William Alston, The Reliability of Sense Perception, Chapter 2: " Track Record and Other Simple...

8/13/2019 William Alston, The Reliability of Sense Perception, Chapter 2: " Track Record and Other Simple Empirical Arguments for Reliability"

1/7

rackRecord nd t h imple rguments 3

Chapter 2T R C K R E CO R D N D O T H E RSIMPLE EMPIRIC L RGUM ENTSFOR RELI BILITY

i. A Track Record Argu ment for the Reliabilityof Sense Perceptual Practice (SP)

Against this background we a re ready to tackle our central ques-tion, whether it is possible to establish the reliability of SP. Let s beginby considering generally what it takes to determine whether a way offorming beliefs is reliable. The inost obvious approach is inductive.Check a suitable sample of outpu ts for truth and take the proportionof truths in that sample as an estimate of the reliability of that mode ofbelief formation. Thou gh, as just pointed ou t, we cannot ul l i fyreliability with a favorable track record, such a record in a suitablylarge and varied spread of cases is the best and most direct evidencefor reliability. Furthermore, it would seem that any other evidencewould be parasitic on this. Suppose, for example, that a perceptualdoxastic mechanism that takes into account background informationof relevant sorts-where the subject is, how situated relative to theobject, etc.-is more reliable than one that does not. How would weknow this to be the case? Presumably, by deterinin ing that mecha-nisms of the for mer sort have a better track record than mechanismsof the latter sort (when tested in comparable situations), or else bycorrelating this difference with other differences that indicate dif-ferential reliability, in which case the assuinption of this latter dif-ferential reliability must be validated by a differential track record,or Thus , although there is no limit to the variety of indicatorsof degree of reliability, the most direct iiidication-favorable track

record-is fundamental to al1 the others. Hence it will be appropriateto focus on it in these preliminary considerations.

If we try to assess the reliability of SP in this way, however, weimmediately run into a roadblock. How do we determine whether theperceptual beliefs in ou r sample are tr ue? If we do so by takinganother look, listen, or whatever, we would seem to be presupposingthe reliability of SP in order to compile evidence for that reliabilityand so to fall into circularity. Moreover, more complex ways of deter-mining the truth value of perceptual beliefs will run into the sameproblem, though less directly. Th e basic point here is that any way wehave of determining perceivable facts in the physical world will de-pend, sooner or later, on what we learn from sense perception. Evenwhen what we appeal to in showing that X is or is not P (that this wineis a Chambolle Musigny) is not itself a perceptual report (that thewine in this bottle carne from a cask used to store wine made fromgrapes grown in the commune of Chambolle Musigny), still, in show-ing that that is the case, we must at some point rely on perceptualreports. We may be relying on various records that detail the historicalorigin of the wine in this bottle. Not only must we use oureyes to readthe reports, but the reports themselves, if valid, were based on obser-vations of the grapes and the wine at various stages of thei r history.don t know of any simple knockdown arguinent for this thesis,tha t any cognitive access we have to the physical world ultinlately restson sense perception. Al1 1 can dot o sup port it is the following. First,can go through a riumber of cases like the one just mentioned andexhibit the plausibility, in each case, of the claim that SP must berelied on at one o r more points. Second, we can survey the resourceswe have for det ermining particular perceivable matters of fact, andverify that they al1 involve reliance on SP at some stage. Thus, inaddition to a direct appeal to perception of the putative fact in ques-tion, there a re the following. 1) Memory of past perceptions, whichis, of course, only as trustworthy as the perceptions remembered. 2 )Inferences from part icular past facts plus general regularities, as inthe Chambolle Musigny case. Here it seems clear that our evidencefor both the past facts and the generalizations depends on sensoryobservation arid hence on SP at sorrie point. (3) Th e use of instru-ineilts. Here we must rely ori SP both Cor reading the instrument andfor our evidence for its reliability. (4) The use of reliable indications,


2/7

4 The Reliabilily o SenseParceptionas when we know from a vapor trail in the sky that a jet plane hasflown by recently. Here we rely on SP both for the knowledge of thevapor trail and for the generalization connectingjet flights and vaportrails. 5) More theoretical inferences, as when the presence of an-other planet was inferred from deviations in the expected orbit ofobserved planets. Here we rely on SP both for the observation of theknown planets and for evidence for the theory used in the derivation.And so it goes. Finaily, there s the general recognition that we haveno a priori access to particular perceivable facts. Even if it were thecase, and this is dubious, that we can establish a priori certain verygeneral principles about the physical world, iike 'Every event has acause' or 'Energy is conserved through every physical transaction',

. it still remains the case that 1 cannot by mere thinking determinewhether your c r is in the garage now or that you are now wearing ared dress.

Let's assume, then, that it is impossible to determine that a particu-lar perceptual belief is tme or false without making use at some pointof what we take ourselves to have learned from SP. It follows that wecannot use an inductive track record argument for the reliability of SPwithout presupposing that reliability, and so falling into circularity.Let's use the term 'basic' for any doxastic practice of which chis is true,that is, any practice for the reliability of which any othenvise effectivetrack record argument would be circular. Such a practice is properlycalled 'basic', since its epistemic claims cannot be validated solely bythe use of other practices. Thus, as shail indicate more fully in thesequel, it has a claim t be at the basis of our cognitive endeavors. Itrepresents a way of forming beliefs that cannot be judged, at least inany simple fashion, on the basis of the output of other, more funda-mental practices.

To be sure, 1 began by saying that our central issue was whether itis possible to determine the reliability of SP that is, to determinewhether r not it is reliable. And in exploring the possibility of a trackrecord argument, al1 1 have shown is that we fa11 into circularity if weuse it to establish reliability. But what about its use to establish unre-libi lity? There we could hardly make a charge of circularity stick. Ifthe conclusion is that SP is unreliable, 1 could not claim that my use ofperceptual premises reflects my practica1 assumption that SP isunreliable Quite the contrary. If such an argument also reties on SP as a

. . . . . . . . .. . . .. . . . . . . . .T > . . . .= _5

nformation, the charge would be different though equallvdisqualifying,namely, that the argument presupposes the contradic-tory of the conclusion. But we are not n a position to claim that anyotherwiseeffective argument for the unreliabilityof SP relies on SP inthis way. For it would be an effective argument for the unreliabilityofa doxasticpractice to exhibit enough inconsistenciesbetween its out-puts. We wouldn't have to determine what's what with respect to itssubjectmatter in order to concludeon those grounds that the practiceis unreliable. In fact, though perceptual reports do sometimescontra-dict each other, the incidenceof that seems clearly not to be of sucha magnitude as to justify taking SP to be significantly unreliable,though it certainly does show that SP is not perfectly reliable. In anyevent, there is little inclination among contemporary philosophers totake seriously the idea that SP is radically unreliable. Hence, in thisdiscussion 1 shall ignore that side of the matter and confine myattention to attempts to establish the reliability of SP.

ii Epistemic CircularityThis is a good place to take a harder look at the kind of circularity

that is involved in the track record argument. It is not the most directkind of logical circularity. We are not using the proposition that senseperception is reliable as one of our premises. Nevertheless, we areassuming the reliability of sense perception in using it, or somesource(s) dependent on it, to generate our premises. If one were tochallenge our premises and continue the challenge long enough, wewould eventually be driven to appeal to the reliability of sense percep-tion in defending our right to those premises. And if were to askmyself why 1 should accept the prernises, 1would, if pushed thereflection far enough, have to make the claim that sense perception isreliable. For f 1 weren't prepared to make that claim on reflection,why would 1, as a rational subject, countenance perceptual beliefs?Since this kind of circularity involves a commitment to the conclusionas a presupposition of our supposing ourselves to be zlstzfied in hold-ing the premises, we can properly term it 'epistemic circularity'.

Epistemically circular arguments would seem to be of no force. Ifwe have to assume the reliability of SP in order to suppose ourselves


3/7

entitled to &e premises, how can an argument from those premises,proposition? But surprisinglyenough, as 1argue in Epistemic Cir-however impecable its logical credentials, provide support for thatcularity, ' that does not prevent ou r using such arguments t s owthat sense perception is reliable or to justz y that thesis. Nor, paripassu, does it prevent us from being ustified in believing sense per-ception to be reliable by virtue of basing that belief on the premises ofa simple track record argument. At least this will be the case if thereare no higher leve r requirements for beingjustified in believing thatp, such as being justified in supposing the practice that yields thebelief to be a reliable one, or being ustified in supposing the groundon which the belief is based to be an adequate one. And, thoughcannot defend the position here, do not think of justification assubject to any such requirements . On my view, a belief is justified ifand only if it is based on an adequate ground; that is, it is necessaryonly that the ground be adequate, not that the subject know orjustifia-bly believe this, much less that the subject know or justifiably believethat al1 requirements for ustification are sat i~ fi ed .~ut then 1can bejustified in accepting the outputs of a certain doxastic practice with-out beingjustified in believing that the practice is reliable. need notalready be justified in supposing SP to be reliable in order to ejustified in various perceptual beliefs. SP must be reliable if am to bejustified in holding perceptual beliefs, but don't have to be justifiedin supposing this to be the case. But then i SP is reliable, can usevario;; (justified) perceptual beli'efs to show that SP is reliable, forneed no; already be jusufied in holding the conclusion in order to bejustified in holding ;he premises. The rgu ment would still be episte-hiullly circular fo i am still assuming a c t k e the reliability of SP informing normal perceptual beliefs. ~e ve rthe les s, don't have to be

'Ahton ig8gb.P or defense of this position see An Intemalisr Extemalism in Alston ig8gb and99 . One reason. and a conclusiveone, for awid ingsuch higher level requirements s thatthey imply that one is required to have an in finite hierarchy ofjusrified beli efs. For if to bejustified in rhe belief that p one has o bejusrified in believing diat l h e p n d ojihejm e jU an adcg uYr u (cal1 h e talicized proposirionq3. then a parallel condiu on wilJ be put on thebelief that q; ro bejustif ied n holdin g it on e must bejusrified in believing thatiheground jihbcliefihal U on aaqua m nd LObe justified in rhat belief, one will have to have a stillhigher level justified belief rhat the p u n d of that belieisadequate. And so on ad infinitum.Since it seems clear that no hum an being can possessan infinite hierarchy of beliers-justifiedor not-to impo se any such requirement will imply that no hu man be ing justifiably believesanything. and our subject matter will have dis ap pe ad .

Trad Record nd t h Simple A'gumcnts 17justified in making that assumption, in order to be ustified in the per-ceptual beliefs that give me my premises. Hence the epistemic circu-larity does not prevent justification from being transmitted from thepremises to a conclusion that would have been unjustified except forthis argument. That applies even to a simple track record ar g ~ m e n t . ~But even if m right about this and it is pos ible to establish thereliability of sense percepuon and other basic sources of belief bysimple track record arguments, these arguments still do not satisfythe usual aspirations of those seeking to determine whether a basicdoxastic practice like SP is reliable. The reason is this. What 1pointedout in the previous paragraph is that i fsmeperceptionis reliabk, a trackrecord argument will suffice to show that it is. Epistemic circularitydoes not in and of itself disqualify the argument. But even grantingthat point, the argument will not do its job unless we rejustified inaccepting its premises; and that is the case only if sense perception isin fact reliable. This is to offer a stone instead of bread. We can say thesame of any belief-forming practice whatever, no matter how disrep-utable. We can just as well say of crystal ball gazing tha t if it s reliable,we can use a track record argument to show that it is reliable. Butwhen we ask whether one or another source of belid is reliable, weare interested in di scr t m i dng those that can reasonably be trustedfrom those that cannot. Hence merely showing that $a given source isreliable it can be shown by its record to be reliable, does nothing tindicate that the source belongs with the sheep rather than with thegoats. have removed an allegedly crippling disability, but have notgiven the argument a clean bill of health. Hence shall disqualifyepistemically circular arguments on the grounds that they do notserve to discriminate between reliable and unreliable doxastic prac-tices.

iii. A Piecemeal ApproachThus, a simple track record argument cannot give us what we arelooking for when we attempt to show that SP is reliable, just becauseit is infected with epistemic circularity. Faced with this result, onemight suggest that the trouble is that we have chosen too iarge andSee Epistemic Circularity (Alston i989 b) for much more demil on bis argument.


4/7

unwieldy a unit for our investigation. Why not consider the questionof whether some subset of perceptual belief-forming mechanisms isreliable? In that case, there is no reason to think that we have topresuppose, even in practice, the reliability of the ways being tested inorder to pile up evidence for their reliability. To explore this, let's startat the other end of the spectrum and consider a very specific belief-forming mechanism, for example, one that takes something's lookinglike a peach as input and yields a belief [That's a peach] as output. Totake a track record approach to establishing reliability here , we con-sider a sufficient number of sufficiently varied cases in which thisinput-output function was operative, cases that are al1 normal in thesense of being in circumstances of the sort we frequently encounter(no laser images, no Cartesian demon). We then determine in whatproportion of those cases the belief engendered was true, that is, inwhat proportion of those cases the object in question was a peach.And how do we do that? Obviously, we have to utilize some other wayof spotting peaches. We can't use this mechanism, whether in theoriginal subject or others, without begging the question. And we dohave other ways of doing this. We can srnell, taste, and feel the object.We can determine whether it was picked from a peach tree. We couldrun biochemical and even rnicrobiological tests. We can make icecream or pie from it and see what that tastes like. And so on. In noneof these cases would we be relying on the move from Il looks like apeachto t i s a peach. To be sure, in these cases we are relying heavily onother perceptual mechanisrns. Even in the sophisticated biochemicaland rnicrobiological tests, the tester has to rely on what she sees,hears, or otherwise perceives at some point in carrying out the test.But that in no way casts any doubt on our capacity to determine thetruth of the outputs of th mechanism without circularity. For at nopoint do we appeal to beliefs generated by that mechanism itself.Thus it is clear that where we are testing a very specific source ofbelief for reliability, there is no difficulty n principie in carrying out atrack record test without falling victim to circularity. And the sourcedoes not have to be nearly this restrictive to achieve this result. Thesame point can be made about testing visual color or shape attribu-tions generally, or visual identification of objects generally, or evenvisual belief formation generally (and so for the other senses). To besure, the more extensive the mechanism (or group of mechanisms)

Track Record and Other Simple rgumantsthe more difficult it is to find an independent check on truth. Forvisual shape attributions (just on the basis of how the object looks) wehave a variety of o ther checks, including how it feels as well asmea-surements of various sorts. But if the g roup is as large as visual beliefsgenerally-even restricting ourselves to primary visual beliefs, be-liefs about objects formed just on the basis of how they look-thingsare more difficult. To carry through a noncircular track record argu-ment here we would have to determine in each case whether is, forexample, a birch tree without relying at any point on how somethinglooks. Testing procedures that rely on appearances only to the othersenses could be devised, suppose, but doing so might require consid-erable ingenuity.

iv. Back to SP s holeThus, whenever it is a question of the reliability of some mode (or

family of modes) of perceptual belief formatioii that is more re-stricted than the whole of that territory, we can find ways of checkingits outputs for truth value that are independent of any reliance on themode(s) under scrutiny. But on further reflection it can hardly fail tostrike us that we will run into circularity if we persist long enough. Tooversimplify, let's say we check the accuracy of visual betiefs by usingaudition and touch in the perceptual part of those tests.4 Now whathappens when we test the reliability of auditorily and tactilely formedbeliefs? Perhaps we can get by with the other two senses (though 1doubt it), but even if we could, when we come to test them we will per-force have to rely on one or more of the modalities already checked.So, to put it schematically, vision is validated by audit ion and touch.Touch is validated by smell and taste. SmeIl and taste are validated byvision and audition. And there we have our circle. Th e details of theprocedure will change the relative position of the modalities on theperimeter of the circle, but they cannot change the general characterof the proceedings. The basic point is that we cannot do anythingtoward determining the truth value of singular statements about

As chis example indicates. am nor supposing and it is noc the case. that in determiningLhe trulh value of a percepcual belief we rely solely on perception. We also will make use o freasoning of various so w and we will rely on memory. Here am concerned only with Lhepoint that perceptual beliefs consutute an essential part oE what we are using and u usting.


5/7

perceivable objects without relying on sense perception for informa-tion somewhere in the proceedings. More precisely, we cannot do sowithout relying on beliefs the ustification of which rests, proximatelyor remotely, on sense perception. I f 1 already know that all the treeson my property are birches, then that confirms a perceptuai reportthat an object in a certain place on my property is a birch tree; 1 don'thave to take another look myself. But, of course, have that antece-den t knowledge only because have seen the trees'rnany times andrecognized them as birches, or because someone who was n a posi-tion t know told me that ail those trees are birches and he knew thatbecause he visually recognized them as birches; or Thus it seemsclear that an basis we have for supposing a perceptuai belief to beBtrue o r faise wiii involve, somewhere along the line, a r el in ce on theaccuracy of perceptual belief formation. And that being the case,though we can support the reliabiiity of certain stretches of the terri-tory by relying on other stretches, we will, since th variety of suchstretches is finite, eventually be involved in a circle in seeking toostablish the reliability of any part of the tenitory.

had better seize this opportunity to make explicit a complicationthat affects many of the points 1 will be making about the epistemiccircularity of various arguments for the reliability of SP. From thestandpoint of coherence theory this circularity ~ not a worry, pro-vided the circle is large enough. Coherence thw ry holds that individ-ual beliefs gain positive epistemic status (justification, rationality,being established, or whatever) by irirtue of being involved in a totalsystem of beliefs that is coherent. And to say that a system is coherentis to say, roughly, that there are strong bonds of mutual supportbetween the constituents. There is no unidirectional support thatpasses from foundations that need no support from other beliefs tothe rest of the system, the superstructure'i. Rather, every componentbelief and every subsystern is involved in both supporting and beingsupported by others. Reciprocal support is the rule rather than theexception. From this standpoint there is nothing disturbing about thecircle involved in using perceptual beliefs to support the principlethat sense perception is reliable. Particular perceptual beliefs, on theone hand, and the belief in the reliability of sense perception, onthe other, support each other, thereby increasing the coherence ofthe system that contains both. However, for purposes of this essay 1

am setting aside coherence theories without a hearing. 1 take it thatthe live possibilityof a multiplicity, perhaps an indefinite multiplicity,of incompatible but equally coherent systems of equal comprehen-siveness is sufficient to show that interna1 coherence cannot be thewhole story of what gives beliefs a positive epistemic status. Andtherefore 1will continue to regard circularity as disqualifying.The above discussionamounts to one way of motivatinga concernwith the reliability of sense perception generally.For if the attempt tojustify a particular group of perceptual doxastic mechanisms by rely-ingon others to do so inevitablyleads to a circleof self-support withinthe sphere of perceptuai belief formation, it looks as if such effortspresuppose the general reliabilityof SP. If perceptual belief isgener-ally reliable, then it is in order to check on any suspicious part thereofby relying on other parts about which suspicions have not beenaroused. But ifweare not already entitled to take sense perception toe mostly reliable, the fact that some of its outputs confirm others

would seem to be of little epistemological significance. For al1 wewould have any reason to suppose, it might be one of those vastcoherent systems of fancy that are regularly thrown up as an objec-tion to the idea that the coherence of a system isa sufficientindicationof its truth. Hence we are dnven to raise the question of reliabilityabout sense perception generally. And once that general question israised it isobvious that wecannot use perceptual beliefs (perceptuallyderived putative information or knowledge) as any part of our casefora positiveanswer to the question. That would be a circularity moreobvious than the one we saw in the piecemeal approach to the episte-mic assessmentof sense perception. Hence we are ineluctably led tothe basic issue of this essay: is it possible to establish the reliabilityofsense perception in a noncircular fashion?

Nor does the interest of this larger question wholly stem from thebackground 1 have just sketched. ns s perception as a whole is, anatural unit about which to raise the question of reliability.Though 1suppose that t is conceivable in the abstract that, say, vision giws usgenerally accurate information but audition or touch does not, andthough there obviously is a difference in the reliability of differentperceptual mechanisms (up close vision is more reliable than distantvision), still it seems overwhelminglyplausible to suppose that senseperception standsor fallsas a wholeat the bar of critica1scrutiny. It is


6/7

22 heRe.? o Sense erceptiadifficult to take seriously the idea that one of our sense modalitiesmight be radical different from others in the reliability of the infor-rnation it provides, or that: one of the senses might vary mi in itsreliability as a source of the beliefs it regularly and confidently en-genders (for exampie, that vision might be reliable for the identifica-tion of trees but not of animalsf. Hence it is quite understandable thatthe question of the reliability of sense perception iibmhaupt has been atraditional preoccupation of philosophers.

Another reason for this prmccupation is that the major reasonsfor doubting the reliability of sense perception have always seerned toapply across the board, not to s o m senses rather than others, or tosome particular kinds of beliefs. This is true of arguments fromperceptual errors.(if one can be mistaken some of the time one can bemistaken al1 of the time), from dreams o r deceiving demons (we can tdistinguish th a s e where we are only dreaming that the Louvre is infront of us, o r where a demon is deceiving us into supposing we seethe Louvre, from the situation in which we are veridicdy seeing theLouvre), and from underdetermination by sensory evidence (oursense experiences are quite compatible with things not being as theyseem to indicate; they do not provide conclusive grounds for thebeliefs we typically base on them). Thu s insofar Sconcern with thereliability of the senses stems from skeptical doubts like these, it isunderstandable that the question should be addressed to the sensesgenerally.

v. Pragmatic Argum ents : Valdation by FruitsWhere are we to turn next? A popular Iine of thought is that SP

proves itself by its fruits, particularly by the way in which it puts usin a position to predict and thereby, to some extent, to control thecourse of events. It provides us with data on the basis of which weestablish lawlike generalizations, which we can then use as the basisfor prediction and control. By taking sense experience as a guide towhat is around us we can learn in each of a number of instantes thatmiIk sours more slowly when cold than when warm. This puts usinto a position to predict that a refrigerated bottle of milk will lastlonger than an unrefrigerated one, and we can use this knowledge

to control the condition of our milk. This is the humblest of exam-ples, and the predictivepower isgreatly increased in scopeand preci-sion as we move further into the higher reaches of theory; but thegeneral point is the same. SP proves itself by what it puts us in aposition to do. If it weren t usually giving us the straight story aboutwhat is happening around us, how could we have so much success inpredictingthe further course of events when we use its deliverancesto build up systems of general beliefs that we use to make thosepredictions?

That sounds right. But how do we know that we are often success-ui in prediction?By induction from particular cases of success, ob-

viously. But how do we know that we are successful in particularcases?By using our senses to determine whether what was predictedactually occurred, or by having recourse to some other method thatsooneror later relies on the deliverances of SP. It is not as i an angeltells us that the prediction isborne out, or as if rationalintuition doesthejob. But then we are back in epistemic circularity.We can mountthis argument for the reliability of SPonly by using SP to get some ofour crucial premises. Once more the argument establishes the re-liability of SP only if SP is in fact reliable. And that still leaves uswondering whether that condition is satisfied.

Another fairly simple argument that quickiy falls victirn to episte-mic circularityisan argument from the (natural world)provenanceofour basic belief-formingdispositions,including SP.5The most promi-nent contemporary form of such an argument, perhaps the onlycurrently prominent form, is the idea that our tendencies to formbeliefs in accordancewith SP, memory, introspection,and inductiveand deductive reasoning have been selected by evolution, that thiswouldn t have happened unless SP and the other basicdoxasticprac-tices confer an adaptive advantage, and that they wouldn t do thisunless they were reliable. jThisiscertainlyan intuitivelyplausible lineof thought, a plausibility indicatedby the followingquotations.

Note that am resuicting rnyselr here t the factors within the world of nature thatinfluence the acquisicion of these dispositions. Larpr we shall consider he very differen targument that od has endow ed us with these dispositions. and that this guararicees theirreliabilitv.See ~ t i c h990 chap 3 fora n assemblage of hints toward such an argumeiit, and foracute criticisrns of any suc h liiie of wasoning.


7/7

4 The eliability o Sm e PerceptimCreatures inveterately wrong in their inductions have a palhetic butpraiseworthy tendency to die out before reproduung their kind.(Quine 1969,p. 26Natural selection guarantees that t of an organism s beliefs will betrue, most of its strategies rational. (Dennett 1981,p. 75

To spell out the argument a bit more, it presumably runs somethinglike this. Our most fundamental belief-forming dispositions havedeveloped in the course of the evolution of the human species, alongwith the rest of ou r innate endowments. But they wouldn t havedeveloped unless they gave us an advantage in the struggle for sur-vival. And belief-forming dispositions couldn t make a positive contri-bution to our ability to survive unless they produce mostly true be-liefs. If they were to yield mostly false beliefs instead, we wouldfrequently be acting on misapprehensions, and if that happenedfrequen tly enough, we wouldn t last long. Hence, given that thehuman species has not only survived but even achieved dominanteon the planet, it must be that our basic doxastic practices are reliable.

Despite its obvious plausibility -this argument is not immune tocriticism. Stephen Stich 1990, hap. 3 subjects it to severe criticismon two points. First, he argues that eyolution does not generally,much less always, produce optimally designed systems in survivingspecies. Second, he argues against the view that natural selection isthe only important factor in the development of ou r basic doxastictenden~ies.~owever, 1 need not get into the question whether thepremises of the argument provide adequate suppo rt for the conclu-sion. The only point need make here is that the argument is ob-viously epistemically circular. The evolutionary theory that the argu-ment presupposes and makes use of obviously rests on empiricalevidence gathered by reliance on SP. And so even if the argumentcomes out totally unscathed from attacks like those of Plantinga andStich, it is still disqualified by virtue of relying for its cogency on theassumption of what it is designed to establish. In the course of ourdiscussion we will find again and again that epistemic circularity rearsits unlovely head, sometimes in the most unexpected places.

vi. The Road AheadHaving warmed up on the simplest arguments for reliability, let s

turn to attempts wecan take more seriously. These can be divided intoempirical and a priori arguments. The latter have often been devel-oped explicitly to avoid epistemic circularity; and if they are wholly apriori they will succeed in that. The empirical arguments are clearlythreatened by epistemiccircularity. Someof their authors make stren-uous efforts to avoid this; with what success will appear in the fullnessof time. Though it might seem most natural to continue the abovediscussion by moving on to the more sophisticated ernpirical argu-ments, it will prove most fitting to save them for later.

'Stich's target is inferential strategies raiher [han doxastic practices. but his argument sare easily adaptable for my purp oses. eealso Plantinga 1999, hap. l o , in which rhere is anallau t asaul ton the assumption hat our doxastic tendenciesc n be adaptiv e only if they arereliable.

William Alston, The Reliability of Sense Perception, Chapter 2: " Track Record and Other Simple...

Documents

Transcript of William Alston, The Reliability of Sense Perception, Chapter 2: " Track Record and Other Simple...