,*APT] #ac 
,ON ,KN[L$GE1 ,]ROR1 & ,PROBABLE ,OP9ION 

#aiabaiab 

,PEOPLE 3 

,AU?OR 3 ,B]TR& ,RUSSELL 

,TEXT 3 

,*,A,P,T,] ,X,I,I,I 

,K,N,[,L,$,G,E1 ,],R,O,R1 ,& ,P,R,O,B,A,B,L,E ,O,P,9,I,O,N 

,! QUE/ION AS TO :AT WE MEAN BY TRU? & FALSEHOOD1 :I* 
WE CONSID]$ 9 ! PREC$+ *APT]1 IS ( MU* LESS 9T]E/ ?AN 
! QUE/ION AS TO H[ WE CAN KN[ :AT IS TRUE & :AT IS FALSE4 
,?IS QUE/ION WILL OCCUPY US 9 ! PRES5T *APT]4 ,!RE CAN BE NO 
D\BT ?AT SOME ( \R BELIEFS >E ]RONE\S2 ?US WE >E 
L$ TO 9QUIRE :AT C]TA9TY WE CAN EV] HAVE ?AT SU* & SU* A 
BELIEF IS NOT ]RONE\S4 ,9 O!R WORDS1 CAN WE EV] KN[ 
ANY?+ AT ALL1 OR DO WE M]ELY SOMETIMES BY GOOD LUCK BELIEVE :AT 
IS TRUE8 ,BE=E WE CAN ATTACK ?IS QUE/ION1 WE MU/1 H[EV]1 FIR/ 
DECIDE :AT WE MEAN BY 'KN[+'1 & ?IS QUE/ION IS NOT SO EASY AS 
MI<T BE SUPPOS$4 

,AT FIR/ SI<T WE MI<T IMAG9E ?AT KN[L$GE C\LD BE 
DEF9$ AS 'TRUE BELIEF'4 ,:5 :AT WE BELIEVE IS TRUE1 IT MI<T BE 
SUPPOS$ ?AT WE HAD A*IEV$ A KN[L$GE ( :AT WE BELIEVE4 ,BUT 
?IS W\LD NOT ACCORD ) ! WAY 9 :I* ! WORD IS COMMONLY 
US$4 ,TO TAKE A V]Y TRIVIAL 9/ANCE3 ,IF A MAN BELIEVES ?AT ! 
LATE ,PRIME ,M9I/]'S LA/ NAME BEGAN ) A ,B1 HE BELIEVES :AT IS 
TRUE1 S9CE ! LATE ,PRIME ,M9I/] WAS ,SIR ,H5RY ,CAMPBELL ,BANN]MAN4 
,BUT IF HE BELIEVES ?AT ,MR4 ,BALF\R WAS ! LATE ,PRIME ,M9I/]1 HE 
WILL /ILL BELIEVE ?AT ! LATE ,PRIME ,M9I/]'S LA/ NAME BEGAN 
) A ,B1 YET ?IS BELIEF1 ?\< TRUE1 W\LD NOT BE ?\<T TO 
CON/ITUTE KN[L$GE4 ,IF A NEWSPAP]1 BY AN 9TELLIG5T ANTICIPATION1  
ANN\NCES ! RESULT ( A BATTLE BE=E ANY TELEGRAM GIV+ ! 
RESULT HAS BE5 RECEIV$1 IT MAY BY GOOD =TUNE ANN\NCE :AT 
AFT]W>DS TURNS \T TO BE ! RI<T RESULT1 & IT MAY PRODUCE 
BELIEF 9 SOME ( ITS LESS EXP]I5C$ READ]S4 ,BUT 9 SPITE ( ! 
TRU? ( !IR BELIEF1 !Y CANNOT BE SAID TO HAVE KN[L$GE4 ,?US IT 
IS CLE> ?AT A TRUE BELIEF IS NOT KN[L$GE :5 IT IS D$UC$ FROM 
A FALSE BELIEF4 

,9 LIKE MANN]1 A TRUE BELIEF CANNOT BE CALL$ KN[L$GE :5 
IT IS D$UC$ BY A FALLACI\S PROCESS ( REASON+1 EV5 IF ! 
PREMISES FROM :I* IT IS D$UC$ >E TRUE4 ,IF ,I KN[ ?AT ALL 
,GREEKS >E M5 & ?AT ,SOCRATES WAS A MAN1 & ,I 9F] ?AT ,SOCRATES 
WAS A ,GREEK1 ,I CANNOT BE SAID TO KN[ ?AT ,SOCRATES WAS A 
,GREEK1 BECAUSE1 AL?\< MY PREMISES & MY CONCLUSION >E TRUE1 ! 
CONCLUSION DOES NOT FOLL[ FROM ! PREMISES4 

,BUT >E WE TO SAY ?AT NO?+ IS KN[L$GE EXCEPT :AT IS 
VALIDLY D$UC$ FROM TRUE PREMISES8 ,OBVI\SLY WE CANNOT SAY ?IS4 
,SU* A DEF9ITION IS AT ONCE TOO WIDE & TOO N>R[4 ,9 ! FIR/ 
PLACE1 IT IS TOO WIDE1 BECAUSE IT IS NOT 5\< ?AT \R PREMISES 
%\LD BE TRUE1 !Y MU/ ALSO BE KN[N4 ,! MAN 
:O BELIEVES ?AT ,MR4 ,BALF\R WAS ! LATE ,PRIME ,M9I/] MAY PROCE$ 
TO DRAW VALID D$UCTIONS FROM ! TRUE PREMISE ?AT ! LATE ,PRIME 
,M9I/]'S NAME BEGAN ) A ,B1 BUT HE CANNOT BE SAID TO 
KN[ ! CONCLUSIONS REA*$ BY !SE D$UCTIONS4 ,?US WE 
%ALL HAVE TO AM5D \R DEF9ITION BY SAY+ ?AT KN[L$GE IS :AT 
IS VALIDLY D$UC$ FROM KN[N PREMISES4 ,?IS1 H[EV]1 IS A 
CIRCUL> DEF9ITION3 IT ASSUMES ?AT WE ALREADY KN[ :AT IS MEANT BY 
'KN[N PREMISES'4 ,IT CAN1 !RE=E1 AT BE/ DEF9E ONE SORT ( 
KN[L$GE1 ! SORT WE CALL D]IVATIVE1 AS OPPOS$ TO 9TUITIVE 
KN[L$GE4 ,WE MAY SAY3 ',D]IVATIVE KN[L$GE IS :AT IS 
VALIDLY D$UC$ FROM PREMISES KN[N 9TUITIVELY'4 ,9 ?IS /ATEM5T 
!RE IS NO =MAL DEFECT1 BUT IT LEAVES ! DEF9ITION ( 
9TUITIVE KN[L$GE /ILL TO SEEK4 

,LEAV+ ON ONE SIDE1 = ! MOM5T1 ! QUE/ION ( 9TUITIVE 
KN[L$GE1 LET US CONSID] ! ABOVE SUGGE/$ DEF9ITION ( 
D]IVATIVE KN[L$GE4 ,! *IEF OBJECTION TO IT IS ?AT IT UNDULY 
LIMITS KN[L$GE4 ,IT CON/ANTLY HAPP5S ?AT PEOPLE 5T]TA9 A TRUE 
BELIEF1 :I* HAS GR[N UP 9 !M BECAUSE ( SOME PIECE ( 9TUITIVE 
KN[L$GE FROM :I* IT IS CAPABLE ( BE+ VALIDLY 9F]R$1 BUT 
FROM :I* IT HAS NOT1 AS A MATT] ( FACT1 BE5 9F]R$ BY ANY 
LOGICAL PROCESS4 

,TAKE1 = EXAMPLE1 ! BELIEFS PRODUC$ BY READ+4 ,IF ! 
NEWSPAP]S ANN\NCE ! DEA? ( ! ,K+1 WE >E FAIRLY WELL 
JU/IFI$ 9 BELIEV+ ?AT ! ,K+ IS DEAD1 S9CE ?IS IS ! SORT 
( ANN\NCEM5T :I* W\LD NOT BE MADE IF IT W]E FALSE4 ,& WE >E 
QUITE AMPLY JU/IFI$ 9 BELIEV+ ?AT ! NEWSPAP] ASS]TS ?AT 
! ,K+ IS DEAD4 ,BUT H]E ! 9TUITIVE KN[L$GE UPON :I* \R 
BELIEF IS BAS$ IS KN[L$GE ( ! EXI/5CE ( S5SE-DATA D]IV$ 
FROM LOOK+ AT ! PR9T :I* GIVES ! NEWS4 ,?IS KN[L$GE 
SC>CELY RISES 9TO CONSCI\SNESS1 EXCEPT 9 A P]SON :O CANNOT READ 
EASILY4 ,A *ILD MAY BE AW>E ( ! %APES ( ! LETT]S1 & PASS 
GRADUALLY & PA9FULLY TO A REALIZATION ( !IR MEAN+4 ,BUT 
ANYBODY ACCU/OM$ TO READ+ PASSES AT ONCE TO :AT ! LETT]S 
MEAN1 & IS NOT AW>E1 EXCEPT ON REFLECTION1 ?AT HE HAS D]IV$ 
?IS KN[L$GE FROM ! S5SE-DATA CALL$ SEE+ ! PR9T$ LETT]S4 
,?US AL?\< A VALID 9F]5CE FROM ! LETT]S TO !IR MEAN+ IS 
POSSIBLE1 & C\LD BE P]=M$ BY ! READ]1 IT S NOT 9 
FACT P]=M$1 S9CE HE DOES NOT 9 FACT P]=M ANY OP]ATION :I* 
CAN BE CALL$ LOGICAL 9F]5CE4 ,YET IT W\LD BE ABSURD TO SAY ?AT 
! READ] DOES NOT KN[ ?AT ! NEWSPAP] ANN\NCES ! 
,K+'S DEA?4 

,WE MU/1 !RE=E1 ADMIT AS D]IVATIVE KN[L$GE :ATEV] IS 
! RESULT ( 9TUITIVE KN[L$GE EV5 IF BY M]E ASSOCIATION1  
PROVID$ !RE IS A VALID LOGICAL CONNECTION1 & ! P]SON 
9 QUE/ION C\LD BECOME AW>E ( ?IS CONNECTION BY REFLECTION4 ,!RE 
>E 9 FACT MANY WAYS1 BESIDES LOGICAL 9F]5CE1 BY :I* WE PASS 
FROM ONE BELIEF TO ANO!R3 ! PASSAGE FROM ! PR9T TO ITS MEAN+ 
ILLU/RATES !SE WAYS4 ,!SE WAYS MAY BE CALL$ 'PSY*OLOGICAL 
9F]5CE'4 ,WE %ALL1 !N1 ADMIT SU* PSY*OLOGICAL 9F]5CE AS A 
MEANS ( OBTA9+ D]IVATIVE KN[L$GE1 PROVID$ !RE IS A 
DISCOV]ABLE LOGICAL 9F]5CE :I* RUNS P>ALLEL TO ! 
PSY*OLOGICAL 9F]5CE4 ,?IS R5D]S \R DEF9ITION ( D]IVATIVE 
KN[L$GE LESS PRECISE ?AN WE C\LD WI%1 S9CE ! WORD 
'DISCOV]ABLE' IS VAGUE3 IT DOES NOT TELL US H[ MU* REFLECTION MAY 
BE NE$$ 9 ORD] TO MAKE ! DISCOV]Y4 ,BUT 9 FACT 'KN[L$GE' IS 
NOT A PRECISE CONCEPTION3 IT M]GES 9TO 'PROBABLE OP9ION'1 AS WE 
%ALL SEE MORE FULLY 9 ! C\RSE ( ! PRES5T *APT]4 ,A V]Y 
PRECISE DEF9ITION1 !RE=E1 %\LD NOT BE S\<T1 S9CE ANY SU* 
DEF9ITION MU/ BE MORE OR LESS MISLEAD+4 

,! *IEF DIFFICULTY 9 REG>D TO KN[L$GE1 H[EV]1 DOES NOT 
>ISE OV] D]IVATIVE KN[L$GE1 BUT OV] 9TUITIVE KN[L$GE4 ,SO 
LONG AS WE >E DEAL+ ) D]IVATIVE KN[L$GE1 WE HAVE ! TE/ ( 
9TUITIVE KN[L$GE TO FALL BACK UPON4 ,BUT 9 REG>D TO 9TUITIVE 
BELIEFS1 IT IS BY NO MEANS EASY TO DISCOV] ANY CRIT]ION BY :I* TO 
DI/+UI% SOME AS TRUE & O!RS AS ]RONE\S4 ,9 ?IS QUE/ION IT 
IS SC>CELY POSSIBLE TO REA* ANY V]Y PRECISE RESULT3 ALL \R 
KN[L$GE ( TRU?S IS 9FECT$ ) SOME DEGREE ( D\BT1  
& A !ORY :I* IGNOR$ ?IS FACT W\LD BE PLA9LY WRONG4 
,SOME?+ MAY BE DONE1 H[EV]1 TO MITIGATE ! DIFFICULTIES ( ! 
QUE/ION4 

,\R !ORY ( TRU?1 TO BEG9 )1 SUPPLIES ! POSSIBILITY 
( DI/+UI%+ C]TA9 TRU?S AS SELF-EVID5T 9 A S5SE 
:I* 5SURES 9FALLIBILITY4 ,:5 A BELIEF IS TRUE1 WE SAID1 !RE IS 
A CORRESPOND+ FACT1 9 :I* ! SEV]AL OBJECTS ( ! BELIEF =M 
A S+LE COMPLEX4 ,! BELIEF IS SAID TO CON/ITUTE KN[L$GE 
( ?IS FACT1 PROVID$ IT FULFILLS ?OSE FUR!R SOME:AT VAGUE 
CONDITIONS :I* WE HAVE BE5 CONSID]+ 9 ! PRES5T *APT]4 ,BUT 
9 REG>D TO ANY FACT1 BESIDES ! KN[L$GE CON/ITUT$ BY BELIEF1  
WE MAY ALSO HAVE ! K9D ( KN[L$GE CON/ITUT$ BY 
P]CEPTION 7TAK+ ?IS WORD 9 ITS WIDE/ POSSIBLE S5SE74 
,= EXAMPLE1 IF Y\ KN[ ! H\R ( ! SUNSET1 Y\ CAN AT ?AT H\R 
KN[ ! FACT ?AT ! SUN IS SETT+3 ?IS IS KN[L$GE ( ! FACT 
BY WAY ( KN[L$GE ( TRU?S2 BUT Y\ CAN ALSO1 IF ! 
WEA!R IS F9E1 LOOK TO ! WE/ & ACTUALLY SEE ! SETT+ SUN3 
Y\ !N KN[ ! SAME FACT BY ! WAY ( KN[L$GE ( 
?+S4 

,?US 9 REG>D TO ANY COMPLEX FACT1 !RE >E1 !ORETICALLY1  
TWO WAYS 9 :I* IT MAY BE KN[N3 7#a7 BY MEANS ( A JUDGM5T1 9 
:I* ITS SEV]AL P>TS >E JUDG$ TO BE RELAT$ AS !Y >E 9 FACT 
RELAT$2 7#b7 BY MEANS ( ACQUA9TANCE ) ! COMPLEX FACT 
ITSELF1 :I* MAY 79 A L>GE S5SE7 BE CALL$ P]CEPTION1 ?\< IT 
IS BY NO MEANS CONF9$ TO OBJECTS ( ! S5SES4 ,N[ IT WILL BE 
OBS]V$ ?AT ! SECOND WAY ( KN[+ A COMPLEX FACT1 ! WAY ( 
ACQUA9TANCE1 IS ONLY POSSIBLE :5 !RE REALLY IS SU* A FACT1  
:ILE ! FIR/ WAY1 LIKE ALL JUDGM5T1 IS LIABLE TO ]ROR4 ,! 
SECOND WAY GIVES US ! COMPLEX :OLE1 & IS !RE=E ONLY POSSIBLE 
:5 ITS P>TS DO ACTUALLY HAVE ?AT RELATION :I* MAKES !M 
COMB9E TO =M SU* A COMPLEX4 ,! FIR/ WAY1 ON ! CONTR>Y1 GIVES 
US ! P>TS & ! RELATION SEV]ALLY1 & DEM&S ONLY ! REALITY 
( ! P>TS & ! RELATION3 ! RELATION MAY NOT RELATE ?OSE 
P>TS 9 ?AT WAY1 & YET ! JUDGM5T MAY OCCUR4 

,IT WILL BE REMEMB]$ ?AT AT ! 5D ( ,*APT] ,X,I WE 
SUGGE/$ ?AT !RE MI<T BE TWO K9DS ( SELF-EVID5CE1 ONE GIV+ 
AN ABSOLUTE GU>ANTEE ( TRU?1 ! O!R ONLY A P>TIAL GU>ANTEE4 
,!SE TWO K9DS CAN N[ BE DI/+UI%$4 

,WE MAY SAY ?AT A TRU? IS SELF-EVID5T1 9 ! FIR/ & MO/ 
ABSOLUTE S5SE1 :5 WE HAVE ACQUA9TANCE ) ! FACT :I* 
CORRESPONDS TO ! TRU?4 ,:5 ,O!LLO BELIEVES ?AT ,DESDEMONA LOVES 
,CASSIO1 ! CORRESPOND+ FACT1 IF HIS BELIEF W]E TRUE1 W\LD BE 
',DESDEMONA'S LOVE = ,CASSIO'4 ,?IS W\LD BE A FACT ) :I* NO ONE 
C\LD HAVE ACQUA9TANCE EXCEPT ,DESDEMONA2 H5CE 9 ! S5SE ( 
SELF-EVID5CE ?AT WE >E CONSID]+1 ! TRU? ?AT ,DESDEMONA LOVES 
,CASSIO 7IF IT W]E A TRU?7 C\LD ONLY BE SELF-EVID5T TO ,DESDEMONA4 
,ALL M5TAL FACTS1 & ALL FACTS CONC]N+ S5SE-DATA1 HAVE ?IS SAME 
PRIVACY3 !RE IS ONLY ONE P]SON TO :OM !Y CAN BE SELF-EVID5T 9 
\R PRES5T S5SE1 S9CE !RE IS ONLY ONE P]SON :O CAN BE 
ACQUA9T$ ) ! M5TAL ?+S OR ! S5SE-DATA CONC]N$4 ,?US 
NO FACT AB\T ANY P>TICUL> EXI/+ ?+ CAN BE SELF-EVID5T TO 
MORE ?AN ONE P]SON4 ,ON ! O!R H&1 FACTS AB\T UNIV]SALS DO 
NOT HAVE ?IS PRIVACY4 ,MANY M9DS MAY BE ACQUA9T$ ) ! SAME 
UNIV]SALS2 H5CE A RELATION BETWE5 UNIV]SALS MAY BE KN[N BY 
ACQUA9TANCE TO MANY DIFF]5T PEOPLE4 ,9 ALL CASES :]E WE KN[ BY 
ACQUA9TANCE A COMPLEX FACT CONSI/+ ( C]TA9 T]MS 9 A C]TA9 
RELATION1 WE SAY ?AT ! TRU? ?AT !SE T]MS >E SO RELAT$ HAS 
! FIR/ OR ABSOLUTE K9D ( SELF-EVID5CE1 & 9 !SE CASES ! 
JUDGM5T ?AT ! T]MS >E SO RELAT$ MU/ BE TRUE4 ,?US 
?IS SORT ( SELF-EVID5CE IS AN ABSOLUTE GU>ANTEE ( TRU?4 

,BUT AL?\< ?IS SORT ( SELF-EVID5CE IS AN ABSOLUTE 
GU>ANTEE ( TRU?1 IT DOES NOT 5ABLE US TO BE ABSOLUTELY 
C]TA91 9 ! CASE ( ANY GIV5 JUDGM5T1 ?AT ! JUDGM5T 9 
QUE/ION IS TRUE4 ,SUPPOSE WE FIR/ P]CEIVE ! SUN %9+1 :I* IS 
A COMPLEX FACT1 & !NCE PROCE$ TO MAKE ! JUDGM5T '! SUN IS 
%9+'4 ,9 PASS+ FROM ! P]CEPTION TO ! JUDGM5T1 IT IS 
NECESS>Y TO ANALYZE ! GIV5 COMPLEX FACT3 WE HAVE TO SEP>ATE \T 
'! SUN' & '%9+' AS CON/ITU5TS ( ! FACT4 ,9 ?IS PROCESS 
IT IS POSSIBLE TO COMMIT AN ]ROR2 H5CE EV5 :]E A FACT 
HAS ! FIR/ OR ABSOLUTE K9D ( SELF-EVID5CE1 A JUDGM5T BELIEV$ 
TO CORRESPOND TO ! FACT IS NOT ABSOLUTELY 9FALLIBLE1 BECAUSE IT MAY 
NOT REALLY CORRESPOND TO ! FACT4 ,BUT IF IT DOES CORRESPOND 79 ! 
S5SE EXPLA9$ 9 ! PREC$+ *APT]71 !N IT MU/ BE 
TRUE4 

,! SECOND SORT ( SELF-EVID5CE WILL BE ?AT :I* BELONGS TO 
JUDGM5TS 9 ! FIR/ 9/ANCE1 & IS NOT D]IV$ FROM DIRECT 
P]CEPTION ( A FACT AS A S+LE COMPLEX :OLE4 ,?IS SECOND K9D ( 
SELF-EVID5CE WILL HAVE DEGREES1 FROM ! V]Y HI<E/ DEGREE D[N TO 
A B>E 9CL9ATION 9 FAVOR ( ! BELIEF4 ,TAKE1 = EXAMPLE1 ! 
CASE ( A HORSE TROTT+ AWAY FROM US ALONG A H>D ROAD4 ,AT FIR/ \R 
C]TA9TY ?AT WE HE> ! HO(S IS COMPLETE2 GRADUALLY1 IF WE LI/5 
9T5TLY1 !RE COMES A MOM5T :5 WE ?9K P]HAPS IT WAS 
IMAG9ATION OR ! BL9D UP/AIRS OR \R [N HE>TBEATS2 AT LA/ WE 
BECOME D\BTFUL :E!R !RE WAS ANY NOISE AT ALL2 !N WE 
?9K WE NO LONG] HE> ANY?+1 & AT LA/ WE 
KN[ WE NO LONG] HE> ANY?+4 ,9 ?IS PROCESS1 !RE IS A 
CONT9UAL GRADATION ( SELF-EVID5CE1 FROM ! HI<E/ DEGREE TO ! 
LEA/1 NOT 9 ! S5SE-DATA !MSELVES1 BUT 9 ! JUDGM5TS BAS$ 
ON !M4 

,OR AGA93 ,SUPPOSE WE >E COMP>+ TWO %ADES ( COLOR1 ONE 
BLUE & ONE GRE54 ,WE CAN BE QUITE SURE !Y >E DIFF]5T %ADES ( 
COLOR2 BUT IF ! GRE5 COLOR IS GRADUALLY ALT]$ TO BE MORE & 
MORE LIKE ! BLUE1 BECOM+ FIR/ A BLUE-GRE51 !N A GRE5Y-BLUE1  
!N BLUE1 !RE WILL COME A MOM5T :5 WE >E D\BTFUL :E!R WE 
CAN SEE ANY DIFF]5CE1 & !N A MOM5T :5 WE KN[ ?AT WE CANNOT 
SEE ANY DIFF]5CE4 ,! SAME ?+ HAPP5S 9 TUN+ A MUSICAL 
9/RUM5T1 OR 9 ANY O!R CASE :]E !RE IS A CONT9U\S 
GRADATION4 ,?US SELF-EVID5CE ( ?IS SORT IS A MATT] ( DEGREE2 & 
IT SEEMS PLA9 ?AT ! HI<] DEGREES >E MORE TO BE TRU/$ ?AN 
! L[] DEGREES4 

,9 D]IVATIVE KN[L$GE \R ULTIMATE PREMISES MU/ HAVE SOME 
DEGREE ( SELF-EVID5CE1 & SO MU/ !IR CONNECTION ) ! 
CONCLUSIONS D$UC$ FROM !M4 ,TAKE = EXAMPLE A PIECE ( REASON+ 
9 GEOMETRY4 ,IT IS NOT 5\< ?AT ! AXIOMS FROM :I* WE />T 
%\LD BE SELF-EVID5T3 IT IS NECESS>Y ALSO ?AT1 AT EA* /EP 9 
! REASON+1 ! CONNECTION ( PREMISE & CONCLUSION %\LD BE 
SELF-EVID5T4 ,9 DIFFICULT REASON+1 ?IS CONNECTION HAS (T5 ONLY A 
V]Y SMALL DEGREE ( SELF-EVID5CE2 H5CE ]RORS ( REASON+ >E NOT 
IMPROBABLE :]E ! DIFFICULTY IS GREAT4 

,FROM :AT HAS BE5 SAID IT IS EVID5T ?AT1 BO? AS REG>DS 
9TUITIVE KN[L$GE & AS REG>DS D]IVATIVE KN[L$GE1 IF WE ASSUME 
?AT 9TUITIVE KN[L$GE IS TRU/WOR?Y 9 PROPORTION TO ! DEGREE 
( ITS SELF-EVID5CE1 !RE WILL BE A GRADATION 9 TRU/WOR?9ESS1  
FROM ! EXI/5CE ( NOTEWOR?Y S5SE-DATA & ! SIMPL] TRU?S ( 
LOGIC & >I?METIC1 :I* MAY BE TAK5 AS QUITE C]TA91 D[N TO 
JUDGM5TS :I* SEEM ONLY JU/ MORE PROBABLE ?AN !IR OPPOSITES4 
,:AT WE FIRMLY BELIEVE1 IF IT IS TRUE1 IS CALL$ KN[L$GE1  
PROVID$ IT IS EI!R 9TUITIVE OR 9F]R$ 7LOGICALLY OR 
PSY*OLOGICALLY7 FROM 9TUITIVE KN[L$GE FROM :I* IT FOLL[S 
LOGICALLY4 ,:AT WE FIRMLY BELIEVE1 IF IT IS NOT TRUE1 IS CALL$ 
]ROR4 ,:AT WE FIRMLY BELIEVE1 IF IT IS NEI!R KN[L$GE 
NOR ]ROR1 & ALSO :AT WE BELIEVE HESITAT+LY1 BECAUSE IT IS1 OR 
IS D]IV$ FROM1 SOME?+ :I* HAS NOT ! HI<E/ DEGREE ( 
SELF-EVID5CE1 MAY BE CALL$ PROBABLE OP9ION4 ,?US ! 
GREAT] P>T ( :AT W\LD COMMONLY PASS AS KN[L$GE IS MORE OR LESS 
PROBABLE OP9ION4 

,9 REG>D TO PROBABLE OP9ION1 WE CAN D]IVE GREAT ASSI/ANCE 
FROM COH]5CE1 :I* WE REJECT$ AS ! DEF9ITION 
( TRU?1 BUT MAY (T5 USE AS A CRIT]ION4 ,A BODY ( 
9DIVIDUALLY PROBABLE OP9IONS1 IF !Y >E MUTUALLY COH]5T1 BECOME 
MORE PROBABLE ?AN ANY ONE ( !M W\LD BE 9DIVIDUALLY4 ,IT IS 9 
?IS WAY ?AT MANY SCI5TIFIC HYPO!SES ACQUIRE !IR PROBABILITY4 
,!Y FIT 9TO A COH]5T SY/EM ( PROBABLE OP9IONS1 & ?US BECOME 
MORE PROBABLE ?AN !Y W\LD BE 9 ISOLATION4 ,! SAME ?+ APPLIES 
TO G5]AL PHILOSOPHICAL HYPO!SES4 ,(T5 9 A S+LE CASE SU* 
HYPO!SES MAY SEEM HI<LY D\BTFUL1 :ILE YET1 :5 WE CONSID] ! 
ORD] & COH]5CE :I* !Y 9TRODUCE 9TO A MASS ( PROBABLE 
OP9ION1 !Y BECOME PRETTY NE>LY C]TA94 ,?IS APPLIES1 9 
P>TICUL>1 TO SU* MATT]S AS ! DI/9CTION BETWE5 DREAMS & 
WAK+ LIFE4 ,IF \R DREAMS1 NI<T AFT] NI<T1 W]E AS COH]5T ONE 
) ANO!R AS \R DAYS1 WE %\LD H>DLY KN[ :E!R TO BELIEVE 
! DREAMS OR ! WAK+ LIFE4 ,AS IT IS1 ! TE/ ( COH]5CE 
CONDEMNS ! DREAMS & CONFIRMS ! WAK+ LIFE4 ,BUT ?IS TE/1  
?\< IT 9CREASES PROBABILITY :]E IT IS SUCCESSFUL1 NEV] GIVES 
ABSOLUTE C]TA9TY1 UNLESS !RE IS C]TA9TY ALREADY AT SOME PO9T 
9 ! COH]5T SY/EM4 ,?US ! M]E ORGANIZATION ( PROBABLE 
OP9ION WILL NEV]1 BY ITSELF1 TRANS=M IT 9TO 9DUBITABLE 
KN[L$GE4 

,*RONOLOGY 3 
,NOVEMB] #c1 #aiaa 3 ,*APT] #ac -- ,PUBLICATION4 
,JANU>Y #a#f1 #bb 3 ,*APT] #ac -- ,ADD$4 

 FILE G5]AT$ FROM 3  
HTTP3_/_/REVOLTLIB4COM_/