in case he advances towardthe moon or tower, he should still beaffected with the like ideas, he is mistaken.But his mistake lies not in what he per-ceives immediately and at present, (it beinga manifest contradiction to suppose heshould err in respect of that) but in thewrong judgment he makes concerning theideas he apprehends to be connected withthose immediately perceived: or concern-ing the ideas that, from what he perceives atpresent, he imagines would be perceived inother circumstances. . . .

Hyl.: I have been a long time distrusting mysenses; methought I saw things by a dimlight, and through false glasses. Now theglasses are removed, and a new light breaksin upon my understanding. I am clearlyconvinced that I see things in their nativeforms; and am no longer in pain about theirunknown natures or absolute existence.This is the state I find myself in at present:though indeed the course that brought me toit, I do not yet thoroughly comprehend. Youset out upon the same principles that Aca-demics, Cartesians, and the like sects, usu-ally do; and for a long time it looked as ifyou were advancing their Philosophicalscepticism; but in the end your conclusionsare directly opposite to theirs.

Phil.: You see, Hylas, the water of yonder foun-tain, how it is forced upwards, in a roundcolumn, to a certain height; at which itbreaks and falls back into the basin from

4039

If the reality of things con-sists only of ideas in minds,

how can we tell the differencebetween real things (those thatexist in God’s mind and are causedby God to exist in our minds) andvarious sorts of illusion? Philo-nous’s response is that any crite-rion for this distinction thatappeals to experience, that is toideas of perception, will be equallyavailable to him.

RR

40

Here is Berkeley’s account ofperceptual error: the ideas in

question are not mistaken inthemselves; rather, the mistake liesin the inferences we draw fromthem as to other ideas that are orcould be perceived.

RR

THOMAS REID: DIRECT REALISM, FROM ESSAYS ON THE INTELLECTUAL POWERS OF MAN 93


1. Some of Berkeley’s arguments rely on the claimthat certain of our sensory ideas are indistinguish-able from ideas of pleasure or pain. Is he rightabout this claim?

2. Spell out the argument from illusion for the caseof the two buckets of water and then for the caseof colors seen with the naked eye and through amicroscope. (Assuming that you cannot look atsomething simultaneously in both ways, the sec-ond argument will have to be more complicated,involving the claim that the object does notchange when you shift from one view to theother). In each case, start with an argument forthe conclusion that at least one of the perceivedqualities is not in the object, and then argue forthe claim that none of them is. (See Annotation18 for a suggestion about this.) How compellingare these arguments? Can you think of an answerto them?

3. Locke invoked the idea of substance because itseemed to him impossible that qualities couldsimply exist on their own, even in groups, withoutbeing qualities of anything. Consider Berkeley’scriticisms of this view, centering around the claimthat we have no intelligible idea of such an under-lying thing. Do these objections also apply toBerkeley’s idea of a mind that perceives ideas?Why or why not? If material objects do after allexist outside the mind, must there also be mate-rial substance?

4. It is a fairly common experience for a given speed,for example while riding in a car, to seem faster orslower depending on how fast one was travelingjust before: 30 mph seems slow after going 60, butfast after sitting still. Construct on this basis anargument that is parallel to the other versions ofthe argument from illusion that Berkeley formu-lates and that leads to the conclusion thatapparent speed (speed as directly perceived orexperienced) exists only in the mind.

5. Both Locke and Berkeley tend to think of ideas asmental images and of representation of things byideas as depending on resemblance between thetwo. Are all ideas images? (See Discussion Ques-tion 1 for the Locke selection.) Do we have imagesof our own minds or selves? If a person does havea mental image, is it automatically a representationof any object that the image resembles (assumingthat such resemblance makes sense)?

6. Locke and Berkeley both appeal to roughly thesame features of our perceptual ideas: their invol-untary character and their systematic regularityand order. Locke can be viewed as arguing thatthese features are best explained by supposingthat ideas are systematically caused by a world ofexternal objects; while Berkeley argues that theyare best explained by supposing that the ideas arecaused in us by an infinite mind or spirit.Whoseexplanation do you think is better? How wouldyou argue for one as against the other?

Thomas ReidThe Scottish philosopher Thomas Reid (1710–1796) was one of the foundersof what became known as the “common sense” school of philosophy. He isbest known for his contributions to epistemology, which are mainly focused ondefending the common-sense view of our knowledge of the material worldagainst what he sees as the skeptical tendencies to be found in his British pred-ecessors, Locke, Berkeley, and Hume, as well as in Descartes and others.

In the following selection, Reid attacks the Lockean view that we perceiveexternal material objects by perceiving in a more immediate way ideas thatrepresent them. His alternative is an early version of direct realism, accordingto which material objects are themselves the most direct objects of our percep-tions. As Reid recognizes, the view that perception is in this way direct does notin itself answer the question of what reason or ground we have for thinking thatour direct perceptual judgments are true. Thus he also argues that the claimthat the “evidence of the senses” is good evidence for the claims we accept onthat basis has the status of an axiom or “first principle”: something that isimmediately known without any need for further justification.

whence it rose: its ascent as well asdescent, proceeding from the same uni-form law or principle of gravitation. Just

so, the same principles which at first viewlead to scepticism, pursued to a certainpoint, bring men back to common sense.


1

This is Reid’s basic accountof what happens when a

person perceives an externalmaterial object. Notice, however,that it so far says nothing at allabout whether and how theresulting belief is justified, that is,whether and how the person has aground or reason for thinking thatit is true.

Direct Realism, from Essays on the IntellectualPowers of Man

From Essays on the Intellectual Powers of Man(Cambridge, Mass.: MIT Press, 1969).

Of Perception…

If . . . we attend to that act of our mind which wecall the perception of an external object of sense,we shall find in it these three things. First, Someconception or notion of the object perceived.Secondly, A strong and irresistible convictionand belief of its present existence. And, thirdly,That this conviction and belief are immediate,and not the effect of reasoning.

1st, It is impossible to perceive an objectwithout having some notion or conception ofthat which we perceive. We may indeed con-ceive an object which we do not perceive; butwhen we perceive the object, we must havesome conception of it at the same time; and wehave commonly a more clear and steady notionof the object while we perceive it, than we havefrom memory or imagination when it is notperceived. . . .

2dly, In perception we not only have a notionmore or less distinct of the object perceived, butalso an irresistible conviction and belief of itsexistence. This is always the case when we arecertain that we perceive it. There may be a per-ception so faint and indistinct, as to leave us indoubt whether we perceive the object or not.Thus, when a star begins to twinkle as the lightof the sun withdraws, one may, for a short time,think he sees it, without being certain, until theperception acquires some strength and steadi-ness. When a ship just begins to appear in theutmost verge of the horizon, we may at first bedubious whether we perceive it or not: but whenthe perception is in any degree clear and steady,there remains no doubt of its reality; and whenthe reality of the perception is ascertained, theexistence of the object perceived can no longerbe doubted.

…

Can any stronger proof be given, that it isthe universal judgment of mankind that the

1

evidence of sense is a kind of evidence whichwe may securely rest upon in the most momen-tous concerns of mankind; that it is a kind ofevidence against which we ought not to admitany reasoning; and therefore, that to reasoneither for or against it, is an insult to commonsense?

The whole conduct of mankind, in the dailyoccurrences of life, as well as the solemn proce-dure of judicatories in the trial of causes, civiland criminal, demonstrates this. I know only oftwo exceptions that may be offered against thisbeing the universal belief of mankind.

The first exception is that of some lunatics,who have been persuaded of things that seem tocontradict the clear testimony of their senses. Itis said there have been lunatics and hypochon-driacal persons, who seriously believed them-selves to be made of glass; and, in consequenceof this, lived in continual terror of having theirbrittle frame shivered into pieces.

…

The other exception that may be made tothe principle we have laid down, is that of somephilosophers who have maintained, that thetestimony of sense is fallacious, and thereforeought never to be trusted. Perhaps it might be asufficient answer to this to say, that there isnothing so absurd which some philosophershave not maintained. It is one thing to professa doctrine of this kind, another seriously tobelieve it, and to be governed by it in the con-duct of life. It is evident, that a man who didnot believe his senses, could not keep out ofharm’s way an hour of his life; yet, in all thehistory of philosophy, we never read of anyskeptic that ever stepped into fire or waterbecause he did not believe his senses, or thatshowed, in the conduct of life, less trust in hissenses than other men have. This gives us justground to apprehend, that philosophy wasnever able to conquer that natural belief whichmen have in their senses; and that all their sub-tile reasonings against this belief were neverable to persuade themselves.

It appears, therefore, that the clear and dis-tinct testimony of our senses carries irresistible

2

2

There are two importantlydifferent claims being made

here: first, that skeptical doubtsabout perceptual knowledge shouldnot be taken seriously because theygo against “the universal judgmentof mankind”; and, second, that thisuniversal judgment shows that “theevidence of sense” does notdepend on or require any sort ofreasoning in its support.


conviction along with it, to every man in hisright judgment.

I observed, 3dly, That this conviction is notonly irresistible, but it is immediate; that is, it isnot by a train of reasoning and argumentationthat we come to be convinced of the existence ofwhat we perceive; we ask no argument for theexistence of the object, but that we perceive it;perception commands our belief upon its ownauthority, and disdains to rest its authority uponany reasoning whatsoever.

The conviction of a truth may be irresistible,and yet not immediate. Thus, my conviction thatthe three angles of every plain triangle, areequal to two right angles, is irresistible, but it isnot immediate: I am convinced of it by demon-strative reasoning. There are other truths inmathematics of which we have not only anirresistible, but an immediate conviction. Suchare the axioms. Our belief of the axioms inmathematics is not grounded upon argument.Arguments are grounded upon them, but theirevidence is discerned immediately by thehuman understanding.

It is, no doubt, one thing to have an immedi-ate conviction of a self evident axiom; it isanother thing to have an immediate convictionof the existence of what we see: but the convic-tion is equally immediate and equally irre-sistible in both cases. No man thinks of seekinga reason to believe what he sees; and, before weare capable of reasoning, we put no less confi-dence in our senses than after. The rudest savageis as fully convinced of what he sees, and hears,and feels, as the most expert logician. . . .

Of the Sentiments of Mr. Locke[Locke] sets out in his Essay with a full convic-tion, common to him with other philosophers,that ideas in the mind are the objects of all ourthoughts in every operation of the understand-ing. This leads him to use the word idea sovery frequently, beyond what was usual in theEnglish language, that he thought it necessary inhis introduction to make [an] apology. . . .

Speaking of the reality of our knowledge, hesays, “It is evident the mind knows not thingsimmediately, but only by the intervention of theideas it has of them. Our knowledge therefore isreal, only so far as there is a conformity betweenour ideas and the reality of things. But what shall

3

be here the criterion? How shall the mind, whenit perceives nothing but its own ideas, know thatthey agree with things themselves? . . .”

We see that Mr. Locke was aware no less thanDes Cartes, that the doctrine of ideas made itnecessary, and at the same time difficult, toprove the existence of a material world withoutus; because the mind, according to that doctrine,perceives nothing but a world of ideas in itself.Not only Des Cartes, but Malebranche, Arnauld,and Norris, had perceived this difficulty, andattempted to remove it with little success.Mr. Locke attempts the same thing; but his argu-ments are feeble. He even seems to be consciousof this: for he concludes his reasoning with thisobservation, “That we have evidence sufficientto direct us in attaining the good and avoidingthe evil, caused by external objects, and that thisis the important concern we have in being madeacquainted with them.” This indeed is saying nomore than will be granted by those who deny theexistence of a material world.

As there is no material difference betweenLocke and Des Cartes with regard to the percep-tion of objects by the senses, there is the lessoccasion, in this place, to take notice of all theirdifferences in other points. . . .

Although no author has more merit thanMr. Locke in pointing out the ambiguity of words,and resolving by that means many knotty ques-tions, which had tortured the wits of the school-men; yet, I apprehend he has been sometimesmisled by the ambiguity of the word idea, whichhe uses so often almost in every page of his Essay.

In the explication given of this word, we tooknotice of two meanings given to it; a popular anda philosophical. In the popular meaning, to havean idea of any thing, signifies nothing more thanto think of it.

. . . When the word idea is taken in this pop-ular sense, there can be no doubt of our havingideas in our minds. To think without ideas wouldbe to think without thought, which is a manifestcontradiction.

But there is another meaning of the word ideapeculiar to philosophers, and grounded upon aphilosophical theory, which the vulgar neverthink of. Philosophers, ancient and modern,have maintained, that the operations of the mind,like the tools of an artificer, can only beemployed upon objects that are present inthe mind, or in the brain, where the mind is

4

3

Reid is clearly right thatperceptual judgments are

not ordinarily arrived at throughan explicit process of reasoning.Does it follow that no reasoning isinvolved, even implicitly, or that noreasoning is required to show thatthe resulting belief is indeed likelyto be true?

4

Thus in one sense, to have anidea of something is just to

think of it, without any implicationthat the idea is a distinct entity thathas to be perceived on its own.

RR


supposed to reside. Therefore, objects that aredistant, in time or place, must have a representa-tive in the mind, or in the brain; some image orpicture of them, which is the object that themind contemplates. . . . Since the time ofDes Cartes, [this representative image] has . . .been called an idea; and every thought is con-ceived to have an idea for its object. As this hasbeen a common opinion among philosophers, asfar back as we can trace philosophy, it is the lessto be wondered at, that they should be apt toconfound the operation of the mind in thinking,with the idea or object of thought, which is sup-posed to be its inseparable concomitant.

…

From this, I think it is evident, that if the wordidea in a work where it occurs in every para-graph, be used without any intimation of theambiguity of the word, sometimes to signifythought, or the operation of the mind in think-ing, sometimes to signify those internal objectsof thought which philosophers suppose, thismust occasion confusion in the thoughts both ofthe author and of the readers. I take this to bethe greatest blemish in the Essay on HumanUnderstanding, I apprehend this is the truesource of several paradoxical opinions in thatexcellent work, which I shall have occasion totake notice of.

Here it is very natural to ask, Whether it wasMr. Locke’s opinion, that ideas are the onlyobjects of thought? or, Whether it is not possi-ble for men to think of things which are notideas in the mind?

To this question it is not easy to give a directanswer. On the one hand, he says often, in dis-tinct and studied expressions, that the termidea stands for whatever is the object of theunderstanding when a man thinks, or whateverit is which the mind can be employed about inthinking: that the mind perceives nothing butits own ideas: that all knowledge consists in theperception of the agreement or disagreementof our ideas: that we can have no knowledgefurther than we have ideas. These, and manyother expressions of the like import, evidentlyimply, that every object of thought must be anidea, and can be nothing else.

On the other hand, I am persuaded thatMr. Locke would have acknowledged, that wemay think of Alexander the Great, or of the

5

planet Jupiter, and of numberless things, whichhe would have owned are not ideas in the mind,but objects which exist independent of the mindthat thinks of them.

How shall we reconcile the two parts of thisapparent contradiction? All I am able to sayupon Mr. Locke’s principles to reconcile them,is this, That we cannot think of Alexander, or ofthe planet Jupiter, unless we have in our mindsan idea, that is, an image or picture of thoseobjects. The idea of Alexander is an image, orpicture, or representation of that hero in mymind; and this idea is the immediate object ofmy thought when I think of Alexander. That thiswas Locke’s opinion, and that it has been gener-ally the opinion of philosophers, there can be nodoubt.

But, instead of giving light to the question pro-posed, it seems to involve it in greater darkness.

When I think of Alexander, I am told there isan image or idea of Alexander in my mind,which is the immediate object of this thought.The necessary consequence of this seems to be,that there are two objects of this thought; theidea, which is in the mind, and the person repre-sented by that idea; the first, the immediateobject of the thought, the last, the object of thesame thought, but not the immediate object.This is a hard saying; for it makes every thoughtof things external to have a double object. Everyman is conscious of his thoughts, and yet, uponattentive reflection, he perceives no such duplic-ity in the object he thinks about. Sometimes mensee objects double, but they always know whenthey do so: and I know of no philosopher whohas expressly owned this duplicity in the objectof thought, though it follows necessarily frommaintaining, that, in the same thought, there isone object that is immediate and in the minditself; and another object, which is not immedi-ate, and which is not in the mind.

…

I apprehend, therefore, that if philosopherswill maintain, that ideas in the mind are the onlyimmediate objects of thought, they will beforced to grant that they are the sole objects ofthought, and that it is impossible for men tothink of any thing else. Yet, surely Mr. Lockebelieved that we can think of many things thatare not ideas in the mind; but he seems not tohave perceived, that the maintaining that ideas in

7

6

5

In contrast to the view inthe previous paragraph, the

Lockean view treats ideas asmental entities that are perceivedon their own and are (somehow)the basis for the perception ofthings outside the mind.

RR

7

This suggests at least that thetwo objects are not per-

ceived in the same way—that inordinary cases the focus of aware-ness is on the external object, withthe idea (if there is one) serving asthe way in which the mind is ableto access that external object.

6

Notice that Locke is therebyattempting to offer an

account of how the mind perceivessomething outside of it. Reid, incontrast, makes no attempt to offerany such account.


the mind are the only immediate objects ofthought, must necessarily draw this conse-quence along with it.

The consequence, however, was seen bybishop Berkeley and Mr. Hume, who ratherchose to admit the consequence than to give upthe principle from which it follows.

…

Reflections on the CommonTheory of Ideas

After so long a detail of the sentiments ofphilosophers, ancient and modern, concerningideas, it may seem presumptuous to call inquestion their existence. But no philosophicalopinion, however ancient, however generallyreceived, ought to rest upon authority.There is nopresumption in requiring evidence for it, or inregulating our belief by the evidence we can find.

To prevent mistakes, the reader must again bereminded, that if by ideas are meant only theacts or operations of our minds in perceiving,remembering, or imagining objects, I am farfrom calling in question the existence of thoseacts; we are conscious of them every day, andevery hour of life; and I believe no man of asound mind ever doubted of the real existence ofthe operations of mind, of which he is con-scious. Nor is it to be doubted, that by the facul-ties which God has given us, we can conceivethings that are absent, as well as perceive thosethat are within the reach of our senses; and thatsuch conceptions may be more or less distinct,and more or less lively and strong. . . . The ideas,of whose existence I require the proof, are notthe operations of any mind, but supposedobjects of those operations. They are not percep-tion, remembrance, or conception, but thingsthat are said to be perceived, or remembered, orimagined.

Nor do I dispute the existence of what the vul-gar call the objects of perception. These, by allwho acknowledge their existence, are calledreal things, not ideas. But philosophers main-tain, that, besides these, there are immediateobjects of perception in the mind itself: that, forinstance, we do not see the sun immediately, butan idea. . . . This idea is said to be the image, theresemblance, the representative of the sun, ifthere be a sun. It is from the existence of the9

8

idea that we must infer the existence of the sun.But the idea being immediately perceived, therecan be no doubt, as philosophers think, of itsexistence.

…

The first reflection I would make on this philo-sophical opinion is, that it is directly contrary tothe universal sense of men who have not beeninstructed in philosophy. When we see the sun ormoon, we have no doubt that the very objectswhich we immediately see, are very far distantfrom us, and from one another. We have not theleast doubt, that this is the sun and moon whichGod created some thousands of years ago, andwhich have continued to perform their revolu-tions in the heavens ever since. But how are weastonished when the philosopher informs us, thatwe are mistaken in all this; that the sun and moonwhich we see, are not, as we imagine, many milesdistant from us, and from each other, but thatthey are in our own mind; that they had no exis-tence before we saw them, and will have nonewhen we cease to perceive and to think of them;because the objects we perceive are only ideas inour own minds, which can have no existence amoment longer than we think of them.

If a plain man, uninstructed in philosophy, hasfaith to receive these mysteries, how great mustbe his astonishment. He is brought into a newworld, where every thing he sees, tastes, ortouches, is an idea; a fleeting kind of beingwhich he can conjure into existence, or canannihilate in the twinkling of an eye.

After his mind is somewhat composed, it willbe natural for him to ask his philosophicalinstructor, pray, sir, are there then no substantialand permanent beings called the sun and moon,which continue to exist whether we think ofthem or not?

Here the philosophers differ. Mr. Locke, andthose that were before him, will answer to thisquestion, that it is very true, there are substantialand permanent beings called the sun and moon;but they never appear to us in their own person,but by their representatives, the ideas in our ownminds, and we know nothing of them but whatwe can gather from those ideas.

Bishop Berkeley and Mr. Hume, would give adifferent answer to the question proposed. Theywould assure the querist, that it is a vulgar error,

11

10

9

The most crucial claim hereis that ideas represent exter-

nal objects.That they do it byresembling them or that they aremental images are further claimsmade in an effort to explain howthe relation of representationworks. (Thus it would be at leastpossible to reject these latterclaims and the account of repre-sentation they embody, while stillholding onto the claim ofrepresentation itself—though thensome other account of represen-tation would be needed.)

10

Here Reid seriously misrep-resents the view he is

criticizing. Locke does not say thatthe sun and moon are in ourminds or cease to exist when weare not thinking of them. Andwhile Berkeley says the first ofthese things, he would not agreeto the second (think about why).

11

This is a more accuratestatement of Locke’s view.What is Reid’s alternative?Do the sun and moon

“appear to us in their own person,”and if so, how do they do this—given that Reid too holds that theyexist outside the mind? Can themind think about somethingoutside it without there beingsome internal feature—whetheror not it is an entity like an idea—that represents the thing inquestion?

RR

8

Berkeley’s view was, ofcourse, idealism. (Hume

holds a somewhat similar view.)


a mere prejudice of the ignorant and unlearned,to think that there are any permanent andsubstantial beings called the sun and moon; thatthe heavenly bodies, our own bodies, and allbodies whatsoever, are nothing but ideas in ourminds; and that there can be nothing like theideas of one mind, but the ideas of another mind.There is nothing in nature but minds and ideas,says the bishop. . . .

In this representation of the theory of ideas,there is nothing exaggerated or misrepresented,as far as I am able to judge; and surely nothingfurther is necessary to show, that, to the unin-structed in philosophy, it must appear extravagantand visionary, and most contrary to the dictates ofcommon understanding.

…

A second reflection upon this subject is, thatthe authors who have treated of ideas, havegenerally taken their existence for granted, as athing that could not be called in question;and such arguments as they have mentionedincidentally, in order to prove it, seem too weakto support the conclusion.

Mr. Locke, in the introduction to his Essay,tells us, that he uses the word idea to signifywhatever is the immediate object of thought;and then adds, “I presume it will be easilygranted me that there are such ideas in men’sminds; every one is conscious of them in him-self, and men’s words and actions will satisfyhim that they are in others.” I am indeed con-scious of perceiving, remembering, imagining;but that the objects of these operations areimages in my mind I am not conscious. I am sat-isfied by men’s words and actions, that theyoften perceive the same objects which I per-ceive, which could not be, if those objects wereideas in their own minds.

…

There remains only one other argument that Ihave been able to find urged against our perceiv-ing external objects immediately. It is proposed byMr. Hume, who, in the Essay already quoted, afteracknowledging that it is an universal and primaryopinion of all men, that we perceive externalobjects immediately, subjoins what follows:

“But this universal and primary opinion of allmen is soon destroyed by the slightest philoso-phy, which teaches us, that nothing can ever be

12

present to the mind but an image or perception;and that the senses are only the inlets throughwhich these images are received, without beingever able to produce any immediate intercoursebetween the mind and the object. The table,which we see, seems to diminish as we removefurther from it; but the real table, whichexists independent of us, suffers no alteration. Itwas therefore nothing but its image which waspresent to the mind. These are the obviousdictates of reason. . . .”

…

The passage now quoted is all I have found inMr. Hume’s writings upon this point; andindeed there is more reasoning in it than I havefound in any other author; I shall thereforeexamine it minutely. . . .

To judge of the strength of this argument, it isnecessary to attend to . . . the distinction betweenreal and apparent magnitude. The real magnitudeof a line is measured by some known measure oflength, as inches, feet, or miles. . . . .

Apparent magnitude is measured by the anglewhich an object subtends at the eye. . . .

From what has been said, it is evident that thereal magnitude of a body must continueunchanged, while the body is unchanged. Thiswe grant. But is it likewise evident, that theapparent magnitude must continue the samewhile the body is unchanged? So far otherwise,that every man who knows any thing of mathe-matics can easily demonstrate, that the sameindividual object, remaining in the same place,and unchanged, must necessarily vary in itsapparent magnitude, according as the pointfrom which it is seen is more or less distant;and that its apparent length or breadth will benearly in a reciprocal proportion to the distanceof the spectator. This is as certain as the princi-ples of geometry.

…

If these things be considered, it will appear,that Mr. Hume’s argument has no force tosupport his conclusion, nay, that it leads to acontrary conclusion. The argument is this. Thetable we see, seems to diminish as we removefurther from it; that is, its apparent magnitude isdiminished; but the real table suffers no alter-ation, to wit, in its real magnitude; therefore itis not the real table we see. I admit both the

13

12

Locke would not, of course,deny that different people

can perceive the same externalobject, though he would insist thatthey do so by each of them havingtheir own separate ideas of it.

(Again, what is Reid’s alterna-tive? How do different people

perceive the same object?)

13

Though Reid is focusing onHume, this is the same basic

line of argument, the argumentfrom illusion (or perceptual relativ-ity), that appears many times inBerkeley’s Three Dialogues.


premises in this syllogism, but I deny theconclusion. . . .

Let us suppose, for a moment, that it is the realtable we see. Must not this real table seem todiminish as we remove further from it? It isdemonstrable that it must. How then can thisapparent diminution be an argument that it is notthe real table? When that which must happen tothe real table, as we remove further from it, doesactually happen to the table we see, it is absurd toconclude from this, that it is not the real table wesee. It is evident, therefore, that this ingeniousauthor has imposed upon himself, by confound-ing real magnitude with apparent magnitude, andthat his argument is a mere sophism.

…

The last reflection I shall make upon thistheory is, that the natural and necessary conse-quences of it furnish a just prejudice against itto every man who pays a due regard to thecommon sense of mankind.

…

It was this theory or means that led DesCartes, and those that followed him, to think itnecessary to prove, by philosophical arguments,the existence of material objects. And who doesnot see that philosophy must make a veryridiculous figure in the eyes of sensible men,while it is employed in mustering up metaphys-ical arguments, to prove that there is a sun and amoon, an earth and a sea?Yet we find these trulygreat men, Des Cartes, Malebranche, Arnauld,and Locke, seriously employing themselves inthis argument.

Surely their principles led them to think, thatall men, from the beginning of the world,believed the existence of these things uponinsufficient grounds, and to think that theywould be able to place upon a more rationalfoundation this universal belief of mankind. Butthe misfortune is, that all the laboured argumentsthey have advanced, to prove the existence ofthose things we see and feel, are mere sophisms.Not one of them will bear examination.

…

However, all these consequences of the doc-trine of ideas were tolerable, compared withthose which came afterward to be discovered byBerkeley and Hume: That there is no material

14

world: No abstract ideas or notions: That themind is only a train of related impressions andideas, without any subject on which they may beimpressed: That there is neither space nortime, body nor mind, but impressions and ideasonly: And, to sum up all, That there is no proba-bility, even in demonstration itself, nor any oneproposition more probable than its contrary.

These are the noble fruits which have grownupon this theory of ideas, since it began to becultivated by skilful hands. It is no wonder thatsensible men should be disgusted at philosophy,when such wild and shocking paradoxes passunder its name. However, as these paradoxeshave, with great acuteness and ingenuity, beendeduced by just reasoning from the theory ofideas, they must at last bring this advantage, thatpositions so shocking to the common sense ofmankind, and so contrary to the decisions of allour intellectual powers, will open men’s eyes,and break the force of the prejudice which hasheld them entangled in that theory.

…

Of the Evidence of Sense, andof Belief in General

The intention of nature in the powers which wecall the external senses, is evident. They areintended to give us that information of externalobjects which the Supreme Being saw to beproper for us in our present state; and they giveto all mankind the information necessary forlife, without reasoning, without any art or inves-tigation on our part.

The most uninstructed peasant has as distincta conception, and as firm a belief of the imme-diate objects of his senses, as the greatestphilosopher; and with this he rests satisfied,giving himself no concern how he came by thisconception and belief. But the philosopher isimpatient to know how his conception of exter-nal objects, and his belief of their existence, isproduced. This, I am afraid, is hid in impenetra-ble darkness. But where there is no knowledge,there is the more room for conjecture: and ofthis philosophers have always been very liberal.

…

We give the name of evidence to whatever is aground of belief. To believe without evidence is

15

14

But is this an adequateresponse to the argument?

What is it for a person to be awareof the apparent magnitude of thetable and for that to diminish whilethe table does not change? Berke-ley (and Hume) would say that ourawareness of this apparent magni-tude involves having something likea mental image that changes (in thesense that the table occupies asmaller part of it) even while thetable undergoes no change, so thatit is this image (and idea) that weare most immediately aware of.Does Reid offer any clear alterna-tive to this view?

15

Is Reid right that the viewthat the existence of the

material world needs to be provedand the skepticism that resultsfrom failed attempts to do soresults only from the theory ofideas? Does rejecting that theorypreclude raising essentially thesame problem—and if so, why?


a weakness which every man is concerned toavoid, and which every man wishes to avoid.Nor is it in a man’s power to believe any thinglonger than he thinks he has evidence.

…

The common occasions of life lead us todistinguish evidence into different kinds, towhich we give names that are well understood;such as the evidence of sense, the evidence ofmemory, the evidence of consciousness, theevidence of testimony, the evidence of axioms,the evidence of reasoning. . . .

I confess that, although I have, as I think, adistinct notion of the different kinds of evidenceabove mentioned, and perhaps of some others,which it is unnecessary here to enumerate, yet Iam not able to find any common nature to whichthey may all be reduced. They seem to me toagree only in this, that they are all fitted bynature to produce belief in the human mind;some of them in the highest degree, which wecall certainty, others in various degrees accord-ing to circumstances.

I shall take it for granted, that the evidence ofsense, when the proper circumstances concur, isgood evidence, and a just ground of belief.My intention in this place is only to compare itwith the other kinds that have been mentioned,that we may judge whether it be reducible to anyof them, or of a nature peculiar to itself.

1st, It seems to be quite different from theevidence of reasoning. All good evidence iscommonly called reasonable evidence, and veryjustly, because it ought to govern our belief asreasonable creatures. And, according to thismeaning, I think the evidence of sense no lessreasonable than that of demonstration. . . .

But when we speak of the evidence of reason-ing as a particular kind of evidence, it means theevidence of propositions that are inferred byreasoning, from propositions already knownand believed. . . .

That the evidence of sense is of a differentkind, needs little proof. No man seeks a reasonfor believing what he sees or feels; and if he did,it would be difficult to find one. But though hecan give no reason for believing his senses, hisbelief remains as firm as if it were grounded ondemonstration.

Many eminent philosophers, thinking itunreasonable to believe, when they could not

17

16

show a reason, have laboured to furnish us withreasons for believing our senses; but theirreasons are very insufficient, and will not bearexamination. Other philosophers haveshown very clearly the fallacy of these reasons,and have, as they imagine, discovered invinciblereasons against this belief; but they have neverbeen able either to shake it in themselves, or toconvince others . . . a man may as soon, byreasoning, pull the moon out of her orbit, asdestroy the belief of the objects of sense.

. . . when I . . . see an object before my eyes,this commands my belief no less than an axiom.But when, as a philosopher, I reflect upon thisbelief, and want to trace it to its origin, I am notable to resolve it into necessary and self-evidentaxioms, or conclusions that are necessarilyconsequent upon them. I seem to want thatevidence which I can best comprehend, andwhich gives perfect satisfaction to an inquisitivemind; yet it is ridiculous to doubt, and I find itis not in my power. An attempt to throw off thisbelief, is like an attempt to fly, equally ridicu-lous and impracticable.

To a philosopher, who has been accustomedto think that the treasure of his knowledge is theacquisition of that reasoning power of which heboasts, it is no doubt humiliating to find, that hisreason can lay no claim to the greater part of it.

By his reason, he can discover certain abstractand necessary relations of things; but his knowl-edge of what really exists, or did exist, comes byanother channel, which is open to those whocannot reason. He is led to it in the dark, andknows not how he came by it.

It is no wonder that the pride of philosophyshould lead some to invent vain theories, inorder to account for this knowledge; and otherswho see this to be impracticable, to spurn at aknowledge they cannot account for, and vainlyattempt to throw it off, as a reproach to theirunderstanding. But the wise and the humble willreceive it as the gift of Heaven, and endeavour tomake the best use of it.

…

Of First Principles in General. . . there are . . . propositions which are nosooner understood than they are believed. Thejudgment follows the apprehension of themnecessarily, and both are equally the work of

19

18

18

Here Reid has Descartes andLocke mainly in mind.

16

This last part is highly ques-tionable. But the main prob-

lem here is that what is thought tobe good evidence need not actuallybe good evidence.

17

This is in keeping with Reid’scommon-sense outlook.But even if it is acceptableto take this for granted, it

can still be asked why and howthis is so.

RR

19

But doubt isn’t really theissue. Even if Reid is right that

doubt is “ridiculous,” a philosophercan (and seemingly should) askwhat makes beliefs of this sortacceptable.


nature, and the result of our original powers.There is no searching for evidence; no weighingof arguments; the proposition is not deduced orinferred from another; it has the light of truth initself, and has no occasion to borrow it fromanother.

Propositions of the last kind . . . are calledfirst principles, principles of common sense,common notions, self-evident truths. . . .

I take it for granted, that there are self-evidentprinciples. Nobody, I think, denies it. And if anyman were so skeptical as to deny that there is anyproposition that is self-evident, I see not how itwould be possible to convince him by reasoning.

But yet there seems to be great difference ofopinions among philosophers about first princi-ples. What one takes to be self-evident, anotherlabours to prove by arguments, and a thirddenies altogether.

Thus, before the time of Des Cartes, it wastaken for a first principle, that there is a sun anda moon, an earth and sea, which really exist,whether we think of them or not. Des Cartesthought that the existence of those things oughtto be proved by argument; and in this he hasbeen followed by Malebranche, Arnauld, andLocke. They have all laboured to prove, by veryweak reasoning, the existence of externalobjects of sense; and Berkeley, and Hume, sen-sible of the weakness of their arguments, havebeen led to deny their existence altogether.

…

The First Principles of Contingent Truths

…

As the minds of men are occupied much moreabout truths that are contingent than about thosethat are necessary, I shall first endeavour topoint out the principles of the former kind.

1st. First, Then, I hold, as a first principle, theexistence of every thing of which I am conscious.

…

When a man is conscious of pain, he is certainof its existence; when he is conscious that hedoubts, or believes, he is certain of the existenceof those operations.

But the irresistible conviction he has of thereality of those operations is not the effect of

20

reasoning; it is immediate and intuitive. Theexistence therefore of those passions and opera-tions of our minds, of which we are conscious,is a first principle, which nature requires us tobelieve upon her authority.

If I am asked to prove that I cannot bedeceived by consciousness, to prove that it is nota fallacious sense; I can find no proof. I cannotfind any antecedent truth from which it isdeduced, or upon which its evidence depends. Itseems to disdain any such derived authority, andto claim my assent in its own right.

If any man could be found so frantic as todeny that he thinks, while he is conscious of it;I may wonder, I may laugh, or I may pity him,but I cannot reason the matter with him. Wehave no common principles from which we mayreason, and therefore can never join issue in anargument.

…

3dly, Another first principle I take to be, Thatthose things did really happen which Idistinctly remember.

This has one of the surest marks of afirst principle; for no man ever pretended toprove it, and yet no man in his wits calls it inquestion; the testimony of memory, like that ofconsciousness, is immediate; it claims ourassent upon its own authority.

…

Indeed the theory concerning ideas, so gener-ally received by philosophers, destroys all theauthority of memory, as well as the authority ofthe senses. Des Cartes, Malebranche, andLocke, were aware that this theory made itnecessary for them to find out arguments toprove the existence of external objects, whichthe vulgar believe upon the bare authority oftheir senses; but those philosophers were notaware, that this theory made it equally necessaryfor them to find arguments to prove the exis-tence of things past, which we remember, and tosupport the authority of memory.

All the arguments they advanced to supportthe authority of our senses, were easily refutedby bishop Berkeley and Mr. Hume, being indeedvery weak and inconclusive. And it would havebeen as easy to answer every argument theycould have brought, consistent with their theory,to support the authority of memory. 22

21 21

Here Reid has in mind con-sciousness of mental entities

and processes, not consciousness ofexternal objects. Is he right thatthere is no reason or argument thatsupports the truth of claims resultingfrom such consciousness?

22

Here is Reid’s response toanother very difficult episte-

mological issue, one that there isno further consideration of in thisbook. Reid is suggesting in partthat the problem of memory andthe problem of perception areparallel, and that the futility oftrying to defend perceptual beliefsvia argument can be made evenmore obvious by reflecting on thisparallel case. (See DiscussionQuestion 3.)

20

There are two different fea-tures indicated here, which

should not be run together (asReid at least sometimes seems todo): first, that the proposition isaccepted as soon as it isunderstood, without any searchingfor evidence or weighing ofarguments; and, second, that thereis some sort of ground or basisfor thinking that it is true (it hasthe “light of truth in itself”).


…

5thly, Another first principle is, That thosethings do really exist which we distinctlyperceive by our senses, and are what we per-ceive them to be.

It is too evident to need proof, that all men areby nature led to give implicit faith to the distincttestimony of their senses, long before they arecapable of any bias from prejudices of educationor of philosophy.

…

We have before examined the reasons givenby philosophers, to prove that ideas, and notexternal objects, are the immediate objects ofperception, and the instances given to prove thesenses fallacious. Without repeating what hasbefore been said upon those points, we shallonly here observe, that if external objects beperceived immediately, we have the samereason to believe their existence as philosophershave to believe the existence of ideas, while theyhold them to be the immediate objects ofperception.

…

7thly, Another first principle is, that thenatural faculties, by which we distinguish truthfrom error, are not fallacious. If any man shoulddemand a proof of this, it is impossible to satisfyhim. For suppose it should be mathematicallydemonstrated, this would signify nothing in thiscase; because, to judge of a demonstration, a

24

23

man must trust his faculties, and take forgranted the very thing in question.

If a man’s honesty were called in question, itwould be ridiculous to refer it to the man’s ownword, whether he be honest or not. The sameabsurdity there is in attempting to prove, by anykind of reasoning, probable or demonstrative, thatour reason is not fallacious, since the very point inquestion is, whether reasoning may be trusted.

If a skeptic should build his skepticism uponthis foundation, that all our reasoning, andjudging powers are fallacious in their nature, orshould resolve at least to withhold assent untilit be proved that they are not; it would beimpossible by argument to beat him out of thisstrong hold, and he must even be left to enjoyhis skepticism.

Des Cartes certainly made a false step in thismatter; for having suggested this doubt amongothers, that whatever evidence he might havefrom his consciousness, his senses, his memory,or his reason; yet possibly some malignantbeing had given him those faculties on purposeto impose upon him; and therefore, that they arenot to be trusted without a proper voucher: toremove this doubt, he endeavours to prove thebeing of a Deity who is no deceiver; whence heconcludes, that the faculties he had given himare true and worthy to be trusted.

It is strange that so acute a reasoner did notperceive, that in this reasoning there is evidentlya begging of the question.

…

25

23

This is Reid’s basic responseto the issue of how percep-

tual beliefs are justified: that thethings we perceive (probably?) existand are as we perceive them to beis a first principle. Is this an ade-quate solution to the problem?(See Discussion Question 2.)

25

Here Reid has in mind thepower of reasoning and

rational thought. He points out,correctly, that there is no way toprove the reliability of this basiccapacity, since any proof would relyon the very capacity in questionand so would beg the question.

RR

24

Reid offers a further argu-ment: once it is granted that

external objects are perceivedimmediately, we have as good areason to accept the perceptualbeliefs that result as we wouldhave for beliefs about ideas (if theyexisted)—or, presumably, as we doin the case of beliefs about statesof mind such as sensations. (SeeDiscussion Question 1.)

RR


1. Consider an introspective belief about one ofyour own conscious states of mind: the beliefthat you are experiencing a pain in your wrist,perhaps, or the belief that you are thinkingabout Reid. What sort of reason or basis do youhave for thinking that such beliefs are true? Isthis just something that has to be accepted as a“first principle,” or is there something more tobe said? (If there is a reason of this sort, thenthe next question is whether anything parallelcan be said about perceptual beliefs concerningexternal objects.)

2. Even if external objects are perceived directly inthe sense that there is no intermediate entitythat is perceived more immediately, it can stillapparently be asked whether the specific beliefor conviction that is the content of a perceptual

judgment is true (or likely to be true). Is itenough of a response to this question to say, asReid does, that it is simply a first principle thatbeliefs of this sort are true—especially if it isadmitted, as he does, that such a claim is not self-evident or necessary? Bertrand Russell remarksthat the method of postulation “has all the advan-tages of theft over honest toil,” by which hemeans that simply postulating that something isso is too easy to be satisfying. Does this sameobjection pertain to Reid’s first principles (and ifnot, why not)?

3. Think about the somewhat analogous problem ofhow, if at all, memory beliefs can be justified(where we are using the term “memory belief” torefer to beliefs that seem to be memories,whether they are actually so or not). Reid is right

LAURENCE BONJOUR: KNOWLEDGE OF THE EXTERNAL WORLD, FROM EPISTEMOLOGY: CLASSIC PROBLEMS AND CONTEMPORARY RESPONSES 103

that philosophers who discuss perception at greatlength often say very little about memory, eventhough very similar issues seem to arise there.Think of how solutions parallel to Locke’s and toBerkeley’s would apply to the case of memory, and

consider the problems that arise in each case.Canmemory beliefs be justified in either of these twoways? If not, is a Reidian solution the only oneavailable? And if this is so, does it, by analogy,support Reid’s account of perception?

Knowledge of the External World, from Epistemology:Classic Problems and Contemporary Responses

Laurence BonJour (1943– ) is an American philosopher who teaches at theUniversity of Washington. He has written or co-written several books inepistemology and is also the coeditor of this volume. In this selection,BonJour recapitulates the problem of the external world, as it grows out ofthe work of Descartes, Locke, and Berkeley; then explains and criticizes thephenomenalist response to this problem, growing out of Berkeley’s view (andHume’s); and finally offers a tentative solution along very roughly Lockeanlines. At the end, he also has a bit to say about direct realist views like theone offered by Reid.

Laurence BonJour

We have so far tentatively accepted the conclu-sion that the immediate object of awareness inperceptual experience is never an external mate-rial object, but is instead something of a quitedifferent sort: . . . a sense-datum. . . . It willbe useful to have a brief label for this disjunctiveresult, and I will refer to it here as perceptualsubjectivism . . . .

We have now to consider the implications ofperceptual subjectivism for the epistemologi-cal issue upon which it bears most directly,which is also arguably the most central issue ofthe modern period of epistemology, whichbegan with Descartes: the issue of whetherand, if so, how beliefs concerning the externalmaterial world and the objects that it allegedlycontains can be justified on the basis of ourimmediate sensory experience, thus under-stood . . . we will first look at the views ofDescartes’s immediate successors, the so-called British Empiricists Locke, Berkeley,and Hume, whose arguments played a majorrole in shaping the subsequent discussion. Wewill then examine the two main alternativeaccounts of “knowledge of the external world”

1

(on the assumption that perceptual subjec-tivism or something like it is indeed true) thathave subsequently emerged, mainly in theforms that they have taken in the twentieth cen-tury: phenomenalism and representationalism.Difficulties with these views will then prompt,in the last part of the chapter, a reconsiderationof whether rejecting perceptual subjectivismmight make available a further, more promis-ing alternative.

Locke, Berkeley, and Hume onPerception and the External WorldAs noted earlier, Locke and Berkeley speak notof sense-data . . . but of “ideas” or “ideas ofsense”—with the former term being appliedalso to contents of thought and indeed appar-ently to conscious contents of any kind. The waythat they use these rather slippery termssuggests in many places something like a sense-datum theory of the immediate objects ofsensory experience. For our purposes, however,it will suffice to take the term “idea” merely torefer to conscious contents of any sort, and“ideas of sense” to the distinctive contents ofsensory experience, without supposing theseterms to indicate any definite metaphysical pic-ture of the nature of such contents.

1

Sense-data (singular:sense-datum) are roughly the samethings that Locke and Berkeleyhave in mind when they speak ofperceptual ideas: entities that havethe qualities that are directly orimmediately experienced and thatexist in the mind.

From Epistemology: Classic Problems and Contempo-rary Responses (Lanham, Md.: Rowman & Littlefield,2002). Some footnotes have been deleted.


will usually experience pain along with it; butif I merely imagine or remember such an expe-rience, there is no pain.† Fourth, “our senses,in many cases, bear witness to the truth of eachother’s report, concerning the existence of sen-sible things without us” [p. 72]. (For example,my visual experience of the appearance of afire close to my body is normally accompaniedby tactile experiences of heat, apparent smellsof burning, the apparent hearing of cracklingsor other distinctive firelike sounds, etc.—thinkhere of other examples of your own.) ButLocke has little to say as to just how these“concurrent reasons” are supposed to showthat our beliefs concerning material objectsthat are arrived on the basis of our immediatesensory experiences are justif ied by thoseexperiences. Does such a conclusion really fol-low, and, if so, how and why? . . .

In fact, Locke’s supposed reasons are of veryunequal weight. The first one is totally worth-less, because it begs the very question at issueand also would require a prior solution ofanother, related epistemological problem. Untilthe problem of justifying belief in externalobjects on the basis of his sensory experiencehas been solved, Locke is obviously not in aposition to appeal to supposed facts about otherpeople’s sense organs, since sense organs arephysical structures and so beliefs about themwould have to be justified in just the way that isin question. Moreover, to invoke this firstreason, he would also have to have justifiedbeliefs about the mental states of other people,specifically concerning whether they do or donot have sensory ideas of the relevant sort. Howthis latter sort of knowledge is possible is aserious problem in itself (the “problem ofother minds” . . .) But it is pretty clear on reflec-tion . . . that knowledge of other people’s mentalstates normally depends on prior knowledge ofthe behavior and condition of their physicalbodies, thus again presupposing the very knowl-edge of the material world that has not yet beenaccounted for. . . .

Locke’s second reason is at least a bit better.The involuntary or spontaneous character of mysensory experience does at least distinguish it

2

*Locke, Essay Concerning Human Understanding, Book IV, ch. XI. Subsequent page references in the text are to theselection in the present book.

Locke’s view is clearly that our beliefs oropinions about material objects existing outsideof our minds are justified by our ideas of sense.*But his discussion of this point is both ratheruncertain and quite guarded. He says that ourassurance on this basis concerning materialobjects “deserves the name of knowledge”[p. 71], thus seeming to suggest that it is notknowledge simply and with no qualification. Healso questions whether anyone can be genuinelyskeptical about the existence of the things thathe sees and feels, and speaks rather vaguely of“the assurance we have from our senses them-selves, that they do not err in the informationthey give us” [p. 72].

But the closest that Locke comes to explain-ing how such beliefs are justified by sensoryexperience is his citing of four “concurrentreasons” that are supposed to further confirmthe assurance derived from the senses: First,we can know that sensory ideas are “producedin us by exterior causes” by observing thatthose lacking a particular sense organ cannever have the corresponding sensory ideas[p. 72]. (Thus, for example, a blind man cannever have immediate sensory experiences ofvisual qualities such as color.) Second, anotherreason for thinking that our sensory ideasresult from external causes is their involuntarycharacter, as contrasted with imagination and,to a lesser extent, memory [p. 72]. (Thus if Ihave my eyes open and am facing in a particu-lar direction, I have no choice as to what appar-ent objects or properties I will experience, thatis, in Locke’s view, what ideas of vision I willexperience—as I look out my study window, Icannot help being aware of a mass of varie-gated green and brown that I take to be a per-ception of trees, branches, and leaves.) Third,another difference between our immediate sen-sory experiences and other sorts of ideas, suchas those of imagination and memory, is thatsensory ideas of certain kinds are accompa-nied by pain, whereas the corresponding ideasof imagination and memory are not [p. 72].(For example, if I have the immediate sensoryexperience of apparently hitting my hand witha hammer while attempting to drive a nail, I

2

How do you know of theexistence and specific

character of other people’s mentalstates? What reason do you havefor thinking that they are not, forexample, just cleverly designedrobots that have no consciousmental states at all?(See Discussion Question 1.)

†There may of course be imagined or remembered pain, but that is obviously not the same thing as really experienced pain.


from other sorts of mental states and experience(albeit perhaps not in a completely sharp way—aren’t many memories and even some imagin-ings similarly involuntary?). But this fact doesnot by itself seem to establish that immediatesensory experiences are, as he claims, caused bysomething external to the person who has them.Why couldn’t my involuntary sensory experi-ences result instead from some subconscious orunconscious faculty of my own mind that is out-side my voluntary control? And, even moreobviously, that the ideas are involuntary tells usnothing at all about whether the external cause,if there is one, has the specific properties thatmy sensory experience seems to portray(whether it “resembles my ideas,” as Lockewould put it). Why couldn’t the external causeof my idea of a green tree, again if there is one,neither be green nor have the other properties ofa tree? Indeed, why couldn’t it, as Berkeley willsuggest, be something utterly different from amaterial object? And the third reason, whileagain perhaps showing that sensory experiencesare importantly different from many other men-tal phenomena, also does not support in anyclear way a conclusion about what is responsiblefor this difference.

What about the fourth reason? Surely it is astriking fact that my various sensory experiencesfit together in an extremely orderly and coherentfashion to depict an ongoing world that is bothextremely complicated and highly regular orlaw-governed. The information or apparentinformation derived at a given time from onesense agrees to a very great degree with both thatderived at that time from other senses and alsowith that derived from both the same sense andothers at other times—allowing, of course, forthe ongoing change and development of theworld, which is also something that is reflectedin regularities within our sensory experience.Thus if I seem to see a chair, I can normally alsohave the experience of touching it, given that Ialso have the experience of moving my body inthe right direction and far enough. And the expe-riences that I have of the furniture and contentsof my office before leaving for a class agree verywell with the similar experiences that I have afterI have apparently returned—allowing, in somecases, for the actions of the janitor or my dog(who is sometimes left there) or my wife (whohas a key). . . . 3

But how exactly is this admittedly strikingfact supposed to support Locke’s intendedconclusion, namely that there is good reason orjustification for thinking that the beliefs aboutthe material world that we arrive at on the basisof our immediate sensory experience are likelyto be true? On this obviously crucial question,Locke has very little to say. . . .

In fact, if you think carefully about it, theorder of my immediate sensory experience andthe seeming agreement between experiencesapparently produced by different senses wouldnot be striking, or at least not nearly so strik-ing, if those ideas were under my voluntarycontrol—for then I could deliberately imaginean orderly world, in something like the way inwhich this is done by an artist or novelist.What makes the order so noteworthy is pre-cisely that it is not voluntarily created, but justoccurs spontaneously and, in many of itsdetails, unexpectedly. Thus we see thatLocke’s fourth “concurrent reason” needs tobe supplemented by his second, and that it isthese two together that might provide at leastthe beginnings of a real argument. Experiencethat was involuntary but chaotic would showvery little, and neither would experience thatwas orderly but voluntarily controlled. It isexperience that is both involuntary and highlyorderly that seems to demand some sort of fur-ther explanation: what is it that produces andsustains the order? Thus it is natural to inter-pret Locke as arguing, admittedly without for-mulating the point very clearly or explicitly,that the best explanation of his involuntary butorderly experience is that it is systematicallycaused by a world of independent materialobjects which it depicts with at least approxi-mate accuracy. (The main way in which thedepiction is only approximately accurate isthat, according to Locke, material objects haveonly primary qualities like size, shape, andmotion, but not secondary qualities like color,smell, taste, and temperature (as felt).)

Does this argument really show that ourbeliefs about the material world that are arrivedat on the basis of our involuntary sensory expe-rience are likely to be true and hence justified?It seems reasonable to think that there must besome explanation for these features of our sen-sory ideas, which is just to say that the sort oforder that they exhibit is extremely unlikely to

3

Think of some furtherexamples of order and

agreement between the percep-tions of different senses orbetween the perceptions of onesense at different times, trying tospecify them in detail.


result from mere chance. But is Locke’s pro-posed explanation the right one?

Berkeley, while appealing to essentially thesame features of our sensory ideas (their beingindependent of our will and their being orderlyand coherent), offers a quite different and in hisview superior explanation: that our sensoryideas are produced in our minds by God, whodetermines and controls their orderly character,so that there is thus no need or justification forsupposing that the independent material realmadvocated by Locke really exists. Berkeley’sGod obviously bears a striking resemblance toDescartes’s evil genius, with the crucial differ-ence that whereas Descartes assumes that theevil genius would be deceiving us, Berkeley’sview is in effect that having sensory ideas sys-tematically produced in us by God (presumablyreflecting God’s ideally complete picture of theworld thus depicted) is just what it is for a worldof ordinary objects to exist. Thus we have atleast two competing explanations for the samefacts concerning our sensory experience, andthe question is how we should decide betweensuch explanations.

Assuming, that is, that we can rationallydecide at all. Hume’s response to the problemis to deny that any such attempt to explain ourexperience by appeal to objects or entitiesexisting outside of that experience could everbe justif ied. An essential ingredient of bothLocke’s and Berkeley’s proposed explanationsis the claim that our immediately experiencedsensory ideas (or “impressions,” as Hume callsthem, in order to distinguish them from otherkinds of ideas) are caused by the externalentities that those explanations invoke—bymaterial objects, according to Locke’s explana-tion, and by God, according to Berkeley’s.Moreover, it seems obvious that any similarattempt to explain experience by appeal tosomething existing outside experience (eventhe person’s own unconscious mind) willrequire a similar causal claim (for how elsewould the explanation work?). But, arguesHume, causal relations can be known only byexperiencing the regular sequence of cause andeffect, something that is impossible in the caseof an alleged causal relation between some-thing outside immediate experience and thatexperience itself. In relation to Locke’sexplanation specifically, the point is that I

5

4

cannot immediately experience material bod-ies causing my sensory ideas because I have noimmediate experience of such bodies at all;and the claim that I indirectly perceive materialbodies presupposes for its justif ication anexplanation relying on the very causal relationin question and so cannot be used to establishthat such a causal relation exists.

Hume’s further discussion of the issue of theexternal world is characteristically muddled byhis general tendency to conflate and confuseissues concerning justif ication with issueshaving to do with the psychological causationor genesis of the beliefs in question. . . . But itis nonetheless easy to see how Hume who wasclearer about the distinction between psycho-logical explanation and epistemic justificationmight have argued that the content of ourclaims about material objects, to the extent thatthis is justified, must have to do solely with fea-tures and patterns of our sensory experience,rather than with genuinely mind-transcendentobjects. . . .

Thus we have initial adumbrations of the twomain views that we will now proceed to discussmore systematically. Locke’s view, according towhich our subjective sensory experience and thebeliefs that we adopt on the basis of it constitutea representation of the external material world,one that is caused by that world and that we arejustified in thinking to be at least approximatelyaccurate, is a version of the more generalposition known as representationalism orrepresentative realism. (So also is Descartes’sview.) The second main view, which Hume’s dis-cussion suggests but never quite arrives at, isthat (i) we can have no knowledge (or perhapseven no intelligible conception) of a realm ofexternal causes of our experience, but also (ii)that our beliefs about the material world can stillbe in general justified and true because theircontent in fact has to do only with the featuresand order of our subjective experience. This isthe view that has come to be known asphenomenalism, a version of idealism. . . .Yet a third possibility would be the essentiallyskeptical view that we can know that our experi-ences are externally caused in some way, but canknow nothing further about the nature of thosecauses. Such a skeptical view would, of course,not be a solution to the problem of the externalworld, but rather a confession that there is no

6

4

Can you think of any way torationally choose between

these two explanations? Can youthink of any other possibleexplanations?

5

For Hume’s discussion ofthis, see the selection in the

induction section.Is Hume right that causalrelations are always known

by repeated experiences of thesequence of cause and effect? Thinkhere of various examples of causalrelations that seem to be known.Can you think of any plausiblecausal claims that don’t seem tobe known in this way?

RR

6

The phenomenalist view isat least similar to Berkeley’s viewof material objects, though withoutthe reference to God: for bothviews, material objects exist only aspatterns of ideas or experiencesor sense-data in minds, and have noexistence outside of all minds (thatwould be “mind-transcendent”existence).


solution; it is thus a view to be adopted only afterthe other two possibilities have clearly failed.

Historically, the objections to the representa-tionalism of Descartes and Locke, especially theHumean one discussed above, were widelytaken to be decisive, with positions in the direc-tion of phenomenalism being viewed as themain nonskeptical alternative, especially in thefirst two-thirds or so of the twentieth century.Thus we will begin our more systematic discus-sion with a consideration of phenomenalism,and then return later to the consideration ofrepresentationalism that was begun in thediscussion of Locke.

PhenomenalismAs just briefly formulated, the phenomenalistview is that the content of propositions aboutmaterial objects and the material world is entirelyconcerned with features and relations of theimmediate objects of our perceptual experience,that is, the features and relations of our sense-data. According to the phenomenalist, to believethat a physical or material object of a certain sortexists just is to believe that sense-data of varioussorts have been experienced, are being experi-enced, will be experienced, and/or would beexperienced under certain specifiable conditions.Thus, for example, to believe that there is a largebrown table in a certain room in the University ofWashington Library is to believe, roughly, (i) thatthe sorts of sense-data that seem from a common-sense standpoint to reflect the presence of such atable either have been, are presently, or will in thefuture be experienced in the context of othersense-data, themselves experienced concurrentlyor immediately before or after, that reflect thelocation as the room in question; and in addi-tion—or instead, if the table has never in factbeen perceived and never in fact will beperceived—(ii) that such sense-data wouldbe experienced if other sense-data that reflect theperceiver’s going to the library and to that roomwere experienced. (This is quite a complicatedspecification, and you will have to think verycarefully about what it is saying.)

In a fairly standard formula, to believe thatsuch a material object exists is, according to thephenomenalist, to believe nothing more thanthat sense-data of the appropriate sort are actual(in the past, present, or future) and/or possible—

7

where to say that certain sense-data are possibleis to say, not just that it is logically possible forthem to be experienced (which would appar-ently always be so as long as the description ofthem is not contradictory), but that they wouldin fact actually be experienced under certainspecifiable circumstances (specifiable in sense-datum terms); thus it would be somewhatclearer to speak of actual and obtainable sense-data. The British philosopher John Stuart Millput this point by saying that material objects arenothing but “permanent possibilities of sensa-tion,” that is, of sense-data—where, of course,the possibilities in question are only relativelypermanent, since objects can change or bedestroyed. The crucial thing to see is that whatMill and the other phenomenalists are saying isthat there are no independently existing objectsthat are responsible for the possibilities ofsensation or the obtainability of sense-data; theactuality and obtainability of sense-data are allthere is to the physical or material world.

Phenomenalism is in fact one of those occa-sional (some would say more than occasional)philosophical views that is so monumentallybizarre and implausible, at least from anythingclose to a common-sense standpoint, as to per-haps make it difficult . . . to believe that it reallysays what it does—and even more difficult tobelieve that such a view has in fact sometimesbeen widely advocated and (apparently)believed, indeed that it was arguably the domi-nant view concerning the problem of the externalworld for a good portion of the twentieth century.The first and most important thing to say aboutthis situation is that you must not, as sometimeshappens, allow it to cause you to fail to under-stand what the view is saying by trying to makeit more reasonable than it is. The phenomenalistreally is saying that there is nothing more to thematerial world (including, of course, our ownphysical bodies! . . . ) than our subjective sensoryexperiences and the possibility, in the senseexplained, of further such experiences (thoughthere is, as we will eventually see, a serious prob-lem about the “our”).

But why should such an obviously implau-sible view be taken seriously, even for amoment? We have already in fact encoun-tered the essential ingredients of the mainargument for phenomenalism, but it will behelpful to reiterate them in a somewhat more

7

Try to spell out the detailsof this example more fully,

and think of some others ofyour own.


explicit fashion. One main premise of theargument is the Humean thesis that causalrelations can be known only via experienceof the causal sequence, so that, as alreadyexplained, there is no way in which a causalrelation between the immediate content ofexperience and something outside that imme-diate content could be known, and hence noway to justif iably invoke such externalcauses as explanations of that experience.This thesis has a good deal of initial plausi-bility, and can be rebutted only by offeringsome other account of how causal relationscan be known. The other main premise issimply the common-sense conviction thatskepticism is false, that we do obviously havejustif ied beliefs and knowledge concerningordinary objects like trees and rocks andbuildings and about the material world inwhich they exist. And the argument is thenjust that the only way that such justif iedbeliefs and knowledge are possible, giventhat no causal or explanatory inference fromimmediate experience to material objects thatare genuinely external to that experiencecould ever even in principle be justified, is ifthe content of our beliefs about the materialworld does not really have to do with objectsexisting outside our immediate experience,but instead pertains just to the objects of thatexperience and the order that they manifest.

Most phenomenalists will admit that thisseems initially very implausible, but will tryto argue that this apparent implausibility is insome way an illusion, one that can beexplained away once the phenomenalist viewand the considerations in favor of it havebeen fully understood.

Objections to PhenomenalismThe foregoing argument, like most argumentsfor implausible philosophical views that arenonetheless widely held, is a serious argument,one not easily dismissed. Neither premise iseasy to rebut, and the conclusion does seem tofollow from these premises. But it is, of course,still abundantly obvious that this conclusioncannot be correct, and so that something musthave gone wrong. For it is obvious upon eventhe slightest unbiased reflection that the contentof propositions about physical or material

8

objects does pertain, whether justifiably or not,to a realm of entities that, if genuine, existoutside of our minds and experiences in anindependent physical realm.

This basic insight seems in fact to constituteby itself a more than adequate reason to rejectphenomenalism. But since it neverthelessamounts to little more than a direct, unargueddenial of the view, it will be useful to see if wecan find further objections and problems of amore articulated sort pertaining to phenomenal-ism. (Considering such objections and theresponses available to the phenomenalist willalso help you to better understand the view.) Infact, there are many such objections and prob-lems that have been advanced. Here we will becontent with a few of the most interesting ones.

Consider, first, what is perhaps the mostobvious question about the phenomenalistview:Why, according to the phenomenalist, arethe orderly sense-data in question obtainableor “permanently possible”? What is theexplanation for the pattern of actual and obtain-able sense experiences that allegedly constitutesthe existence of a material object or of thematerial world as a whole, if this is not to beexplained by appeal to genuinely externalobjects? The only possible phenomenalistresponse to this question is to say that the factthat sensory experience reflects this sort oforder is simply the most fundamental fact aboutreality, not further explainable in terms of any-thing else. For any attempted further explana-tion, since it would obviously have to appeal tosomething outside of that experience, would be(for the reasons already discussed) unjustifiedand unknowable. The phenomenalist will addthat it is obvious anyway that not everything canbe explained, since each explanation just intro-duces some further fact for which an explana-tion might be demanded.

But while this last point seems correct(doesn’t it?), it seems quite implausible tosuppose that something as large and compli-cated as the total order of our immediate expe-rience has no explanation at all—and also veryobvious that common sense (at least if itaccepted perceptual subjectivism) wouldregard claims about material objects asproviding such an explanation, rather than asjust a redescription of the experiential orderitself (as the phenomenalist claims them to be).

8

An underlying additionalpremise is the thesis of

perceptual subjectivity.Thisargument is similar to Berkeley’sargument, except that Berkeleyclaims only that causal relationsbetween experience and externalmaterial objects are unknowable.Does Berkeley have any goodreasons for thinking that causationof experience by God is moreknowable?


Perhaps, for all we have seen so far, the phe-nomenalist is right that we cannot ever knowthat any such explanation is correct, but this, ifso, is an argument for skepticism about thematerial world, not a justif ication for per-versely reinterpreting the meaning or contentof claims about material objects. (Here it isimportant to be very clear that phenomenalismis not supposed to be a skeptical view, butrather an account of how beliefs about mate-rial objects are indeed justified and do consti-tute knowledge—given the phenomenalistaccount of the content of such beliefs.)

A second problem (or rather a set of relatedproblems) has to do with the specification ofthe conditions under which the various sense-data that (according to phenomenalism) arewhat a material-object proposition is abouteither are or would be experienced. It is clearthat such conditions must be specified to haveeven a hope of capturing the content of at leastmost such propositions in sense-datum terms.To recur to our earlier example, to say merelythat the sense-data that are characteristic of abrown table are actual or obtainable in somecircumstances or other may perhaps capturethe content of the claim that the world containsat least one brown table (though even that isdoubtful), but surely not of any more specificclaim, such as the one about such a table beingin a particular room in the University of Wash-ington Library. For that, as we saw briefly,conditions must be specif ied that say, as itwere, that it is in relation to that particularroom that the sense-data are or would be expe-rienced. (But remember here that for thephenomenalist, the room does not exist as amind-external place; talk of a room or of anyphysical location is to be understood merely asa way of indicating one aspect of the order ofimmediate experience, namely that the varioussense-data that reflect the various featuresascribed to the room tend to be experiencedtogether or in close succession, with thiswhole “cluster” of sense-data standing insimilar relations to the further sense-data thatpertain to the surrounding area.)

What makes this problem extremely difficultat best is that for phenomenalism to be aviable position, the conditions under whichsense-data are experienced or obtainable mustthemselves (as just in effect indicated) be

9

specifiable in terms of other sense-data, not interms of material objects and structures such asthe library or room in question. For the essentialclaim of phenomenalism is that the content ofpropositions about material objects can beentirely given in terms of sense-data. If in spec-ifying the conditions under which the actual andobtainable sense-data relevant to one material-object proposition would occur, it were neces-sary to make reference to other material objects,then the account of the content of the firstproposition would not yet be completely insense-datum terms. And if in specifying theconditions relevant to claims about those othermaterial objects, still other material objectswould have to be mentioned, and so on, thenthe phenomenalist account would never becomplete. If the content of propositions aboutmaterial objects cannot be given entirely interms of sense-data, if that content involvesessential and ineliminable reference to furthersuch objects, then phenomenalism fails.

There are in fact many problems here, but wemay continue to focus on the one suggested bythe example of the table in the library room. Howcan the idea that sense-data are or would beobserved in a certain location be adequatelycaptured in purely sense-datum terms? Thenatural response, which was in effect invokedwhen the example was originally discussed, is toappeal to the idea of a sensory route: a series ofjuxtaposed and often overlapping sense-data thatwould be experienced in what we think of intu-itively as moving to the location in question.(But, to reiterate, there is not supposed to be anyreal mind-external location or bodily movement;according to the phenomenalist, claims aboutthis sort of experienced movement have to doonly with sequences of sense-data that are expe-rienced or could be experienced—includingthose that we think of intuitively as the feelingsassociated with bodily movements like walking.)

There are at least two serious problemspertaining to this answer, however. One is thatthere are normally many different sensory routesto a given location, depending on where onestarts and how one approaches it; and if thestarting location is itself determined by aprevious sensory route, then a regress threatens,in which the sensory conditions must go furtherand further back in time without ever reaching aplace from which they can begin. A second

9

The phenomenalist, like

Berkeley, means to be giving

an account of what the existence of

ordinary objects amounts to, not to

be denying their existence.

RR


problem is that it seems clear that we can oftenunderstand the claim that a certain materialobject or set of objects exists at a certain physi-cal location without having any clear idea of therelevant sensory route: for example, I under-stand the claim that there are penguins at theSouth Pole, but have no clear idea of the sensoryroute that I would have to follow to guarantee oreven make it likely that I have reached the SouthPole. (Note that it is a guarantee that is actuallyrequired, for otherwise the content of the claimin question as not been fully captured.)

…

And there is also the related, but still muchmore difficult problem of what the phenomenal-ist can say about the content of propositions aboutmaterial objects and events in the past, perhapsthe very distant past. Consider this one carefullyon your own, focusing on the most difficult case:past events that were not observed by anyone atthe time in question. Under what sensory condi-tions would sense-data of a tree have to have beenobtainable to make it true that there was a pinetree in the place now occupied by my house in1000 B.C.? It is thus very doubtful that the sort ofspecification of conditions that the phenomenal-ist needs is possible in general.

…

A . . . final objection to phenomenalism, onethat is, thankfully, much simpler and morestraightforward, concerns what the phenomenal-ist must apparently say about the knowledge ofthe mental states of people other than myself (orother than whoever is thinking about the issue—for reasons that will become clear, each of youwill have to formulate this issue for yourselves).The whole thrust of the phenomenalist position,as we have seen, is that any inference beyondimmediate experience is impossible, that claimsthat might seem to be about things outside ofexperience must, if they are to be justified andknowable, be understood as pertaining only tofeatures and orderly patterns of that experience.But the mental states of other people, their expe-riences and feelings and conscious thoughts, aresurely outside of my immediate experience.Indeed, to reach justified conclusions about what

people distinct from me are genuinely thinkingand experiencing would apparently require twoinferences: first, an inference from my immedi-ate experience of sense-data pertaining to theirphysical bodies to conclusions about thosebodies; and then, second, an inference from thefacts about those bodies thus arrived at to furtherconclusions about the minds and mental states ofthe people in question. Both of these inferencesdepend on causal relations that are, according tothe phenomenalist, unknowable, because wecannot experience both sides, or in the secondcase even one side, of the relation; and thus nei-ther inference, construed in that way, is justifiedaccording to the basic phenomenalist outlook.

What phenomenalism must apparently sayhere, in order to be consistent, is (i) that thecontent of propositions about the conditionsand behavior of other people’s bodies (like thatof all other material object propositions)pertains only to facts about my immediate expe-rience; and (ii) that the content of further claimsabout the mental states associated with thosebodies is only a further, more complicated andless direct description of, once again, my expe-rience. Though the phenomenalist would per-haps resist putting it this way, the upshot is thatmy mind and mental states, including my imme-diate experience, is the only mind and the onlycollection of mental states that genuinely exist,with claims that are apparently about otherminds amounting only to further descriptions ofthis one mind and its experiences. This is theview known as solipsism—which each of youmust obviously formulate for yourselves(assuming that any of you are really out there!).It seems clearly to be an absurd consequence,thus yielding a really decisive objection, if onewere still needed, to phenomenalism.*

Back to RepresentationalismIf phenomenalism is indeed untenable, andassuming that we continue to accept per-ceptual subjectivism, then the only nonskepticalalternative apparently left is representationalism:the view, restating it a bit, that our immediatelyexperienced sense-data, together with the furtherbeliefs that we arrive at on the basis of them,

*A woman once allegedly wrote to Bertrand Russell that she found solipsism to be so obviously a correct view that shecouldn’t understand why it wasn’t more widely accepted. (Think about it!)


constitute a representation or depiction of anindependent realm of material objects—one thatwe are in general, according to the representation-alist, justified in believing to be true.

Defenses of representationalism have taken avariety of forms, but I will assume here that thebest sort of defense for such a view is one alongthe general lines that we found to be suggested,albeit not very explicitly, in Locke (and indeedalso, though even less explicitly, in Descartes).The central idea is, first, that (contrary to theclaim of the phenomenalist) some explanationis needed for the complicated and intricateorder that we find in our involuntarily experi-enced sense-data . . . ; and, second, that the bestexplanation, that is, the one most likely to becorrect, is that those experiences are caused byand, with certain qualifications, systematicallyreflect the character of a world of genuinelyindependent material objects, which we accord-ingly have good reasons for believing to exist.

I have already remarked that representational-ism was widely repudiated as untenable duringmost of the period between Locke and recenttimes, with the main argument being the onethat we found in Hume about the unknowabilityof any causal relation between something out-side experience and experience itself. We willbegin by looking further at that argument andconsidering in a general way how it might beanswered. Having argued that representational-ism cannot be simply ruled out as impossible inthe way that Hume tries to do, we will thenconsider the further issue of whether and howthe specific explanation of experience that therepresentationalist proposes can be defendedagainst other alternatives, such as Berkeley’s.Finally, we will look at the significant qualifica-tion, already briefly mentioned, advocated byDescartes, Locke, and many others with regardto the accuracy with which our experience rep-resents the true character of material objects: theone having to do with the distinction betweenprimary and secondary qualities.

A Response to Hume’s Argument:Theoretical or Explanatory

InferenceTo recall, Hume’s objection to representation-alism rests on the premise that causal relationscan be known only by experiencing the regular

sequence between cause and effect, whichrequires experiencing both sides of the causalrelation. This, he argues, is impossible for analleged causal relation between somethingoutside of direct experience and the experi-ence itself, so that the claim that such acausal relation exists can never be justified orknown. And therefore, he concludes, neithercan the representationalist’s proposed expla-nation of the order of our experience, sincethat depends essentially on such an unknow-able and unjustifiable causal claim.

If Hume’s initial premise is accepted, then therest of his argument seems to follow. But shouldthat premise be accepted? One way to approachthis issue is to consider examples where we seemto reason in ways that conflict with that premisebut which still seem intuitively cogent. Here Iwill consider [one example] of this kind, . . .having to do with knowledge concerning unob-servable entities and events, such as electrons orquarks or radioactivity, in theoretical physics. In[this case], we seem intuitively to have justifiedbelief and knowledge pertaining to causalrelations that could not be arrived at in the waythat Hume’s premise, if correct, would require.(In considering [this example], we adopt thestandpoint of common sense, thus assumingthat the problem of the external world has—somehow—been solved.)

…

[In the] case of unobservable scientificentities and events[,] . . . we seem to have justi-fied belief and knowledge concerning causalrelations among such entities and events andbetween them and various sorts of observableresults, even though the entities and eventsthemselves cannot be experienced in even theindirect sense: knowledge, for example, thatradioactivity results from the splitting or decayof various sorts of atoms and that it produces acrackling sound in a Geiger counter. Obviouslybeliefs concerning relations of these kindscannot be justified by experiencing both sidesof the causal relation in the way that Hume’spremise would require.

Notice carefully that the claim so far is notthat these alleged cases of causal knowledge aregenuine, so that Hume’s premise would have tobe mistaken. It is possible for a proponent ofHume’s view to respond by claiming either that


we do not really have the causal knowledgein question. . . . Thus defenders of Hume’sview have often also been advocates . . . offictionalism (the view that seemingly unobserv-able scientific entities do not really exist, butonly reflect ways of talking that help to system-atically describe observations). But [this viewseems] desperately implausible, so that if a rea-sonably plausible general account can be givenof how such causal knowledge can be justifi-ably arrived at, this would be enough to warrantthe rejection of Hume’s premise and theargument that results from it.

The account that has been offered . . . holdsthat knowledge of the sort in question dependson a fundamental and sometimes unrecognizedmode of reasoning, one that is quite distinctfrom both deductive reasoning and . . . induc-tive reasoning [:] . . . theoretical or explanatoryreasoning. In reasoning of this sort, a hypothe-sis is advanced to explain some relevant set ofdata and is justif ied simply on the basis ofbeing the best explanation of the data in ques-tion.* Exactly what makes an explanationthe best is a difficult and complicated issue, aswe will see to some extent below, but the pointfor the moment is that if such an assessmentcan be defended, then it allegedly becomes jus-tif iable to accept the entire explanatoryhypothesis, including any causal claims that itmay involve, on that basis—without anyrequirement that there be experiential evidenceof the sort that Hume’s premise would requirefor those causal claims by themselves. Thus forexample when the entire physical theory ofradioactive isotopes and their decay into otherkinds of atoms is justified as the best explana-tion of a variety of observed phenomena,including the fogging of photographic film,changes in the composition of samples, tracksin cloud chambers, etc., the causal relationsbetween the various kinds of atoms andparticles and also between these unobservableentities and processes and their observablemanifestations are justified as part of the totalpackage, with accordingly no need for them tobe justified separately.

A full defense of the idea of theoretical orexplanatory reasoning is obviously not possible[here].The suggestion for the moment is only that

11

10

the idea is plausible enough, especially in light ofexamples like those given, to make it reasonableto reject Hume’s thesis about knowledge of causalrelations, at least tentatively, thus opening thedoor to the possibility that the representationalistposition on the problem of the external worldmight be defensible after all.

The RepresentationalistExplanation

But this only opens the door. We still need toworry about whether the representationalist’sproposed explanation of our experience is reallythe best one. And before we can do that, we needto consider in substantially more detail what therationale for that explanation might be and howit is supposed to work.

The place to start is to ask what it is about thecharacter of our immediate sensory experiencethat points to or perhaps even seems to demandsuch an explanation. As we saw earlier, Lockepoints to two features of our experience in thisconnection: its involuntary character and itssystematic order. But while these features mayindeed demand some sort of explanation, theydo not, at least when described at that level ofabstraction, seem to point at all clearly at thespecific one that the representationalist favors(which is why the door is seemingly open toBerkeley’s alternative). If anything about expe-rience does this, it will thus have to be, I wouldsuggest, more specific features than any thatLocke explicitly mentions.

Here is a question for you to think hard about,preferably before reading beyond thisparagraph—one that is both historically andsubstantively as fundamental as any in thewhole field of epistemology. Think as carefullyas you can about your immediate sensory expe-rience: your experience of qualities like colorsand shapes and apparent spatial relations andapparent sounds and tactile qualities and so on.You are presently experiencing patterns of blackand white marks that according to the represen-tationalist are caused by and represent the pagesof this book, along with other colors reflectingthe immediate surrounding environment; yourauditory sensations might be those allegedlyreflecting the steps of people in the library or the

12

12

Theoretical or explanatoryreasoning is in fact quite

common.Try to think of otherexamples, drawn from outsideof science.

*Thus the sort of reasoning in question is also sometimes referred to as “inference to the best explanation.”

10

This is a counter-example toHume’s general claim about

knowledge of causality: a casewhere the sort of knowledge thathe rejects as impossible seemsnonetheless to exist. How clear isit that such knowledge is genuine(and hence that fictionalism ismistaken)?

11

Deductive reasoning isreasoning that is logically conclu-sive and entirely a priori. Inductivereasoning, as BonJour is using theterm, is reasoning that involvessimple generalization fromobserved patterns in experience:from the observation of manyravens that are black (and nonethat are any other color) to theconclusion that all ravens are black.Theoretical reasoning also dependson observations, but concludes toan explanation in other termsrather than just a generalization.(Sometimes the term “induction” isused for both of these last sorts ofreasoning, but they are nonethelessimportantly different.)


music that you listen to while you read; youhave tactile sensations allegedly reflectingthings like the book in your hand, the chair orcouch you are sitting on, and so on; perhapsthere is an odd smell of some sort as well. What,if anything, about those experienced qualitiestaken in themselves suggests that their source orcause is an independent realm of materialobjects of the sort that the representationalistadvocates? Why, apart from mere familiarity,does such an explanation of experience seem sonatural and compelling?

My suggestion is that the answer to this ques-tion has two main parts. The first points to thepresence in immediate experience of repeatablesequences of experienced qualities, qualities thatoverlap and often shade gradually into oneanother. Here I have in mind something like the“sensory routes” that are, as discussed earlier,invoked by the phenomenalist. While these“sensory routes” cannot ultimately do the job thatthe phenomenalist needs them to do, for the rea-sons given there, they are nonetheless very realand pervasive. Think of the ways in which such“sensory routes” can be experienced in oppositeorders (imagine here what common sense wouldregard as walking from one place to another andthen returning to the first place by the sameroute—perhaps even walking backwards, so as tomake the two sequences as similar as possible).Think of the ways in which such “sensory routes”intersect with each other, thus, for example,allowing one to get from one end to the otherwithout going through the “route” itself, therebydelineating a sensory loop. Think of the resultingstructure of a whole set of overlapping and inter-secting “sensory routes.”

Here it may be helpful, as a kind of analogy, tothink from a common-sense standpoint of howyou would go about programming a computergame to simulate a “space” containing “objects”through which the computer character can move.You would program successive “screens” of visu-ally observable colors and shapes in such a wayas to mimic the appearance of objects that aregradually approached and passed, perhaps withconcomitant sound qualities that get louder andthen softer and imaginably even other systemati-cally varying qualities like smells or tempera-tures. (Perhaps the game is played in an enclosedbooth that can be heated or cooled.) You wouldalso include some controllable way in which the

character can be made to face in different direc-tions, move at different rates, and stand still. Inthese terms, my suggestion is that our actualimmediate experience has more or less exactlythe features that an ideal program of this sortwould create. (Again, you will have to ponder thispoint, “chew on” it, in relation to a range of yourown examples, in order to fully understand it.)

The idea is then that at least the most obviousand natural explanation of these features of ourexperience is that we are located in a spatial realmof objects through which we move and of whichwe can perceive at any given moment only thelimited portion that is close enough to be accessi-ble to our various senses (what this requiresdiffers from sense to sense)—a kind of experien-tial “tunnel.” Our experience reflects both thequalities of these objects and the differentperspectives from which they are perceived as wegradually approach them from different direc-tions, at different speeds, under different condi-tions of perception, etc. Thus the relativelypermanent structure of this spatial array of objectsis partially reflected in the much more temporaryand variable, but broadly repeatable features ofour immediate experience. . . .

The second part of the answer to the questionof what it is about the character of immediateexperience that points to the representationalistexplanation cites the fact, already noticed in ourdiscussion of phenomenalism, that the experien-tial order just described, though undeniablyimpressive, is in fact incomplete or fragmentaryin a number of related ways. The easiest way toindicate these is by reference to the sorts of sit-uations that, from a common-sense standpoint,produce and explain them (though the represen-tationalist cannot, of course, assume at thisstage, without begging the question, that thesethings are what is actually going on). Imaginethen traversing a “sensory route” of the sort justindicated, but doing so (i) with one’s eyes closed(or one’s ears plugged, etc.) during some of thetime required, or perhaps while asleep duringpart of the time (traveling in a car or train); or(ii) while the conditions of perception, includ-ing those pertaining to the functioning of yoursense-organs and to your mental “processing,”are changing or being varied (involving suchthings as changing lighting, including completedarkness; jaundice and similar diseases thataffect perception; objects and conditions that

13 13

Think carefully aboutpatterns of experience of

the sort indicated, making them asdetailed as possible.Think aboutwhy the hypothesis that the experi-ences are caused by a stable worldof external objects seems toexplain those experiences quitewell. Can you think of any alterna-tive explanation that doesn’t invokea being like Berkeley’s God or aCartesian evil genius?


temporarily block or interfere with perception;even something as simple as turning one’s headin a different direction, blinking, or wiping one’seyes). If you think about it carefully, you will seethat interfering factors of these various kindsmake the sensory sequences that define the var-ious “routes” far less regular and dependablethan they might at first seem. . . .

Thus the basic claim is that the realm of imme-diate sensory experience, of sense-data . . . , isboth too orderly not to demand an explanation andnot orderly enough for that explanation to be thatthe sense-data have an intrinsic order of their own.What this strongly suggests, the representational-ist will argue, is an independent realm of objectsoutside our experience, one that has its ownpatterns of (mainly spatial) order, with the partialand fragmentary order of our experience resultingfrom our partial and intermittent perceptualcontact with that larger and more stable realm.

The discussion so far provides only an initialand highly schematic picture of the representa-tionalist’s proposed explanation. It would haveto be filled out in a number of ways in order tobe even approximately complete. Here I will becontent with three further points. First, the mainfocus of the discussion so far has been onspatial properties of material objects and thefeatures of immediate experience that seem tosuggest them. Thus the result to this point is atbest only a kind of skeletal picture of the mate-rial world, one that would have to be “fleshedout” in various ways in order to even approxi-mate the common-sense picture of the world. Infact, it is useful to think of the representational-ist explanation as starting with spatial propertiesas a first and most fundamental stage and thenadding further refinements to that starting point.

Second, the main addition to this initial spatialpicture of the world would be various sorts ofcausal relations among material objects andbetween such objects and perceivers, togetherwith the causal and dispositional properties ofobjects (flammability, solubility, malleability,brittleness, toxicity, etc.) that underlie such rela-tions. These are, from the representationaliststandpoint, basically added in order to explainapparent changes in material objects that arereflected in relatively permanent changes in theotherwise stable “sensory routes.” . . .

Third, there is the issue of primary and second-ary qualities. As already noted, Locke’s view is

14

that material objects have primary qualities likesize, shape, and motion through space, but notsecondary qualities like color, smell, taste, andfelt temperature, a view with which most otherrepresentationalists have tended to agree. Here itwill suffice to focus on color, surely the mostapparently pervasive and interesting of the sec-ondary qualities. Clearly to deny that materialobjects are genuinely colored complicates therepresentationalist’s proposed explanation bymaking the relation between material objects andour immediate experiences much less straightfor-ward than it would otherwise be: according tosuch a view, whereas our immediate experiencesof spatial properties are caused more or lessdirectly by closely related spatial properties ofobjects (allowing, importantly, for perspective),our immediate experiences of color propertiesare caused by utterly different properties ofmaterial objects, primarily by how their surfacesdifferentially reflect wavelengths of light.

Locke offers little real argument for this view,but the argument he seems to have in mind isthat as the causal account of the material worlddevelops, it turns out that ascribing a propertylike color (construed as the “sensuous” propertythat is present in immediate visual experience)to material objects is in fact quite useless forexplaining our experiences of colors. What col-ors we experience depends on the properties ofthe light that strikes our eyes and this in turn, inthe most standard cases, depends on howmaterial objects reflect and absorb light, whichyet in turn depends on the structure of theirsurfaces as constituted by primary and causalproperties . . . if [this] is correct, then the denialthat material objects are really colored simplyfollows from the basic logic of the representa-tionalist position: according to representational-ism, the only justification for ascribing anyproperty to the material world is that it bestexplains some aspect of our immediate experi-ence, so that the ascription of properties that donot figure in such explanations is automaticallyunjustified.

Alternatives to theRepresentationalist Explanation

The discussion so far has perhaps made a reason-able case, though of course nothing like a conclu-sive one, first, that the representationalist’s pro-

15

14

Try to think of specificexamples of fragmentary or

interrupted order. How commonare they?

15

Is this a good reason forthinking that color is not

a genuine feature of externalmaterial objects? (See DiscussionQuestion 2.)


posed explanation of the order of our immediateexperience cannot be ruled out on Humeangrounds; and, second, that this explanation has agood deal of plausibility in relation to that expe-rience. But this is still not enough to show that itis the best explanation and hence the one, evenassuming the general acceptability of theoreticalreasoning, whose acceptance is thereby justified.Here we are essentially back to the questionposed very early on in this chapter: why, if at all,should the explanation of our experiencethat invokes external, mind-independent materialobjects be preferred to other possible explana-tions such as Berkeley’s (or the very similar ifnot identical one that appeals to Descartes’sevil genius)?

It should be obvious that Berkeley’s explana-tory hypothesis is capable of explaining the verysame features of immediate experience that therepresentationalist appeals to. All that is needed,as suggested earlier, is for God to have an ideallycomplete conception or picture of the representa-tionalist’s material world and then to systemati-cally cause experiences in perceivers that reflecttheir apparent location in and movement throughsuch a world. (This assumes that God can recog-nize intentions to “move” in various directionsand adjust the person’s perceptions accordingly;of course, no genuine movement really takesplace, nor does the perceiver really have a physi-cal location.)* A different, but essentially parallelexplanatory hypothesis, is provided by a sciencefiction scenario: the perceiver is a disembodiedbrain floating in a vat of brain nutrients andreceiving electrical impulses from a computerthat again contains an ideally complete model orrepresentation of a material world and generatesthe impulses accordingly, taking account ofmotor impulses received from the brain thatreflect the person’s intended movements. Andfurther explanatory hypotheses can be generatedaccording to the same basic formula: there mustbe some sort of a representation or model of amaterial world together with some sort of mech-anism (which need not be mechanical in theordinary sense) that systematically producesexperience in perceivers, allowing for theirsubjectively intended movements. Any pattern of

immediate experience that can be explained bythe representationalist’s explanatory hypothesiscan thus automatically be also explained byexplanatory hypotheses of this latter sort, proba-bly indefinitely many of them, with no possibleexperiential basis for deciding between them orbetween any one of them and the representation-alist hypothesis.

If there is to be a reason for favoring therepresentationalist hypothesis, therefore, itwill have to be a priori in character, and it ismore than a little difficult to see what it mightbe. Here I will limit myself to one fairly tenta-tive suggestion. . . .

One striking contrast between the representa-tionalist’s explanatory hypothesis and theothers we have looked at is that under the repre-sentationalist view there is a clear intuitivesense in which the qualities of the objects thatexplain our immediate experience are directlyreflected in the character of that experienceitself, so that the latter can be said to be, allow-ing for perspective and perhaps other sorts ofdistortion, experiences of the former, albeitindirect ones. Once again this applies moststraightforwardly to spatial properties: thus, forexample, the rectangular or trapezoidal shapethat is immediately experienced can be said tobe an indirect perception of a rectangular faceof the material object that causes that experi-ence. In contrast, the features of the elements inthe other explanatory hypotheses that areresponsible for the various features of our expe-rience are not directly reflected in that experi-ence. For example, what is responsible in theseother hypotheses for the rectangular or trape-zoidal shape in my immediate experience is oneaspect of God’s total picture or conception of amaterial world, or perhaps one aspect of a rep-resentation of such a world stored in a com-puter. This aspect has in itself no shape of anysort (or at least, in the case of the computer,none that is at all relevant to the shape that Iexperience); it is merely a representation of arelated shape, according to some system ofrepresentation or coding. Thus its relation tothe character of the experience that it is sup-posed to explain is inherently less direct, more

16

*Here again a computer game provides a helpful analogy. In playing many such games, you control the “movement” of a com-puter character through the “world,” often in a fairly realistic way, even though there is really no movement of that sort nor anyworld of that sort in which it might take place. In Berkeley’s explanatory hypothesis, God plays the role of the computer.

16

Because all of theexplanations are equally

compatible with the relevantexperience.


convoluted than in the case of the representa-tionalist’s explanation.

My suggestion is that the inherently lessdirect, more convoluted character of the waythat these competing explanatory hypothesesaccount for the features of our immediateexperience may yield a reason for preferring themore direct and thus in a sense simpler repre-sentationalist explanatory hypothesis, forregarding it as more likely to be true. But how,exactly? The idea is that an explanatory hypoth-esis like Berkeley’s, at least as we haveconstrued it, depends for its explanatorysuccess on the truth of two equally essentialclaims: first, the claim that a material world ofthe sort postulated by the representationalistcould account for the features of our experi-ence, for it is precisely by emulating ormimicking the action of such a world that God(or the computer) decides just what experiencesto produce in us; and, second, that God (or thecomputer) can indeed successfully produce therequired emulation. But the representationalistview requires only the truth of the first of thesetwo claims. It is thus, I suggest, inherently lessvulnerable to problems and challenges and somore likely to be true. And this is an apparentreason for regarding the representationalist’sexplanatory hypothesis as providing the best ofthese competing explanations. . . .

Is this a successful argument for representa-tionalism? There are at least two questionsabout it that need to be considered. First, theargument assumes that the competitors torepresentationalism are all parasitic upon therepresentationalist explanatory hypothesis in theway indicated, and it is worth asking whether thisis really so. Is there an explanation of our imme-diate experience that does not in this way rely onan emulation of the way in which a materialworld would produce that experience? It will notdo to say simply that God causes our experiencewithout saying how and why he produces thespecific results that he does, for that is not reallyto give a complete explanation. But is there someother way of filling out Berkeley’s explanatoryhypothesis or one of the parallel ones that doesnot invoke a conception or model of a materialworld? Second, even if the argument succeeds toa degree, how probable or likely does it make thematerial world hypothesis in comparison to theseothers? Is the resulting degree of probability or

17

likelihood high enough to agree approximatelywith our common-sense convictions inthis regard (or to yield knowledge . . .)? I willleave these further difficult questions for you tothink about.

Is There a Better Alternative?:Direct Realism

The upshot of our discussion so far is thatphenomenalism appears entirely untenable, andthat at least a better defense than many havesupposed possible can be offered for representa-tionalism. Many recent philosophers, however,have thought that there is a third alternative thatis superior to either of these: one usuallyreferred to as direct realism. The central idea ofdirect realism is that the view we have calledperceptual subjectivism is false, that is, thatinstead of immediately experiencing . . . sense-data . . . , we instead directly experience externalmaterial objects, without the mediation of theseother sorts of entities or states. And the sugges-tion often seems to be, though this is usually notexplained very fully, that such a view can sim-ply bypass the representationalist’s problem ofjustifying an inference from immediate experi-ence to the material world and do so withouthaving to advocate anything as outlandish asphenomenalism.

For anyone who has struggled with the ideaof sense-data . . . and with the difficulties andcomplexities of representationalism and phe-nomenalism, the apparent simplicity of directrealism, the way in which it seems to makeextremely difficult or even intractable problemssimply vanish, may be difficult to resist. Wemust be cautious, however. What does such aview amount to, and can it really deliver theresults that it promises?

We may begin with a point that is oftenadvanced in arguments for direct realism, onethat, while correct as far as it goes, turns out infact to be of much less help than has sometimesbeen thought in either defending or evenexplaining the view. Think about an ordinaryexample of perceptual experience: standing inmy back yard, I watch my dogs chasing eachother in a large circle around some bushes, weav-ing in and out of the sunshine and shadows, as acar drives by on the street. The direct realist’sclaim is that in such a case (assuming that I am

17

Is this a good reason forfavoring the representation-

alist’s explanation over Berkeley’s?Can you think of any reason forpreferring Berkeley’s explanationto the representationalist’s?


in a normal, nonphilosophical frame of mind),the picture that it is easy to find in or read intosome representationalists, according to which Ifirst have thoughts or occurrent beliefs aboutthe character of my experience (. . . in sense-datum . . . terms) and then infer explicitly fromthese to thoughts or beliefs about materialobjects is simply and flatly wrong as a descrip-tion of my actual conscious state. In fact, theonly things that I think about at all directly andexplicitly in such a case are things like dogs andbushes and cars and sunlight, not anything assubtle and abstruse as sense-data. . . . The directrealist need not deny (though some have seemedto) that my sensory experience somehowinvolves the various qualities, such as compli-cated patterns of shape and color, that these otherviews have spoken of, or even that I am in someway aware or conscious of these. His point is thatwhatever may be said about these other matters,from an intuitive standpoint it is material objectsand nothing else that are “directly before mymind”—and that any view that denies this obvi-ous truth is simply mistaken about the facts.

I have already said that I think that thedirect realist is at least mostly right about this.What happens most centrally in perceptualexperience is that we have explicit thoughts or“perceptual judgments” about what we areperceiving; and in normal cases (apart fromvery special artistic or perhaps philosophicalcontexts), these perceptual judgments aredirectly and entirely about things (andprocesses and qualities) in the external mate-rial world. Philosophers speak of that which apropositional state of mind is directly about asits intentional object, and we can accordinglysay that the intentional objects of our basicperceptual judgments are normally alleged orapparent material objects. In this way, therelation of such judgments to material objectsis, it might be said, intentionally direct.

But what bearing, if any, does this intentionaldirectness have on the central epistemologicalquestion of what reason or justification we havefor thinking that such perceptual judgmentsabout the material world are true? Perhaps thesort of direct presence to the mind that isinvolved in the idea of immediate experience . . .yields the result that one’s beliefs or awarenessesconcerning the objects of such experienceare automatically justified, simply because

18

there is no room for error to creep in.But is there any way in which it follows from themere fact that perceptual judgments aboutmaterial objects are intentionally direct thatthey are also justified? It still seems obviousthat both a perceptual judgment and the totalstate of mind of which it is a part are quitedistinct from the material object, if any, that is itsintentionally direct object. This is shown bythe fact that in cases like hallucination, theobject in question need not exist at all, but itwould be clear enough even without suchcases—phenomenalist views having beenrejected, the material object does not somehowliterally enter the mind. Thus even thoughperceptual judgments are directly about suchobjects in the intentional sense, the question ofwhether they represent them correctly still arisesin exactly the same way that it does for therepresentationalist. And this question mustapparently still be answered, if at all, by appealto the immediately experienced featuresinvolved in the perceiver’s state of mind, with thespecific character of the sensory experiencebeing the only obvious thing to invoke.

Thus while the idea of intentional directnesscan be used to present a somewhat more accu-rate picture of a normal perceiver’s state ofmind, the view that results is still fundamen-tally a version of representationalism in that itfaces the same essential problem of justifyingthe transition (whether it is an explicit infer-ence or not) from the character of the person’sexperience to beliefs or judgments about thematerial world. If this is all that direct realismamounts to, then it is not a genuinely distinctthird alternative with regard to the basic issueof how perceptual beliefs or judgments arejustified.

…

My tentative conclusion . . . is that the ideathat direct realism represents a further alterna-tive on the present issue is a chimera. Thus, oncephenomenalism is rejected as hopeless, the onlyalternatives with regard to knowledge of theexternal world appear to be skepticism andsome version of representationalism, perhapsone that recognizes and incorporates the viewthat perceptual judgments about the materialworld are intentionally direct.

20

19

18

This does not mean thatideas or sense-data play no

role, but only that they are not theobject of the perceptual judgmentthat results. Instead, they might beessential vehicles for our percep-tual consciousness, essential toexplaining how we are able tothink “directly” of objects outsidethe mind. (See the introduction tothe Locke selection.)

19

The suggestion is that onlyideas or sense-data and

other states of mind are immedi-ately experienced in this sense.

20

The claim is that the way inwhich we are directly aware

of external objects (the fact thatsuch objects and not our experi-ences or ideas or sense-data arewhat we are thinking about in suchexperience) really does nothing atall to explain why our perceptualclaims are justified, why we havegood reasons to think that theyare true. Our state of mind is stilldistinct from the object, and so theissue of whether it represents suchan object correctly can still beraised—in exactly the same waythat it arises for Locke.



1. Can you think of a good argument for the exis-tence of mental states other than your own inwhat appear to be other people? The obviousstarting point is the behavior of other bodies.Tryto construct such an argument, and evaluate itsstrength.

2. Think of how a scientific explanation of colorexperience might go. Are Locke and BonJour rightthat color as a feature of material objects plays norole in such an explanation? If so, does it followthat we have no reason to believe that suchobjects are really colored, or is there some otherreason (besides a role in explanation) for thinkingthey are colored?

3. Whose explanation of the patterns of experienceis better: the representationalist’s or Berkeley’s?

Why? Can you think of other explanations thatare significantly different from either of these?What general standards are the right ones forchoosing among such explanations, and why?

4. BonJour offers what is admittedly only a sketchof an argument for the existence of materialobjects. How hard would it be to fill this outinto a complete argument (where a large part ofthe job would be giving a much fuller descrip-tion of the relevant features of experience)? Isit plausible to suppose that ordinary peoplewho have beliefs about material objects havethis argument or anything like it in mind? If not,does this mean that they have no good reasonsto think that their beliefs are true? How implau-sible is such a result?

Sextus EmpiricusSextus Empiricus (third century A.D.) was a Greek philosopher who seems tohave served as a kind of historian of the Skeptical school of philosophy. Verylittle is known about his life, though he was apparently a medical doctor whotaught in a Skeptical academy, perhaps in Alexandria. The following selectioncomes from his explanation of the Ten Modes of argument, via which skepticscan reach suspension of belief—viewed by them as an intrinsically desirablestate in which one is cured of dogmatism.

From Outlines of Pyrrhonism

The Fourth ModeBut we can also reach suspension by basing ourargument on each sense separately, or even bydisregarding the senses. To this end we employthe fourth mode of suspension, which we callthe mode based on the circumstances. Weunderstand by “circumstances” the states inwhich we are. This mode, we say, is seen incases of natural or unnatural states, in states ofwaking or sleeping, in cases where age, motionor rest . . . are involved; or where the determin-ing factor is a state of want or satiety. . . . Forexample, things appear dissimilar according towhether we are in a natural or unnatural state;delirious people, and those who are possessedby a god, think that they hear divine voices,

while we do not. Often they claim that they per-ceive, among a number of other things, the odorof . . . frankincense, or something of that sort,where we perceive nothing. And the same waterthat seems hot to a person when poured oninflamed parts seems lukewarm to us. The coatwhich appears yellowish-orange to men withbloodshot eyes does not appear so to me, yet itis the same coat. And the same honey thatappears sweet to me appears bitter to those suf-fering from jaundice.

Now, one might object that in those whosecondition is unnatural it is the intermingling ofcertain humors that causes them to get unnat-ural impressions from the external objects.Our reply to this would be that it is possiblethat the external objects actually are in realitysuch as they appear to those who are said to bein an unnatural state; and that since persons ina state of good health also have mixed humors,it may be that it is these humors that make the

11

Examples of these kinds andthe ones that follow are also

the basis for the argument fromillusion, as formulated repeatedlyby Berkeley in the earlier selec-tion. Sextus is probably assumingsomething like Berkeley’sconclusion that what we are mostimmediately aware of is an idea(he uses the term “sense-impression”) in the mind. But he isalso aiming at a different, moreclearly skeptical conclusion as well.

From Selections from the Major Writings on Skepticism,Man, and God, translated by Sanford G. Etheridge(Indianapolis: Hackett, 1985).

SEXTUS EMPIRICUS: FROM OUTLINES OF PYRRHONISM 119

2

Given that what we assume(on what basis?) to be the

same object or situation gives riseto different sense-impressions,depending on the “circumstances”that the perceiver is in, the issuearises of how to decide which suchsense-impressions (if any) revealthe true nature of that object.

objects appear different to them. For it wouldbe a fabrication to attribute to the humors ofsick people a power to change externalobjects, and to deny this power to the humorsof the healthy. After all, it is natural for thehealthy to be in a healthy state, and unnaturalfor them to be in a sick state. By the sametoken it is unnatural for the sick to be in ahealthy state, but natural for them to be in asick state. Consequently, the sick warrant cre-dence also, since they too are in some respectin a natural state. Whether one is in a sleepingor a waking state also makes a difference inthe sense-impressions, since our manner ofperception while awake differs from the per-ception we have in sleep; and our manner ofperception in sleep is not like our wakingperception. As a result, the existence or non-existence of our sense-impressions is notabsolute but relative, since they bear a relationto our sleeping or waking state. It is probable,therefore, that although our dream-images areunreal in our waking state, they are neverthe-less not absolutely unreal, for they do exist inour dreams. In the same manner the realities ofthe waking state, even if they do not exist indreams, nevertheless exist. Age also makes adifference. Old men, for example, may thinkthe air is cold, but the same air seems mild tothose who are in the prime of life. The samecolor appears dim to older persons but full tothose in their prime. And a sound, likewise thesame, seems faint to the former but quite audi-ble to the latter. . . . From this it follows thatdifferences of age also can cause the sense-impressions to be different where the externalobjects are the same. Motion or rest may alsobe the determining factor when objects appeardifferent. For instance, things which we see tobe motionless when we are stationary seem tobe moving when we are sailing past, them. . . .. Hunger or satiety may also be the cause, asthe same food can seem very tasty to the hun-gry and yet disagreeable to those who havehad their fill. . . . Predispositions are anothercause, since the same wine seems to be sour ifyou have eaten dates or dried figs beforehand,and sweet if you have eaten nuts or chick-peas.Also, the vestibule of the bath-house iswarming to those who come in from the out-side, but chills those who are coming out, ifthey tarry in it. . . .

Now, considering the fact that so muchdiscrepancy is due to the states we are in, andthat men are in different states at different times,it is easy, perhaps, to state the nature of eachobject as it appears to this or that person, but dif-ficult to say further what its real nature is. Thisis because the discrepancy does not lend itself tojudgment. In fact, whoever attempts toresolve this discrepancy will find himself eitherin one or the other of the aforesaid states or elsein no state at all. But now to say that he is in nostate at all, that he is neither healthy nor sick,neither in motion nor at rest, that he is not of anyparticular age, and that he is free from the otherstates, is perfectly absurd. On the other hand, thefact of his being in some state or other whileattempting to pass judgment will make him aparty to the controversy. And moreover, he willbe confused by the states in which he finds him-self, and this will prevent him from being anabsolute judge in the matter. A person, there-fore, who is in the waking state cannot comparethe impressions of a sleeping person with thoseof waking persons, and a healthy person cannotcompare the impressions of sick people withthose of the healthy. We do, after all, tend to giveour assent to those things which are present andhave a present influence over us rather than tothings which are not present.

The discrepancy between such impressions isirresolvable on other grounds also, for if a personprefers one sense-impression to another, and onecircumstance to another, he does so eitherwithout judging and without proof or by judgingand offering proof. But he cannot do so withoutjudgment and proof, for then he will be discred-ited. Nor can he do so even with judgment andproof, for if he judges the impressions, he mustat all events use a criterion in judging them. Andthis criterion he will declare to be either true orfalse. If false, he will not be worthy of belief; butif he claims it is true, then his statement that thecriterion is true will be offered either withoutproof or with proof. If without proof, again hewill not be worthy of belief; but if he offers prooffor his statement, the proof must in any case be atrue one, otherwise he will not be worthy ofbelief. Now, if he says that the proof employedfor the confirmation of his criterion is true, willhe say this after having passed judgment on theproof, or without having judged it? If he has notjudged it, he will not be worthy of belief, but if

3

2

3

Sextus argues that there is no reason to view the sense-

impressions of the healthy as moreaccurate than those of the sick andpoints out that neither a healthynor a sick person is in a position tomake such an assessment in anunbiased way. And he would saythe same thing about the otherdifferences in “circumstances.”

RR


he has, obviously he will say he has used a crite-rion in his judgment. We shall ask for a proof forthis criterion, and for this proof another criterion.For the proof always needs a criterion to confirmit, and the criterion needs a proof to show that itis true. A proof cannot be sound without the pre-existence of a true criterion, and a criterioncannot be true either without prior confirmationof the proof. And so both the criterion and theproof fall into circular argument, in which bothare found to be untrustworthy. The fact that eachexpects confirmation from the other makes bothof them equally untrustworthy. It is impossible,then, for a person to give the preference to onesense-impression over another. This being so,such differences in sense-impressions as arisefrom a disparity of states will be irresolvable. Asa result, this mode also serves to introducesuspension of judgment with regard to the natureof external objects.

The Fifth ModeThe fifth argument is the one based on posi-tions, distances, and places. Each of these cancause the same objects to appear different. Takea portico for example. When seen from eitherend, it appears tapered, yet the same porticoviewed from the middle appears symmetrical onall sides. Also, the same boat appears small andstationary from a distance, and large and mov-ing from close by. And the same tower appearsround from afar but square from near by.

These differences are the result of the dis-tances involved. Then there are differences dueto the places involved. For example, the light ofa lamp appears dim in sunlight but bright in the

4

dark, the same oar appears broken in the waterbut straight when out of the water. . . .

Position also may be the cause of differentappearances. The same picture appears smoothwhen inclined backwards, but seems to havehollows and prominences when inclined for-ward to a certain angle. Also, the necks ofpigeons appear different in color according todifferences in inclination.

Thus all objects appearing to us are seen asbeing in some place or other, at a certain dis-tance, and in a certain position, and each of thesefactors makes a great difference in the sense-impressions, as we have mentioned. Hence, bythis mode also we shall be compelled to haverecourse to suspension of judgment. In point offact, anyone who wishes to give any of thesesense-impressions the preference over the otherswill be undertaking an impossible task. For if hemakes his judgment simply and without proof,he will be discredited; and if he wishes to employa proof, then says his proof is false, he will berefuting himself; if he says the proof is true, hewill be asked for a proof of its truth, and anotherproof for that one, and so on ad infinitum. But itis impossible to present an infinite series ofproofs; therefore he will not be able, even by theuse of proofs, to prefer one sense-impression toanother. And if a person is unable to pass judg-ment on the above-mentioned impressions eitherwithout proof or with proof, then the necessaryresult is suspension of judgment. Thus, while weare perhaps able to state of what nature eachobject appears to be in one particular position, atone particular distance, and in one location, it isnot in our power, for the reasons just given, todeclare what its true nature is. 6

5

4

Here we have a version of afamous skeptical argument,

one whose implications extend farbeyond the specific issue thatSextus is discussing.The underlyingissue is how any criterion orstandard of rational acceptabilitycan be established as correct: if bya proof, then what criterion orstandard can be used to show thatthe proof is itself cogent? (Sextussays here that the upshot is circu-larity, in which the very criterionwhose correctness is at issue isused to validate the proof; butthere is another possibility,discussed in the next section, thatis equally unacceptable.)

5

Here we have a somewhatdifferent set of examples,

based on the general phenomenonof perceptual relativity: the way inwhich the sense-impressions aperson receives from an objectdepend on the perspective fromwhich it is perceived.Again theissue arises of how to decidewhich of the various, seeminglyconflicting sense-impressions revealthe true nature of the object.

RR

6

And again the argument isthat there is no basis available

for choosing one sense-impressionover another as revealing theobject’s true nature because thereis no criterion for such a choicethat can be established as the rightone to appeal to. (Notice that theobjection at the end of the FourthMode was that the appeal to such acriterion leads to circularity, whilehere the objection is that it leads toan infinite regress of further criteriaand further proofs. Obviouslyeither result is possible.Thus theclearest version of the argumentwould say that it must lead to oneor the other of these two equallyunacceptable results.)

RR


1. Imagine a detailed example in which a particularobject seems to different people to have differentand conflicting qualities, depending on their“circumstances”—expanding the category of “cir-cumstances” to include also the differences of posi-tion discussed in the Fifth Mode.Try to formulate acriterion or standard for judging that some of thesense-impressions in question reveal the truenature of the object and others do not.Then try togive a justification or proof for this criterion and askwhat established this justification or proof as itself

correct. Is there any way to avoid falling into eithercircularity or an infinite regress?

2. One possible way to respond to Sextus is to say thata correct account of the nature of such an objectmust be able to explain all of the different appear-ances or sense-impressions, rather than just select-ing among them. (See BonJour’s account of how arepresentative realist might appeal to explanatoryinference.) In relation to a suitable example, try todecide whether this approach helps to any signifi-cant degree in solving the problem raised by Sextus.

CONCLUDING DIALOGUE ON THE PROBLEM OF THE EXTERNAL WORLD 121

Concluding Dialogue on the Problem of theExternal World

As I think you realize, this issue is sometimes hard to get people to take seriously. That wehave lots of knowledge about material objects—trees and buildings and cars and moun-tains—just seems so utterly obvious and undeniable that philosophers who raise doubtsabout it look like they need some sort of therapy—or maybe just need to be locked up sothey won’t hurt themselves!

I agree that the existence of the material world seems utterly obvious. But the philosophicalproblem is to say—in a sufficiently clear and perspicuous way—just why and how it reallyis obvious. If no one, after many, many tries, is able to do this, then it’s hard to dismiss thepossibility that it isn’t really so obvious after all—that perhaps common sense is ignoringor failing to notice problems that are perfectly genuine.

If we take the problem seriously, many recent epistemologists seem to think, following Reid,that the right response is direct realism: we apprehend material objects directly or imme-diately, and so the issue of trying to justify beliefs about the material world on the basisof subjective ideas or sense experience, in the way that Descartes or Locke or BonJour tryto do, just doesn’t arise. We don’t have to somehow reason our way to the existence of theexternal world because that’s where we start.

Direct realism, like almost any view that makes an otherwise very difficult problem just go away,is initially very appealing. But does it really deal adequately with the underlying issue?

Well, even BonJour concedes that the direct realist is right about at least one thing: that werarely if ever think very explicitly about our experiences, but instead just leap directly tobeliefs or judgments about material objects. Surely even Descartes or Locke wouldn’tseriously disagree about that?

I think that’s right, though one could wish that they were clearer about this issue. But the prob-lem is that this point doesn’t really speak in any clear way to the issue of justification.Even if we don’t think explicitly about our experience, it still seems hard to deny thatmaterial objects and the conscious experiences on the basis of which we seem somehowto know them are, from a metaphysical standpoint, quite distinct from each other: theoccurrence of an experience (over here, in me) is one state of affairs and the existence ofthe object it seems to depict (over there, outside of me) is a separate, quite different stateof affairs. Our mind doesn’t somehow literally reach out to grasp the world, nor do mate-rial objects somehow literally enter the mind. But then it still seems possible to askwhether those experiences provide good reasons or evidence for our claims about thoseexternal objects, and there is apparently nothing but facts about those experiences andour other mental states to appeal to in trying to answer this question.

So you’re saying that a successful answer to this question would still amount to some sort of arepresentative realist view. Maybe we don’t have to first think about our experience andthen explicitly infer to the world. But features of that experience or at least of ourconscious states of mind generally are the only place to look for good reasons for ourbeliefs about material objects—so we still seem to need the sort of inference thatDescartes and Locke and BonJour try to give an account of.

Yes. And if that is right, then direct realism really doesn’t speak to the main issue in any veryhelpful way.

But the representative realist project seems so desperately difficult! It’s very hard to give aclear and detailed description of the specific features of experience that might seem tobe relevant and even harder to say why they are relevant in the right way—why thosefeatures really do yield good reasons for thinking that a world containing just such-and-such sorts of material objects exists.

I agree that it’s difficult and that no one has done either of those things in a detailed and con-vincing way—at least not so far. But it’s a peculiar fact about the history of this issue thatrelatively little philosophical effort has been put into this project, with the vast majority


of philosophers either adopting some sort of idealism or else embracing skepticism.(Until recently they wouldn’t have been tempted by direct realism—what they wouldhave called “naïve realism.”)

I guess I’ve always had a soft spot in my heart for Berkeley’s view—or maybe forphenomenalism. If all that we have direct knowledge of is our experiences, why supposethat there is anything more? Why not just view ordinary objects as just patterns ofexperience of the right sort? What more do we really need?

Are you entirely sure that soft spot isn’t in your head?? In any case, it’s really important to dis-tinguish phenomenalism from Berkeley’s view, even though the two are often treated aspretty much the same thing. People who do that aren’t taking Berkeley’s appeal to Godseriously, but Berkeley surely took it very seriously.

I suppose I see what you mean. While the phenomenalist says that there is no further expla-nation for our experience and its patterns (or at least none that is knowable), Berkeleyis proposing a quite definite further explanation—one that is completely different fromLocke’s, but still an explanation—and one that appeals to something that is also outsideof experience!

Exactly. And, strange as it may seem at first, this makes Berkeley in effect an odd sort of rep-resentative realist—only what our sensory ideas represent or at least correspond to isn’tindependent material objects of the sort Locke advocates, but rather something aboutGod: the specific picture of the world that he has in mind and conveys to us. But Berke-ley still faces a quite parallel problem of justifying an inference from our experience tosomething outside of our experience—the sort of inference that Hume (as BonJourreports) thinks can never be justified.

So, as BonJour more or less suggests, the central issue between Berkeley and Locke is whoseexplanation is better, with both of them agreeing in opposition to the phenomenalist thatsome explanation is needed. They surely seem right on this last point: it would be betterto say that our experience must have an explanation, even if we don’t or even can’t knowwhat it is, than to say that all of that order and complexity is merely a brute fact that hasno explanation at all. And, as BonJour points out, phenomenalism faces lots of other prob-lems that no one has ever come close to solving.

Thus the crucial issue is whether there is any way to argue that one of these explanations—ormaybe yet some further explanation—is clearly better than any other in a way that makesit more likely to be true. BonJour’s tentative suggestion is that the Lockean material objectexplanation is to be preferred because it faces less risk of failure than the others by appeal-ing directly to the causation of experience by a material world rather than to some furtherbeing or mechanism that emulates such causation.

But that assumes that all of the other explanations must take that form. That isn’t clearlywrong. At the very least, it isn’t easy to think of an explanation not fitting this pattern thatreally explains why we have just the specific experiences that we do. But maybe (alwaysa problem with inferences to the best explanation!) there is some possibility that wehaven’t thought of.

That concern can, of course, be raised about almost any scientific explanation, so BonJour’sargument would be in good company if that were the worst problem. But such an argu-ment also needs a lot of detailed elaboration, both with respect to the patterns of experi-ence in question and with regard to the details of the explanation, elaboration that won’tbe easy to provide—if it can be done at all.

So maybe the lesson to be learned from all this is that we should lower our expectations. Thoughit isn’t in the readings, one important view about lowered expectations is the one thatImmanuel Kant defends in his famous book Critique of Pure Reason. Without going into thepretty convoluted details of Kant’s actual position, what he says is essentially that while wecan’t know what the external world that causes our experience is really like “in itself,” we canknow that it will regularly and reliably cause various sorts of experiential appearances. Whyisn’t that good enough? Why aren’t those regular and reliable appearances, whatever theirultimate source may be, an adequate basis for choices and actions and human life generally?

DAVID HUME: FROM AN ENQUIRY CONCERNING HUMAN UNDERSTANDING 123

But the problem is that “regular and reliable” part. If we could know about the external causeor source of our experience in the way that a successful Lockean or even Berkeleyan viewwould make possible, then perhaps we could know not just that experience has been “reg-ular and reliable” so far, but also that this will continue to be so. But if all we can know isthat the appearances have been generally regular and reliable up to now, it’s hard to seewhat good reason there is for thinking that this regularity and reliability will persist. Whycouldn’t the familiar patterns of experience suddenly change, in ways that would make ournormal choices and actions no longer yield satisfactory results? (Here we are pretty closeto the issue of induction, but we’ll talk about that in a separate discussion.)

So the point is that if, for example, our experience is really being produced by something-we-know-not-what—rather than by a knowable material world or even by a Berkeleyan Godwho is somehow known to be reliable—then we have no reason for thinking that such acause won’t suddenly and arbitrarily produce entirely different patterns of experiences.Maybe the experience of a solid-looking floor, instead of being followed by the experienceof being supported as we try to walk on it, will instead be followed by experiences offalling through, being injured, and so on.

Right! And if we can’t find any good reason why such an alteration isn’t just as likely as con-tinued “regularity and reliability,” then merely knowing appearances (so far) in the waythat Kant suggests isn’t very satisfactory from even a practical standpoint.

I guess you’ve convinced me, if I needed convincing, that this problem has to be taken seri-ously—and also that there isn’t any easy solution. It does seem as though there really isn’tany very good alternative to the representative realist’s attempt to find a justification forthe claim that the best explanation of our experience is some definite sort of independentreality. But I don’t think we should assume too quickly that the best such explanation willappeal to material objects—especially material objects as conceived by common sense.

I think that’s right. The issue of secondary qualities already seems to show that the best expla-nation is likely to depart from common sense in important ways: common sense seems toattribute qualities like the colors we experience to material objects, but it’s pretty hard tosee how such properties will have any serious explanatory role—and therefore hard to jus-tify this part of the common-sense view via an explanatory argument. What will explainour color experience is much more likely to be properties that are not at all like the onesthat we seem to experience—structural properties of the surfaces of objects in virtue ofwhich they reflect one sort of light rather than another.

But I had a more radical departure from common sense in mind. Maybe something like Berke-ley’s God is still in the running, after all?

There you go again.

Is Induction Justified?

David HumeDavid Hume (1711–1776), a Scottish philosopher and historian, is usuallyregarded as one of the most important and influential philosophers of all time.He did very influential work in epistemology, metaphysics, ethics, and thephilosophy of religion, producing a comprehensive philosophical position thatis most distinctive for its skeptical tendencies—tendencies that are reflected inthe present selection.

In the first section (Section IV) included in this selection, Hume presentswhat is perhaps the first reasonably clear formulation of the problem of induc-tion. His initial concern is with knowledge of causal relations, which he claimsto be the only basis for knowledge of matters of fact that go beyond directexperience and memory. His view is that causal relations can themselves be


From An Enquiry Concerning Human Understanding

From An Enquiry Concerning Human Understand-ing, 1748.

Section IV: Skeptical DoubtsConcerning the Operations of the

UnderstandingPart I

All the objects of human reason or inquiry maynaturally be divided into two kinds, to wit,relations of ideas, and matters of fact. Of the firstkind are the sciences of geometry, algebra, andarithmetic; and in short, every affirmation whichis either intuitively or demonstratively certain.That the square of the hypotenuse is equal to thesquares of the two sides, is a proposition whichexpresses a relation between these figures. Thatthree times five is equal to the half of thirty,expresses a relation between these numbers.Propositions of this kind are discoverable by themere operation of thought, without dependenceon what is anywhere existent in the universe.Though there never was a circle or triangle innature, the truths demonstrated by Euclid wouldfor ever retain their certainty and evidence.

Matters of fact, which are the second objectsof human reason, are not ascertained in thesame manner; nor is our evidence of theirtruth, however great, of a like nature with theforegoing. The contrary of every matter of factis still possible; because it can never imply acontradiction, and is conceived by the mindwith the same facility and distinctness, as ifever so conformable to reality. That the sunwill not rise tomorrow is no less intelligible a

1

known only through repeated experience of the causal sequence. But thisraises in turn the more general problem of how repeated experience of thing orproperty A always being followed by thing or property B can justify theconclusion that A will probably always be followed by B (which is at least partof the meaning of the claim that A causes B, in the sense of being a causallysufficient condition for B). Hume’s skeptical thesis is that there is no cogentreasoning that leads from such an experiential premise to this conclusion, sothat inductive reasoning of this sort is in fact not justified. (It is important torealize that he is not claiming merely that inductive conclusions cannot beknown with certainty: his stronger and much more startling claim is that thereis no good reason at all to accept them even as probable.)

In the second of the two sections (Section V—only briefly excerpted here),Hume offers a fundamentally psychological account of how and why wereason inductively, according to which such inferences rely on custom or habit(rather than on reason).

proposition, and implies no more contradictionthan the affirmation, that it will rise. Weshould in vain, therefore, attempt to demon-strate its falsehood. Were it demonstrativelyfalse, it would imply a contradiction, and couldnever be distinctly conceived by the mind.

It may, therefore, be a subject worthy ofcuriosity, to inquire what is the nature of thatevidence which assures us of any real existenceand matter of fact, beyond the present testimonyof our senses, or the records of our memory. . . .

All reasonings concerning matter of fact seemto be founded on the relation of cause and effect.By means of that relation alone we can gobeyond the evidence of our memory and senses.

If you were to ask a man, why he believesany matter of fact, which is absent; for instance,that his friend is in the country, or in France; hewould give you a reason; and this reason wouldbe some other fact; as a letter received from him,or the knowledge of his former resolutions andpromises. A man finding a watch or any othermachine in a desert island, would conclude thatthere had once been men in that island. All ourreasonings concerning fact are of the samenature. And here it is constantly supposed thatthere is a connection between the present factand that which is inferred from it. Were therenothing to bind them together, the inferencewould be entirely precarious. The hearing of anarticulate voice and rational discourse in thedark assures us of the presence of some person:Why? because these are the effects of the humanmake and fabric, and closely connected with it.If we anatomize all the other reasonings of this

3

2

2

The primary meaning of“matters of fact” seems to

be claims that can be deniedwithout contradiction and so arecontingent rather than necessary.But Hume is also saying that theseare not knowable a priori, and alsonot relations of ideas and so notanalytic, but rather synthetic.The overall implication is again thatall three distinctions (a priori/aposteriori, necessary/contingent,analytic/synthetic) coincide, thoughHume gives no real argument forthis claim (a version of moderateempiricism).

3

Thus the claim is that anycontingent fact that goes

beyond direct sense perception orthe memory of such perceptioncan be known only through causalreasoning.

RR

1

This distinction is the mainbasis for the subsequent argumentand needs to be considered care-fully.“Relations of ideas” suggeststhe idea of analyticity (see theAppendix to the chapter introduc-tion), but Hume is also claimingthat anything knowable a priori hasthis status—and also, in the nextparagraph, that all such claims arenecessary, since nothing contingenthas this status.Thus Hume is assum-ing in effect that the three maindistinctions pertaining to a prioriknowledge (see again the Appendixto the chapter introduction)coincide (in which case, amongother things, there would be nosynthetic a priori truths).


nature, we shall find that they are founded onthe relation of cause and effect, and that thisrelation is either near or remote, direct or collat-eral. Heat and light are collateral effects of fire,and the one effect may justly be inferred fromthe other.

If we would satisfy ourselves, therefore,concerning the nature of that evidence, whichassures us of matters of fact, we must inquire howwe arrive at the knowledge of cause and effect.

I shall venture to affirm, as a general proposi-tion, which admits of no exception, that theknowledge of this relation is not, in anyinstance, attained by reasonings a priori; butarises entirely from experience, when we findthat any particular objects are constantlyconjoined with each other. Let an object bepresented to a man of ever so strong naturalreason and abilities; if that object be entirelynew to him, he will not be able, by the mostaccurate examination of its sensible qualities, todiscover any of its causes or effects. Adam,though his rational faculties be supposed, at thevery first, entirely perfect, could not haveinferred from the fluidity and transparency ofwater that it would suffocate him, or from thelight and warmth of fire that it would consumehim. No object ever discovers, by the qualitieswhich appear to the senses, either the causeswhich produced it, or the effects which will arisefrom it; nor can our reason, unassisted by expe-rience, ever draw any inference concerning realexistence and matter of fact.

This proposition, that causes and effects arediscoverable, not by reason but by experience,will readily be admitted with regard to suchobjects, as we remember to have once beenaltogether unknown to us; since we must beconscious of the utter inability, which we then layunder, of foretelling what would arise from them.Present two smooth pieces of marble to a manwho has no tincture of natural philosophy; he willnever discover that they will adhere together insuch a manner as to require great force to sepa-rate them in a direct line, while they make sosmall a resistance to a lateral pressure. Suchevents, as bear little analogy to the commoncourse of nature, are also readily confessed to beknown only by experience; nor does any manimagine that the explosion of gunpowder, or theattraction of a loadstone, could ever be discov-ered by arguments a priori. In like manner, when

4

an effect is supposed to depend upon an intricatemachinery or secret structure of parts, we makeno difficulty in attributing all our knowledge of itto experience. Who will assert that he can givethe ultimate reason, why milk or bread is propernourishment for a man, not for a lion or a tiger?

But the same truth may not appear, at firstsight, to have the same evidence with regard toevents, which have become familiar to us fromour first appearance in the world, which bear aclose analogy to the whole course of nature, andwhich are supposed to depend on the simplequalities of objects, without any secret structureof parts. We are apt to imagine that we coulddiscover these effects by the mere operation ofour reason, without experience. We fancy, thatwere we brought on a sudden into this world, wecould at first have inferred that one billiard ballwould communicate motion to another uponimpulse; and that we needed not to have waitedfor the event, in order to pronounce with cer-tainty concerning it. Such is the influence ofcustom, that, where it is strongest, it not onlycovers our natural ignorance, but even concealsitself, and seems not to take place, merelybecause it is found in the highest degree.

But to convince us that all the laws of nature,and all the operations of bodies without excep-tion, are known only by experience, the followingreflections may, perhaps, suffice. Were any objectpresented to us, and were we required topronounce concerning the effect, which willresult from it, without consulting past observa-tion; after what manner, I beseech you, must themind proceed in this operation? It must invent orimagine some event, which it ascribes to theobject as its effect; and it is plain that thisinvention must be entirely arbitrary. The mindcan never possibly find the effect in the supposedcause, by the most accurate scrutiny andexamination. For the effect is totally differentfrom the cause, and consequently can never bediscovered in it. Motion in the second billiard ballis a quite distinct event from motion in the first:nor is there anything in the one to suggest thesmallest hint of the other. A stone or piece ofmetal raised into the air, and left without anysupport, immediately falls: but to consider thematter a priori, is there anything we discover inthis situation which can beget the idea of a down-ward, rather than an upward, or any other motion,in the stone or metal?

5 5

In the case of events that aresimilar enough to those with

which we are familiar, we may seemto ourselves to have direct insightsinto what will cause what that donot depend on experience. Humeis saying that this is an illusion.

RR

4

Thus A can be known tobe the cause of B only by

finding in experience that the twoare constantly conjoined (in theright order), that is that A isalways followed by B.

RR


And as the first imagination or invention of aparticular effect, in all natural operations, isarbitrary, where we consult not experience; somust we also esteem the supposed tie or connec-tion between the cause and effect, which bindsthem together, and renders it impossible that anyother effect could result from the operation ofthat cause. When I see, for instance, a billiardball moving in a straight line towards another;even suppose motion in the second ball shouldby accident be suggested to me, as the result oftheir contact or impulse; may I not conceive,that a hundred different events might as well fol-low from that cause? May not both these ballsremain at absolute rest? May not the first ballreturn in a straight line, or leap off from the sec-ond in any line or direction? All these supposi-tions are consistent and conceivable. Why thenshould we give the preference to one, which isno more consistent or conceivable than the rest?All our reasonings a priori will never be able toshow us any foundation for this preference.

In a word, then, every effect is a distinct eventfrom its cause. It could not, therefore, be discov-ered in the cause, and the first invention or con-ception of it, a priori, must be entirely arbitrary.And even after it is suggested, the conjunctionof it with the cause must appear equally arbi-trary; since there are always many other effects,which, to reason, must seem fully as consistentand natural. In vain, therefore, should we pre-tend to determine any single event, or infer anycause or effect, without the assistance of obser-vation and experience. . . .

Part II

But we have not yet attained any tolerablesatisfaction with regard to the question firstproposed. Each solution still gives rise to a newquestion as difficult as the foregoing, and leadsus on to farther inquiries. When it is asked, Whatis the nature of all our reasonings concerningmatter of fact? the proper answer seems to be,that they are founded on the relation of causeand effect. When again it is asked, What is thefoundation of all our reasonings and conclu-sions concerning that relation? it may be repliedin one word, experience. But if we still carry onour sifting humor, and ask, What is the founda-tion of all conclusions from experience? thisimplies a new question, which may be of moredifficult solution and explication. . . .

6

I shall content myself, in this section, with aneasy task, and shall pretend only to give anegative answer to the question here proposed.I say then, that, even after we have experience ofthe operations of cause and effect, our conclu-sions from that experience are not founded onreasoning, or any process of the understanding.This answer we must endeavor both to explainand to defend.

It must certainly be allowed, that nature haskept us at a great distance from all her secrets,and has afforded us only the knowledge of a fewsuperficial qualities of objects; while sheconceals from us those powers and principles onwhich the influence of those objects entirelydepends. Our senses inform us of the color,weight, and consistence of bread; but neithersense nor reason can ever inform us of those qual-ities which fit it for the nourishment and supportof a human body. Sight or feeling conveys an ideaof the actual motion of bodies; but as to that won-derful force or power, which would carry on amoving body for ever in a continued change ofplace, and which bodies never lose but bycommunicating it to others; of this we cannotform the most distant conception. But notwith-standing this ignorance of natural powers andprinciples, we always presume, when we see likesensible qualities, that they have like secret pow-ers, and expect that effects, similar to those whichwe have experienced, will follow from them. If abody of like color and consistence with thatbread, which we have formerly eat, be presentedto us, we make no scruple of repeating the exper-iment, and foresee, with certainty, like nourish-ment and support. Now this is a process of themind or thought, of which I would willinglyknow the foundation. It is allowed on all handsthat there is no known connection between thesensible qualities and the secret powers; and con-sequently, that the mind is not led to form such aconclusion concerning their constant and regularconjunction, by anything which it knows of theirnature. As to past experience, it can beallowed to give direct and certain information ofthose precise objects only, and that precise periodof time, which fell under its cognizance: but whythis experience should be extended to futuretimes, and to other objects, which, for aught weknow, may be only in appearance similar; this isthe main question on which I would insist. Thebread, which I formerly eat, nourished me; that is,

7

6

One important sort of causalrelation that this does not

seem to account for is that whichinvolves unobservable entities, as intheoretical science. See theBonJour selection (in the externalworld section of this chapter) for adiscussion of how such causalrelations might be known.

7

Hume might seem here to besuggesting that causal knowl-

edge would be easier to obtain if itwere somehow possible to directlyobserve these “secret powers.”But this is misleading. Even if wecould somehow observe the innernature of things (by which Humeseems to have in mind somethinglike their molecular or atomicstructure) and not merely their“superficial qualities,” the causalrelations between those qualitiesand any further effects would stillhave to be established throughexperience in essentially thesame way.


a body of such sensible qualities was, at that time,endued with such secret powers: but does it fol-low, that other bread must also nourish me atanother time, and that like sensible qualities mustalways be attended with like secret powers? Theconsequence seems no wise necessary. At least, itmust be acknowledged that there is here a conse-quence drawn by the mind; that there is a certainstep taken; a process of thought, and an inference,which wants to be explained. These two proposi-tions are far from being the same, I have foundthat such an object has always been attendedwith such an effect, and I foresee, that otherobjects, which are, in appearance, similar, will beattended with similar effects. I shall allow, if youplease, that the one proposition may justly beinferred from the other; I know, in fact, that italways is inferred. But if you insist that the infer-ence is made by a chain of reasoning, I desire youto produce that reasoning. The connectionbetween these propositions is not intuitive. Thereis required a medium, which may enable the mindto draw such an inference, if indeed it be drawnby reasoning and argument. What that medium is,I must confess, passes my comprehension; and itis incumbent on those to produce it, who assertthat it really exists, and is the origin of all ourconclusions concerning matter of fact.

This negative argument must certainly, inprocess of time, become altogether convincing,if many penetrating and able philosophers shallturn their inquiries this way and no one be everable to discover any connecting proposition orintermediate step, which supports the under-standing in this conclusion. But as the questionis yet new, every reader may not trust so far tohis own penetration, as to conclude, because anargument escapes his inquiry, that therefore itdoes not really exist. For this reason it may berequisite to venture upon a more difficult task:and enumerating all the branches of humanknowledge, endeavor to show that none of themcan afford such an argument.

All reasonings may be divided into two kinds,namely demonstrative reasoning, or that concern-ing relations of ideas, and moral reasoning, orthat concerning matter of fact and existence.That there are no demonstrative arguments in thecase seems evident; since it implies no contradic-tion that the course of nature may change, andthat an object, seemingly like those which wehave experienced, may be attended with different

9

8

or contrary effects. May I not clearly and dis-tinctly conceive that a body, falling from theclouds, and which, in all other respects, resem-bles snow, has yet the taste of salt or feeling offire? Is there any more intelligible propositionthan to affirm, that all the trees will flourish inDecember and January, and decay in May andJune? Now whatever is intelligible, and can bedistinctly conceived, implies no contradiction,and can never be proved false by any demonstra-tive argument or abstract reasoning a priori.

If we be, therefore, engaged by arguments toput trust in past experience, and make it the stan-dard of our future judgment, these argumentsmust be probable only, or such as regard matter offact and real existence, according to the divisionabove mentioned. But that there is no argument ofthis kind, must appear, if our explication of thatspecies of reasoning be admitted as solid andsatisfactory. We have said that all argumentsconcerning existence are founded on the relationof cause and effect; that our knowledge of thatrelation is derived entirely from experience; andthat all our experimental conclusions proceedupon the supposition that the future will be con-formable to the past. To endeavor, therefore, theproof of this last supposition by probable argu-ments, or arguments regarding existence, must beevidently going in a circle, and taking that forgranted, which is the very point in question.

In reality, all arguments from experience arefounded on the similarity which we discoveramong natural objects, and by which we areinduced to expect effects similar to those whichwe have found to follow from such objects. Andthough none but a fool or madman will everpretend to dispute the authority of experience,or to reject that great guide of human life, it maysurely be allowed a philosopher to have so muchcuriosity at least as to examine the principle ofhuman nature, which gives this mighty authorityto experience, and makes us draw advantagefrom that similarity which nature has placedamong different objects. From causes whichappear similar we expect similar effects. This isthe sum of all our experimental conclusions.Now it seems evident that, if this conclusionwere formed by reason, it would be as perfect atfirst, and upon one instance, as after ever solong a course of experience. But the case is farotherwise. . . . It is only after a long course ofuniform experiments in any kind, that we attain

11

10

8

Hume is claiming that there isno cogent reasoning of any

sort from the observational premisethat constant conjunction has beenobserved to the conclusion that thesame sequence will occur in othercases. His first reason for this is achallenge to his opponent toproduce such reasoning, to spell outin detail how an argument from thatsort of premise to that sort ofconclusion would go—somethingHume is confident cannot be done.

RR

10

A priori reasoning will notwork because such reasoning

depends in Hume’s view on theavoidance of contradiction, andthere is no contradiction in sup-posing that “the course of naturemay change” in such a way that asequence that has been experi-enced so far may cease to hold.

RR

9

His second and more impor-tant argument takes the

form of a dilemma, relying on thedistinction explained at the begin-ning of the selection. Accordingto this distinction, there are onlytwo possible sorts of reasoning: apriori reasoning and reasoning thatappeals to experience.

(Hume’s use of the term“moral” to refer to the latter is anarchaic usage that has nothing todo with the modern use of theterm to refer to matters having todo with ethics.)

RR

11

And reasoning that appeals toexperience will not work,

since the only way that experiencecan support a general conclusion ofthe sort in question is by generalizingfrom repeated sequences in exactlythe way that is at issue.Thus toappeal to any such reasoning (suchas by arguing that inductive reasoninghas been observed in the past toyield true conclusions and so is likelyto do so in the future as well) wouldthus beg the question.

RR


a firm reliance and security with regard to aparticular event. Now where is that process ofreasoning which, from one instance, draws aconclusion, so different from that which it infersfrom a hundred instances that are nowisedifferent from that single one? This question Ipropose as much for the sake of information, aswith an intention of raising difficulties. I cannotfind, I cannot imagine any such reasoning. But Ikeep my mind still open to instruction, if anyonewill vouchsafe to bestow it on me.

Should it be said that, from a number ofuniform experiments, we infer a connectionbetween the sensible qualities and the secret pow-ers; this, I must confess, seems the samedifficulty, couched in different terms. The ques-tion still recurs, on what process of argument thisinference is founded? Where is the medium, theinterposing ideas, which join propositions sovery wide of each other? It is confessed that thecolor, consistence, and other sensible qualities ofbread appear not, of themselves, to have any con-nection with the secret powers of nourishmentand support. For otherwise we could infer thesesecret powers from the first appearance of thesesensible qualities, without the aid of experience;contrary to the sentiment of all philosophers, andcontrary to plain matter of fact. Here, then, is ournatural state of ignorance with regard to thepowers and influence of all objects. How is thisremedied by experience? It only shows us a num-ber of uniform effects, resulting from certainobjects, and teaches us that those particularobjects, at that particular time, were endowedwith such powers and forces. When a new object,endowed with similar sensible qualities, is pro-duced, we expect similar powers and forces, andlook for a like effect. From a body of like colorand consistence with bread we expect like nour-ishment and support. But this surely is a step orprogress of the mind, which wants to beexplained. When a man says, I have found, in allpast instances, such sensible qualities conjoinedwith such secret powers: And when he says,Similar sensible qualities will always beconjoined with similar secret powers, he is notguilty of a tautology, nor are these propositions inany respect the same. You say that the one propo-sition is an inference from the other. But youmust confess that the inference is not intuitive;neither is it demonstrative: Of what nature isit, then? To say it is experimental, is begging the

12

question. For all inferences from experience sup-pose, as their foundation, that the future willresemble the past, and that similar powers will beconjoined with similar sensible qualities. If therebe any suspicion that the course of nature maychange, and that the past may be no rule for thefuture, all experience becomes useless, and cangive rise to no inference or conclusion. It isimpossible, therefore, that any arguments fromexperience can prove this resemblance of the pastto the future; since all these arguments arefounded on the supposition of that resemblance.Let the course of things be allowed hitherto everso regular; that alone, without some new argu-ment or inference, proves not that, for the future,it will continue so. In vain do you pretend to havelearned the nature of bodies from your past expe-rience. Their secret nature, and consequently alltheir effects and influence, may change, withoutany change in their sensible qualities. Thishappens sometimes, and with regard to someobjects: Why may it not happen always, and withregard to all objects? What logic, what process ofargument secures you against this supposition?My practice, you say, refutes my doubts. But youmistake the purport of my question. As an agent,I am quite satisfied in the point; but as a philoso-pher, who has some share of curiosity, I will notsay scepticism, I want to learn the foundation ofthis inference. No reading, no inquiry has yetbeen able to remove my difficulty, or give mesatisfaction in a matter of such importance. Can Ido better than propose the difficulty to the public,even though, perhaps, I have small hopes ofobtaining a solution? We shall, at least, by thismeans, be sensible of our ignorance, if we do notaugment our knowledge.

I must confess that a man is guilty of unpar-donable arrogance who concludes, because anargument has escaped his own investigation, thattherefore it does not really exist. I must also con-fess that, though all the learned, for several ages,should have employed themselves in fruitlesssearch upon any subject, it may still, perhaps, berash to conclude positively that the subject must,therefore, pass all human comprehension. Eventhough we examine all the sources of our knowl-edge, and conclude them unfit for such a subject,there may still remain a suspicion, that the enu-meration is not complete, or the examination notaccurate. But with regard to the present subject,there are some considerations which seem to

12

To say that the inference is“intuitive” would be to say we havea direct a priori insight that theconclusion follows; to say it is“demonstrative” would be to saythere is a more extended a prioriargument involving intermediatesteps. But both of these ways of jus-tifying the inference depend, Humethinks, on it being contradictory toaccept the experiential premise andreject the conclusion—which, hehas argued, is not the case here.


remove all this accusation of arrogance or suspi-cion of mistake.

It is certain that the most ignorant and stupidpeasants—nay infants, nay even brute beasts—improve by experience, and learn the qualities ofnatural objects, by observing the effects whichresult from them. When a child has felt the sen-sation of pain from touching the flame of a can-dle, he will be careful not to put his hand nearany candle; but will expect a similar effect froma cause which is similar in its sensible qualitiesand appearance. If you assert, therefore, that theunderstanding of the child is led into thisconclusion by any process of argument or ratio-cination, I may justly require you to produce thatargument; nor have you any pretense to refuse soequitable a demand. You cannot say that the argu-ment is abstruse, and may possibly escape yourinquiry; since you confess that it is obvious to thecapacity of a mere infant. If you hesitate, there-fore, a moment, or if, after reflection, you pro-duce any intricate or profound argument, you, ina manner, give up the question, and confess thatit is not reasoning which engages us to supposethe past resembling the future, and to expectsimilar effects from causes which are, to appear-ance, similar. This is the proposition which Iintended to enforce in the present section. If I beright, I pretend not to have made any mightydiscovery. And if I be wrong, I must acknowl-edge myself to be indeed a very backwardscholar; since I cannot now discover an argumentwhich, it seems, was perfectly familiar to melong before I was out of my cradle.

Section V: Skeptical Solution of TheseDoubts

. . . Though we should conclude, for instance, asin the foregoing section, that, in all reasoningsfrom experience, there is a step taken by themind which is not supported by any argument orprocess of the understanding; there is no dangerthat these reasonings, on which almost allknowledge depends, will ever be affected bysuch a discovery. If the mind be not engaged byargument to make this step, it must be inducedby some other principle of equal weight andauthority; and that principle will preserve itsinfluence as long as human nature remains thesame. What that principle is may well be worththe pains of inquiry.

13

Suppose a person, though endowed with thestrongest faculties of reason and reflection, to bebrought on a sudden into this world; he would,indeed, immediately observe a continual succes-sion of objects, and one event followinganother; but he would not be able to discoveranything farther. He would not, at first, by anyreasoning, be able to reach the idea of cause andeffect; since the particular powers, by which allnatural operations are performed, never appearto the senses; nor is it reasonable to conclude,merely because one event, in one instance,precedes another, that the one is the cause, theother the effect. Their connection may bearbitrary and casual. There may be no reason toinfer the existence of one from the appearanceof the other. And in a word, such a person, with-out more experience, could never employ hisconjecture or reasoning concerning any matterof fact, or be assured of anything beyond whatwas immediately present to his senses.

Suppose again, that he has acquired moreexperience and has lived so long in the world asto have observed familiar objects or events to beconstantly conjoined together; what is the con-sequence of this experience? He immediatelyinfers the existence of one object from theappearance of the other.Yet he has not, by all hisexperience, acquired any idea or knowledge ofthe secret power by which the one object pro-duces the other ; nor is it, by any process ofreasoning, he is engaged to draw this inference.But still he finds himself determined to draw it:And though he should be convinced that hisunderstanding has no part in the operation, hewould nevertheless continue in the same courseof thinking. There is some other principle whichdetermines him to form such a conclusion.

This principle is custom or habit. For whereverthe repetition of any particular act or operationproduces a propensity to renew the same act oroperation, without being impelled by any reason-ing or process of the understanding, we alwayssay, that this propensity is the effect of custom. Byemploying that word, we pretend not to have giventhe ultimate reason of such a propensity. We onlypoint out a principle of human nature, which isuniversally acknowledged, and which is wellknown by its effects. Perhaps we can push ourinquiries no farther, or pretend to give the cause ofthis cause; but must rest contented with it as theultimate principle which we can assign, of all our

14

13

Hume reinforces the chal-lenge to his opponent by

claiming that the reasoning inquestion, if it existed, could notbe very difficult to specify, since itwould have to be familiar even toyoung children (who obviouslydraw such conclusions).

RR

14

As already noted (seeAnnotation 7), the reference

to “secret powers” is highlymisleading, in that there is no reasonto think that we would not face thesame essential problem no matterhow much knowledge of suchpowers we might have.


conclusions from experience. It is sufficient satis-faction, that we can go so far, without repining atthe narrowness of our faculties because they willcarry us no farther. And it is certain we hereadvance a very intelligible proposition at least, ifnot a true one, when we assert that, after the con-stant conjunction of two objects—heat and flame,for instance, weight and solidity—we are deter-mined by custom alone to expect the one from theappearance of the other. This hypothesis seemseven the one which explains the difficulty, why wedraw from a thousand instances, an inference thatwe are not able to draw from one instance, that is,in no respect, different from them. Reason is inca-pable of any such variation. The conclusions thatit draws from considering one circle are the samewhich it would form surveying all the circles in theuniverse. But no man, having seen only one bodymove after being impelled by another, could inferthat every other body will move after a likeimpulse. All inferences from experience, there-fore, are effects of custom, not of reasoning.

Custom, then, is the great guide of human life.It is that principle alone which renders our experi-ence useful to us, and makes us expect, for thefuture, a similar train of events with those whichhave appeared in the past. Without the influenceof custom, we should be entirely ignorant of everymatter of fact beyond what is immediately presentto the memory and senses. We should never knowhow to adjust means to ends, or to employ our nat-ural powers in the production of any effect.There would be an end at once of all action, aswell as of the chief part of speculation.

…

What, then, is the conclusion of the wholematter? A simple one; though, it must be con-fessed, pretty remote from the common theories ofphilosophy. All belief of matter of fact or real exis-tence is derived merely from some object, presentto the memory or senses, and a customary con-junction between that and some other object. Or inother words; having found, in many instances, thatany two kinds of objects—flame and heat, snowand cold—have always been conjoined together; ifflame or snow be presented anew to the senses, themind is carried by custom to expect heat or cold,and to believe that such a quality does exist, andwill discover itself on a nearer approach. Thisbelief is the necessary result of placing the mind insuch circumstances. . . . All these operations are a

16

15

species of natural instincts, which no reasoning orprocess of the thought and understanding is ableeither to produce or to prevent.

…

Here, then, is a kind of pre-established harmonybetween the course of nature and the succession ofour ideas; and though the powers and forces, bywhich the former is governed, be wholly unknownto us; yet our thoughts and conceptions have still,we find, gone on in the same train with the otherworks of nature. Custom is that principle, bywhich this correspondence has been effected; sonecessary to the subsistence of our species, andthe regulation of our conduct, in every circum-stance and occurrence of human life. Had not thepresence of an object, instantly excited the idea ofthose objects, commonly conjoined with it, all ourknowledge must have been limited to the narrowsphere of our memory and senses; and we shouldnever have been able to adjust means to ends, oremploy our natural powers, either to the produc-ing of good, or avoiding of evil. Those who delightin the discovery and contemplation of finalcauses, have here ample subject to employ theirwonder and admiration.

I shall add, for a further confirmation of theforegoing theory, that, as this operation of themind, by which we infer like effects from likecauses, and vice versa, is so essential to the sub-sistence of all human creatures, it is not probable,that it could be trusted to the fallacious deductionsof our reason, which is slow in its operations;appears not, in any degree, during the first yearsof infancy; and at best is, in every age and periodof human life, extremely liable to error and mis-take. It is more conformable to the ordinary wis-dom of nature to secure so necessary an act of themind, by some instinct or mechanical tendency,which may be infallible in its operations, may dis-cover itself at the first appearance of life andthought, and may be independent of all thelabored deductions of the understanding. Asnature as taught us the use of our limbs, withoutgiving us the knowledge of the muscles andnerves, by which they are actuated; so has sheimplanted in us an instinct, which carries forwardthe thought in a correspondent course to thatwhich she has established among external objects;though we are ignorant of those powers andforces, on which this regular course and succes-sion of objects totally depends. 18

17

15

Here Hume seems to beassuming that the fact that

there are many repeated instances(and no conflicting ones) is itselfrationally irrelevant, that onlythe character of the individualinstances makes any differenceto reason, which is why the sameconclusion would follow from oneinstance as from many. But mightnot the fact of repetition supporta general conclusion even ifindividual instances do not?

16

But that we have knowledgerather than ignorance on

this basis seems to depend onthere being some reason forthinking that the results arrived atin this way are at least likely to betrue.According to Hume, there isno such reason.

18

Here the unjustified assump-tion that customary or

habitual inductive reasoning willnonetheless lead to the truth iseven clearer. Such reasoning maybe “infallible” in the sense thatit never fails to function, but ifHume’s earlier argument wascorrect, there is no reason tothink that its results will be infalli-ble in the sense of producingtrue conclusions.

17

This is a very puzzlingpassage. Hume seems to be

assuming that the general conclu-sions arrived at through the opera-tion of custom will in fact be inaccordance with “the other worksof nature”—that is, that they will betrue. (This is the “pre-establishedharmony,” though it is not “pre-established” in any very clear way.)But by his own earlier argument,he has no good reason to think thatthis will be so in general or evensome of the time.

WESLEY SALMON: THE PROBLEM OF INDUCTION, FROM THE FOUNDATIONS OF SCIENTIFIC INFERENCE 131


1. Attempts are sometimes made to justify inductivereasoning by adding a further premise or prin-ciple to the argument: the premise that thefuture will probably resemble the past, or perhapsthat unobserved cases will probably resembleobserved cases. What would be Hume’s likelyresponse to such an attempt?

2. Hume seems to assume that any case in which onesort of event A was always followed (in the futureas well) by another sort of event B would be a caseof causation,but this is too simple a picture.Can youthink of one or more common-sense examples inwhich it is plausible (apart from worries aboutinduction) to suppose that A is always followed byB, but in which A is not the cause or even part ofthe cause of B? (Hint: think of cases where both Aand B are caused by some third sort of event C.)

3. Look back at the discussion of theoretical orexplanatory reasoning in the BonJour selection (inthe external world section of this chapter). Couldthe acceptance of an inductive conclusion bejustified by this sort of reasoning on the groundsthat the existence of an objective regularity innature is the best explanation of the observedregular sequence? What other explanations arepossible for such an observed sequence? Doessuch an argument show that the inductive conclu-sion is probably true, given the truth of thepremise? What do you think Hume would sayabout this sort of defense of induction? Is he per-haps even tacitly assuming something like this atthe end of his discussion? (This question is muchharder and more complicated than most of thediscussion questions.)

Wesley Salmon (1925–2001) was an American philosopher who taught atIndiana University, the University of Arizona, and the University of Pittsburgh.Salmon wrote widely in the philosophy of science, with his work on explana-tion and the related topic of causality being especially influential. In thefollowing selection, Salmon restates the problem of induction as it grows outof Hume’s discussion and then considers and evaluates a variety of possiblesolutions (some of which have been omitted here). In the process of doingthis, he also offers a version of the moderate empiricist view of a priori knowl-edge (see the chapter introduction). This is relevant because Salmon wantsto argue against the possibility of an a priori justification of induction. (Seethe following selection from Ewing for a contrasting view.) This is one of themore challenging selections in this book, and you will have to work hard totry to understand it.

Wesley Salmon

The Problem of Induction, from The Foundationsof Scientific Inference

We all believe that we have knowledge of factsextending far beyond those we directly perceive.The scope of our senses is severely limited inspace and time; our immediate perceptual knowl-edge does not reach to events that happenedbefore we were born, to events that are happeningnow in certain other places, or to any futureevents. We believe, nevertheless, that we havesome kind of indirect knowledge of such facts.

We know that a glacier once covered a large partof North America, that the sun continues to existat night, and that the tides will rise and fall tomor-row. Science and common sense have at least thisone thing in common: Each embodies knowledgeof matters of fact that are not open to our directinspection. Indeed, science purports to establishgeneral laws or theories that apply to all parts ofspace and time without restriction. A “science”that consisted of no more than a mere summaryof the results of direct observation would notdeserve the name.

From The Foundations of Scientific Inference(Pittsburgh: University of Pittsburgh Press, 1967).


Hume’s profound critique of induction beginswith a simple and apparently innocent question:How do we acquire knowledge of the unob-served?* This question, as posed, may seem tocall for an empirical answer. We observe thathuman beings utilize what may be roughlycharacterized as inductive or scientific methodsof extending knowledge from the observed tothe unobserved. The sciences, in fact, embodythe most powerful and highly developed meth-ods known, and we may make an empiricalinvestigation of scientific methods much as wemight for any other sort of human behavior. Wemay consider the historical development ofscience. We may study the psychological, socio-logical, and political factors relevant to thepursuit of science. We may try to give an exactcharacterization of the behavior of scientists. Indoing all these things, however, important andinteresting as they are, we will have ignored thephilosophical aspect of the problem Humeraised. Putting the matter very simply, theseempirical investigations may enable us todescribe the ways in which people arrive atbeliefs about unobserved facts, but they leaveopen the question of whether beliefs arrived atin this way actually constitute knowledge. It isone thing to describe how people go about seek-ing to extend their knowledge; it is quite anotherto claim that the methods employed actually doyield knowledge.

One of the basic differences between knowl-edge and belief is that knowledge must befounded upon evidence—i.e., it must be belieffounded upon some rational justification. To saythat certain methods yield knowledge of theunobserved is to make a cognitive claim for them.Hume called into question the justification ofsuch cognitive claims. The answer cannot befound entirely within an empirical study ofhuman behavior, for a logical problem has beenraised. It is the problem of understanding thelogical relationship between evidence andconclusion in logically correct inferences. It is theproblem of determining whether the inferencesby which we attempt to make the transition fromknowledge of the observed to knowledge of theunobserved are logically correct. The fact that

people do or do not use a certain type of inferenceis irrelevant to its justifiability. Whether peoplehave confidence in the correctness of a certaintype of inference has nothing to do with whethersuch confidence is justified. If we should adopt alogically incorrect method for inferring one factfrom others, these facts would not actuallyconstitute evidence for the conclusion we havedrawn. The problem of induction is the problemof explicating the very concept of inductiveevidence.

…

Consider a simple and highly artificial situa-tion. Suppose a number of balls have beendrawn from an urn, and that all of the black onesthat have been drawn are licorice-flavored. . . .The problem—Hume’s basic philosophicalproblem—is this: Given that all of the observedblack balls have been licorice-flavored, andgiven that somehow the conclusion has beenentertained that the unobserved black balls inthe urn are also licorice-flavored, do theobserved facts constitute sound evidence forthat conclusion? Would we be justified inaccepting that conclusion on the basis of thefacts alleged to be evidence for it?

As a first answer to this question we maypoint out that the inference does conform to anaccepted inductive principle, a principle sayingroughly that observed instances conforming to ageneralization constitute evidence for it. Itis, however, a very small step to the next ques-tion: What grounds have we for accepting this orany other inductive principle? Is there anyreason or justification for placing confidence inthe conclusions of inferences of this type?Given that the premises of this inference aretrue, and given that the inference conforms to acertain rule, can we provide any rational justifi-cation for accepting its conclusion rather than,for instance, the conclusion that black balls yetto be drawn will taste like quinine?

It is well known that Hume’s answer to thisproblem was essentially skeptical. It was hisgreat merit to have shown that a justification ofinduction, if possible at all, is by no means easyto provide. In order to appreciate the force of his

3

2

1

*David Hume, Enquiry Concerning Human Understanding.

1

It would be somewhat moreperspicuous to say that the

problem is explaining why inductiveevidence, as standardly understood(see the chapter introduction), isgood evidence—why it genuinelysupports the sort of conclusion thatis standardly drawn from it.

2

This is not the best example.There is no reason to think

that drawing from an urn reflectsany real regularity in nature, whichis what the problem of induction ismainly concerned with.

3

That is, that observations ofcases of A’s that are also B’s

constitute evidence for the gener-alization that all A’s are B’s.

RR


necessarily truth-preserving inferences that arealso ampliative. Is there any type of inferencewhose conclusion must, of necessity, be true ifthe premises are true, but whose conclusion sayssomething not stated by the premises? Humebelieved that the answer is negative and so doI, but it is not easy to produce an adequatedefense of this answer. Let us see, however, whatan affirmative answer would amount to.

Suppose there were an ampliative inferencethat is also necessarily truth-preserving.Consider the implication from its premises,P1, . . . , Pk, to its conclusion C. If the inferencewere an ordinary nonampliative deduction, thisimplication would be analytic and empty; butsince the argument is supposed to be amplia-tive, the implication must be synthetic. At thesame time, because the argument is supposedto be necessarily truth-preserving, this impli-cation must be not only true but necessarilytrue. Thus, to maintain that there are inferencesthat are both ampliative and necessarily truth-preserving is tantamount to asserting that thereare synthetic a priori truths. . . .

If we agree that there are no synthetic apriori truths, then we must identify necessarilytruth-preserving inference with nonampliativeinference. All ampliative inference is non-demonstrative. This leads to an exhaustivetrichotomy of inferences: valid deductive infer-ence, correct inductive inference, and assortedfallacies. The first question is, however, whetherthe second category is empty or whether thereare such things as correct inductive inferences.This is Hume’s problem of induction. Can weshow that any particular type of ampliative infer-ence can be justified in any way? If so, it willqualify as correct induction.

Consider, then, any ampliative inferencewhatever. The example of the licorice-flavoredblack balls illustrates the point. We cannotshow deductively that this inference will have atrue conclusion given true premises. If wecould, we would have proved that the conclu-sion must be true if the premises are. Thatwould make it necessarily truth-preserving,hence, demonstrative. This, in turn, wouldmean that it was nonampliative, contrary to ourhypothesis. Thus, if an ampliative inferencecould be justified deductively it would not beampliative. It follows that ampliative inferencecannot be justified deductively.

6

argument it is first necessary to clarify someterminological points. This is particularlyimportant because the word induction has beenused in a wide variety of ways.

For purposes of systematic discussion onedistinction is fundamental, namely, the distinctionbetween demonstrative and nondemonstrativeinference. A demonstrative inference is onewhose premises necessitate its conclusion; theconclusion cannot be false if the premises are true.All valid deductions are demonstrative inferences.A nondemonstrative inference is simply one thatfails to be demonstrative. Its conclusion is notnecessitated by its premises; the conclusion couldbe false even if the premises are true. A demon-strative inference is necessarily truth-preserving;a nondemonstrative inference is not.

The category of nondemonstrative inferences,as I have characterized it, contains, among otherthings perhaps, all kinds of fallacious inferences.If, however, there is any kind of inference whosepremises, although not necessitating the conclu-sion, do lend it weight, support it, or make itprobable, then such inferences possess a certainkind of logical rectitude. It is not deductive valid-ity, but it is important anyway. Inferencespossessing it are correct inductive inferences.

Since demonstrative inferences have beencharacterized in terms of their basic property ofnecessary truth preservation, it is natural to askhow they achieve this very desirable trait. For alarge group of demonstrative inferences, includ-ing those discussed under “valid deduction” inmost logic texts, the answer is rather easy.Inferences of this type purchase necessary truthpreservation by sacrificing any extension ofcontent. The conclusion of such an inference saysno more than do the premises—often less. Theconclusion cannot be false if the premises are truebecause the conclusion says nothing that was notalready stated in the premises. The conclusion isa mere reformulation of all or part of the contentof the premises. In some cases the reformulationis unanticipated and therefore psychologicallysurprising, but the conclusion cannot augmentthe content of the premises. Such inferences arenonampliative; an ampliative inference, then, hasa conclusion with content not present eitherexplicitly or implicitly in the premises.

While it is easy to understand why nonamplia-tive inferences are necessarily truth-preserving,the further question arises whether there are any

5

4

5

Salmon’s view is that alllogically valid arguments are

“nonampliative” because thecontent of the conclusion is pres-ent, at least “implicitly,” in thecontent of the premises—wherethis is supposed to explain why theconclusion must be true if thepremises are true. (See thefollowing selection by Ewing formore discussion of this issue.)

RR

6

The statement that all caseswhere the premises are true

are cases where the conclusion istrue would be a necessary truth(and, Salmon is assuming, therebyknowable a priori), but would besynthetic (in Kant’s sense) becausethe predicate is not included in thesubject. (See the discussion of apriori knowledge in the Appendixto the chapter introduction.)

RR

4

It is also important to beclear that a demonstrative argu-ment guarantees only that the con-clusion will be true if the premisesare true.

Notice that this so far saysnothing about why particular

arguments have the feature ofbeing necessarily truth-preserving.


At the same time, we cannot justify any sortof ampliative inference inductively. To do sowould require the use of some sort of non-demonstrative inference. But the question atissue is the justification of nondemonstrativeinference, so the procedure would be questionbegging. Before we can properly employ anondemonstrative inference in a justifyingargument, we must already have justified thatnondemonstrative inference.

Hume’s position can be summarized suc-cinctly: We cannot justify any kind of ampliativeinference. If it could be justified deductively itwould not be ampliative. It cannot be justifiednondemonstratively because that would beviciously circular. It seems, then, that there is noway in which we can extend our knowledge to theunobserved. We have, to be sure, many beliefsabout the unobserved, and in some of them weplace great confidence. Nevertheless, they arewithout rational justification of any kind!

This is a harsh conclusion, yet it seems to besupported by impeccable arguments. It might becalled “Hume’s paradox,” for the conclusion,although ingeniously argued, is utterly repugnantto common sense and our deepest convictions.We know (“in our hearts”) that we have knowl-edge of unobserved fact. The challenge is toshow how this is possible.

II. Attempted SolutionsIt hardly needs remarking that philosophershave attempted to meet Hume’s intriguingchallenge in a wide variety of ways. . . . In thissection I shall survey what seem to me to bethe most important efforts to deal with theproblem.

1. Inductive Justification. If Hume’s argu-ments had never been propounded and we wereasked why we accept the methods of science,the most natural answer would be, I think, thatthese methods have proved themselves by theirresults. We can point to astonishing technolog-ical advances, to vastly increased comprehen-sion, and to impressive predictions. Science hasprovided us with foresight, control, and under-standing. No other method can claim a compa-rable record of successful accomplishment. Ifmethods are to be judged by their fruits, there isno doubt that the scientific method will comeout on top.

8

7

Unfortunately, Hume examined this argumentand showed that it is viciously circular. It is anexample of an attempt to justify inductivemethods inductively. From the premise thatscience has had considerable predictive successin the past, we conclude that it will continue tohave substantial predictive success in the future.Observed cases of the application of scientificmethod have yielded successful prediction;therefore, as yet unobserved cases of the applica-tion of scientific method will yield successfulpredictions. This argument has the same structureas our black-balls-in-the-urn example; it isprecisely the sort of ampliative inference fromthe observed to the unobserved whose justifiabil-ity is in question.

Consider the parallel case for a radicallydifferent sort of method. A crystal gazerclaims that his method is the appropriatemethod for making predictions. When wequestion his claim he says, “Wait a moment; Iwill f ind out whether the method of crystalgazing is the best method for making predic-tions.” He looks into his crystal ball andannounces that future cases of crystal gazingwill yield predictive success. If we shouldprotest that his method has not been especiallysuccessful in the past, he might well makecertain remarks about parity of reasoning.“Since you have used your method to justifyyour method, why shouldn’t I use my methodto justify my method? If you insist upon judg-ing my method by using your method, whyshouldn’t I use my method to evaluate yourmethod? By the way, I note by gazing into mycrystal ball that the scientific method is now infor a very bad run of luck.”

The trouble with circular arguments is obvious:with an appropriate circular argument you canprove anything. . . .

4. Synthetic a priori Principles. A longphilosophical tradition, dating back to antiquity,denies the empiricist claim that knowledge ofthe world rests solely upon observationalevidence—that factual knowledge is limited towhat we can observe and what we can infertherefrom. In the modern period, this rationalis-tic tradition is represented by men likeDescartes and Leibniz who took their inspira-tion from the abstract aspect of modern physics.After Hume’s devastating criticism of induction,Kant provided a more precise formulation, a

9

7

This paragraph and thepreceding one contain

Salmon’s version of Hume’sdilemma argument.

8

To see just how “harsh” or“repugnant” this conclusion

is, ask yourself how much of ourknowledge would be underminedif induction were rejected,and how much would be left.(See Discussion Question 1.)

9

Not all “methods” will auto-matically be self-justifying

in this way: if crystal balls dosometimes give clear and definiteanswers to questions, then someof them might say that crystal ballsare not reliable or might give nodefinite answer either way to thisparticular question. But while amethod that undermined itself inthis way would be even less defen-sible, this does nothing to showthat self-justification is enoughfor genuine justification.


fuller elaboration, and a more subtle defense ofrationalism than any that had been given earlier(or, quite possibly, subsequently). As Kant him-self testified, it was Hume who awakened himfrom his “dogmatic slumbers” and thereuponstimulated the Critique of Pure Reason.

The doctrine that there are synthetic a prioritruths is, as I explained above, tantamount to theview that there are necessarily truth-preservingampliative inferences. If we could find a bonafide demonstrative ampliative inference wewould have a solution to Hume’s problem of theground of inference from the observed to theunobserved. This solution could be presented ineither of two ways. First, one could assert thatthere are factual propositions that can be estab-lished by pure reason—without the aid ofempirical evidence—and that these synthetic apriori propositions, in conjunction with prem-ises established by observation, make it possibleto deduce (nonampliatively) conclusions per-taining to unobserved matters of fact. Second,one could claim that these synthetic a prioripropositions, although not added as premises toampliative inferences to render them nonam-pliative, do instead provide a warrant for gen-uinely ampliative inferences from the observedto the unobserved. . . .

In order to appreciate the philosophicalissues involved in the attempt to justify induc-tion by means of a synthetic a priori principle,we must introduce some reasonably precisedefinitions of key terms. Two pairs of conceptsare involved: first, the distinction betweenanalytic and synthetic statements, and second,the distinction between a priori and a posteri-ori statements. . . .

Although it is reasonable to maintain, I think,that all logical truths are analytic, there seem to beanalytic statements that are not logical truths.

For instance,

All bachelors are unmarried

is not [a logical truth]. However, given thedefinition,

Bachelor = unmarried adult male,

the foregoing statement can be transformed intoa logical truth, for the definition gives license tosubstitute the definiens, “unmarried adultmale,” for the definiendum “bachelor.” Thissubstitution yields,

11

10

All unmarried adult males are unmarried,

which is a logical truth. To incorporate cases ofthis sort, we may define an analytic statement asone that is a logical truth or can be transformedinto a logical truth by definitional substitutionof definiens for definiendum. The negationof an analytic truth is a self-contradiction. Anystatement that is neither analytic nor self-contradictory is synthetic. . . .

…

Analytic statements are often said . . . to bedevoid of any factual content. Although there aredifficulties in giving an adequate account of theconcept of factual content, enough can be said toilluminate its relation to analytic statements. Thebasic feature seems to be that factual content of astatement is a measure of the capacity of thatstatement to rule out possibilities. In this respect,it is a negative concept. In a state of total igno-rance all possible states of affairs are live possibil-ities; any possible state of affairs might, for all weknow, be the actual state of things. As knowledgeaccumulates, we realize that some of the possibil-ities are not actualized. The statements expressingour knowledge are incompatible with descriptionsof various possible worlds, so we know that thesepossibilities are ruled out—our actual world doesnot coincide with any of these possibilities that areincompatible with what we know. Generallyspeaking, moreover, the greater our knowledge—the greater the factual content of the statementswe know—the more possibilities are disqualifiedfrom being actual. Imagine, for instance, theinhabitants of Plato’s famous cave, who are totallyignorant of the nature of the external world. Theycan imagine birds of all sorts, including ravens ofvarious colors. When the emissary to the outerworld returns and reports that all ravens are black,those who remained in the cave can rule out allpossibilities that had room for ravens of other col-ors. The statement, “All ravens are black,” has fac-tual content because of the descriptions of possi-ble worlds with which it is incompatible. If,however, the emissary should return and remarkthat every raven is either black or nonblack, hisstatement would be totally lacking in content, andthe permanent inhabitants of the cave—anxiousfor knowledge of the external world—would bejustly furious with him for his empty report. Hisstatement would lack content because it is com-

12

11

A logical truth is one that isguaranteed to be true by virtue ofits form. (We have omitted Salmon’smore technical account, whichamounts to the same thing.)

10

The first of these alternativeshas no plausibility at all. No

one has ever claimed that the con-clusion of an inductive argument isguaranteed to be true, since it isbeyond question that such conclu-sions are occasionally false—forexample, when the observedregularity was just an accident.The second alternative can avoidthis problem because the principlein question might say only that theconclusion of an inductive argu-ment is probably true when theobservational premises are true.

12

This is Salmon’s maindefinition of “analytic.” You

should compare it to Kant’s, asdiscussed in the Appendix to thechapter introduction. (“All un-married adult males are unmarried”is a logical truth because it is of theform “All FGH are F,” and any claimof that form is necessarily true.)(See Discussion Question 2.)


patible with every possibility. It is . . . a logicaltruth. . . . Since it is true under any possible cir-cumstances and is not incompatible with anydescription of a possible world, its content is zero.Any analytic statement will, as we have seenabove, share this characteristic. We are, therefore,entitled to assert that analytic statements have nofactual content. . . .

Let us now turn to the other distinctionrequired for our discussion. A statement is apriori if its truth or falsity can be establishedwithout recourse to observational evidence; it isa posteriori if observational evidence is neededto establish its truth or falsity. The distinctionbetween a priori and a posteriori statementsrefers exclusively to the justification of state-ments and has nothing to do with discovery.The statements of arithmetic, for example, areregarded by most philosophers as a priori; thefact that children may learn arithmetic by count-ing physical objects (e.g., fingers) has nothingto do with the issue. Arithmetic statements canbe established . . . without the aid of empiricalobservation or experiment, and this qualifiesthem as a priori. It is evident, moreover, thatanalytic statements, as they have been describedabove, are a priori. Since their truth followsfrom logical truths and definitions alone—thatis, from syntactical and semantical considera-tions alone—observation and experiment arenot required for their proof.

Most philosophers would acknowledge thatmany synthetic statements are a posteriori. Itwould seem that no amount of pure ratiocinationwould reveal whether I had eggs for breakfastthis morning or whether there is a typewriter onthe desk in the next office. Some sort of obser-vation would seem to be indispensable. How-ever, it is not nearly as evident that all syntheticstatements are a posteriori. The doctrine thatthere are synthetic a priori statements is, I takeit, the thesis of rationalism. It was maintained byKant, as well as by many other philosophersboth before and after him. The doctrine that alla priori statements are either analytic or self-contradictory is the thesis of empiricism as Iunderstand it.

I know of no easy way to argue the questionof whether there are any synthetic a priori

14

13

statements. The history of human thought hasprovided many attempts to establish synthetic apriori truths, with a notable lack of success inmy opinion. . . .

Descartes provides an especially clear exampleof the use of synthetic a priori principles to jus-tify ampliative inference. Starting from hisfamous cogito, he set out to deduce a completeaccount of the real world. He never supposed thatnonampliative deduction would be equal to thetask; instead, he appealed to principles he consid-ered evident to the natural light of reason: “Nowit is manifest by the natural light that there mustat least be as much reality in the efficient andtotal cause as in its effect. For, pray, whence canthe effect derive its reality, if not from itscause?”* The man who thought he could not becertain that 2 + 2 = 4 or that he had hands unlesshe could prove that God is not a deceiver foundthe foregoing principle so clear and distinct thatit is impossible to conceive its falsity!

…

Kant’s approach to the question of synthetic apriori principles is profoundly instructive. Soconvinced was he that geometry providedexamples of synthetic a priori propositions thathe did not need to tarry long over the question ofwhether there are any such things. Instead, hemoved on the question of how they are possible.Synthetic a priori knowledge (if there is such)does exhibit a genuine epistemological mystery.After some exposure to formal logic one can seewithout much difficulty how linguistic stipula-tions can yield analytic statements that hold inany possible world. It is easy to see that “Snow iswhite or snow is not white” is true simplybecause of the meanings we attach to “or” and“not.” Analytic a priori statements are no greatmystery. Likewise, it is not too difficult to seehow our senses can provide clues to the nature ofphysical reality, helping us to establish proposi-tions that are true in some but not all possibleworlds. . . . But how could we conceivably estab-lish by pure thought that some logically consis-tent picture of the real world is false? How couldwe, without any aid of experience whatever, findout anything about our world in contradistinctionto other possible worlds? Given a logically

15

* René Descartes, “Of God: That He Exists,” Meditations, III [p. 160]; the translation in this book varies slightly.

13

Does the fact that a claimhas no factual content in

this sense help to explain howit can be justified or known apriori (as Salmon seems to besuggesting)? How would suchan explanation go?

14

That is, moderate empiricism.RR15

Salmon is right that this iswhat Descartes’s claim

amounts to.Think back to Descartesand try to see clearly why this is so.(Would there be any plausibility tothe claim that Descartes’s principleis analytic under either Kant’s orSalmon’s definition?)


contingent formula—one that admits of true aswell as false interpretations—how could wehope to decide on a completely a priori basiswhich of its interpretations are true and whichfalse? The empiricist says it is impossible todo so, and in this I think he is correct. Neverthe-less, it is tempting to endow various principleswith the status of synthetic a priori truths. It wasto Kant’s great credit that he saw the urgency ofthe question: How is this possible?

Various causal principles, as we have seen,have been accorded the status of synthetic apriori truths—for example, the traditionalexnihilo principle, the principle of sufficientreason, and Descartes’ principle that the causemust be as great as the effect. Kant also, inaddition to claiming that the propositions ofarithmetic and geometry are synthetic a priori,maintained that the principle of universal cau-sation—everything that happens presupposessomething from which it follows according toa rule—is synthetic a priori. It is by means ofthis principle that he hoped to dispose of theproblem of induction. However, Kant’sattempt to explain the possibility of synthetica priori propositions is unsatisfactory. Thepropositions of Euclidean geometry do notenjoy epistemological primacy; the proposi-tions of arithmetic lack synthetic content,and the physical world can be made intelligi-ble in nondeterministic terms. Human powersof conception and visualization far exceed thelimits Kant saw as necessary constraints uponthe human intellect and as the source ofsynthetic a priori truths.

It is unfortunate that subsequent philosophershave paid little attention to Kant’s centralquestion: How are synthetic a priori proposi-tions possible? Instead, the category of synthetica priori propositions has too often become aconvenient wastebasket for statements not read-ily classifiable as analytic or a posteriori. . . . Itseems to me that all such statements can beshown, on careful examination, to be analytic ora posteriori, and that no convincing example ofa synthetic a priori proposition has yet beenproduced. Even if this is so, of course, it doesnot prove that there are no synthetic a prioristatements. It should, however, give us pause,

18

17

16

and it does relieve us of any obligation to acceptthe positive rationalistic thesis that there aresynthetic a priori propositions. It does place theburden of proof upon those who hope to escapeHume’s problem of induction by way of asynthetic a priori principle. Moreover, even if arecalcitrant example were given—one thatseemed to defy all analysis as either analytic ora posteriori—it might still be reasonable tosuppose that we had not exercised sufficientpenetration in dealing with it. If we are left witha total epistemological mystery on the questionof how synthetic a priori propositions are possi-ble, it might be wise to suppose it more likelythat our analytic acumen is deficient than that anepistemological miracle has occurred.

5. The Principle of Uniformity of Nature.A substantial part of Hume’s critique of induc-tion rested upon his attack on the principle of theuniformity of nature. He argued definitively thatthe customary forms of inductive inferencecannot be expected to yield correct predictionsif nature fails to be uniform—if the future is notlike the past—if like sensible qualities are notaccompanied by like results.

All inferences from experience suppose, as theirfoundation, that the future will resemble thepast, and that similar powers will be conjoinedwith similar sensible qualities. If there be anysuspicion that the course of nature may change,and that the past may be no rule for the future, allexperience becomes useless, and can give rise tono inference or conclusion.*

He argued, moreover, that there is no logicalcontradiction in the supposition that nature isnot uniform—that the regularities we haveobserved up to the present will fail in wholesalefashion in the future. . . . He argues, in addition,that the principle of uniformity of nature cannotbe established by an inference from experience:“It is impossible, therefore, that any argumentsfrom experience can prove this resemblance ofthe past to the future; since all these argumentsare founded on the supposition of that resem-blance.”‡ Throughout Hume’s discussionthere is, however, a strong suggestion that wemight have full confidence in the customaryinductive methods if nature were known to beuniform.

20

19

18

See Ewing for a discussion ofthe issue of whether arith-

metical propositions such as 5 + 7 =12 are analytic or synthetic.

16

Here Salmon does not get theissue quite right: A rationalist

proponent of synthetic a prioriknowledge does think that somelogically contingent or consistentclaims—that is claims that are notnecessary solely by virtue of theirform—can be known to be true apriori. But he does not think thatsuch claims are really contingentor consistent, but instead that theyare necessary, even if not formallynecessary. (See the Ewing selection,especially Annotation 1 and thecorresponding text.) Thus a syn-thetic a priori claim is supposed tobe true in every possible worldand so would not tell us something“about our world in contradistinc-tion to other possible worlds.”

19

This is probably Salmon’sstrongest argument against the

existence of synthetic a priori claims.But it depends on the claim that theidea of analyticity can explain all clearcases of a priori justification andknowledge with no such mystery.(See Discussion Questions 2 and 3.)

20

To be plausible, as we havealready seen, the claim would

have to be that the future resem-bles the past to a substantialdegree, not that it does so perfectlyin the sense that all observed regu-larities also obtain in the future—since we know that this isn’t true.

*David Hume, Human Understanding [p. 290].‡Ibid. [p. 290].

17

The “ex nihilo principle” isthe principle that something can-not come from nothing, and theprinciple of sufficient reason is theclaim that everything that existsmust have a cause or explanation.Both of these “principles” are oftenappealed to in discussions ofthe cosmological argument (seeChapter 7).


Kant attempted to deal with the problem ofinduction in just this way, by establishing aprinciple of uniformity of nature, in the form ofthe principle of universal causation, as asynthetic a priori truth. Kant claimed, in otherwords, that every occurrence is governed bycausal regularities, and this general characteris-tic of the universe can be established by purereason, without the aid of any empiricalevidence. He did not try to show that the princi-ple of universal causation is a principle of logic,for to do so would have been to show that it wasanalytic—not synthetic—and thus lacking infactual content. He did not reject Hume’s claimthat there is no logical contradiction in the state-ment that nature is not uniform; he did not try toprove his principle of universal causation bydeducing a contradiction from its denial. He didbelieve, however, that this principle, while not aproposition of pure logic, is necessarily truenevertheless. Hume, of course, arguedagainst this alternative as well. He maintainednot only that the uniformity of nature is not alogical or analytic truth, but also that it cannotbe any other kind of a priori truth either. Evenbefore Kant had enunciated the doctrine of syn-thetic a priori principles, Hume had offeredstrong arguments against them:

I shall venture to affirm, as a general proposition,which admits of no exception, that the knowl-edge of this relation [of cause and effect] is not,in any instance, attained by reasonings a priori.*

Adam, though his rational faculties be sup-posed, at the very first, entirely perfect, couldnot have inferred from the fluidity and trans-parency of water that it would suffocate him,or from the light and warmth of f ire that itwould consume him.†

. . .

Now whatever is intelligible, and can bedistinctly conceived . . . can never be provedfalse by any . . . abstract reasoning a priori.§

Hume argues, by persuasive example andgeneral principle, that nothing about the causalstructure of reality can be established by purereason. He poses an incisive challenge to thosewho would claim the ability to establish a priori

21

knowledge of a particular causal relation or ofthe principle of universal causation. In the fore-going discussion of synthetic a priori statements,I have given reasons for believing that Kantfailed to overcome Hume’s previous objections.

…

7.A Probabilistic Approach. It may seemstrange in the extreme that this discussion ofthe problem of induction has proceeded atsuch great length without seriously bringing inthe concept of probability. It is very temptingto react immediately to Hume’s argument withthe admission that we do not have knowledgeof the unobserved. Scientific results are notestablished with absolute certainty. At best wecan make probabilistic statements about unob-served matters of fact, and at best we can claimthat scientific generalizations and theories arehighly confirmed. We who live in an age of sci-entif ic empiricism can accept with perfectequanimity the fact that the quest for certaintyis futile; indeed, our thanks go to Hume forhelping to destroy false hopes for certainty inscience.

Hume’s search for a justification of induction,it might be continued, was fundamentallymisconceived. He tried to find a way of provingthat inductive inferences with true premiseswould have true conclusions. He properly failedto find any such justification precisely becauseit is the function of deduction to prove the truthof conclusions, given true premises. Inductionhas a different function. An inductive inferencewith true premises establishes its conclusions asprobable. No wonder Hume failed to find a jus-tification of induction. He was trying to makeinduction into deduction, and he succeeded onlyin proving the platitude that induction is notdeduction. If we want to justify induction, wemust show that inductive inferences establishtheir conclusions as probable, not as true.

The foregoing sort of criticism of Hume’sarguments is extremely appealing, and it hasgiven rise to the most popular sort of attempt,currently, to deal with the problem. In order toexamine this approach, we must consider, at

21

Kant’s claim is roughly thatthe mind orders experience

so as to make the causal principletrue within the realm of experi-ence, though not necessarily trueof things as they exist in them-selves outside of experience.

*Ibid. [p. 286].†Ibid. [p. 286].§Ibid. [pp. 288–89].


least superficially, the meaning of the concept ofprobability. Two basic meanings must be takeninto account at present.

One leading probability concept identifiesprobability with frequency—roughly, the prob-able is that which happens often, and theimprobable is that which happens seldom. Letus see what becomes of Hume’s argument underthis interpretation of probability. If we were toclaim that inductive conclusions are probable inthis sense, we would be claiming that inductiveinferences with true premises often have trueconclusions, although not always. Hume’s argu-ment shows, unhappily, that this claim cannotbe substantiated. It was recognized long beforeHume that inductive inferences cannot beexpected always to lead to the truth. Hume’sargument shows, not only that we cannot justifythe claim that every inductive inference withtrue premises will have a true conclusion, butalso, that we cannot justify the claim that anyinductive inference with true premises will havea true conclusion. Hume’s argument shows that,for all we can know, every inductive inferencemade from now on might have a false conclu-sion despite true premises. Thus, Hume hasproved, we can show neither that inductiveinferences establish their conclusions as truenor that they establish their conclusions asprobable in the frequency sense. The intro-duction of the frequency concept of probabilitygives no help whatever in circumventing theproblem of induction, but this is no surprise, forwe should not have expected it to be suitable forthis purpose.

A more promising probability conceptidentifies probability with degree of rationalbelief. To say that a statement is probable inthis sense means that one would be rationallyjustified in believing it; the degree of proba-bility is the degree of assent a person wouldbe rationally justified in giving. We are not, ofcourse, referring to the degree to whichanyone actually believes in the statement, butrather to the degree to which one couldrationally believe it. . . . To say that a state-ment is probable in this sense means that it issupported by evidence. But, so the argumentgoes, if a statement is the conclusion of aninductive inference with true premises, it issupported by evidence—by inductiveevidence—this is part of what it means to be

22

supported by evidence. The very concept ofevidence depends upon the nature of induc-tion, and it becomes incoherent if we try todivorce the two. Trivially, then, the conclusionof an inductive inference is probable underthis concept of probability. To ask, withHume, if we should accept inductive conclu-sions is tantamount to asking if we shouldfashion our beliefs in terms of the evidence,and this, in turn, is tantamount to askingwhether we should be rational. In this way wearrive at an “ordinary language dissolution”of the problem of induction. Once we under-stand clearly the meanings of such key termsas “rational,” “probable,” and “evidence,” wesee that the problem arose out of linguisticconfusion and evaporates into the question ofwhether it is rational to be rational. Such tau-tological questions, if meaningful at all,demand affirmative answers.

Unfortunately, the dissolution is not satisfac-tory. . . . The fundamental difficulty arises fromthe fact that the very notion of inductiveevidence is determined by the rules of inductiveinference. If a conclusion is to be supported byinductive evidence, it must be the conclusion ofa correct inductive inference with true premises.Whether the inductive inference is correctdepends upon whether the rule governing thatinference is correct. . . . In order to be able to saywhether a given statement is supported byinductive evidence we must be able to say whichinductive rules are correct.

For example, suppose that a die has beenthrown a large number of times, and we haveobserved that the side two came up in one sixthof the tosses. This is our “evidence” e. Let h bethe conclusion that, “in the long run,” side twowill come up one sixth of the times. Considerthe following three rules:

1. (Induction by enumeration) Given m/n ofobserved A are B, to infer that the “long run”relative frequency of B among A is m/n.

2. (A priori rule) Regardless of observedfrequencies, to infer that the “long run” rela-tive frequency of B among A is 1/k, where k isthe number of possible outcomes—six in thecase of the die.

3. (Counterinductive rule) Given m/n of observedA are B, to infer that the “long run” relativefrequency of B among A is (n – m)/n. 24

23

22

The proof is just thedilemma argument (see

Annotations 9, 10, and 11 to theHume selection and the corre-sponding text).Think carefullyabout how that argument wouldapply to the probabilistic claim.

23

The claim is that reasoninginductively is just what we

mean by “rational” in this sort ofcase, so that inductive reasoning isguaranteed to be rational.

RR

24

Here the first rule is just thestandard conception of

inductive reasoning, generalizedslightly to cover also cases wheresome fraction of observed A’s areB’s (in the case where all observedA’s are B’s, m and n are the samenumber).The second rule is the so-called “principle of indifference”—which might be badly wrong if thedice are “loaded.” The third rule isthe intuitively bizarre rule whichsays that the truth is just the oppo-site of what the evidence so farseems to show.


. . . In order to say which conclusions aresupported by what evidence, it is necessary toarrive at a decision as to what inductive rulesare acceptable. If Rule 1 is correct, the evidencee supports the conclusion h. If Rule 2 is correct,we are justified in drawing the conclusion h,but this is entirely independent of the observa-tional evidence e; the same conclusions wouldhave been sanctioned by Rule 2 regardless ofobservational evidence. If Rule 3 is correct, weare not only prohibited from drawing the con-clusion h, but also we are permitted to draw aconclusion h9 which is logically incompatiblewith h. Whether a given conclusion issupported by evidence—whether it would berational to believe it on the basis of givenevidence—whether it is made probable byvirtue of its relation to given evidence—depends upon selection of the correct rule orrules from among the infinitely many rules wemight conceivably adopt.

The problem of induction can now be refor-mulated as a problem about evidence. Whatrules ought we to adopt to determine the natureof inductive evidence? . . . If we take the custom-ary inductive rules to define the concept ofinductive evidence, have we adopted a properconcept of evidence? Would the adoption ofsome alternative inductive rules provide a moresuitable concept of evidence? These are genuinequestions which need to be answered.

We find, moreover, that what appearedearlier as a pointless question now becomessignif icant and diff icult. If we take thecustomary rules of inductive inference toprovide a suitable definition of the relation ofinductive evidential support, it makes consid-erable sense to ask whether it is rational tobelieve on the basis of evidence as thus definedrather than to believe on the basis of evidenceas defined according to other rules. . . .

There is danger of being taken in by an easyequivocation. One meaning we may assign tothe concept of inductive evidence is, roughly,the basis on which we ought to fashion ourbeliefs. Another meaning results from therelation of evidential support determined bywhatever rule of inductive inference we adopt.It is only by supposing that these two concepts

are the same that we suppose the problem ofinduction to have vanished. The problemof induction is still there; it is the problem ofproviding adequate grounds for the selectionof inductive rules. We want the relation of evi-dential support determined by these rules toyield a concept of inductive evidence which is,in fact, the basis on which we ought to fashionour beliefs.

…

8. Pragmatic Justification. Of all thesolutions and dissolutions proposed to dealwith Hume’s problem of induction, HansReichenbach’s attempt to provide a pragmaticjustification seems to me the most fruitful andpromising.* This approach accepts Hume’sarguments up to the point of agreeing that it isimpossible to establish, either deductively orinductively, that any inductive inferences willever again have true conclusions. Neverthe-less, Reichenbach claims, the standard methodof inductive generalization can be justif ied.Although its success as a method of predictioncannot be established in advance, it can beshown to be superior to any alternative methodof prediction.

The argument can be put rather simply.Nature may be sufficiently uniform in suitablerespects for us to make successful inductiveinferences from the observed to the unob-served. On the other hand, for all we know, shemay not. Hume has shown that we cannot provein advance which case holds. All we can say isthat nature may or may not be uniform—if sheis, induction works; if she is not, inductionfails. Even in the face of our ignorance aboutthe uniformity of nature, we can ask whatwould happen if we adopted some radically dif-ferent method of inference. Consider, forinstance, the method of the crystal gazer. Sincewe do not know whether nature is uniform ornot, we must consider both possibilities. Ifnature is uniform, the method of crystal gazingmight work successfully, or it might fail. Wecannot prove a priori that it will not work. At thesame time, we cannot prove a priori that it willwork, even if nature exhibits a high degree ofuniformity. Thus, in case nature is reasonably

25

*Hans Reichenbach, Experience and Prediction (Chicago: U. of Chicago Press, 1938), Chap. 5, and The Theory of Probability(Berkeley: U. of California Press, 1949), Chap. 11.

25

Thus if reasoning inductivelyjust means whatever way of

reasoning is rational in this sort ofcase, it will not be automaticallytrue that this is following the firstrule rather than one of the othersor some further rule—so somefurther defense will be needed,which is just the problem of induc-tion all over again.Whereas ifreasoning inductively meansfollowing the first rule, then itcannot just be assumed that this isthe rational way to proceed(where “rational” means roughlythat a method will probably lead totrue or reliable results).

RR


uniform, the standard inductive method mustwork while the alternative method of crystalgazing may or may not work. In this case, thesuperiority of the standard inductive method isevident. Now, suppose nature lacks uniformityto such a degree that the standard inductivemethod is a complete failure. In this case,Reichenbach argues, the alternative methodmust likewise fail. Suppose it did not fail—suppose, for instance, that the method of crys-tal gazing worked consistently. This would con-stitute an important relevant uniformity thatcould be exploited inductively. If a crystal gazerhad consistently predicted future occurrences,we could infer inductively that he has a methodof prediction that will enjoy continued success.The inductive method would, in this way, sharethe success of the method of crystal gazing, andwould therefore be, contrary to hypothesis,successful. Hence, Reichenbach concludes,the standard inductive method will be success-ful if any other method could succeed. As aresult, we have everything to gain and nothingto lose by adopting the inductive method. If anymethod works, induction works. If we adoptthe inductive method and it fails, we have lostnothing, for any other method we might haveadopted would likewise have failed. Reichen-bach does not claim to prove that nature isuniform, or that the standard inductive methodwill be successful. He does not postulatethe uniformity of nature. He tries to show thatthe inductive method is the best method forampliative inference, whether it turns out to besuccessful or not.

This ingenious argument, althoughextremely suggestive, is ultimately unsatisfac-tory. As I have just presented it, it is impossibly

26

vague. I have not specified the nature of thestandard inductive method. I have not statedwith any exactness what constitutes success forthe inductive method or any other. Moreover,the uniformity of nature is not an all-or-noneaffair. Nature appears to be uniform to someextent and also to be lacking in uniformity tosome degree. . . . The vagueness of the forego-ing argument is not, however, its fundamentaldrawback. It can be made precise. . . . When itis made precise, as we shall see, it suffers theserious defect of equally justifying too wide avariety of rules for ampliative inference.

I have presented Reichenbach’s argumentrather loosely in order to make intuitivelyclear its basic strategy. The sense in which it isa pragmatic justif ication should be clear.Unlike many authors who have sought a justi-fication of induction, Reichenbach does nottry to prove the truth of any synthetic proposi-tion. He recognizes that the problem concernsthe justification of a rule, and rules are neithertrue nor false. Hence, he tries to show that theadoption of a standard inductive rule is practi-cally useful in the attempt to learn about anddeal with the unobserved. He maintains thatthis can be shown even though we cannotprove the truth of the assertion that inductivemethods will lead to predictive success. Thispragmatic aspect is, it seems to me, the sourceof the fertility of Reichenbach’s approach.Even though his argument does not constitutean adequate justif ication of induction, itseems to me to provide a valid core fromwhich we may attempt to develop a moresatisfactory justification.

…

27

established regularities pertaining to various sortsof evidence. (How much would it help to be toldthat although we have no good reason for thinkingthat inductive conclusions are true, the inductivemethod is still guaranteed to succeed in findingthe truth if success is possible, which is whatReichenbach’s pragmatic justification—see Dis-cussion Question 4—amounts to?)

2. Are all plausible a priori justifiable claims analyticin Salmon’s sense? Consider, for example, the

1. If Hume’s skepticism about induction is correct,then we have no reason at all to think that anyconclusion that depends on inductive reasoning istrue. How implausible a result is this from thestandpoint of common sense? Notice in thinkingabout this that at least most of the results oftheoretical science are arrived at as the bestexplanations of inductively established observa-tional regularities. Notice also that evidence forhistorical claims also depends on inductively


26

Reichenbach’s claim is notthat reasoning inductively will

even probably lead to true conclu-sions, but only that it will do so ifany method can succeed in this way.

This is some justification, butis it enough to really solve

the problem? (See DiscussionQuestion 1.)

RR

27

Reichenbach’s basic claim isthat the method of general-

izing from experience (“positing”that the proportion of all A’s thatare B’s is the same as the currentlyobserved proportion and thencorrecting that posit as newobservations come in) is guaran-teed to succeed in the long run ifsuccess is possible, that is, if thereis an objective regularity to befound.The problem Salmon isindicating is that there are manyother methods that can be justifiedin the same way. (See DiscussionQuestion 3.)


claim that nothing can be red all over and green allover at the same time. Is this something that canbe justified a priori? It is not a logical truth as itstands (since many claims of the form “nothing canbe F all over and G all over at the same time” arefalse—think about this).Can it be made into a log-ical truth by giving definitions of one or more ofits terms, presumably “red” and/or “green”?(Remember that to be acceptable, a definitionmust actually capture the meaning of the term.)

3. Think about the rationale for Reichenbach’spragmatic justification of induction. Can you seewhy the method described in Annotation 27 isguaranteed to eventually find an objective regu-larity in nature if one actually exists? Why is suchsuccess only guaranteed in the long run? Can you

think of another method that is equally guaran-teed to succeed in the long run, but that gives asignificantly different answer in the short run(think here of cases where the observed propor-tion of A’s that are B’s is some fraction ratherthan all—see Annotation 24 and the associatedtext). Hint: what you want is a rule that deviatesfrom the observed proportion but where thedeviation gets smaller and smaller as the numberof cases gets larger, so that in the long run—butnot in the short run, which is where the methodis always actually applied—the result arrived atwill be the same as that of the inductive method.A further issue: how long a run is necessary forthe inductive method to be guaranteed to suc-ceed if success is possible?

A. C. EwingAlfred Cyril Ewing (1899–1973) was a British philosopher who taught atCambridge University and made important contributions to ethics and meta-physics. In this selection, he defends a rationalist view of a priori justification,according to which the human mind has a fundamental capacity for insightinto necessary truths. Along the way, he replies to many popular argumentsfor moderate empiricism and also suggests an a priori solution to the problemof induction, though without really developing this idea very far.

The “A Priori” and the Empirical, from TheFundamental Questions of Philosophy

Meaning of the Distinction;“A Priori” Character of

MathematicsIn the theory of knowledge, the first point thatconfronts us is the sharp distinction betweentwo kinds of knowledge which have beencalled respectively a priori and empirical. Mostof our knowledge we obtain by observation ofthe external world (sense-perception) and ofourselves (introspection). This is called empir-ical knowledge. But some knowledge we canobtain by simply thinking. That kind of knowl-edge is called a priori. Its chief exemplifica-tions are to be found in logic and mathematics.In order to see that 5 + 7 = 12 we do not needto take five things and seven things, put themtogether, and then count the total number. We

can know what the total number will be simplyby thinking.

Another important difference between a prioriand empirical knowledge is that in the case ofthe former we do not see merely that something,S, is in fact P, but that it must be P and why it isP. I can discover that a flower is yellow (or atleast produces sensations of yellow) by lookingat it, but I cannot thereby see why it is yellow orthat it must be yellow. For anything I can tell itmight equally well have been a red flower. Butwith a truth such as that 5 + 7 = 12 I do not seemerely that it is a fact but that it must be a fact.It would be quite absurd to suppose that 5 + 7might have been equal to 11 and just happenedto be equal to 12, and I can see that the nature of5 and 7 constitutes a fully adequate and intelli-gible reason why their sum should be 12 and notsome other number. It is indeed conceivable thatsome of the things which make the two groupsof 5 and 7 might, when they were put together,

1

From The Fundamental Questions of Philosophy (NewYork: Collier, 1962). Some footnotes have been omitted.

1

This is an important claimabout the apparent charac-

ter of a priori insight: it seems toreveal not only that the claim inquestion is true, but also that theclaim is necessary (in the logicalor metaphysical sense) and whythe claim is true.

RR

A. C. EWING: THE “A PRIORI” AND THE EMPIRICAL, FROM THE FUNDAMENTAL QUESTIONS OF PHILOSOPHY 143

fuse like drops of water, or even vanish, so thatthere were no longer 12 things; but what isinconceivable is that there could at the sametime be 5 + 7 things of a certain kind at once ina certain place and yet less than 12 things of thatkind in that place. Before some of the thingsfused or vanished they would be 5 + 7 in num-ber and also 12 in number, and after the fusionor disappearance they would be neither 5 + 7 nor12. When I say in this connection that somethingis inconceivable, I do not mean merely or prima-rily that we cannot conceive it—this is not a caseof a mere psychological inability like the inabil-ity to understand higher mathematics. It is apositive insight: we definitely see it to be impos-sible that certain things could happen. This wedo not see in the case of empirical propositionswhich are false: they are not true but might foranything we know have been true. It is even con-ceivable, so far as we can see, that the funda-mental laws of motion might have been quitedifferent from what they are, but we can see thatthere could not have been a world which contra-dicted the laws of arithmetic. This is expressedby saying that empirical propositions arecontingent, but true a priori propositions necess-ary. . . .

We must not be misled here by the fact thatin order to come to understand arithmetic weoriginally required examples. Once we havelearnt the beginnings of arithmetic in thekindergarten with the help of examples, we donot need examples any more to grasp it, andwe can see the truth of many arithmeticalpropositions, e.g. that 3112 + 2467 = 5579, ofwhich we have never had examples. We haveprobably never taken 3112 things and 2467things, put them together and counted theresulting set, but we still know that this is whatthe result of the counting would be. If it wereempirical knowledge, we could not know itwithout counting. The examples are needed,not to prove anything, but only in order toenable us to come to understand in the firstinstance what is meant by number.

…

The “A Priori” in LogicAnother important f ield for a priori knowl-edge is logic. The laws of logic must be knowna priori or not at all. They certainly are not a

2

matter for empirical observation, and thefunction of logical argument is just to give usconclusions which we have not discovered byobservation. The argument would be superflu-ous if we had observed them already. We areable to make inferences because there issometimes a logical connection between oneor more propositions (the premise or prem-ises) and another proposition, the conclusion,such that the latter must be true if the formeris. Then, if we know the former, we can assertthe latter on the strength of it, thus anticipat-ing any experience. To take an example, thereis a story that Mr. X., a man of high reputationand great social standing, had been asked topreside at a big social function. He was late incoming, and so a Roman Catholic priest wasasked to make a speech to pass the time till hisarrival. The priest told various anecdotes,including one which recorded his embarrass-ment when as confessor he had to deal with hisf irst penitent and the latter confessed to aparticularly atrocious murder. Shortly after-wards Mr. X. arrived, and in his own speech hesaid: “I see Father _____ is here. Now, thoughhe may not recognize me, he is an old friend ofmine, in fact I was his first penitent.” It is plainthat such an episode would enable one to inferthat Mr. X. had committed a murder withouthaving observed the crime. . . .

Other Cases of the “A Priori”A priori knowledge, while most prominent inmathematics and logic, is not limited to thesesubjects. For instance, we can see a priori thatthe same surface cannot have two differentcolours all over at the same time, or that athought cannot have a shape. Philosophershave been divided into rationalists and empiri-cists according to whether they stressed the apriori or the empirical element more. Thepossibility of metaphysics depends on a prioriknowledge, for our experience is quiteinadequate to enable us to make on merelyempirical grounds any sweeping generaliza-tions of the kind the metaphysician desires.The term a priori covers both self-evidentpropositions, i.e. those which are seen to betrue in their own right, and those which arederived by inference from propositions them-selves self-evident.

3

2

Could a contingent claim, onethat is false in some possible

worlds, ever be known a priori tobe true in the actual world? Ewingis denying that this could happen,but some recent philosophers havequestioned this.The standardexample (due to Saul Kripke) is“the standard meter stick (theplatinum-iridium bar in Paris) isone meter long.” Is this a counter-example to Ewing’s claim?

3

The claim that Mr. X is amurderer follows logically

from the two premises indicated. Isit contained in them in any sensethat would help to explain why itfollows logically? (In Salmon’sterms, is this an ampliative or anon-ampliative inference?)


The Linguistic Theory of the“A Priori” and the Denial That

“A Priori” Propositions orInferences Can Give New

KnowledgeAt the present time even empiricist philosophersrecognize the impossibility of explaining away apriori propositions as merely empirical general-izations, but they are inclined to the view that apriori propositions and a priori reasoning aremerely concerned with language, and so cannottell us anything new about the real world. Thus itis said that, when we make an inference, theconclusion is just part of the premises expressedin different language. If so, inference wouldbe of use merely for clarifying our language andwould involve no real advance in knowledge.Some inferences are of this type, e.g. A is afather, therefore A is male. But are they all? Thatwould be hard indeed to square with the primafacie novelty of many conclusions. Take, forinstance, the proposition that the square on thehypotenuse of a right-angled triangle is equal tothe sum of the squares on the other two sides.Such a proposition can be inferred from theaxioms and postulates of Euclid, but it certainlydoes not seem to be included in their meaning.Otherwise we should know it as soon as weunderstood the axioms and postulates. Theexample I gave of the murder discovered by alogical argument seems to be another case of afact not known at all beforehand by the reasonerwhich is discovered by his reasoning. Extremeempiricist philosophers contend that this appear-ance of novelty is really illusory, and that in somesense we knew the conclusion all along; but theyhave never succeeded in making clear in whatsense we did so. It is not enough to say that theconclusion is implicit in the premises. “Implicit”means “implied by,” and of course a conclusionis implied by its premises, if the inference is cor-rect at all. But this admission leaves quite openthe question whether or not a proposition can fol-low from a different one which does not containit as part of itself; and since we obviously can bydeductive inference come to know things whichwe did not know before in any ordinary sense of“know,” we must treat the empiricist’s claim asunjustified till he has produced a clearly definedsense of “implicit in” or “contained in” whichleaves room for that novelty in inference which

4

we all cannot help really admitting. In any ordi-nary sense of “know” the conclusion is not in thecases I have mentioned known prior to the infer-ence, and since the premises are and indeed mustbe known before we know the conclusion, it istherefore in no ordinary sense of “part” part ofthe premises.

…

Nevertheless, the view that inference cannotyield new conclusions dies hard, and so it willnot be superfluous to bring further arguments.(1) “This has shape” admittedly follows logicallyfrom “this has size” and vice versa. If the view Iam criticizing were true, “this has size” would,therefore, have to include in its meaning “thishas shape,” and “this has shape” would also haveto include in its meaning “this has size.” But thiswould only be possible if the two sentencesmeant exactly the same thing, which they obvi-ously do not. (2) Take an argument suchas—Montreal is to the north of New York, NewYork is to the north of Washington, thereforeMontreal is to the north of Washington. If theview I am discussing is true, the conclusion ispart of the premises. But it is not part of eitherpremise by itself, otherwise both premises wouldnot be needed. So the only way in which it couldbe part of both together would be if it were divis-ible into two propositions one of which was partof the first and the other part of the second. I defyanybody to divide it in this way. (3) The proposi-tion “Socrates was a philosopher” certainlyentails the proposition “if Socrates had measlessome philosophers have had measles,” but itcannot be that the second proposition is includedin the first. For the first proposition certainlydoes not include the notion of measles.

What is really the same view is often expressedby saying that all a priori propositions are“analytic.” A distinction has commonly beendrawn between analytic propositions, in whichthe predicate is in the notion of the subjectalready formed before the proposition is asserted,so that the proposition gives no new information,and synthetic propositions in which the predicateis not so contained and which are thus capable ofgiving new information. Analytic propositionsare essentially verbal, being all true by definition,e.g. all fathers are male. As an example of a syn-thetic proposition we could take any propositionestablished by experience such as “I am cold” or

6

5

6

An analogy: if the member-ship of group A is contained

in that of group B, and vice versa,then the two groups must haveexactly the same membership.Thusif all the elements included in themeaning of one claim are includedin that of the other and vice versa,the total set of elements must beexactly the same.

4

That is, as Salmon puts it, allsuch inferences are

non-ampliative.

5

If the claim of containmentjust means that the conclu-

sion follows from the premises,then it cannot explain why this isso (or how we are able to see thatit is).

RR


“It is snowing,” but empiricists often assert thatthere are no synthetic a priori propositions. Thatthis view cannot be justified may be shown atonce. The proposition that there are no synthetica priori propositions, since it cannot be estab-lished by empirical observations, would be, ifjustified, itself a synthetic a priori proposition,and we cannot affirm it as a synthetic a prioriproposition that there are no synthetic a prioripropositions. We may therefore dismiss off-hand any arguments for the theory. Sucharguments, whatever they were, would have toinvolve synthetic a priori propositions. . . .

The analytic view seems plausible when we areconcerned with the simplest propositions of logicand arithmetic, but we must not assume that aproposition is analytic because it is obvious.Though it may be very difficult to determineprecisely where analytic propositions end andsynthetic propositions begin, we cannot use this asa ground for denying the latter. It is very difficultto say precisely where blue ends and green begins,since the different shades run into each otherimperceptibly, but we cannot therefore argue thatall blue is really green. Taking arithmetic, even ifthere is a good deal of plausibility in saying that2 + 2 is included in the meaning of “4,” there isnone in saying 95 – 91 or > 216 > ÷ 2 – (> 287 +25) > ÷ 3 are so included. Yet, if the analyticview were true, all the infinite numerical combina-tions which could be seen a priori to be equal to 4would have to be included in the meaning of “4.”

…

People have been inclined to deny synthetic apriori propositions because they could not see howone characteristic could necessarily involveanother, but that this could not happen wouldbe itself a synthetic a priori metaphysical proposi-tion. People have also thought that it was neces-sary to give some sort of explanation of a prioriknowledge, and could not see how this could bedone except in terms of language. To this I shouldreply that there is no reason to suppose that a pri-ori knowledge requires some special explanationany more than does our ability to attain knowledgeempirically by observation. Why not take it as anultimate fact? Human beings certainly cannotexplain everything, whether there is ultimately anexplanation for it or not.

…

8

7

InductionThe chief use of a priori knowledge is toenable us to make inferences. Whenever wecan make an inference from p to q, we mustknow, or at least be justified in believing, thehypothetical “if p, then q.” For q to follow nec-essarily from p that hypothetical propositionmust be a priori; necessary connection is not amatter of empirical observation. An importantdistinction is, however, drawn between deduc-tion and induction. . . . Induction in all its morecharacteristic forms is a matter of empiricalgeneralization, i.e. we argue that, becausesomething has proved true in a number ofobserved cases, it will be likely to prove truein similar cases that have not yet beenobserved. The conclusion is not (except invery special cases) certain, but it may behighly probable, and all our rational predic-tions about the future depend on an inferenceof this kind. Induction has presented very seri-ous problems to logicians and philosophersgenerally. The inference in induction iscertainly not a merely empirical matter: weuse it to enable us to forecast the future, but wehave not observed the future empirically. Thewhole use of induction is indeed to enable usto infer what we have never observed. There-fore some a priori principle about the world isrequired if induction is to be justified. Theprinciple must be of such a nature as to justifyus in supposing that what has happened inobserved cases is likely to recur in unobservedones; but logicians have certainly not hit upona principle which is both self-evident and ade-quate to justify inductive inference. The onewhich has been most commonly put forwardas supplying what is needed is the principlethat every change has a cause, but it would bemuch disputed nowadays whether this waseither necessary or sufficient to justify induc-tion. The meaning of “cause” itself is also verymuch a subject on which there are varyingviews. So the justification of induction is oneof the worst problems of logic. The fact thatwe must use induction if we are to have anyscience at all is a proof that science cannot bemerely empirical (even where it does not usemathematics), but we cannot say that the con-clusion of an inductive inference follows fromits premises with the same necessity as doesthe conclusion of a deductive one.

9

7

This is a tricky point.To thinkabout it clearly, you need to

restate the claim in question tomake it explicitly of subject-predicate form. (See DiscussionQuestion 1.)

8

Notice also that it would alsohave to be claimed that the

meaning of each of these morecomplicated expressions isincluded in that of the other—andhence that they also have the samemeaning.

9

There is an implicit argumenthere: if an inductive argument

provides a good reason or justifica-tion for thinking that its conclusionis true, then, since the transitionfrom the inductive evidence to theconclusion cannot be justified onempirical or experiential grounds(since we have not observed thefuture), this justification must be apriori. Salmon would deny thisconclusion, but can he do sowithout conceding that the conclu-sion of an inductive argument isn’treally justified after all?


Most of the propositions we call empirical arenot justified by mere observation but by thattogether with induction. This applies to allconclusions of science, for these are neverstatements of single observed facts but general-izations about what usually happens or infer-ences from such generalizations. It applies alsoto all our ordinary judgments about physicalobjects, for we always read in more than is actu-ally observed and the only justification for whatwe read in must be found in induction (i.e. weassume that objects will have besides thecharacteristics we actually observe at the timethose characteristics which we observed in thepast in similar objects, e.g. that the ice we see iscold though we are not feeling the cold). Thisdoes not mean that, whenever we observe, weare consciously making inferences. But we canspeak of “implicit inferences,” meaning thatwithout making conscious inferences we holdbeliefs on the strength of experiences whichcould be used as premises to justify the beliefsby explicit inferences.

IntuitionA necessary, though not always recognized,presupposition of inference is intuition. Toargue validly—A ... B ... C—we must see aconnection between A and B and a connectionbetween B and C. But how do we know thatthat connection itself holds? We may be able tointerpolate further terms, D and E, and say A, ...D, ... B, ... E, ... C; but we obviously cannot goon in this way ad infinitum. Sooner or later wemust come to a point where we just see immedi-ately that there is a connection, but cannot provethis to be so. When we see immediately some-thing to be true otherwise than by observing itempirically, we are said to have an intuition.No doubt apparent intuitions are often to beexplained as really implicit or suppressedinferences, but the argument I have just givenshows that this cannot be so with all cases ofintuition. Even when we have made explicit allthe omitted steps, the logical point remains thatfor it to be possible to infer any one propositionfrom any other we must see the connectionbetween them, and this connection cannotalways be proved by interpolating fresh proposi-tions, yet it must be known somehow. Theconnection may be said to be derived from the

11

10

fundamental laws of logic, but the laws them-selves are known intuitively. Further, before wecan see their validity in the abstract, we musthave seen their validity in particular cases. Allof us have used syllogisms long before we knewanything about the rules of syllogistic inference.That was because we could see that a conclusionfollowed from its premises without knowing thegeneral principle on account of which itfollowed, as we can move our arms effectively indefinite purposive ways without knowing thelaws of physiology which govern our motions orthe laws of physics which make the selectedmotions the most effective ones for our practicalpurpose. The same applies whether the connec-tion is such as to make the conclusion followwith certainty given the premises, or whether itis, as in most inductive arguments, such that thepremises make the conclusion only probable.But, while logicians have easily agreed on thegeneral principles governing deduction and wecan see these to be completely self-evident,neither is the case as regards the general princi-ples governing induction. Yet though neither theplain man nor even the logician is clear as to thefundamental laws behind the process, we cansee in particular cases of induction that the truthof the premises makes the truth of the conclu-sion probable, as we can see that the truth of thepremises makes the truth of the conclusioncertain in particular cases of deduction.“Intuition” as a source of knowledge or justifiedbelief is often brushed aside with contempt, butthe argument I have given at any rate shows that,if there are to be any valid inferences, there mustbe some intuitions.

…

The main argument of those who attack thenotion of intuition is that apparent intuitions areliable to conflict with each other and there isthen no means of deciding which is right.But this is a mistake; we can in fact test them invarious ways. We can consider whether theyare capable of any clear and internally consis-tent statement. We can ask whether they fitinto a coherent system with the rest of our well-established beliefs. We can also ask whetherintuitions of the same kind have been confirmedin the past. We can ask whether an intuitionstands or falls by itself or is a presupposition ofa whole number of other beliefs which we

13

12

10

The symbol ... means“therefore.”

11

What Ewing is describinghere is a priori intuition,

sometimes referred to as a prioriinsight. (This is not intuition in thesense of a hunch or premonition.)Empiricists typically deny thatthere is any such thing.

12

Ewing is saying that particularinductive arguments can be

seen a priori to be rationallycogent, even if we cannot state ageneral principle in relation towhich they are so. (See DiscussionQuestion 3.)

RR

13

This is one standard objec-tion to a priori intuition.


cannot help holding, as some (though confused)intuition of the occurrence of causation or theuniformity of nature seems a necessary presup-position of all inductive beliefs. We can considerthe plausibility of giving an alternative explana-tion of the intuitive belief. The result may thenbe negative or positive. It may be that our appar-ent intuition will evaporate when we think of theexplanation, and then the latter is probably acorrect one at least as to why we held the belief.Or it may be that the intuition will persistunshaken, in which case the explanation is prob-ably at least inadequate. We can again considerwhether the intuition repeats itself when consid-ered in different contexts and different moods orwith different examples.

So when two people have conflicting intu-itions we need not suppose that there is just anirreducible difference of intuitive facultybetween them and that there is nothing more tobe done about it. Arguments may well be avail-able which without strictly proving either sideto be wrong put a disputant into a position inwhich he can see better for himself whether heis right or wrong or at least partially confirmor cast doubt on the truth of his view. In gen-eral, the clearer we have made ourselves abouta subject by inferential thought, by analysing

the different factors involved and by clearingup our terminology, the more likely are we tohave correct intuitions on the matter if suchare available at all. Again, intellectual confu-sions may be revealed which were responsiblefor the truth of the belief in question. Thus aperson who really sees that A is B may confuseB with C and will then think he sees intuitivelythat A is C. Some such conflicts may be causedsimply or mainly by ambiguities of terminol-ogy or the attaching of different meanings tosome word. And of course we need not denythat differences of intuition may sometimes bedue on one side or even on both to “wishfulthinking” or to the kind of cause which it is thebusiness of the psycho-analyst (or of a patientand tactful friend) to remove. . . . We cannot ofcourse settle all disputes in these ways, butneither can we in practice settle all disputes inscience. The most we can say is that they aresoluble in principle, though we may not havethe ability to hit on the right way of solving aparticular dispute. Similarly, there is no reasonto believe that conflicts between rival intu-itions would not all be capable of a solution ifthese methods were applied aright and withgood will on both sides, though in fact wecannot so apply them. * . . .14

14

Ewing’s response is thatthere are many ways to

attempt to resolve the disputebetween opposed intuitions, eventhough none of these is guaranteedto succeed. (Try to think howthese would work in particularcases.)

RR

*Similarly, the phrase “is contained in” is sometimes used just to mean “follows from” or “is implied by” and need not con-note that the conclusion is actually part of the premises, as would be the case on the literal meaning of “contained.”


1. How might the claim that there are no synthetic apriori truths—that is, that no synthetic proposi-tion is capable of being justified a priori—berestated in the form “all A’s are B’s”? One attempt,albeit a bit clumsy, would be “all synthetic propo-sitions are propositions incapable of being justifieda priori.” Is the predicate of this claim included inits subject in any clear or obvious way? Is there anyway that its truth can be seen to follow from thedefinitions of the terms involved? Assume that“synthetic” is defined as “nonanalytic.”

2. Is 5 + 7 = 12 knowable a priori? If so, is it analyticor synthetic (as Kant defines these concepts—seethe Appendix to the introduction to this chapter)?Here it is important to get the issue into veryclear focus:The issue is not merely whether theclaim is one such that anyone who understands itcan see it to be true. It is rather whether the wayit is seen to be true is by analyzing the concept of7 + 5 and finding 12 to be contained in it, in the way

in which unmarried is contained in bachelor.Another way to put this question: is the denial of7 + 5 = 12 a contradiction in the way that theclaim that some bachelors are married is? It is falseand indeed necessarily false, but is it contradictoryin itself—without bringing in other mathematicalprinciples or axioms? (This is a hard question.)

3. Is 18,697 + 23,849 = 42,546 analytic or synthetic?Clearly most people can understand the conceptof the sum of 18,697 and 23,849 without therebyknowing that it has the property of being equal to42,546, in a way that they could not understandthe concept of bachelor without thereby knowingthat the property of unmarried is contained in it.Does this show that the claim is synthetic?Why orwhy not?

4. Consider the claim that nothing can be red allover and green all over at the same time. Is thisclaim knowable a priori? Is it analytic or synthetic?This claim is not clearly of subject-predicate form


as it stands, but here is a somewhat convolutedrewording that is of subject-predicate form and isparallel to “all bachelors are unmarried”: all thingsthat are red all over at a given time have theproperty of not being green all over at that sametime. Is the predicate of this claim included in itssubject—which amounts to asking whethernot being green is included in red? (One rele-vant question here: could one understand theconcept of red without yet having learned theconcept of green?)

5. Think of an actual instance of inductive reasoning:a case in which there have been many observa-tions of A’s that are B’s under varied conditionsand circumstances and no observations of A’s thatare not B’s, and in which it is inferred that all A’sare B’s. How plausible is it that we have an a prioriintuition (or at least “ostensible intuition”) thatthe conclusion of the argument is probably true—that the premises of the argument support theconclusion in a way analogous to that whichoccurs in a deductive argument?

Concluding Dialogue on the Problem of InductionHere again we have a problem that many people have trouble taking very seriously—perhaps

that’s a feature of most epistemological issues. It again just seems so obvious that induc-tive reasoning is justified, that it does make its conclusions likely to be true, even if notguaranteed. Even Hume, who challenged whether that is really so, admitted that hecouldn’t help believing it as soon as he left his study and went out into the world.

But again, as you know, the problem is to explain how and why inductive reasoning is justified—why inductive arguments are genuinely cogent. I doubt whether anyone, in their heart ofhearts, really doubts that they are. But that fact still doesn’t constitute a solution to thephilosophical problem.

So what are the possibilities here? Salmon favors Reichenbach’s “pragmatic” solution (hisnumber 8), while admitting that there are still serious problems about it that have yet to besolved. Ewing favors an a priori solution (Salmon’s number 4)—but I can’t help pointingout that a lot of people regard that as hopeless, a non-starter (which is pretty clearlySalmon’s view). And then there is the “ordinary language” solution that Salmon explainsand argues against (his number 7). These really seem to me to be the only significantcandidates. Appeals to a Principle of Uniformity only make sense as versions of an apriori solution, which is the way that Salmon views them (his number 5).

I think that you’re basically right about those being the main alternatives. But I suppose thatwe should at least mention the so-called “inductive justification of induction” (Salmon’snumber 1). This seems, as he argues, obviously circular or question begging, because ituses the very sort of argument whose justification is in question to try to show that argu-ments of that sort are justified. But people have now and then been tempted by this idea.I think that’s because inductive reasoning really does seem intuitively cogent so that anyargument that employs it seems cogent too—even the inductive defense of induction. Butthe thing to say, I think, is that even if such an argument really is in fact cogent—so thatit really does in a way give a good reason for thinking that the conclusions of inductivearguments are likely to be true—offering this argument does nothing to explain why anyinductive argument—including that one—is cogent. Thus it doesn’t really speak to themain issue.

Right. What we want is a perspicuous account of why inductive reasoning is justified, not justan argument that presupposes—even if correctly—that it is. So what about Reichenbach’sview? Salmon points to one problem: that the same sort of defense will work for lots ofother rules for inferring from observed instances to general conclusions. This, as I under-stand it, is a fairly tricky technical point. The other rules in question are all pretty con-trived ones that will yield a result very close to that yielded by the inductive rule in thelong run—as the number of observed cases becomes very large (that’s why they will also

CONCLUDING DIALOGUE ON THE PROBLEM OF INDUCTION 149

succeed if success is possible). But they can support results that will deviate quite widelyfrom the standard inductive conclusion in the short run—with a smaller number of cases.

Yes, that’s the basic idea. You just need a rule that “corrects” or “adjusts” the inductive conclu-sion in a way that will diminish to zero as the number of cases becomes very large, in thelong run, but where the “correction” can still be very substantial in the short run. And thereason that this problem is really serious is that any actual use of induction will always bein the short run in relation to at least some of these rules, so that some alternative rule thatis equally justified on Reichenbachian grounds could yield more or less any other conclu-sion you like about the proportion of A’s that are B’s.

That’s pretty bad! But Reichenbach can still say—can’t he?—that if we reason inductively weare following a rule that is guaranteed to succeed eventually in finding the true regularity innature, if there is one to be found, even if only in the infinitely long run. Why isn’t that stillat least some justification for reasoning in that way, even if we don’t quite know what to sayabout these other contrived rules (ones that no one really takes seriously anyway)?

Unfortunately, there is another problem lurking here that seems to me even more serious.Reichenbach says that induction will succeed in finding a regularity in nature if there isone there to be found. But he is in effect admitting that inductive evidence in any partic-ular case, no matter how much of it there may be or how completely without exceptions itmay be, still gives us no reason at all for thinking that there is in fact a regularity in nature.And this amounts to admitting that such evidence gives us no reason at all for thinkingthat the conclusion of the corresponding inductive argument, which of course asserts thatthere is such a regularity, is true. His view may still yield some sort of justification—induction is better than guessing at random, which isn’t likely to succeed even if there isa regularity—but not the sort of justification that is needed for knowledge: a reason forthinking that the claim or belief in question is true.

It’s really hard to believe that inductive conclusions, and all of the further scientific and tech-nological claims that ultimately depend on them, have no more justification than that.What about the “ordinary language” justification? Is it maybe better than Salmon thinks?

Unfortunately, Salmon seems to be right about that one. It may be true that in ordinary con-texts what we have mainly in mind—and in that sense intend to convey—by speaking of“good evidence” is inductive evidence. Certainly that’s a plausible interpretation of whatsomeone might intend by asking whether there is good evidence for, say, the view thattaking vitamin C will prevent colds or similar sorts of claims. But what we reallyultimately mean, in the strictest sense, by “good evidence” is precisely evidence thatgenuinely shows that the claim in question is likely to be true, and the fact that weordinarily believe and even take for granted that inductive evidence does this doesn’testablish that it really does.

I see. If “good evidence” is just whatever common sense regards as good, then inductive evi-dence surely qualifies. But if we take “good evidence” to mean evidence that really isgood in the sense of genuinely showing that the conclusions it is taken to support are true,then what common sense thinks is irrelevant—or at least there is just the same problem allover again of trying to say why common sense is correct in thinking what it thinks.

Yes. The “ordinary language” solution is really just a disguised appeal to common sense, andthus (like all such appeals) fails to really address the underlying philosophical issue. Here,as elsewhere, common sense may of course be right, but we want to understand why it’sright—and, of course, the answer can’t be that it is right because it agrees with commonsense, that is, with itself.

So if all of this so far is correct, we seem to be left with only the a priori solution as apossibility. I know that you are sympathetic with that view, but I need a better explanationof how and why it is supposed to work. Certainly Ewing’s discussion of it doesn’t get usvery far. If we can’t articulate any sort of general principle, then just saying that inductiveinferences are justified according to a priori “intuition” seems like nothing more thansaying that we are sure that they are somehow justified and that we don’t see how the


justification can be anything but a priori, but that we really have no idea of how it actuallygoes. Surely that doesn’t help very much with the problem!

I have to agree that Ewing’s account of the a priori solution isn’t very illuminating in thatregard. But I think that he is still making a very important point. Suppose that we agree(as we must) that the empirical evidence collected in the inductive premise doesn’t in andof itself contain the more general conclusion—that is, that the conclusion genuinely goesbeyond the evidence (so that the inference is, as Salmon says, “ampliative”). Suppose alsothat we think that inductive inferences really do establish that their conclusions are likelyto be true—and we do all think that, after all. Then the hypothetical claim that says that ifthe premise of such an argument is true, then the conclusion is likely to be true must itselfbe justified somehow. And to say that it isn’t justified empirically (because the empiricalevidence establishes only the “if ” part) seems to leave only a priori justification as apossibility. That is a significant result because so many (like Salmon) have wanted toreject it, while still seeming to say that induction is nonetheless justified. If Ewing is right,that simply isn’t a tenable position.

But even if you and Ewing are right about that (and I admit that I can’t see any way around it),it still doesn’t take us very far. Aren’t a priori truths supposed to be rationally self-evidentor something like that? And if there is a general self-evident truth here, shouldn’t we beable to say what it is? Ewing’s appeal to the intuitive cogency of individual arguments justisn’t satisfactory—he can’t be saying that they are each justified a priori in a way that isindependent of all the others, which would be the only way to avoid the demand for a gen-eral principle or at least some sort of general rationale.

I agree that if the a priori solution is right, then there must be a general principle or rationaleto be found. The place to look, I think (though we can’t go very far into this here), is therepeated regularity that is involved in inductive evidence—something that Hume at onepoint claims to have no rational relevance at all. The question to ask, and I think it’s anobvious one, is why this general pattern of evidence persists: what the best explanation forthis persistence is. The underlying rationale for inductive reasoning might be then that, ingeneral and in the absence of specific reasons to the contrary, the best explanation of sucha pattern of evidence is that it corresponds to and is produced by an objective regularityin nature—which would then mean that the inductive conclusion is true.

That’s a little too fast. I can agree that if the pattern of evidence is quite extensive, then it isvery unlikely to be due to chance—which is to say that some explanation is needed. Butwhy should we think that the objective regularity explanation is the best one, where thatmeans the one that is most likely to be true?

I agree that a lot more elaboration is needed. That would be a long story, and we really don’thave time for it here. But the way to think about it is to try to think of other possibleexplanations for a persisting pattern of evidence of this sort and then try to find reasonswhy they are less likely—or maybe more likely—than the objective regularity explana-tion. Have a good time with that!

Documents

Bonjour Knowledge and Skepticism