Our Entitlement to Self-Knowledge: I.

Tyler Burge Author(s): Tyler Burge and Christopher Peacocke Reviewed work(s): Source: Proceedings of the Aristotelian Society, New Series, Vol. 96 (1996), pp. 91-116 Published by: Blackwell Publishing on behalf of The Aristotelian Society Stable URL: http://www.jstor.org/stable/4545230 . Accessed: 17/12/2011 19:35
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

The Aristotelian Society and Blackwell Publishing are collaborating with JSTOR to digitize, preserve and extend access to Proceedings of the Aristotelian Society.

http://www.jstor.org

TO OUR ENTITLEMENT SELF-KNOWLEDGE Peacocke TylerBurge and Christopher I-Tyler Burge

our for want to understand epistemic warrant a certainrangeof I ourown thoughtsand attitudes. am guidedby judgmentsabout two hypotheses.Oneis thattherearecertainsortsof self-knowledge that are epistemicallyspecial. The otheris thatthe epistemic right or warrantwe have to these sorts of self-knowledgeis, in a sense, neutral.I want to understand specialnessand this environmentally this environmental neutrality. The hypothesisof epistemicspecialnesswill be arguedfor in this paper.The hypothesisof environmental neutralityis relevantto a projectthat deals with scepticism and the natureand functionsof reason.I will not develop this latterhypothesisin depthhere, but I will commenton it for the sake of orientation. Most of our empirical thoughts and our thoughts about our empirical thoughts depend for their individuationconditions on relations that we bear to a particularenvironment.But, on my for guidinghypothesis,ourepistemicwarrant ourjudgmentsabout our thoughtsdoes not dependon particular relationsto a particular It environment. is commonto anyenvironment derivesfromthe and natureof the thinkeras a criticalreasoner.This point is relevantto showingthatcertainclaimsto self-knowledgewhich areamongthe do premisesin a certainanti-sceptical argument notbeg thequestion by depending on presumptionsabout the environmentthat the scepticcalls intoquestion.In thispaperI will notdiscussscepticism. But I begin with the sort of cogito-like judgmentsthat figured in traditional anti-sceptical arguments. I believe these judgments relevantnot only to scepticism,but to the epistemic specialnessof some self-knowledge. Although some striking features of cogito-likejudgmentsare not sharedby all membersof the wider range of judgments about one's thoughtswhose epistemic status interestsme, cogito-likejudgmentsprovidea useful paradigmfor reflection.
I

92

I-TYLER BURGE

So I begin with some remarksabouta judgmentthat: (1) I am thinkingthatthereare physical entities. Thisjudgmentis an instanceof cogito-likethoughts,an elaboration of Descartes' I am thinking. Let us construe 'thinking' in (1) minimally-as engagingin thoughtor havinga thought,regardless a of whetherit is merelyentertaining thought,makinga judgment, In or whatever. this sense, one 'thinks'all propositional components of any thoughtone thinks (includingnegatedones, antecedentsof conditionals,andso on). (1) is the contentof my judgment.I accept it as true.To be true, (1) requiresonly thatI am engaging in some thoughtwhose contentis thatthere are physicalentities. We do not rest thisjudgmentuponany observationor perception is suchas was traditionally called 'innersense'. Thejudgment direct, the judgment requires sufficient based on nothing else. Making to understanding think (1). But once one makes the judgment,or indeedjust engages in the thought,one makesit true.The thoughtis contextuallyself-verifying.One cannoterrif one does not thinkit, andif one does thinkit one cannoterr.In this sense, such thinkings are infallible. I do not claim thatjudgmentslike (1) are indubitable. The scope for humanperversityis very wide. One could be so far gone as to thinkto oneself: 'I do not know whetherI am now thinkingor not; maybeI am deador unconscious;my mantra mayhavefinallymade me blissfully free of thought'.Such mistakendoubtwould evince cognitive pathology,but I thinkit possible. It is an error,however, thatmost people would avoid withoutswerving. Key featuresof (1) are sharedby judgmentsof (2) I judge, herewith,thatthereare physical entities. When judge in (2) is used to execute not merely describe a judgment,judgments of (2) are contextuallyself-verifying. (2) is not made trueby the mere thinkingof it, nor does it have quitethe same quasi-logical self-evident status that (1) does. These are subtletiesthatI will have to discuss on anotheroccasion. (1) and(2) arenot merephilosophicalcuriosities.I thinkthatthey the represent form of manyordinary self-awarejudgments(at least when (1) is taken to have the 'herewith'reflexivity of (2).) When one makesajudgmentandis conceptuallyawareof one's so doing,

OUR ENTITLEMENT SELF-KNOWLEDGE TO

93

whether or not one spells out this conceptual awareness, one's judgmentshave a reflexive form like thatof (2). Such conceptual self-awarenessgoes beyondsimply consciouslythinkinga thought, but it is not an unusualphenomenonamong people with normal second-orderabilities. Thus I believe that cogito-like judgments constitutea significantsegment of our everydaymentalactivity. To remarkthat (1) and (2) are contextuallyself-verifying is to remark theirtruth on conditions,noton ourjustification epistemic or warrantin thinking them. It does seem that understanding(1) suffices for knowing thatit is true.And the relevantunderstanding requiresno great perspicacity.But noting that it is self-evidently self-verifying (supposing that this needed no more commentwhich of course it would) would not capturefully whatis involved in its epistemic status.For I thinkthatcogito-likejudgmentsshare an interesting epistemic status with a number of types of selfknowledgethatarenot contextuallyself-verifyingor infallible,and thatlack the quasi-logicalstatusof (1). I have in minda widerclass of judgments about states, not just reflexive occurrencesjudgmentsaboutwhatone believes, wants,intends. When we makejudgmentsaboutmany of our mentalstates and events, our judgments commonly constitute knowledge. I know very well thatI believe that there are physical entities-if I judge thatI do. Suchjudgmentsdo not merelyevince an innerstatein the way thata yelp evinces a pain;norarethey avowalsor conventional practiceswithout cognitive value. What is the epistemic status of such judgments? What epistemic warrantdo we have to make them? I take the notion of epistemic warrantto be broaderthan the ordinarynotion of justification.An individual'sepistemic warrant may consist in a justificationthatthe individualhas for a belief or otherepistemic act or state. But it may also be an entitlementthat consistsin a statusof operating anappropriate in accordwith in way normsof reason, even when these normscannotbe articulated by the individualwho has thatstatus.Wehave an entitlement certain to perceptualbeliefs or to certainlogical inferenceseven thoughwe may lack reasonsor justificationsfor them. The entitlementcould in principle presumably-though often only with extreme philosophicaldifficulty-be articulated someone.Butthisarticulation by

94

I-TYLER BURGE

need not be part of the repertoireof the individual that has the entitlement. Our epistemic warrant much of our self-knowledgeis of this to sort. Most of us have no justifying argumentor evidence backing the relevant judgments. The judgments are immediate, non-inferential. Althoughcogito-likejudgmentsmay countas selfevident or self-verifying, most judgmentsthat interestme do not. Whereinare we being reasonable-in the sense of operatingunder normssanctionedby reason-in makingjudgmentsaboutourown minds? As I have intimated,the remarksaboutself-verificationsuggest an initialanalogybetweencogito-likejudgmentsandknowledgeof simple logical truths.The truthof judgmentsof (1) and (2) is, in a broadsense, presentin the form and logic of the thought.Thereis something of the same self-evident and obvious featureshere as there are in simple logical truths.The main differences are that cogito-like judgments are dependenton being thought for being true,and are in theirspecially directway self-verifying. Anotheranalogyto knowledgeof simplelogicaltruths this:The is key to the epistemicstatusof cogito-likejudgmentsseems to reside in ordinaryunderstanding, in some mechanismconnectingthe not knowerwith a sensed object. This point will be one of the key elements in my accountof the environmental neutrality and specialness of self-knowledge. Perceptual experiencesparticular a given environment to inevitably figure in the acquisition of understanding almost any given of content.But one's epistemicwarrant believing the contentmay for not incorporatethe perceptualexperiences or beliefs that go into it. understanding Thisis the traditional view of knowledgeof logical or mathematicaltruths.One may need perceptualexperience to come to understand simplelogical orarithmetical notionsandtruths. (This is surelythe case with such logical truthsas 'nothingis-botha dog and not a dog'.) But on the traditional view such experienceis not a constituentof one's justificationor entitlementin believing simple logical or arithmetical truths. I am not arguingfor the traditional view-just recallingit. The element in it relevantto ourpurposesis the following. The account of epistemic justification or entitlementmay presuppose understanding, which may be dependent on particular perceptual

TO OUR ENTITLEMENT SELF-KNOWLEDGE

95

But relationsto a given environment. the accountneed not include perceptualbeliefs or experiencesas constituentsin the individual's of justification or entitlement.The account can allow attribution conceptsto the individualwhichcould be acquiredonly in a limited range of possible environments,while itself taking a form that is applicable to any critical reasoner, regardless of the particular dependentcontentsof his or her thought. environmentally I wantto illustratethe relevanceof this idea to ourdiscussionby reconsidering the scenario of one's being switched between differentenvironments unawares-a scenarioI discussedin a paper some years back.1Let us assume for the sake of argumentthatmy thinkingthattherearephysicalentities(hencemy thinkingthatI am thinking that there are physical entities) is the thought that it is becauseof relevantcausalrelationsI bearto actualphysicalobjects in my environment.Let us also assume that an individualwith a chemicallyidenticalbody could have beenbroughtup in a situation in which such relationswere lacking-and in which the conceptof physical object could not be acquired-but in which different, counterpart thoughts occurred. (I doubt that physical object is a but conceptuniversalto all possiblecriticalreasoners; if one did not doubt,anotherconceptcould be chosen.) Finally,let us grantthatif at any time one were switchedunawaresfromone's actualsituation situation,one would have no resourcesthat into such a counterpart would tip one off to the difference. Unless memoryandlearningconnectionsto the originalenvironmentwere broken,it is hardto describea switchof actualsituations thatwould producea new twin set of the concepts,with no residue from the past experiences. So in the case I am imagining one's thoughtsdo not switch to twin thoughts.Because of a switch one's thoughts might, however, change content, broadening their extensions withoutone's being awareof theirdoing so. I takeit thatthis observationis sufficientto promptthe following question. Given that we are insensitive to such alleged possible changes in content,how can we know what we are thinking?
1 'Individualismand Self-Knowledge' Journal of Philosophy 85 (1988), pp. 649-663. Reprintedin Quassim Cassam ed. Self-Knowledge(Oxford, Oxford University Press, 1994).

96

I-TYLER BURGE

I will nottryto dealwith thisquestionin all its ramifications here. But as I noted some years back, some of the negative force of the There question can be shown to be illusory by this consideration: is no way for one to make a mistake about the content,of one's present-tensed thoughtin the relevantcases. Suppose that I think that I am engaging in a thoughtthat there are physical objects. In thinkingthis, I have to engage in the very thoughtI am referringto and ascribingto myself. The referenceto the content-expressed in the that-clause-cannot be carriedout unless I actually engage in the thought. The intentionalcontent mentionedin the that-clauseis not merelyan objectof referenceor cognition;it is partof the cognitionitself. It is thoughtandthought aboutin the sameact. If background conditionsaredifferentenough so that I am thinking different thoughts, they will be different enough so thatthe objectsof referenceand self-ascriptionwill also be different.So no matterhow my thoughtsare affected,no matter how I am switched around,I will be correctin self-ascriptionsof contentthatare correctlyexpressedin cogito-that-clause form. It wouldbe a mistaketo replythatbecauseone's correctreference does notgive one any graspof whatone is referring thisreference to, is empty. For to self-ascribe thoughts in the way expressed by that-clauses,one has to understand thoughtsone is referringto the well enoughto thinkthem.One need nothave any moreexplicatory of understanding one's thoughtsthanis necessaryto thinkthem.One need not masteranti-individualism, much less have an empirical masteryof the conditionsthat have establishedthe identityof the thoughtsone thinks.Such masteryis emphaticallynot guaranteed by masteryof cogito-self-ascriptions. one is guaranteed one But that ascribes something of which one has the ordinaryunderstanding involved in using conceptsandthinkingthoughts. This understanding presupposesthe causal-perceptual relations to a particularenvironmentthat help determinewhat content is available for being understood. What one can think is partly dependent on relations to one's environment. And one's second-order self-ascriptions inherit both the content and the backgroundenvironmentalcontent-determining conditions from one's first-order understanding. I have grantedthatone need not be sensitiveto actualor counterfactualchanges in whatone understands undertransportations into

OUR ENTITLEMENT SELF-KNOWLEDGE TO

97

environments where the contentof one's understanding changesor would be different. One need not be capable of detecting such in changes.Butin any situation whicha personcanthinktherelevant cogito-like judgments, the person would think them with understanding-andto all appearances, knowledgeably. anyrate, At thereis no obviousreasonwhy knowledgein suchjudgments would be preventedby suchchanges,muchless suchpossiblechanges.2In any such twin situation, the person would understandthe selfascribedcontents and would self-ascribe them with a justice and reliability that is equal to that in any ordinary situation. Some entitlementattachingto understanding seems to be what the selfknowledge depends upon, not on some knowledge of what the consists in, or whetherit differsfromunderstanding understanding thatis past or possible. The person's epistemic entitlementto the self-ascriptionspresupposes understanding. Understanding is, as I have noted, dependenton and local to causal-perceptual relations to a given environment.But the entitlement that underlies knowledgeable cogito-like thoughtsandotherself-ascriptionsdoes not seem local andseems to survivesuch switches.It seems to be carriedsomehow by the fact that we correctly self-ascribe any content at all with understanding. Wheredoes the entitlementderivefrom?And what makes it capableof survivingsuch environmental switches?

2 Are thereswitching situationsin which one would have reasonablegroundfor doubting what contents one is thinking, so that a cogito-type judgment would not constitute knowledge?This is very complex, butI will makea few remarks here.The self-ascription in the that-clauseway cannot involve a mistake about the intentionalcontent. So the possibilityof a switchdoes not threaten mistake.I thinkthereforethatsuch possibilities a pose no relevant alternativethreat to one's entitlement to one's judgment about the that-clausecontentof one's thoughts.I believe that the relevantminimalunderstanding suffices for knowledge in cogito-like judgments. Even in non-cogito-likejudgments, switches in content cannot, for the same reason, undermineknowledgeabilityof the contentof self-ascriptions.Cf. my 'Individualism Self-Knowledge'op. cit., p. 659. and A fuller story has to be told aboutthe propositional-attitude concepts in non-cogito-like judgments.I thinkthe possibilityof switching,or of errorsof incompleteunderstanding, do not by themselves undermineknowledge; but I will have to discuss these matters furtherelsewhere. Some worriesaboutswitchingsituationshave focused on memory.I thinkthatthey tend to confuse preservativememory with memory of objects or with comparisonswithin memory,andto overratethe extent to which the contentretrievedin memoryis sensitive to immediateenvironmental context. For a discussion of preservativememory,see my 'ContentPreservation' PhilosophicalReview 102 (October1993), pp. 457-488. The

98

I-TYLER

BURGE

I thinkthattherelevantentitlement derivesnotfromthereliability relationbetweencognitionandits object. of some causal-perceptual judgmentsin It has two othersources.One is the role of the relevant criticalreasoning.The otheris a constitutiverelationbetween the judgments and their subject matter-or between the judgments aboutone's thoughtsandthejudgments'being true.Understanding and making suchjudgmentsis constitutivelyassociatedboth with being reasonableand with gettingthemright. Briefly drawn,my line of thoughtwill be this. To be capableof critical reasoning, and to be subject to certain rational norms necessarily associatedwith such reasoning,some mental acts and states must be knowledgeablyreviewable.3The specific character of this knowledgeablereviewabilityrequiresthat it be associated with an epistemic entitlementthat is distinctive. The entitlement judgments.There mustbe strongerthanthatinvolvedin perceptual rational relation,of a sortto be explained, mustbe a non-contingent, judgmentsandtheirsubjectmatteror betweenrelevantfirst-person truth. All of us, even scepticsamongus, recognizea practiceof critical Criticalreasoningis reasoningthatinvolves an ability reasoning.4 to recognizeandeffectively employreasonablecriticismor support for reasonsandreasoning.It is reasoningguidedby an appreciation, use, and assessmentof reasonsandreasoningas such. As a critical reasoner,one not only reasons.One recognizesreasonsas reasons. One evaluates, checks, weighs, criticizes, supplements one's ability reasonsand reasoning.Clearly,this requiresa second-order to think about thought contents or propositions, and rational relationsamongthem.
present 3 I thinkthat the following necessity also holds:To thinkthe relevantfirst-person tense thoughts about one's thoughts and attitudes, one must be capable of critical conceptor fully formed reasoning.Indeed,I thinkthatto have a fully formedfirst-person attitudes,one must be capableof criticalreasoning.To master concepts of propositional concepts of propositionalattitudes in a suitably rich sense, one must be capable of appreciatingthe force and relevance of reasons to attitudesas such, which amountsto being able to reasoncriticallyaboutreasonsand reasoning.And to mastera fully formed first-personconcept, one must have concepts of propositionalattitudes. 4 In actual practice, critical reasoning approximateswhat I call reflective reasoning, Reflective reasoningmakesuse of all the mainconceptsnecessaryto a full understanding of essential or fundamental elementsin reasoning.Criticalreasoningis simply reasoning that is sufficiently articulateto appreciatereasonsas reasonsand to employ articulated criticismof reasonsand reasoning(as reasonsand reasoning).

OURENTITLEMENT SELF-KNOWLEDGE TO

99

When one carries out a proof, one checks the steps of the reasoning,making sure that the inferences are valid. Any activity of proof requiressome conception of validity, which requiresan abilityto thinkof the propositionsin a proofas constitutingreasons for what follows from them. Indeed, it is arguable that use of therefore in reasoning-deductive or otherwise-constitutes an exercise of this meta-cognitive ability. When one engages in practicaldeliberation, articulates weighs considerations one and on each side, goes over possible sources of bias, thinks through consequences.Essentialto carryingout criticalreasoningis using one's knowledge of what constitutesgood reasons to guide one's actualfirst-order reasoning. A non-critical reasoner reasons blind, without appreciating reasonsas reasons.Animals and small childrenreasonin this way. But reasoning under rational control of the reasoner is critical reasoning.Not all reasoningby criticalreasoners critical.Muchof is our reasoningis blind,poorly accessible, and unaware.We change attitudes rational in ways withouthavingmuchsense of whatwe are doing. Often we are poor at saying what our reasoningis. Still, the abilityto takerationalcontrolof one's reasoningis crucialin many enterprises-in giving a proof, in thinkingthrougha plan, in constructing theory,in engagingin debate.Forreasoningto be critical, a it must sometimesinvolve actualawarenessandreview of reasons; andsuch a reviewingstandpoint mustnormallybe available.5 Criticalreasoninginvolves an ability not merely to assess truth, falsity, evidential support,entailment,and non-entailment among propositionsor thoughtcontents.It also involves anabilityto assess thetruth reasonability reasoning-hence attitudes.Thisis not and of to say thatcriticalreasoningmustfocus on attitudes,as opposedto their subject matter.Normally we reason not about ourselves but aboutthe world or aboutpracticalgoods. But to be fully a critical reasoner,one must be able to-and sometimes actually-identify, distinguish,evaluatepropositions asserted,denied,hypothesized as
5 I think Kantneglected distinctionsbetween reasoning,critical reasoning,and reflective reasoning.But he clearly saw that it is the possibility of applicationsof 'I think'to our thoughts -not ourbeing self-awarein this way all the time-that is basic to full reflective rationality.Of course, the form of 'I think' does not by itself make the relevantcontributionto reflective rationality. One could dreamcogito-thoughts.It is the abilityto be conceptuallyawareof oneself as thinkingwith a certaincontrolandagencythatis crucial.

100

I-TYLER BURGE

or merely considered.6Such abilities and activities are central to argumentation. Similarly,in critical practicalreasoning,one must be able to-and sometimes actually-evaluate propositions conceptualized as expressing pro-attitudes,to distinguish them explicitly from those thatexpressbeliefs, andto evaluaterelations of reason among such propositions as so conceptualized. Such evaluation constitutes minimal evaluations of propositional attitudes. one To be a criticalreasoner, mustalso be able to, andsometimes actually,use one's knowledgeof reasonsto make,criticize,change, confirmcommitments regarding propositions-to engageexplicitly in reason-induced changesof mind.Criticalreasoninghereinvolves an ability to distinguish subjectivities from more objectively commitments to explicitlyaltertheformerin favour and supportable of the latter.Its pointis reasonablyto confirmand correctattitudes and reasoning (not merely assess propositionalconnections), by referenceto rationalstandards. Critical reasoning must be exercised on itself. Any critical reasoning,even aboutabstractpropositional relationsor aboutthe reasoningof others, involves commitmentsby the reasoner.And genuinely critical reasoning requires an application of rational to standards thosecommitments. Abeing thatassessedgood andbad reasoningin othersor in the abstract, hadno inclinationto apply but such standards the commitmentsinvolved in those very assessto ments, would not be a critical reasoner.To reason critically-to consider reasons bearingon the truthof some matter,to suspend belief or desire, to weigh values undera conception of the good -one must treat one's own commitments as matters to be considered and evaluated. Critical evaluation of one's own commitmentsis centralto formingthemandto rationallychanging one's mind or standingfast. So criticalreasoningrequiresthinkingaboutone's thoughts.But it further requiresthatthatthinkingbe normallyknowledgeable.To appreciateone's reasons as reasons-to check, weigh, criticize,
6 In effect, Frege's use of the assertionsign is an acknowledgementof a minimaluse of these abilities. Withoutan abilityto recognizethata propositionshouldbe andis judged to be true, one cannot reason critically.Having a concept of judgment and using it in reasoningmeets my requirement.

TO OUR ENTITLEMENT SELF-KNOWLEDGE

101

confirm one's reasons-one must know what one's reasons, thoughts,and reasoningare. One need not always be knowledgeable, or even right. But being knowledgeablemust be the normal situationwhenone reflectson one's reasonsin thecourseof carrying outreasonableinquiryordeliberation. interesthereis less in the The requirementof normal knowledgeability-which is shared with othercognitive activities,such as perception. The interestlies in the groundof the requirement. Why must we be normallyknowledgeable aboutour thoughtswhen we reflectuponthem? I will answerthis questionin three stages. First,I want to show that to evaluate reasons critically, one must have an epistemic entitlementto one's judgmentsaboutone's thoughts,reasons,and reasoning.Second,I wantto supportthe strongerthesis thatcritical reasoning requires that one know one's thoughts, reasons, and reasoning.Third,I will try to show thatthis knowledge must take a distinctive,non-observational form. So I beginwith thematter entitlement. basicideais simple. of The Put crudely:since one's beliefs orjudgmentsaboutone's thoughts, reasons,andreasoningarean integralpartof the overallprocedures of critical reasoning, one must have an epistemic right to those beliefs orjudgments.To be reasonablein the whole enterprise, one must be reasonablein thatessential aspect of it. Less crudely,considerthe process of reasoningwhich involves the confirming and weighing of one's reasons. One must make judgmentsaboutone's attitudesandinferences.If one's judgments aboutone's attitudes inferenceswere notreasonable-if one had or no epistemic entitlement to them-one's reflection on one's attitudesand their interrelations could add no rationalelement to the reasonabilityof the whole process. But reflection does add a rationalelementto the reasonability reasoning.It gives one some of rationalcontrolover one's reasoning. To put the point somewhatmore fully: if one lackedentitlement to judgments about one's attitudes,there could be no norms of reason governinghow one ought check, weigh, overturn,confirm reasons or reasoning.For if one lacked entitlementto judgments about one's attitudes,one could not be subject to rationalnorms governinghow one oughtto alterthose attitudesgiven thatone had reflected on them. If reflection provided no reason-endorsed judgmentsaboutthe attitudes,the rationalconnectionbetweenthe

102

I-TYLER BURGE

attitudesreflected upon and the reflection would be broken. So reasons could not apply to how the attitudesshould be changed, suspended,or confirmedon the basis of reasoningdependingon such reflection. But critical reasoningjust is reasoning in which normsof reasonapplyto how attitudesshouldbe affectedpartlyon the basis of reasoning that derives from judgments about one's attitudes. So one must have an epistemic entitlement to one's judgmentsaboutone's attitudes. I turnnow to the strongerthesis. One might imagine some gap between epistemicentitlementand knowledge.Might one have an epistemic entitlementbut be systematicallymistaken?Or might failure of some third Gettier-typecondition (beyond truth and epistemic entitlement)undermineknowledge? It is possible in given cases for reflection to be disconnectedin these ways from the attitudespurportedly reflectedupon. But both possibilitiesif generalizedareincompatible withourhavingthe sort of entitlementto the reflectionjust argued for. That entitlement restedon the assumptionthatreflectionaddeda rationalelementto the reasonabilityof the whole process of critical reasoning-a process wherebyobject-level attitudesare guided by reflectionon theirreasonability.If reflectivejudgmentswere not normallytrue, reflectioncould not add to the rationalcoherenceor add a rational componentto the reasonabilityof the whole process. It could not rationallycontrolandguidethe attitudes beingreflectedupon(even though one could imagine situationsin which such disconnected reflection would be mechanicallyor instrumentally beneficial in formingtrueor rationalbeliefs). So reflectionwould not addin the relevant way to the reasonabilityof the process, and therefore would not have the source of entitlement just arguedfor. The samepointappliesto thepossiblefailureof some Gettier-type condition. Again, if reflection were connected to the truthof our judgments about our thoughts in an accidental or nonknowledge-yielding way, the reason-guiding and rationalcoherence-makingfunctions of rationalreview would be broken. Since part of our entitlement to reflective judgments about our attitudes derives from their functions in critical reasoning, the entitlementitself would be undermined. Not only the relevant entitlementto reflective judgments that derives from their functions within critical reasoning,but critical

OUR ENTITLEMENT SELF-KNOWLEDGE TO

103

reasoningitself is constitutivelydependenton the truth-andGettier conditionsbeing met. If a being had an epistemicentitlementto its judgments about its attitudes but were systematically mistaken about them-never got them right-it would not be a critical reasoner.Or if our entitlementwere always connectedto the truth of our judgments about our thoughts in an accidental or nonknowledge-yieldingway, criticalreasoningwould not be possible. Forcriticalreasonrequires rational of integration one's higher-order evaluationswith one's first-order, object-oriented reasoning.The formermust be reason-guidedand reason-guiding. And they must cement the rationalcoherence between the two levels. If the two came radicallyapart,or were only accidentallyconnected,critical reasoningwould not occur. So if we failed normallyto know our thoughtsand attitudes,in ordinary reasoningaboutreasons,eitherthroughsystematicfalsity of our judgments or through systematic mismatch between our entitlementandtruth,criticalreasoningwould not occuramongus. Indeed, the entitlementto reflective judgments that derives from thosejudgments' placein criticalreasoningwouldlapse.Butcritical reasoningdoes occur among us; and we are entitled to reflective judgments by virtue of their contributionto the reasonabilityof critical reasoning. So as critical reasoners we must know our thoughtsand attitudes. Symptomatic of the connection I have noted between the rationality of reflection in critical reasoning and the truth of reflectivejudgmentsis thefactthereareseverelimitson bruteerrors injudgmentsaboutone's presentordinary, accessiblepropositional attitudes.A bruteerroris an errorthatindicatesno rationalfailure and no malfunctionin the mistakenindividual.7Brute perceptual errors commonly result from misleading natural conditions or look-alikesubstitutes. One canbe perceptually wrongwithoutthere
7 1 introducedthe notion of bruteerrorin 'Individualismand Self-Knowledge' op. cit., p. 657. I intendrationalfailuresto includeany failureof entitlementorjustification,not just ones that are epistemically culpable. I intend malfunctions to cover not only mechanicalor biological failures in, say, the individual'sperceptualapparatus, also but failuresof normalunderstanding-as for example when an individualbelieves arthritis can occur in the thigh. The idea is that a bruteerrorwould have occurredeven if the individual's epistemic warrantswere in order and the individual's perception and ordinaryunderstanding were functioningoptimally.

104

I-TYLER BURGE

being anythingwrong with one. Such brute perceptualerrorsare unremarkable. errorsaboutwhat one's thoughtsand attitudes But are normally seem to involve some malfunction or rational deficiency. There are exceptions-the cases of unconscious, modularattitudesthat are not accessible to reflection. One could easily make brute errorsabout these. There arejudgments about one's emotions, character, deep motives, that seem hardto get or right. I leave open whetherthese might sometimes involve brute errors.But it seems thatwe makemistakesaboutmanyattitudes that are accessible to reflectionprimarilywhen we are subjectto some failureof rationality defect in ourcognitivepowers.8 or I statedthatI would arguethatthe specific role of knowledgeof our thoughtsin criticalreasoningrequiresthatit be associatedwith a distinctive sort of epistemic entitlement that necessitates a non-contingent,rationalrelationbetween the relevantfirst-person present-tense judgmentsandtheirsubjectmatter truth. or Whyneed self-knowledgebe in any way special?Why is it not enoughthatit be prettyreliableobservation? Some knowledgeof ourown mental states and events is empiricalin the sense thatit is based eitheron imaging,remembering, reasoningaboutsensed inner-goings-on, or or on observingourown behaviour hearingaboutit fromothers. and Simplicitytemptssome to hold thatall self-knowledgeis like that. Let me elaboratethistemptation. is commonlyheld thatbeliefs It aboutothers'attitudesmustbe based on inferencesfromor criteria for observation.9On the model at issue, beliefs about one's own attitudesdiffer only in thatone need not always inferthose beliefs,
8 Ourepistemicentitlementto judgmentsaboutourpresentattitudesis a generalrightand is compatiblewith ourmakingvariousmistakesaboutour attitudeseven in the courseof critical reasoning.(Of course, then we are, in a sense, not critically reasoningwith the attitudeswe are mistakenabout.)We make mistakesof haste, bias, and self-deception. Some attitudesare hardto get at, except with discipline,and even maturation therapy. or In some cases, otherpeople arebetterat knowingourattitudesthanwe are.So one might demandfurtherspecificationof our entitlement.One mightask underwhatconditionsit is overturned insufficientto give us knowledge.And one mightinquirein moredepth or into the conditionsunderwhich errorsarise.These issues arecomplex. I thinkthatwhen ourjudgmentsabout a certainclass of our thoughtsand attitudesare in a certainsense immediate (which entails that they are neitherinferrednor otherwisebiased by other attitudes),and when our minds are not subjectto malfunction,we do not make errors. But thereis no recipefor insuringthatourjudgmentsareimmediateor thatthey areabout the relevantclass. There is no internalrecipe for avoidingerror. 9 I do not accept this view, but I need not questionit here. Cf. 'ContentPreservation' op. cit. Certainlyone's beliefs aboutothers'thoughtsare often based this way.

OUR ENTITLEMENT SELF-KNOWLEDGE TO

105

because one is the closest witness. There is no authority in self-knowledge, other than the authorityof inner observational presence,practice,andfamiliarity. This simple observationalmodel does not accountplausiblyfor cogito-like thoughts. Such thoughts are logically special in their self-verificationandepistemicallyspecialin theircleardependence for entitlementon intellectionand understanding, on any sort not of observation.But cogito thoughtsdo not constitutethe full range of thoughtsthatenteressentially into criticalreasoning. The simpleobservational modelis encumbered withtheobscurity of the notion of inner observation as applied to thoughts and attitudes.Unlike sensationsor images, thoughtsand attitudeslack distinctivepresentations phenomenologies.The model is phenoor menologically implausiblefor many immediatejudgments about one's own beliefs or currentthoughts.But I wantto show thatthere is a deeperproblemif the model is takento cover all cases.10 Before presenting the argument,I will say what I take to be fundamental the simpleobservational to model.The modelneed not claim any phenomenological presentation in self-knowledge, thoughwaiving such a claim weakens the analogyto observation. The fundamental claim is that one's epistemic warrant for self-knowledgealways rests partlyon the existence of a patternof veridical,but brute,contingent,non-rational relations-which are plausiblyalways causalrelations-between the subjectmatter(the attitudesunderreview) andthejudgmentsaboutthe attitudes.This claim is compatiblewith holding that from the point of view of epistemology, observationaljudgments are often immediate and non-inferential, requiringno background causal hypothesison the of the individualabouttheirsource. part
10 Humeis, I think,a proponent the simpleobservational of model.A morerecentproponent is D. M. Armstrong, MaterialistTheoryof theMind(London,Routledge& KeganPaul, A 1968), pp. 323-338. The rationalist tradition, in its emphasis on the role of selfknowledge in rationality,and the role of understanding (not sensory observation)in self-knowledge,is the source of my view. Kantdevelops this traditionin a particularly deep way, although his epistemology left him with what was, in my opinion, an implausibly restrictive account of cognition of one's own thoughts, one indeed that overrates the role of inner sense. A more recent non-observational account that emphasizesthe role of self-knowledgein reasoningmay be foundin Sidney Shoemaker, 'On KnowingOne's Own Mind' PhilosophicalPerspectives2, (1988), pp. 183-209.

106

I-TYLER BURGE

My view aboutperceptualentitlementis more specific thanthe model.I believe that fundamental claim of the simpleobservational our entitlementrests partlyon our being perceivers,which entails that we-or our species-perceptual systems-are or have been in but brute,contingent,non-rational veridicalrelationsto objects of perception,andthe kindsthatourperceptual 11 judgmentsspecify. It is necessarily constitutiveof the content of our observationalor perceptualbeliefs aboutphysical objects, andof the very natureof our perceptual systems, that we be veridically attuned to the environmentthroughcausal relationsto it-either in our learning historiesor indirectlyin the evolutionof our perceptualsystems.12 Entitlement observational to physical objectbeliefs rests partlyon
11 A temptingoversimplification to claim thatthese constitutiveveridicalcausalrelations is are always reliably veridical. Such a claim is temptingbecause in so many cases our perceptionsare reliable. Perhapsmany types of perceptionmust be. But the claim is oversimplifiedbecausesome perceptual intentionaltypes in some perceptual systemsare establishedthroughthe systems'reliableavoidanceof false negativesrather thanthrough theirreliableachievementof truepositives.Itis morecriticalto a hare'sperceptual system that it not fail to register a predatorwhen one is there than that it be reliable in its registrationof predators.So the system could commonly indicate the presence of predators falsely-and be broadlyunreliablein its perceptions-as long as it was reliable in correctlyindicatingpresentpredators. remains,however,constitutiveof the systems' It perceiving predatorsas predatorsthat some veridical perceptionsplayed a role in the evolution-fashioned function or in the actualuse of the system. Of course, reliabilityis more important perceptionsof safety thanfor perceptionsof danger. for These qualificationson reliability,of course, complicate any account of the relation between perceptual-content constitution and perceptualentitlement. For presumably epistemic entitlementsareprimafacie comprisedby constitutivelyunreliableperceptual deliverances.I believe thata perceptualsystem in any agent, however,is constitutively associatedwith reliableperceptionsin a rangeof cases. But these are issues for another occasion. 12 I am inclined to think that it is a conceptualnecessity that there be causal relationsin perception. But for purposes of my argument, the fundamentalfeature is that the entitlementto observational beliefs necessarilyrestson some pattern brute,contingent, of non-rational relations between observed and observer, regardless of whether the contingentre!ationsare causal. It is common to my view and the opposed observational view of self-knowledge that in many of the cases under dispute, there is a causal mechanismthatrelatesattitudesto judgmentsaboutthem.Whatis in disputeis the nature of the epistemic entitlement that one has to such judgments, not the existence of a psychological mechanism. On the simple observational model, our entitlement to self-knowledgealways restspartlyon the brute,contingent,non-rational causalrelations. On my view, in some important cases, it does not: Christopher Peacocke has pointedout to me thatin some, thoughI thinknot all, cases of special self-knowledge,the entitlement may specify some causalrelationbetweensubjectmatter judgment.But notall causal and relations are brute, contingent, non-rationalones. (For example those involved in a person'sdeductiveinferencearenot.) Wherea causalrelationis not merelya background enablingcondition,but an element in the relevantentitlementto self-knowledge, it will on my view never be a brute,contingent,non-rationalone. It will be associated in the entitlementwith normsfor transferof reasons.

OUR ENTITLEMENT SELF-KNOWLEDGE TO

107

the this necessity.But in particular instancesof perception, relations on which one's entitlementto perceptual judgmentsrests arebrute, It contingent,non-rational.'3 is this claim thatforms the paradigm for the simple observational model. The existence of veridicalbeliefs involving de re causallybased relations to the environmentis necessary to and constitutive of something'sbeing a perceptual system.But the individualrelations are brute,contingent,non-rational. brutecontingencyof these The relationsin individualinstancesof veridicalperceptual judgmentis fundamental observation. to Differentconditionscouldhave caused a perceptual judgmentthatwas internally-indistinguishable (indeed I thinkthe samejudgment-type)but non-veridical,withoutloss of entitlement,even as the system functionedoptimallywell.'4 Thus
13 The simple observational model is inspired by a comparison of self-knowledge to observational judgmentsaboutphysical objects. I operatewith a commonsensicalconception of such judgments.There are non-common-sensicalconceptions that take the basic relationthatunderliesourepistemicrightto be one betweenthe observerandsome I sort of mentalitem, a sense datumor an appearance. ignore such theoriesnot because I regardthemas mistaken(althoughI do regardthem as mistaken).I ignorethembecause they model observationof physical objects on knowledge of one's mental events. The model I am attacking proposes to illumine self-knowledge through an independent model. I do thinkthatknowledge of our pains and othersensations-as contrastedwith knowledge of our propositionalstates and events-is empirical in the sense that it depends for its entitlementon sensory experience or sensory beliefs. Judgmentsthat constitutesuch knowledgejust are sensory beliefs. AlthoughI believe thatbruteerroris possible in certainjudgmentsof this sort, such cases are marginal.Understanding even these empiricaljudgments will, I think, owe more to the kinds of considerationsI am elaboratingthan to reflection on ordinaryperceptionsof physical objects. But I regard from knowledge of one's knowledge of one's sensationsas requiringseparatetreatment thoughtsand attitudes. It is worthnoting thata view thatwe must 'inferentially' basejudgmentsaboutphysical objects on observationsof sense datawould also normallybe committedto holding that one's entitlementto thosejudgmentsrestson brute,contingent,non-rational relationsto the physical objects that always allow for brute error. The same point applies to inference-to-the-best-explanation views of our warrantsfor perceptualbeliefs about
physical objects.

14 This gloss on the brutecontingencyof the relations,apartfromthe parenthetical remark, is less committal than my own view of the contingency involved in observational relations. I think the same perceptualobject could, with different external auxiliary conditions,have caused a differentnon-veridical judgment.And I thinkthata different perceptualobject, or perhapsnone at all, could have combined with differentexternal auxiliary conditions to cause a perceptionor perceptualjudgment of the same type, though perhaps one with a different token demonstrativeelement, making it nonveridical. (I do not depend on these views in my argumenthere.) These different conditions, in individual cases, need not affect the individual's entitlement to the perceptualjudgment; nor need they affect the well-functioning of the individual's perceptual-cognitiveapparatus.The sense in which the relevant relations are nonrational is complex. Perhapsit can suffice here to note thatsince in the case of ordinary perceptionthe perceptualobjects arephysical kinds or physical individuals,therecan in

108

I-TYLER

BURGE

The objector conditions perceptionis always subjectto bruteerror. of perceptioncould lead us into misperception withouttherebeing any failureof entitlementand withouttherebeing any malfunction of ourcognitive or perceptualsystems. The objectivity of perception depends on the possibility of epistemically entitled misperception.Perceptualjustificationand criticism necessarilypresupposea distinctionbetween a person's cognitive perspective and the objective, physical subject matter. this They further presuppose unremarkable possibilityof contingent mismatches in individual cases that in no way impugn the funcindividual's epistemic entitlementsor perceptual-cognitive tioning.Rationalandepistemicevaluationfixes on the individual's perceptual judgmentsandperspective,not on theirphysicalsubject matter.For this is only brute contingently related, in individual cases, to epistemicentitlement. A consequenceof interpreting self-knowledgeon the simple all observationalmodel is thatin any given case bruteerrors-errors that do not reflect on the rationalityor sound functioningof the reviewing judgment-are possible. I intimatedearlier that brute errorsdo not seem to threatensome instancesof judgmentsabout attitudes.I proposeto show why this mustbe so. Not all one's knowledge of one's propositionalattitudescan fit the simple observationalmodel. For general applicationof the model is incompatiblewiththe functionof knowledgeof one's own in attitudes criticalreasoning.The mainidea is thatsuchapplication would entail a dissociation between cognitive review and the thoughtsreviewed that is incompatiblewith norms of epistemic reasonability that are basic to all critical inquiry, including empirical,mathematical, philosophical,andpracticalinquiry. Rationalevaluationof attitudescommonlyappliesto and within a perspectiveor pointof view. The argument makereferenceto will this fact. Differentpeople have differentpoints of view. My judgment that your beliefs are irrationalmay be reasonablefrom my point of view. But it does not follow thatthereis reasonfrom your perspective to change your beliefs. I may have made some brute
that case be no questionof a rationalrelationbetweenthem-which have no intentional contentat all-and perceptionsor perceptual judgments.

TO OUR ENTITLEMENT SELF-KNOWLEDGE

109

errorabout what your beliefs are, or your perspective may have information from mine. differentassociatedreasonsor background There can be differentperspectivesor points of view within a given person.What is reasonablefor a personat a given time may be different from what is reasonablefrom the perspectiveof the person's memory back on that time. What is reasonable on reflectionmay differ from what is reasonablein modularcognitive processes, or in an instantpracticalreasoning,or in subconscious reasoning. My argumenthinges on how reasons transferacross points of view. Supposethat all one's knowledge of one's propositionalmental events and states fit the simple observationalmodel. Then one's entitlementto instances of such knowledge would always rest on purely contingent relations between any given judgment about one's mentalstatesandthe subjectmatterof thejudgment.Whatis more, bruteerrorwould be possible in any given case. Normative evaluationsof reasonabilityand epistemic entitlementin critical reasoning-in checking and evaluating one's reasoning-would applywithinthe perspectiveof thejudgments,butnot immediately withinthe perspectiveof the subjectmatter thejudgments,except of insofaras it contingentlyconformedto thosejudgments,andexcept insofar as it happenedto be embeddedin a perspectiverelevantly similar to the perspectivefrom which the judgmentswere made. For the subjectmattermight, in any given case, fail to conformto the judgmentsthroughno failure of justificationor entitlementin the judge, and throughno malfunctionof the relevantfaculties. But this pictureis nonsenseif it is appliedto alljudgmentsabout one's own propositionalattitudes.For it is constitutiveof critical reasoning that if the reasons or assumptionsbeing reviewed are justifiably found wanting by the reviewer, it rationallyfollows immediatelythat there is primafacie reason for changing or supplementingthem,wherethis reasonapplieswithinthepointof view of the reviewedmaterial just withinthereviewingperspective). (not If the relationbetweenthe reviewingpointof view andthe reasons or assumptionsbeing reviewed always fit the simple observational model, there would never be an immediaterationallynecessary connectionbetweenjustifiedrationalevaluationwithinthe review, on one hand, and its being primafacie reasonablewithin the reviewed perspectiveto shapeattitudes accordwiththatevaluation, in

110

I-TYLER BURGE

on the other.For the relationbetween the perspectiveof the review and that of the reviewed attitudeswould always be purely contingent,even undercanonicaldescriptionsof them,for purposesof reviewedwouldbe to thereviews rationalevaluation.(The attitudes to ourobservational as physicalobjectsare judgments.They would be purely'objects'of one's inquiry, partof the perspectiveof the not It for inquiry.) wouldbe reasonable thepersonfromthepointof view of the review thata change in the reviewed materialbe made. But thisreasonwouldnotnecessarilytransfer withinthepointof view to of the attitudesunderreview,even thoughthatis a pointof view of the sameperson.Itstransferring woulddependon brute,contingent, non-rational relationsbetweenthe two pointsof view. In critical reasoning, however, the connection is rationally immediateand necessary.Justifiablyfinding one's reasonsinvalid or one's thoughtsunjustified,is normallyin itself a paradigmatic reason,from the point of view of the thoughtsbeing reviewed (as well as fromthe perspectiveof the review), to alterthem. If in the course of criticalreasoningI reasonablyconclude that my belief thata given personis guiltyrestsentirelyon unreasonable premises or bad reasoning,then it normallyfollows immediately bothfor the perspectiveof the review andfor the perspectiveof the reviewedbelief thatit is reasonableto give up my belief aboutguilt or look for new groundsfor it. In such second-order reasoning,I am not normallyreasonablein alteringmy first-order views aboutguilt orinnocenceonly withtheprovisothattheyareembeddedin reasons that contingently match those associated with my reviewing perspective.I do not normallyhave the sort of excusing condition thatallows for rationalerrorthathinges on the contingentrelation thatthe subjectmatterbears to my judgmentsaboutit. Rathermy checking my belief and finding it wantingnormallyitself provides immediate prima facie reason to change it from within the perspectiveof thereview.This is becausethefirst-andsecond-order perspectivesarethe same pointof view. The reviewing of reasons that is integral to critical reasoning includes the review and the reviewed attitudesin a single point of view. The simple observationalmodel treats the review and the system being reviewed as dissociatedin a way incompatiblewith the norms of critical reasoning.It makes the reviewed system an object of investigation,but not partof the investigation'spoint of

OUR ENTITLEMENT SELF-KNOWLEDGE TO

111

view. So the model fails to account for the norms of critical

reasoning.

A closely relatedpoint centres on epistemic responsibility.We are epistemically responsible only because we are capable of reviewingourreasonsandreasoning.And we areparadigmatically responsiblefor our reasonswhen we check andreview them in the course of critical reasoning. But the simple observationalmodel implies that in carrying out reviews of one's reasoning, one is epistemically responsible not primarily for the thoughts being reviewed but primarilyfor the review. The model implies thatwe are in reviewing our reasons only derivatively responsible for objects of review, as one might be responsiblefor the actions of one's child or dog-but fully andprimarily responsibleonly where one's knowledge and control contingently matched what one is justified in believing aboutthem. But one is not epistemically responsible for the thoughts one reflects upon in criticalreasoningin the way one is responsiblefor somethingone owns or parents.One's responsibilityin reflecting on one's thoughtsis immediatelyfor the whole point of view. The model fails to accountfor the fact thatcritical simple observational reasoningis carriedout withina single multi-levelpoint of view. Of course, we are sometimes disunified.Sometimesto our own good and efficiency, sometimesto our misfortune,we fail to know ourmotives or reasons,or know themonly through observationand empirical reasoning. Sometimes from the point of view of our self-conscious reviewing selves, we are indeed epistemically responsible only derivatively for attitudes that we know only empirically.But in these cases, we arenot reasoningcriticallywith those aspectsof ourselvesthatwe know only in these ways. Theoreticalknowledge of one's modularattitudesis one sort of purely observationally-based'self-knowledge'. When attitudes cannotbe knownin a directnon-observational way, one commonly enterssomequalification thesense in whichtheattitudes one's on are own. Inthesecases failureto knowtheattitudes non-observationally is no sign of dissociation.But whenone knowsonly observationally unconscious attitudes which are in principle accessible to non-observationalself-knowledge, there is some dissociation of self, constitutedby a divide between the point of view of one's criticalreasoningand the attitudeknown only observationally.

112

I-TYLER BURGE

Knowledge throughtherapyof one's unconscious, before full integrationof that knowledge, provides one sort of example. One may know the attitudes on the basis of observationallybased therapy,butthe unconsciousattitudesmay providea point of view of their own into which the meta-evaluationsof them may not transfer. Those evaluations may not speak to unconscious considerations areintegralto the unconsciouspathology;or the that unconsciouspoint of view may not have 'takenin' mattersthatare integralto the rationality the meta-,therapeutic of pointof view. Psychoanalytic cases are not the only sort that illustrate the relevant dissociation. One may know from experience or theory thatone will act a certainway, andyet rationalmeta-considerations to that may not penetrate the systemof attitudes motivatethe action. Some self-admittedcompulsions provide examples. One knows one has or will have the relevantintention,but knows the intention only as object; it is then not the productof critical deliberation. Sometimes rational considerationsfrom the meta-pointof view may not have the samerationalforce andrelevancewithinthe point of view thatincludes the observationally known attitudes. There are cases of knowledge of one's beliefs like this as well. One may know from self-observation that 'underneath' one believes somethingbecause one needs to believe it, while feeling sincere rational urges to assert the contrary. The system of underlyingpracticalbeliefs that motivate the needed belief may form a point of view that does not recognize as sufficient the rationalmeta-reasonsthat one can offer oneself for giving up the need-based belief. The person's meta-perspectivemay correctly condemnthe need-basedbelief as epistemicallyirrational. the But belief may be dissociated from the point of view of his observational knowledge of thatbelief. It may be rational'overall' for the person to give up the belief, but the practicalrationalityof the limited, need-drivenperspectivemay exclude or outweigh the considerationsthatcount againstthe belief. The relevant psychological dissociation is, I think, sometimes partlyto be explained in terms of the fact that a second 'point of view', or system of attitudeswith its own internalcoherence,has gotten set up withinthe person,in such a way thatreasonsfromthe pointof view of theperson'scriticalrationality notautomatically do transferto withinthe second point of view, ratheras reasonsfrom

OUR ENTITLEMENT SELF-KNOWLEDGE TO

113

my perspectivedo not always apply as reasonsfrom yours. Being known from the perspective of the critical reasoner only as an object, on the basis of observationand theory, is sufficient for an attitudeto be dissociatedin this way. Attitudesthatarepartof such a dissociated point of view may provide us with reasons, even operativeones, for doing things.But insofaras we know themonly observationally, they are not partof our criticalreasoning. Where we know our thoughtsor attitudesonly by observation, thequestionof meansof control-ofeffective application reasons of to them-arises, at least from the perspectiveof our observational knowledge.Wherewe know ourattitudes only as empiricalobjects, not only areourrationalevaluationsof those attitudes relativizedto contingencies associated with the knowledge. But our ability to apply our reasons (those associated with the point of view of the observationalknowledge) must acknowledge the contingency of ourrationalcontrolover those attitudes. mustface a questionof We how, by what means,to makethose reasonseffective in view of the contingentrelationbetweenthe pointof view of the self-knowledge andrational known evaluation,on one hand,andthe observationally on theother.Again,thisis not criticalreasoning.In critical attitudes, reasoning,such questionsof means and controldo not arise, since one's relationto the known attitudesis rationallyimmediate:they arepartof the perspectiveof the review itself. Sometimes observation-basedself-knowledge enables one to assimilatean attitudeinto one's criticalpoint of view, and to take directcriticalcontrolandresponsibilityfor the attitude. may learn I throughobserving my behaviouror throughreasoningin therapy thatI believe thata friendis untrustworthy. may 'internalize'this I belief so that it is no longer merely an object of observational knowledge.This processis sometimesimmediate,sometimeshard, requiringdeep personalchange. Althoughmuchreasoningandrationalattitudeformationoccurs outside the purviewof criticalself-knowledge,or indeed any selfknowledge, critical reasoningremainscentral to our identities as persons.So no reasonableaccountof self-knowledgecan ignorethe role andentitlementscriticalreasoninggives to self-knowledge. The argumentI have given against the simple observational model indicatesthatthe relationsbetween knowledge and subject matteron which one's entitlementrests cannotalways be causally

114

I-TYLER BURGE

brute,contingent,non-rationalones. In some particular instances of self-knowledge, the connection must be a rational one. For conclusions about the reasonabilityof one's thoughts based on self-review directlyyield reasonswithinthe point of view of those reviewed thoughtsto alter or confirmthem. The relationbetween self-knowledge and subjectmatteris that they must normallyand rationallybe part of the same theoretical and practical point of view-elements of a single theoryor plan. Connectionsbetweenreviews andthoughtsunderreview thatare fully open to reason and that allow immediate transmissionof reasonsare necessaryto the rationalcoherenceof a point of view. A merelyobservation-based relationbetween attitudesinsuresthat they are partsof differentpoints of view. Indeed,it is constitutive of a pointof view thatfailureto follow orunderstand connections its by the holderof the point of view is a failureof rationality. Thus a point of view is not closed under deductive consequence, and containsrationalconnectionsotherthandeductiveones. So entitlementto knowledgeof one's own thoughtsandattitudes is not purely a matterof what one does. It has to do with who one is. One's statusas a person and criticalreasonerentails epistemic entitlementto some judgmentsaboutone's propositional attitudes. It entails some non-observational knowledge of them. Cogito-like thoughts illustrate one non-contingent rational relationbetween knowledge and subjectmatter. those cases, the In reviewed thought is simply a logical part of the review. But cogito-like thoughtsare in many ways special cases. If we are to understand criticalreasoning,the entitlementthatI have discussed must apply more broadly.It must includejudgmentsaboutbeliefs, intentions,wants,as well as occurrent thoughts. So far, I have put little weight on the first-personpresenttense form of the relevant pieces of self-knowledge. Clearly, for the review andthe reviewedthoughtsto be partof a practiceof critical reasoning,the reviewedthoughtsmustbe capableof becomingpart of the reasoner'spresentarrayof attitudes.And the special features of cogito cases do dependon presenttense. But muchof whatI have said about the dependenceof an entitlementon its role in critical reasoning, and about the non-observational character of this entitlement,appliesto preservative memory-that type of memory that preserves propositions and our commitments to them in

OUR ENTITLEMENT SELF-KNOWLEDGE TO

115

I reasoning.15 thinkthatthis sortof memoryprovidesus with some non-observational knowledgeof our past mentalstates andevents, and is epistemicallyunderwritten its role in criticalreasoning. by The first-person point of view is clearlybasic to self-knowledge differsfrom in criticalreasoning.The self-knowledgethatI featured observational knowledgeof physicalobjectsin thatthe first-person point of view is deeply relevant to the epistemic status of the knowledge.In observationsof physicalobjects,anyonecould have the withequalright,if the same madesubstantially sameobservation angle of perception had been available at the same time. But self-ascriptionsconstitutean epistemicangle in themselves. come to? If thereviewingknowledgeis Whatdoes this metaphor to be integralto critical reasoning, if it is to provide immediate the rational groundforchangein thereviewedmaterial, reviewmust takeup the sameperspectiveor pointof view as the actunderreview -the reasoner'sown object-level point of view. The first-person pointof view bearsa distinctiverelationto the relevanceof rational norms to rationalactivity.For a review of a propositionalmental eventor stateto yield animmediate rational groundto defendoralter the attitude,the point of view of the review andthatof the attitude reviewed mustbe the same and mustbe first-personal. In evaluatingreasoningcritically,one must make commitments to attitudespartlyon the basis of criticalevaluationsof them.If one is to fully articulate rationalbasis for the application rational the of normswithin criticalreasoning,the commitmentsto both reviews and reviewed attitudes must be conceptually acknowledged as one's own. For acknowledgingthem as one's own is takingthem as attitudes one couldrationally directlychangeor confirm. that and Acknowledging first-orderattitudesas one's own is necessary to the articulating directrationalrelevanceof one's criticalreasonsto first-orderreasoning (or more generally, reasonable activity). I intendto say more aboutthis matteron anotheroccasion. I have sketchedthe environmental neutralityof our entitlement to self-knowledge.The entitlement remainsconstantunderpossible unnoticeable variations in environmental circumstances or cognitive content.For it does not dependon the empiricalcontent
15 Cf. 'ContentPreservation',op. cit.

116

I-TYLER BURGE

of the judgments. It does not depend on checking whether our judgments meet certainconditions.It depends on the judgments' being instances of a kind essential to critical reasoning. Critical reasoningpresupposesthatpeople are entitled to suchjudgments. Since we are criticalreasoners,we are so entitled. Epistemicentitlementderivesfromjurisdiction-from the place of thejudgmentsin reasoning.In cogito-like thoughts,this place is coded in the contentof thejudgmentsthemselves.In otherrelevant sorts of self-knowledge,which are fallible, the entitlement,indeed one's knowledge,dependsonly on one's notmisusingthejudgments 16 andon one's remaininga sane criticalreasoner.

16 Versionsof this paperwere given as the thirdof six Locke Lecturesat Oxfordin 1993, in and the first of two WhiteheadLecturesat Harvard 1994, as well as on several other occasions. The key idea and first draftof the paperdates from 1985. I am gratefulfor helpful comments on draftsor talks based on this paperto RobertAdams, Kent Bach, Phil Clark,David Kaplan,Christopher Peacocke,MarleenRozemond, Tony Brueckner, HilaryPutnam,NathanSalmon,HoustonSmit, BarryStroud,PatrickSuppes,andCorliss Swain.