Sindbad~EG File Manager

Current Path : /usr/home/beeson/public_html/michaelbeeson/research/papers/
Upload File :
Current File : /usr/home/beeson/public_html/michaelbeeson/research/papers/reality.dvi

����;� TeX output 1999.02.18:1119������l�����'���&��t8��D��tG�G�cmr17�Realit��qy�7tand�T���Vruth�in�Mathematics��������û��X�Qcmr12�Mic��rhael��Beeson������g3�Departmen��rt��of�Mathematics�and�Computer�Science���������San��Jose�State�Univ��rersit�y��������lSan��Jose,�California�95192�������m�USA�������g�email:�8�b�S�eeson@mathcs.sjsu.edu������]�����<cF��Vebruary��18,�1999��(K����?�.��N�ffcmbx12�1��WL�In���tro�s3duction�����?�K�`y

cmr10�In��the�rst�part�of�the�t���w�en�tieth��cen�tury��*�,��wthere�w�as�a�crisis�in�the�foundations�of����?mathematics,���precipitated�W�in�part�b���y�the�disco�v�ery�of�parado�xes�suc�h�as�Russell's����?parado���x��Kab�Gout�the�set�of�all�sets,��Hand�in�part�b�y�the�w�ork�of�Zermelo,��Hwho����?pro���v�ed���the�coun���terin�tuitiv�e���theorem�that�ev���ery�set�can�b�Ge�w�ell-ordered.����^��ٓ�Rcmr7�1���
���The����?usual�/Dordering�on�the�set�of�real�n���um�b�Gers�/Dis�not�a�w���ell-ordering����^��2���|s�,�6�but�Zermelo's����?theorem���implies�that�there�m���ust�b�Ge�a�w�ell-ordering�of�the�reals.�E�Since�nob�Go�dy����?could���nd�suc���h�a�w�ell-ordering,��Tthe�pro�Gof�w�as�closely�examined,��Tand�led�to�m�uc�h����?discussion.����^��3������N�What�7udo�Ges�it�mean�to�sa���y�that�a�w�ell-ordering�of�the�reals�exists�if�w�e�can't����?nd��one?�W�The�discussion�of�this�issue�led�to�Zermelo's�form���ulation�of�the�axiom����?of��2c���hoice,���and�probably�con�tributed�to�Brou�w�er's�dev�elopmen�t�of�in�tuitionism.����?Brou���w�er�XHmain�tained�that�it�means�nothing�at�all�to�sa�y�that�a�w�ell-ordering����?of��the�reals�exists�if�w���e�can't�nd�one.�hOther�mathematicians,�notably�Da�vid����?Hilb�Gert,��gdisagreed,�and��1the�con���tro�v�ersy��1has�nev���er�b�een�denitiv���ely�settled.�([In-����?stead,�%<what���happ�Gened�w���as�that�axioms�for�set�theory��*�,�originally�dev���elop�Ged�to��?�e�ff��v�	J=�����"5��-:��Aa�cmr6�1����L��|{Ycmr8�A��A�"#�f�cmti8�wel�p[l-or���dering��of�a�set���2cmmi8�X�s�is�an�ordering�relation��x�\t<�y���on��Athe�set�suc�Îh�that�(1)�it�ob�<reys��	��the�?usual�axioms�of�order,��namely:�"�x���<�y�L��and�?�y��p<�z�YĻimplies��x�<�z�V��;�2�x�<�x��is�imp�<rossible;�2and���if����x�����K�cmsy8�6�=��y��$�then��x�<�y��$�or��y��!<�x�,���and�(2)�ev�Îery�subset�of��X�*{�has�a�least�elemen�t;��Qthat�is,���for���ev�Îery�Ƌsubset��A��of��X��һ,�Ɂthere�is�an�elemen�t��b��of��A��suc�h�that��b�\t<�c�Ƌ�whenev�er��c��is�a�mem�b�<rer�of��A����dieren�Ît��Xfrom��b�.��	�>�����"5��-:�2����LܻThe��Xset�of�p�<rositiv�Îe�n�um�b�<rers,�for�example,�has�no�least�elemen�t.�������"5��-:�3����LܻI���do���not�mean�to�imply�that�the�\crisis�in�foundations"�b�<regan��suddenly��with�Russell's��	��parado�Îx�0�and�Zermelo's�theorem;�^�Can�tor�had�already�deriv�ed�parado�xes�in�set�theory��J�,�G�whic�h���he�2solv�Îed�b�y�rejecting�the�Absolute�Innit�y��J�,�I/and�there�had�b�<reen�con�tro�v�ersies�b�<ret�w�een�Kro-���nec�Îk�er�N�and�Dedekind,�m0and�b�<ret�Îw�een�N�Gordan�and�Hilb�ert.�3�But�Zermelo's�theorem�dealt�with���the�M�con�Îtin�uum,�k�not�some�strange�sets�on�the�fringe�of�mathematics,�and�while�it�w�Îas�not�a���con�Îtradiction,��Xit�certainly�seemed�coun�terin�tuitiv�e.�������1����*�l�����'������?�clarify�Rthe�issues�in���v�olv�ed�Rin�Zermelo's�pro�Gof,�R�b�ecame�Rregarded�as�the�\founda-����?tion"�<Dof�mathematics,�AGso�that�when�mathematicians�nd�themselv���es�w�ondering����?or��disagreeing�ab�Gout�the�ultimate�meaning�of�a�piece�of�mathematics,�Z
they����?usually�0�stop�arguing�when�they�ha���v�e�0�seen�ho���w�to�reduce�the�mathematics�in����?question��uto�set�theory��*�.�L(While�some�philosophers�con���tin�ued��uto�examine�the�basic����?issues,�8!most�0�mathematicians�preferred�to�get�on�with�the�mathematics.�e�Just�to����?b�Ge�UUexplicit,�here�is�a�list�of�the�\basic�issues"�in�question:����N�
!",�

cmsy10��UU�What�is�real,�and�ho���w�do�w�e�kno�w�it?����N��UU�What�do�Ges�it�mean�to�sa���y�a�thing�exists?����N��UU�Can�things�exist�that�w���e�can't�kno�w�ab�Gout?����N��UU�Can�things�exist�that�w���e�don't�kno�w�ho�w�to�nd?����N��UU�What�do�Ges�it�mean�to�sa���y�something�is�true?����N��UU�Ho���w�can�w�e�kno�w�whether�something�is�true?����N��UU�Can�things�b�Ge�true�that�w���e�can't�ev�er�kno�w�to�b�Ge�true?����NWithin��fteen�y���ears�of�Russell's�parado�x�and�the�furor�o�v�er�Zermelo's�pro�Gof,����?L.�%�E.�J.�Brou���w�er�%�had�giv���en�answ�ers�to�all�of�the�questions�ab�Go�v�e,�Y�at�least�in����?the���realm�of�mathematics,���in�his�philosoph���y�kno�wn�as�in�tuitionism.�xBrou�w�er's����?answ���ers��)w�ere�as�follo�ws:�1What�is�real,���in�mathematics�at�least,�is�that�whic���h�can����?b�Ge���constructed.��What�it�means�to�sa���y�a�thing�exists,��]is�that�w�e�kno�w�ho�w�to����?construct�Ait.�kConsequen���tly��*�,�Eif�a�thing�exists,�w���e�at�least�kno�w�ho�w�to�construct����?it,��Vso���there�can't�b�Ge�things�w���e�don't�kno�w�ab�Gout�or�can't�nd.��What�it�means����?to��sa���y�a�thing�is�true�is�that�w�e�ha�v�e�a�pro�Gof�of�it����^��4���|s�,���and�that�is�also�the�w�a�y��*�,����?and��the�only�w���a�y��*�,��that��w�e�can�kno�w�a�thing�is�true.�I
Consequen�tly�there�can't����?b�Ge�UUtrue�things�that�w���e�can't�ev�er�kno�w�to�b�Ge�true.����NBrou���w�er's�P�answ�ers�w�ere�rejected�b�y�a�ma��8jorit�y�of�mathematicians,���along�with����?the�ȍphilosophical�system�that�supp�Gorted�those�answ���ers.�B�F��*�or�example,��Brou�w�er's����?famous��copp�Gonen���t,�ĦDa�vid�Hilb�Gert,�Ħstrongly�disagreed�with�Brou�w�er�ab�Gout�b�oth����?realit���y�b5and�truth.��iF��*�or�him�realit�y�did�not�dep�Gend�on�our�constructions.��iHe����?w���ould��sha�v�e�agreed�with�the�statemen�t�attributed�to�F��*�rege,��:that�mathemati-����?cians�$�disco���v�er�theorems�in�the�same�w�a�y�that�explorers�disco�v�er�islands.�߲Y��*�et����?Hilb�Gert's��Mansw���er�to�the�last�question,��Kengra�v�ed�on�his�tom�bstone����^��5���|s�,��Kis�the�same����?as�?�Brou���w�er's:�Gwhatev�er�is�true�can�b�Ge�pro�v�ed.�While�for�Brou�w�er�this�w�as�a����?consequence�?�of�the�meanings�of�the�terms,�Dfor�Hilb�Gert�it�w���as�an�article�of�faith.����NIn�F&spite�of�their�rejection,��ZBrou���w�er's�F&ideas�con���tin�ued�F&to�dra���w�the�in�terest����?of��csome�philosophers,���logicians,�and��ca�few�mathematicians,���and�ha���v�e��cnot�b�Geen����?denitiv���ely��pro�v�ed�wrong.�M�No�w�the�t�w�en�tieth�cen�tury�is�dra�wing�to�a�close;��and��?�X-�ff��v�	J=�����"5��-:�4����LܻPro�<rof��Xfor�Brou�Îw�er��Xmean�t�men�tal�construction,�not�a�linguistic�or�sym�b�<rolic�ob�x�ject.��	�>�����"5��-:�5����LܻHis�ۜepitaph�(see�the�last�paragraph�of�[�47���])�is�literally��J�,��-\Wir�m�<r����ussen�wissen.���Wir�w�Îerden��	��wissen."�� (W��J�e��Xm�Îust�kno�w.�� W��J�e�will�kno�w.)�������2����
 l�����'������?�during�{�the�in���terv�ening�{�decades,��,there�has�b�Geen�relev��q�an���t�scien�tic�and�mathemat-����?ical���progress.�<�The�question�to�b�Ge�addressed�in�this�pap�er�is�the�follo���wing:�!��8�':

cmti10�What,����?if�?anything,�i�has�twentieth-c��}'entury�pr�o�gr�ess�in�physics,�i�mathematics,�and�?lo�gic����?c��}'ontribute�d���to�our�understanding�of�the�philosophic��}'al�questions�liste�d�ab�ove?����^��6������?�W��*�e��will�see�whether�Brou���w�er's��answ�ers�stand�up�under�the�ligh�t�of�a�cen�tury�of����?scien���tic�UUprogress,�and�what�answ�ers,�if�an�y��*�,�our�progress�has�supplied.����NT��*�o���start�with,��I���will�list�the�main�ac���hiev�emen�ts���of�the�cen���tury�whic�h�ma�y�b�Ge����?relev��q�an���t,�UUmore�or�less�c�hronologically��*�.����N���1�The�dev���elopmen�t��1of�formal�logic�and�axiomatic�systems,���b���y�Russell,�Zer-����?melo,�UUHilb�Gert,�Heyting,�and�others����N��8�The�reorganization�of�mathematics�along�set-theoretic�lines,�q�b���y�the�Bourbaki����^��7������?�and�UUothers.����N����The�scien���tic�analysis�of�the�notion�of�\algorithm"�b�y�Ch�urc�h,���Kleene,�and����?T��*�uring�UUin�the�thirties����N����The�understanding�of�the�dierence�b�Get���w�een��syn�tax�and�seman�tics,�H!and����?their�UUconnection�through�the�completeness�theorem�for�rst-order�logic����N��UU�Quan���tum�mec�hanics����N��UU�The�recursiv���e�unsolv��q�abilit�y�of�the�halting�problem����N��UU�The�incompleteness�theorems����N��UU�T��*�arski's�denition�of�truth�and�the�related�theorem�on�undenabilit���y����N��UU�Kleene's�recursiv���e�realizabilit�y�in�terpretation����N��UU�The�indep�Gendence�pro�ofs�in�set�theory����NI�[�will�[�set�ground�rules�for�the�discussion�to�follo���w.��rFirst,�]�I�will�assume�that����?there���is�suc���h�a�thing�as�realit�y;��-that�is,�ԿI��pprop�Gose�to�dismiss�the�solipsist�p�osition����?without��discussion.����^��8���	�L�Second,��qas�used�in�this�pap�Ger,�existence�is�a�binary�prop-����?ert���y��*�.�H�Just��6as�y�ou�can't�b�Ge�\a�little�bit�pregnan�t"�or�\a�little�bit�dead",���y�ou�can't����?b�Ge�8\a�little�bit�real".�XEither�y���ou�exist,��or�y�ou�don't.�XThird,��existence�cannot�b�Ge��?�X-�ff��v�	J=�����"5��-:�6����LܻThe�2�title�of�this�pap�<rer�w�Îould�ha�v�e�a�m�uc�h�nicer�ring�in�Dutc�h:�u��Werkelijkheid�{@en�Waarheid��	��in�^jWiskunde�.���All�:{three�of�these�w�Îords�ha�v�e�meaningful�o�v�ertones:��f�werkelijkheid��implies���that�m�realit�Îy�is�that�whic�h�is�w�ork��able,�� that�with�whic�h�w�e�can�w�ork.���Waarheid��carries�the���implication�w�that�truth�is�testable;��something�is�true�if�w�Îe�can�nd�a�w�arran�t�for�it,���a�reason�to���assert�uit.��And��wiskunde�,��[the�Dutc�Îh�w�ord�for�mathematics,��[means�literally�\exact�kno�wledge".���None��jof�the�English�w�Îords�ha�v�e�suc�h�ric�h�connotations,���and�b�<resides,�in�Dutc�Îh�the�title�is���alliterativ�Îe.���As�=�it�turns�out,���Brou�w�er�published�a�Dutc�h�translation�of�his�famous�1908���pap�<rer�I�The�?�unr���eliability�of�the�lo�gic�al�principles�I�in�1919�under�the�title��Wiskunde,�KZwaarheid,���werkelijkheid�.�`,(See�\the�table�of�con�Îten�ts�\of�[�9��@].)�I�Odid�not�kno�Îw�that�when�I�ga�Îv�e�\this�pap�<rer���a��Xtitle.��	�>�����"5��-:�7����LܻThe��Bourbaki�w�Îere�a�group�of�F��J�renc�h�mathematicians�whose�co�<rordinated�eorts�at�writing���mathematics��Xin�a�more�formal�w�Îa�y��Xw�ere�v�ery�in
uen�tial.�������"5��-:�8����LܻThat��XBrou�Îw�er�w�as�not�a�solipsist�is�eviden�t,�for�example,�from�the�rst�page�of�[�8��@].�������3������l�����'������?�dened�)�in�terms�of�something�simpler.�c.Creation,�2Lexistence,�and�)�destruction�are����?elemen���tal�UUand�fundamen�tal.����^��9������N�Disclaimer���:�Giv���en�q�the�stated�aim�of�the�pap�Ger,��]a�certain�amoun�t�of�historical����?discussion��is�necessary��*�,��but�I��qwish�to�emphasize�that�the�primary�sub��8ject�of�this����?pap�Ger���is�philosoph���y�of�mathematics,�Bonot�its�history��*�.�	9A��jcomplete�history�of����?w���ork�5�relev��q�an�t�to�the�questions�listed�ab�Go�v�e�w�ould�require�a�m�uc�h�longer�and����?more�(�sc���holarly�w�ork,�1�and�w�ould�certainly�go�bac�k�at�least�half�a�cen�tury�b�Gefore����?Brou���w�er,�UUwhose�w���ork�is�tak�en�as�the�starting�p�Goin�t�here.����N�A��}'cknow���le�dgements��!�:�@nBoris��Kushner,�`Wilfried�Sieg,�and�Ric���hard�Tieszen�read����?drafts�UUof�this�pap�Ger�and�supplied�v��q�aluable�criticisms,�for�whic���h�I�thank�them.��!č��?�2��WL�The�ffmeaning�of�existence�����?�The�wconcept�of�existence�is�cen���tral�to�sev�eral�of�the�\basic�issues".�'�It�is�a�concept����?that��came�explicitly�to�the�fore�when�Zermelo�claimed�to�pro���v�e��the�existence�of����?a�p�w���ell-ordering�that�nob�Go�dy�p�could�exhibit.���Existence�is�in�v�olv�ed�in�the�Russell����?parado���x,�)�whic�h��sho�ws�that�a�certain�set��
�b>

cmmi10�R��߲=���f�x��:��x��62��x�g��cannot�exist,�)�although����?one��Ais�surprised�b���y�this�fact,���since�its�denition�seems�simple�enough�un�til�one����?sees�
the�parado���x.�]Existence�is�in�v�olv�ed�in�the�solution�put�forth�to�the�parado�x����?b���y���Russell�and�Whitehead,��yin��Principia���Mathematic��}'a�,�whic���h�in�v�olv�ed�building����?a�bhierarc���h�y�of�sets,�e>sets�of�sets,�sets�of�sets�of�sets,�eac���h�one�of�whose�existence����?could�UUb�Ge�safely�assumed�or�pro���v�ed,�UUand�whic���h�w�ould�not�include�the�set��R�Dz.����NBrou���w�er's��cen�tral�p�Goin�t�w�as�that�no�sensible�meaning�could�b�Ge�attac�hed�to����?the��phrase�\there�exists"�other�than�\w���e�can�nd",��-and�that�therefore�mathe-����?matics�UUshould�b�Ge�done�using�that�meaning,�let�the�c���hips�fall�where�they�ma�y��*�.����NExistence�_�is�a�philosophical�issue�that�far�transcends�mathematics,�bRand�has����?b�Geen��"considered�for�millenia�b���y�philosophers.�^/It�has�also�receiv�ed�considerable����?atten���tion�1�in�this�cen�tury�from�ph�ysicists�seeking�to�understand�the�m�ysteries����?of���the�subatomic�w���orld�and�the�origin�of�the�univ�erse,���in�conditions�where�the����?quan���tum���nature�of�realit�y�is�not�mask�ed�b�y�statistical�a�v�erages.��,F��*�or�example,����?one��of�the�fundamen���tal�questions�ab�Gout�existence�is�whether�things�that�exist,����?ha���v�e�g�an�existence�indep�Genden���t�of�the�mind,��or�whether�the�mind�is�someho�w�nec-����?essarily�!�in���v�olv�ed�in�existence.�`�This�question�came�up�in�one�form�in�connection����?with���Brou���w�er's�in�tuitionism,��and�in�another�form�in�connection�with�quan�tum����?mec���hanics,��6and��=w�as�considered�already�thousands�of�y�ears�ago.�IIn�this�sec-����?tion,���therefore,�w���e��	consider�the�meaning�of�existence,���and�sp�Gecically��*�,�whether����?and���what�t���w�en�tieth-cen�tury���progress�has�con���tributed�to�our�understanding�of����?existence.��?�ff�ff��v�	J=�����"5��-:�9����LܻAll��zreligions�address�these�three�pro�<rcesses.���Hinduism�in�particular�giv�Îes�Go�d�three�ma�x�jor��	��asp�<rects:�� Brahma��Xthe�Creator,�Vishn�Îu�the�Preserv�er,�Shiv��a�the�Destro�y�er.�������4����+��l�����'�������?���N�cmbx12�2.1��]�Examples��uT��?�It���is�not�ob���vious�that�w�e�alw�a�ys�mean�the�same�thing�when�w�e�use�the�w�ord�\ex-����?istence"�k�or�the�w���ords�\there�exists".��^Consider�the�follo�wing�examples,�q~and�ask����?y���ourself�r�what�it�means�for�these�things�to�exist:���tables,�zc�hairs,�and�r�b�Geer�m�ugs;����?electrons;�delds���(as�the�electromagnetic�eld);�space�(not�\outer�space",��3but�the����?space��in�whic���h�things�are�placed�and�lo�Gcated);�/�w�a�v�e�functions;�/�abstractions�lik�e����?justice,�UUtruth,�b�Geaut���y;�Ideas�in�the�Platonic�sense;�minds�and�\I";�Go�d.����NIn��Gthe�realm�of�mathematics,��ask�y���ourself�what�it�means�for�the�follo�wing����?things�L�to�exist,���and�whether�it�means�the�same�thing�in�eac���h�case:�`�in�tegers;����?functions��from�in���tegers�to�in�tegers�(sequences�of�in�tegers);�*3real�n�um�b�Gers;�*3sets�of����?n���um�b�Gers;��Winnitesimals��and�non-standard�reals;�measurable�cardinals;�the�set����?or�UUclass�of�all�sets�(the�\set-theoretic�univ���erse").����NCommonplace�+ob��8jects�lik���e�tables,�2ac�hairs,�and�+b�Geer�m�ugs�seem�to�exist�b�Ge-����?cause�Bthey�are�tangible,�%Fi.e.,�they�Bpro�Gduce�sense-impressions.�]�The�philosophical����?view�pkno���wn�as��naive���r��}'e�alism�p�sa�ys�that�realit�y�is�that�whic�h�corresp�Gonds�to�our����?sense-impressions.����^��10���2ڲThis�mdview�w���ould�require�us�to�reject�as�unreal�(that�is,����?nonexisten���t)���all�the�other�items�in�the�list.��rElectrons,��for�example,�cannot�b�Ge����?sensed�.]directly;���their�existence�is�inferred�from�their�eects�on�measuring�in-����?strumen���ts.�Y�But��in�appropriate�exp�Gerimen�ts,�_w�e�can�determine�their�size,�_shap�Ge,����?mass,���diameter,�and��felectrical�c���harge,���whic�h��fenables�us�to�form�a�go�Go�d��fmen���tal����?mo�Gdel���of�\an�electron".�<�Ev���en�so,�Ջtheir�existence�is�not�quite�on�the�same�fo�oting����?as���tables,���c���hairs,�and�b�Geer�m���ugs,�b�Gecause�according�to�ph���ysics,�an�y���t�w�o�electrons����?are���absolutely�in���terc�hangeable.�{nIf���someone�sneakily�switc���hed�y�our�electron�for����?another���one�while�y���ou�w�eren't�lo�Goking,��	y�ou�couldn't�tell,��	and�not�just�in�prac-����?tice,��in����principle��y���ou�couldn't�tell.��2This�seems�to�run�coun�ter�to�our�in�tuitiv�e����?feeling�9�that�when�a�thing�exists,�sit�can�b�Ge�distinguished�from�the�rest�of�the����?univ���erse.���Nev�ertheless,�m�nob�Go�dy�h�denies�(no���w�ada�ys�h�at�least)�that�electrons�exist;����?w���e���seem�to�feel�that�an�ything�with�a�long�list�of�prop�Gerties�m�ust�exist{�something����?�m���ust�UUb�Ge�there�to�ha�v�e�the�prop�Gerties.����NWhen��rw���e�come�to�elds,�Ĺsuc�h�as�the�electromagnetic�or�gra�vitational�eld,����?opinion��ma���y�not�b�Ge�quite�unanimous.�1�Some�ma�y�insist�that�the�gra�vitational����?eld�_�is�merely�a�men���tal�construction,���a�gmen�t�of�the�imagination.��In�our�minds,����?w���e��assign�a�n�um�b�Ger�to�eac�h�p�Goin�t�in�space,��and�w�e�call�that�the�gra�vitational����?eld.�bHOn�&�the�other�hand,�0$according�to�Einstein,�the�eld�has�mass�and�attracts����?other��Vmatter,�Vjust�as�an�electron�do�Ges,�and�exerts�its�eects�on�measuring�in-����?strumen���ts.���Quan�tum��Dtheory�has�sho���wn�us�that�the�truth�is�more�complicated:����?neither�}elds�nor�the�particles�whic���h�\carry�the�eld"�can�exist�without�the����?other!�%=The��'photon�is�the�carrier�of�the�electromagnetic�eld,��for�example,�and����?the���gra���viton�is�the�carrier�of�the�gra�vitational�eld.��[Ev�en�though�no�gra�viton����?has�*ev���er�b�Geen�detected,�L�ev�ery�ph�ysicist�b�Geliev�es�gra�vitons�exist�and�are�ev�ery����?bit�dbas�real�as�photons.���The�classical�eld�has�b�Geen�sho���wn��not��to�exist�after�all;����?it�UUw���as�only�a�gmen�t�of�our�imaginations�used�in�an�appro�ximate�theory��*�.��?�@o�ff��v�	J=�����w���-:�10����LܻF��J�or��Xa�classic�refutation�of�naiv�Îe�realism,�see�Russell[�50���].�������5����:,�l�����'������N�While��xthe�classical�eld�has�b�Geen�sho���wn�not�to�exist�in�m�y�lifetime,�E�on����?the��lother�hand,��2the�quark�has�b�Geen�sho���wn�to�exist.��
While�I��Jw�as�a�studen�t�in����?the���1960's,���quarks�w���ere�regarded�as�mathematical�ctions�that�could�nev�er�b�Ge����?observ���ed.�q�No�w�UUthey�are�as�real�as�electrons�[�41��
].����NMo���ving���further�do�wn�the�list,��unanimit�y�of�opinion�breaks�do�wn�completely:����?there�Pfare�dozens�of�opinions�ab�Gout�the�nature�of�the�existence�of�\I",�including����?the�?Buddhist�p�Gosition�that�the�\I"�>�do�es�not�really�exist����^��11���x�,�v�and�the�Hindu�p�osition����?that���it�do�Ges�not�exist�separately�from�Go�d.����^��12���
��Other�fundamen���tal�questions�on����?whic���h���there�is�no�unanimit�y�are:��whether�the�existence�of�the�\I"��`is�indep�Genden�t����?of�UUthe�existence�of�the�b�Go�dy��*�,�UUand�whether�Go�Gd�exists.����NIn��Ythe�mathematical�univ���erse,��%there�is�a�similar�progression�from�the�familiar����?and�$nconcrete�things,�.6whose�existence�is�univ���ersally�recognized,�to�the�\fringes",����?where���unanimit���y�breaks�do�wn,��Tbut�the�issues�of�existence�are�nev�ertheless�sig-����?nican���t��ones.�W�Kronec�k�er�made�the�famous�remark�that�\Go�Gd�made�the�p�ositiv���e����?in���tegers,�n{the�6Arest�is�due�to�man."��This�seems�to�sym�b�Golize�the�fact�that�the����?in���tegers���are�\almost�tangible."���Although�of�course�they�are�not�sense-ob��8jects,����?w���e�kseem�to�ha�v�e�direct�men�tal�exp�Gerience�of�the�p�ositiv���e�in�tegers.��It�is�no�w�an����?elemen���tary�[�exercise�to�construct�the�negativ�e�and�then�the�rational�n�um�b�Gers����?from�.�the�in���tegers,�6�so�the�next�question�of�existence�is�the�sequences�of�in�tegers.����?Here���there�is�already�a�div���ergence�of�opinion:���m�ust�a�sequence�b�Ge�giv�en�b�y�a����?rule,�L;or��can�it�b�Ge�determined�b���y�an�arbitrary�set�of�pairs��h�n;���m�i��pro�vided�w�e����?can�
�pro���v�e�single-v��q�aluedness�(only�one��m��for�eac�h��n�)?�Y�Cauc�h�y�sho�w�ed�us�ho�w�to����?construct��ereal�n���um�b�Gers��efrom�sequences�of�rationals,�Еso�the�existence�question�for����?sequences�e�of�in���tegers�is�the�same�as�for�reals.��Sets�of�n�um�b�Gers�can�certainly�b�e����?determined��<b���y�a�dening�prop�Gert�y��*�,��6but�do�there�exist�sets�of�n�um�b�Gers�without����?an���y���dening�prop�Gert�y?�\iThis�is�one�of�the�questions�on�whic�h�denite�progress����?has�UUb�Geen�made�in�this�cen���tury��*�,�as�w�e�will�see�b�Gelo�w.����NNon-standard��%and�innitesimal�reals,�bwhic���h�w�ere�used�at�the�end�of�the�sev-����?en���teen�th��cen�tury�b�y�the�founders�of�calculus,�$&w�ere�resuscitated�and�placed�on�a����?mo�Gdern��fo�oting�b���y�Abraham�Robinson.�A�But�most�mathematicians�probably�feel����?that�S0they�don't�really�exist.�qThey�tak���e�a�view�m�uc�h�lik�e�the�view�that�gra�vita-����?tional��'elds�don't�really�exist;���these�n���um�b�Gers��'are�men���tal�creations,���gmen�ts�of����?the��Oimagination.�KT��*�rue,���y���ou�can�use�them,�but�they�aren't�real�in�the�same�sense����?as�ε�[ٲ.�VEA��go�Go�d�theory�of�mathematical�existence�should�oer�to�ols�to�answ���er�this����?question:�lldo�J�innitesimals�exist,�L�or�not?�n5It�seems�that�hardly�an���y�one�J�is�willing����?to��defend�their�existence�(see�[�44��
],[�45��]��for�one;�Cbsee�also�[�49��
],��page�267��)�but�I��?��f�ff��v�	J=�����w���-:�11����LܻF��J�or��example:��\Kno�Îw�that�there�is�nothing�whic�h�is�not�a�re
ection,��>there,�y�et��nothing."��	��[�38���],���p.��;357.�The���v��arious�branc�Îhes�of�Buddhism�tak�e�dieren�t�p�<rositions�on�some�fundamen�tal���on�Îtological�n�and�epistemological�issues,��so�the�statemen�t�in�the�text�is�not�only�an�o�v�ersimpli-���cation�v�of�a�complex�p�<rosition,���but�as�so�on�as�one�tries�to�elab�orate�it,���it�b�ecomes�con�Îtro�v�ersial.��	�>�����w���-:�12����Lܻ\This��9is�the�truth:��the�sparks,�Hrthough�of�one�nature�with�the�re,�leap�from�it;���un-���coun�Îted�T}b�<reings�leap�from�the�Ev�erlasting,�nCbut�these,�m�Îy�son,�merge�in�Îto�It�again."��-(Mundak��a-���Upanishad,���Bo�<rok���2;���p.��>52�in�[�55���].�This�accoun�Ît�ma�y�mak�e�it�seem�as�if�the�\coun�tless�b�<reings"���ha�Îv�e��a�separate�existence,�(but�there�are�h�Îundreds�of�passages�with�this�sense:�K�\Go�<rd�liv�es�in���the��Xhollo�Îw�of�the�heart,�lling�it�with�immortalit�y��J�,�ligh�t,�in�telligence."�� ([�55���],�p.�66)�������6����Jߠl�����'������?�ha���v�en't�UUheard�a�con���vincing�dispro�Gof.����NMeasurable��Acardinals�and�other,�)�ev���en�more�esoteric,�large�cardinals,�cause����?the��same�kind�of�breakdo���wn�in�unanimit�y�of�opinion�that�religious�issues�do.����?There�6%is�no�ma��8jorit���y�opinion�and�most�p�Geople�agree�not�to�discuss�these�things����?at�UUthe�dinner�table.���6���?�2.2��]�Prop�`erties��of�Existence��uT��?�Is�oRit�p�Gossible�to�sa���y�an�ything�sensible�ab�Gout�the�meaning�of�existence,��Sthat�w�ould����?cast�[gsome�ligh���t�on�the�examples�ab�Go�v�e,��dand�giv�e�us�some�to�Gols�for�analyzing�these����?questions?�!In�c view�of�the�in���tractabilit�y�c of�these�problems�o���v�er�c the�past�cen���turies,����?ev���en�ƙa�little�progress�w�ould�b�Ge�in�teresting.�ŕSince,���as�men�tioned�at�the�outset,����?it�uis�not�p�Gossible�to��dene��existence,�?=let�us�b�egin�b���y�en�umerating�some�of�its����?prop�Gerties.����NW��*�e��Ycan't�sp�Geak�meaningfully�of�existence�without�sa���ying�something�exists.����?W��*�e��oare�isolating�or�selecting�some�\thing"�(or�concept,��idea,�etc.)��and��osa���ying����?that���it�is�an�en���tit�y��*�.�3�The���idea�is�similar�to�that�put�forw���ard�b�y�the�Gestalt����?psyc���hologists.�"�Nearly��ev�ery�one�has�seen�those�images�whic�h�can�b�Ge�view�ed�in����?t���w�o�UUdieren�t�w�a�ys,�for�example�as�a�v��q�ase�or�as�t�w�o�faces.����NW��*�e�UUthen�ha���v�e�UUto�distinguish:����N��UU�sp��}'e�cic���existenc�e�:�q�this�dog�\Harley"�exists����N��UU�gener��}'al���existenc�e�:�q�dogs�exist,�ghosts�exist����NIn�UUb�Goth�cases,�to�mak���e�the�claim�clear,�w�e�ha�v�e�to����N(a)��explain�what�it�means�to�b�Ge�giv���en�the,�Zor�one�of�the,�things�whose����?existence� �is�asserted.�`/What�do�Ges�it�mean�to�b�e�giv���en�(presen�ted�with)�a�dog?�`/a����?ghost?�nsThis�KWm���uc�h�is�necessary�to�mak�e�the�prop�Gosition�\�X�9�exists"�meaningful.����?W��*�e�UUm���ust�also�explain����N(b)�P�ho���w�to�recognize�something�as�an��X���,���whic�h�generally�will�in�v�olv�e�kno�wing����?what��;y���ou�can�do�with�or�to�an��X���.�ay(It's�a�dog�if�it�has�a�certain�c�haracteristic����?app�Gearances�>�and�b�eha���viors;���but�note�that�no��one��suc�h�c�haracteristic�is�absolutely����?essen���tial,�e.g.�U�there���are�three-legged�dogs,�hairless�dogs,�dogs�that�don't�pla���y����?F��*�risb�Gee,�UUetc.)����NNo���w,�UUto�establish�it�as�true�that��X�7�exists,�or��X���'s�exist,�w�e�migh�t����N(c)��pro�Gduce�a�sp�ecic��X���(in�the�case�of�Harley)�or�evidence�that�a�sp�ecic����?�X���has�&3b�Geen�pro�duced�(in�the�case�of�aliens,�/�miracles,�or�&3subatomic�particles�for����?instance)����N(d)��sho���w�ho�w�to�pro�Gduce�an��X�^��on�demand�(in�the�case�of�subatomic�particles����?but�d�not�aliens�or�miracles).��(F��*�or�example,�h�electricit���y�exists�b�Gecause�y�ou�can�get����?it�UUfrom�a�w���all�so�Gc�k�et.�q�Originally�it�w�as�harder�to�pro�v�e�electricit�y�existed.����NLet's��.test�these�ideas.��TDo�Ges�there�exist�a�monk���ey�that�can�pla�y��Til���l��Ther��}'e����?Was�9�Y��;�ou��?�on�the�accordion?�QIt's�hard�to�imagine�what�w���ould�mak�e�us�b�Geliev�e�in����?the�+�existence�of�suc���h�a�monk�ey�other�than�seeing�and�hearing�the�p�Gerformance,������7����]f�l�����'������?�or��$the�v���ery�trusted�secondhand�rep�Gort�of�the�p�erformance.��4The�ab�o���v�e��$criteria����?seem�UUto�explain�the�meaning�of�this�sen���tence�quite�adequately��*�.����NNo���w�[�consider�the�question,��^\Do�monk�eys�exist?"��This�ma�y�seem�at�rst����?glance�^.to�b�Ge�a�dieren���t�kind�of�question.��THo�w�ev�er,�`ein�b�Goth�cases�w�e're�singling����?out�<ua�certain�kind�of�en���tit�y;�D�the�<ufact�that�there�is�a�single�w���ord��monkey��for�one����?kind�T�and�no�single�w���ord��ac��}'c�or�dion-monkey�T��for�the�other�is�a�linguistic�acciden�t����?(or�5�non-acciden���tal�consequence�of�the�relativ�e�rarit�y�of�accordion-monk�eys,�<but����?a��,linguistic�phenomenon�rather�than�an�essen���tial�philosophical�dierence).�>�If�w�e����?hadn't��}ev���er�seen�a�monk�ey��*�,��w�e�w�ould�w�an�t�visual�pro�Gof�of�their�existence�just����?as�UUw���e�no�w�w�an�t�it�for�aliens.����^��13�����6���?�2.3��]�Non-constructiv��e��existence�of�concrete�ob��jects��uT��?�The��Fp�Goin���t�will�not�b�e�lost�on�those�educated�in�mathematical�philosoph���y��*�,��that����?the��Cab�Go���v�e�criteria�for�existence�are�closely�related�to�those�put�forw�ard�b�y�the����?constructivists.�@�As��Sa�test�case,���then,�let��Sus�try�to�construct�non-mathematical����?examples�UUof�non-constructiv���e�existence.����N�Example�t�1�[�:�`�Do�Ges�3athere�exist�an�animal�whic���h�is�a�dog�if�Goldbac�h's�conjec-����?ture�s�is�true�and�a�cat�if�not?�&�(Here�the�only�signicance�of�Goldbac���h's�conjecture����?is�UUthat�it�is�an�unsolv���ed�mathematical�problem.)����NPresumably��the�v��q�ast�ma��8jorit���y�of�mathematicians�will�answ�er�y�es,�H!suc�h�an����?animal�?exists.�0FF��*�or�either�Goldbac���h's�conjecture�is�true�or�not;���in�the�former����?case��|m���y�dog�Harley�will�do�for�the�animal,��and�in�the�latter�case�m�y�brother's����?cat�UUSimon�will�do.����NW��*�e�UUcan�get�mathematics�completely�out�of�the�picture:����N�Example��2�[�:�°Do�Ges�}�there�exist�a�dog�whic���h�is�a�Dalmatian�if�global�w�arming����?is�UUa�serious�danger�and�a�Chih���uah�ua�UUif�not?����NThis��?assumes�that�the�prop�Gosition�\global�w���arming�is�a�serious�danger"�is����?w���ell-dened;���if���y�ou�don't�think�so,��,substitute�some�other�w�ell-dened�question����?to�z�whic���h�w�e�don't�kno�w�the�answ�er.��uNo�w,��HI�z�ask�ed�sev�eral�non-mathematicians����?this��question,��and�they�all�said,�of�course�not.�M�What�a�silly�question!�Of�course,����?w���e�'�don't�ha�v�e�to�tak�e�their�opinion�seriously��*�,�0�p�Gerhaps�they�are�the�same�p�eople����?who�J�think�aliens�exist.�Q�But�the�essence�of�their�opinion�seems�to�b�Ge�that�in����?Example�z2,��two�G�do��}'gs�exist�,�one�zof�whic���h�has�the�desired�prop�Gerties�but�w�e�don't����?kno���w�QVwhic�h�one.�prIt�is�the�fact�that�the�question�asks�ab�Gout��a��:do��}'g��whic�h�mak�es����?it���seem�nonsensical;�S�and�when�the�meaning�is�un���tangled,�)[it�seems�to�ha�v�e�no����?con���ten�t����^��14���x�,�UUwhic�h�is�wh�y�the�question�is�\silly��*�."��?�X-�ff��v�	J=�����w���-:�13����LܻA�7surprisingly�72high�p�<rercen�Îtage�of�the�p�opulation�do�es�not�ha�Îv�e�72suc�h�high�standards�for��	��b�<relieving��in�the�existence�of�aliens!��I��tak�Îe�this�to�sho�w�simply�that�these�p�<reople�are�neither���philosophers��Xnor�scien�Îtists.��	�>�����w���-:�14����LܻThis���disen�Îtanglemen�t�is�related�to�Herbrand's�theorem�[�32���],[�61��],��"page�89,�whic�Îh�sa�ys�that���when�x$�9�x�(�x�)�is�pro�Îv��able�in�classical�rst-order�logic,���then�there�are�sp�<recic�terms��t��8:�;�cmmi6�i�����suc�h�that���the��Xdisjunction�of�the���(�t��8:�i��,r�)�is�also�pro�Îv��able.�������8����	k�l�����'������N�My���p�Gosition�is�that�examples�lik���e�these�sho�w�that�the�concept�of�existence����?as�used�in��9�x�(�x�)�when����is�a�complex�prop�Gert���y�is�not�a�primitiv�e�concept.�_VW��*�e����?should�Hsconsider�as�primitiv���e,�Kthe�concept�of�existence�as�used�in��9�x�(�x�),�where����?��(�x�)��is�a�prop�Gert���y�that�can�b�e�v���eried�on�demand�of�an�y�particular��x��that�is����?presen���ted.���6���?�2.4��]�Platonism���!���@cmti12�vs.��Constructivism��uT��?�It���is�the�thrust�of�this�section�that�the�distinction�b�Get���w�een���Platonist�and�non-����?Platonist�8�philosophies�of�mathematics�has�some�subtleties�that�ha���v�e�8�b�Geen�o���v�erlo�ok�ed.����^��15������N�The��preceding�principles�used�the�w���ord�\pro�Gduce".��VW��*�e�pro�duce�a�dog�b���y����?presen���ting�vhhim�for�sense-p�Gerception,���but�ho�w�do�w�e�pro�Gduce�an�in�teger�or�a����?set,��or��an�abstraction�lik���e�justice?�T��*�o�pro�Gduce�suc�h�a�thing�means�to�presen�t����?it�?�to�the�mind.�0�On�the�mo�Gdern�Platonist�view�held�b���y�G����odel,�z.mathematical����?ob��8jects��ha���v�e�an�existence�indep�Genden�t�of�the�mind,�and�the�mind�(or�part�of�it)����?functions��similarly�to�a�sense�organ�in�allo���wing�us�to�apprehend�mathematical����?ob��8jects.��	This��is�in�con���trast�to�the�view�of�Brou�w�er,��in�whic�h�mathematics�is�a����?series�	�of�men���tal�constructions.�X�According�to�this�view,��m�y�in�tegers�are�dieren�t����?from�e,y���our�in�tegers{mine�are�in�m�y�mind,�i"and�y�ours�are�in�y�our�mind.����^��16���
4�When����?I�:�tell�:�y���ou�that�4�W+�3��=�7,�?�I�predict�:�the�result�y���ou�will�get�b�y�p�Gerforming�certain����?constructions���on�y���our�in�tegers.�	e�On�this�view,�e�w�e�can�comm�unicate�certain����?prop�Gerties��Gthat�our�in���tegers�ha�v�e�in�common,��but�ultimately�w�e�m�ust�rely�on����?�intuition�UU�for�the�certain���t�y�UUthat�y���our�in�tegers�are�isomorphic�to�m�y�in�tegers.����NOn��the�Brou���w�erian��view,�w�e�aren't�\giv�en"�a�mathematical�ob��8ject,�rather����?w���e��.are�told�ho�w�to�construct�it�in�our�o�wn�mind.�BeThe�principle�that�if��X���'s�exist,����?w���e���should�kno�w�what�it�means�to�b�Ge�giv�en�an��X���,��is�replaced�b�y��*�,��if��X���'s�exist,�w���e����?should�LAkno���w�what�it�means�to�construct�an��X�#�(in�our�mind).�n�On�the�G����odelian����?view,�� mathematical�Փob��8jects�are�treated�just�lik���e�dogs�and�cats:�1�w�e�presen�t�them����?to�|�eac���h�other�for�apprehension�b�y�the�mind.���Note�that�ev�en�with�dogs�and����?cats�A�the�mind,�E�not�only�the�sense�organs,�is�in���v�olv�ed:�hdamage�A�to�certain�neural����?areas���can�cause�p�Geople�to�not�recognize�dogs�and�cats�ev���en�though�their�ey�es�are����?functioning���correctly��*�.���W�e�are�using�dogs�and�cats�here�as�examples�of�ob��8jects����?whic���h�B�are�not�men�tal�constructions,�FLi.e.,�they�B�ha�v�e�an�existence�indep�Genden�t�of����?the�UUp�Gerceiving�mind.��?�X-�ff��v�	J=�����w���-:�15����LܻThe�Qsection�en�Îtitled��Philosophic���al�R�R�emarks�,�+�pp.��332-347�Qin�[�24���],�is�recommended�to�the��	��reader���who�needs�orien�Îtation�in�this�long-running�debate.��QIn�the�terminology�of�that�section,���Brou�Îw�er��Xw�ould�b�<re�a�neo-conceptualist.��	�>�����w���-:�16����LܻStrictly�|xsp�<reaking,��>the�conclusion�that�m�Îy�in�tegers�are�dieren�t�from�y�ours�do�<res�not�follo�w���so�v~easily:�	lThe�w�Îord��in��as�used�in��in��Cmy�mind�v~�do�<resn't�necessarily�imply�a�spatial�lo�cation;���indeed��w�Îe�usually�don't�think�of�in�tegers�as�ha�ving�a�spatial�lo�<rcation.��FP�erhaps�the�in�tegers�I���construct���are�literally�the�same�in�Îtegers�as�y�ou�construct,��hev�en�though�w�e�do�it�\in"�dieren�t���minds.��UThe��isame�in�Îtegers�could�b�<re�in�b�oth�of�our�minds.��U(I�am�indebted�to�Amanda�Beeson���for�ۡp�<roin�Îting�this�out.)��Ho�w�ev�er,�3I��^b�<reliev�e�the�conclusion�follo�ws�b�y�another�argumen�t:�Ӳif���w�Îe�U�eac�h�mak�e�a�separate�act�of�creation�(construction�is�creation)�then�the�results�of�those���dieren�Ît��Xacts�of�creation�m�ust�b�<re�dieren�t.�������9����
za�l�����'������N�T��*�raditionally�,���the��Platonist�view�of�existence�is�illustrated�b���y�considering�the����?existence���of�abstractions�suc���h�as�justice,���rather�than�b�y�mathematical�ob��8jects.����?Let��us�consider�whether�the�dierence�b�Get���w�een��the�Platonist�and�Aristotelian����?view��of�existence�is�the�same�as�the�dierence�b�Get���w�een��the�G����odelian�and�Brou���w-����?erian�ƥviews.�B8On�the�Platonist�view,��/an�abstraction�lik���e�justice�really�exists.�The����?opp�Gosite��mview,���whic���h�I��]will�call�Aristotelian,�is�that�justice�is�just�an�attribute����?of�bVsituations;�h�some�situations�are�just�and�some�are�not,�e�or�some�are�more�just����?than��others.�j&T��*�ransferring�this�question�to�mathematical�philosoph���y�,���w�e��can����?ask�g_whether�the�set�of�all�ev���en�in�tegers�(or�the�prop�Gert�y�of�ev�enness)�exists,����?or�f~whether�ev���enness�is�just�an�attribute�of�in�tegers,�j�without�an�existence�of�its����?o���wn.�c6The�)�dierence�b�Get�w�een�the�G����odelian�and�Brou�w�erian�view�is�\where"�the����?mathematical�|>ob��8jects�liv���e:���in�our�minds,���or�outside�our�minds?��The�dierence����?b�Get���w�een���the�Platonist�and�Aristotelian�view�is�whether�prop�erties�ha���v�e���an�ex-����?istence��Lof�their�o���wn.���The�w�ord�\reication"�means�making�an�ob��8ject�out�of�a����?prop�Gert���y;���let��"us�call�a�\reist"�a�p�erson�who�b�eliev���es�this�can�alw�a�ys�b�Ge�done.�CA����?reist��b�Geliev���es����^��17������\ev�ery�adjectiv�e�is�also�a�noun."�^8G����odel�(and�b�Gefore�him�Plato,����?Can���tor�Oeand�Dedekind)�w�ere�\reists",�P�while�Aristotle,�Kronec���k�er�Oeand�the�early����?Brou���w�er�
�w�ere�not.���On�the�other�hand,�;�Brou�w�er�w�as�a�\men�talist",�;�b�Gelieving����?that��mathematical�ob��8jects�exist�within�the�mind,��and�G����odel�w���as�b�y�con�trast�an����?\ob��8jectivist".�K�Can���tor,��Dedekind,�and��DG����odel�w�ere�reist�ob��8jectivists.����^��18���Ĩ�Brou�w�er����?w���as�ߟa�non-reist�men�talist,�1at�least�in�his�early�pap�Gers.��Later,�when�Brou���w�er����?in���tro�Gduced�*�higher-order�sp�ecies�(sp�ecies�of�sp�ecies),�3he�b�ecame�a�reist�men���tal-����?ist.��The�{�p�Gosition�of�Kronec���k�er�{�seems�to�ha���v�e�{�b�een�non-reist�ob��8jectivist;���since����?he�1though���t�Go�Gd�made�the�in�tegers,�hpresumably�they�w�ere�not�in�eac�h�of�our����?minds���separately��*�,�'�but�simply�apprehended�b���y�our�minds.�kLSo�all�four�p�Gossible����?p�Gositions�UUha���v�e�famous�represen�tativ�es.����^��19������N�W��*�e���ha���v�e�to�ask�wh�y��*�,��if�these�distinctions�are�meaningful,�they�ha���v�e�remained����?unresolv���ed�SSfor�so�long.�k�What�exp�Gerimen�t�could�distinguish�them?�k�W��*�ell,���the����?answ���er�#�is�clear:�X�if�w�e�destro�y�all�minds�capable�of�mathematics,�-�do�the�in�tegers����?still�^�exist?��2If�so,�a,then�the�ob��8jectivists�are�righ���t.�If�not,�a,they're�wrong.�Nob�Go�dy����?seems��"to�doubt�that�Mt.�@�Ev���erest�w�ould�still�exist,�ߓfor�example,�but�the�outcome����?of��the�exp�Gerimen���t�ab�out�the�in���tegers�is�still�in�doubt,�(�and�in�addition�to�ethical����?questions,�
�there���seems�to�b�Ge�a�fundamen���tal�con�tradiction�in�the�instructions�for����?the�׏exp�Gerimen���t,��since�after�the�destruction�of�all�minds�capable�of�mathematics,����?w���e�&�still�need�suc�h�a�mind�to��test��for�the�existence�of�the�in�tegers,�Z�ev�en�if�w�e����?b�Geliev���e�|�them�to�exist��outside��that�mind.���So�here�w�e�ha�v�e�a�meaningful�but��?�X-�ff��v�	J=�����w���-:�17����LܻI��Xam�indebted�to�Nathan�Hellerstein�for�this�phrase.��	�>�����w���-:�18����LܻDedekind�J�w�Îas�ev�en�willing�to�consider�the�collection�of�\all�the�p�<roten�tial�ob�x�jects�of�m�y��	��though�Ît"��as�an�ob�x�ject;���he�used�it�to�\pro�v�e"�the�existence�of�an�innite�set�([�14���],���Theorem�66,���p.�� 64).�������w���-:�19����LܻI�m�could�m�not�nd�a�famous�mathematician�who�adv�Îo�<rcated�the�p�osition�that�the�in�Îtegers���don't���exist.�ŨThat�is,���the�reication�of�the�adjectiv�Îe�\sev�en"�as�in�\sev�en�ngers"�to�the�noun���\sev�Îen"���seems�to�b�<re�accepted�b�y�all�mathematicians.��On�the�other�hand,���it�w�as�p�<roin�ted�out���to��qme�b�Îy�R.�Tieszen�that�some�\nominalist"�philosophers�ha�v�e�made�suc�h�a�claim,�we.g.�?l[�21���],���[�31���].��������10�����v�l�����'������?�un���testable�UUdistinction.�q�This�is�denitely�philosoph�y��*�,�not�science.����NNote�ʦthat�these�diculties�are�completely�indep�Genden���t�of�the�old�\mind����?of��Go�Gd"�ob��8jection,���according�to�whic���h�the�destruction�of�all�(mathematical)����?minds��is�imp�Gossible�b�ecause�among�them�is�the�Mind�of�Go�d,��whic���h�con�tains����?the��hwhole�univ���erse.��There�are�enough�diculties�in�the�theory�of�existence����?without�UUconsidering�this�one.���6���?�2.5��]�Existence,��creation,�and�destruction��uT��?�Although��sexistence�cannot�b�Ge�dened�in�terms�of�something�simpler,��:it�ma���y�help����?to��{examine�the�pro�Gcesses�of�creation�and�destruction,��b���y�whic�h�a�thing�comes����?to���exist.�O�W��*�e�ha���v�e���already�said�that�existence�is�not�partial:�>�a�table�either�exists����?or�Zit�do�Ges�not�exist.��But�as�the�table�is�b�eing�created,�[*there�is�a�momen���t�when����?three��legs�are�attac���hed�and�the�fourth�leg�is�ab�Gout�to�b�e�attac���hed,�lwhen�it�is����?not��y���et�a�table;�'follo�w�ed�so�Gon�b�y�the�attac�hmen�t�of�the�fourth�leg�and�the�table����?is�f�b�Gorn.��tOne�migh���t�argue�that�the�exact�momen�t�of�creation�is�fuzzy:���ho�w�dry����?m���ust�rthe�glue�b�Ge�b�efore�the�leg�is�really�attac���hed,��turning�the�table�parts�in�to�a����?table?��wThe�n�truth�is,�t�though,�that�n�it�is�the�concept�of�\table"�that�is�fuzzy��*�,�t�not����?the�UUexistence.����NThe��momen���t�of�creation�of�a�h�uman�b�Geing�is�at�the�heart�of�the�ab�ortion����?debate.�!�As�:�with�the�table,�s�it�is�the�concept�of�\h���uman�b�Geing"�that�is�fuzzy��*�,����?not�Xthe�concept�of�existence.�z%In�mathematics,���w���e�ha�v�e�to�w�atc�h�out�for�the����?same�'=phenomenon.��The�Russell�parado���x,�[�for�example,�exp�Goses�a�fuzziness�in����?our�UUconcept�of��class�.��!č��?�3��WL�Existence�ffin�mathematics�����?�It�L�is�time�to�return�to�our�theme:�m�the�con���tributions�of�scien�tic�progress�in�the����?t���w�en�tieth���cen�tury�to�the�solution�of�the�problems�of�existence�(realit�y)�and�truth����?in��mathematics.�>
In�the�rst�decades�of�this�cen���tury��*�,��(the�relation�b�Get�w�een�syn�tax����?and���seman���tics�w�as�not�y�et�clear.�-�F��*�or�instance,���there�is�no�discussion�in��Principia����?Mathematic��}'a�+�of�the�seman���tics�of�the�system.�YdAlthough�the�main�purp�Gose�of�the����?w���ork��w�as�to�establish�a�consisten�t�framew�ork�for�mathematics,�!�and�the�ob�vious����?consistency�pro�Gof�(to�da���y)�is�to�exhibit�a�mo�del�of�the�theory��*�,��no�suc���h�argumen�t����?is�Tto�b�Ge�found�in��Principia�.�q[Another�piece�of�evidence�is�that�the�completeness����?theorem��for�rst-order�logic�w���as�not�form�ulated�in�prin�t�un�til�1928.����^��20������During����?the���1920's,���the�fo�Gcus�of�atten���tion�w�as�the�decision�problem�for�rst-order�logic����?(a�l-primarily�seman���tic�question:��xwhic�h�form�ulas�are�true�in�all�mo�Gdels?)��Prather����?than�qcompleteness�(whic���h�adds�a�syn�tactic�comp�Gonen�t:��6whic�h�form�ulas�ha�v�e����?pro�Gofs?)��Sometime���b�et���w�een�1900�and�1930,���the�mo�Gdern�distinction�b�et���w�een��?��f�ff��v�	J=�����w���-:�20����LܻSee�$�the�historical�notes�in�[�27���],�Hpp.��D47-48,�whic�Îh�$�suggest�that�the�1928�form�ulation�w�as�not��	��y�Îet��:correct;���but�see�also�[�56���],��Awhere�it�is�argued�that�indeed�the�question�w�as�w�ell-understo�<ro�d���at��Xthat�time.��������11�������l�����'������?�syn���tax�uand�seman�tics�dev�elop�Ged�and�b�ecame�understo�o�d;��it�seems�lik���ely�that����?it���w���as�in�the�late�1920's.�YThe�historical�evidence�is�not�conclusiv�e,��Rbut�for�our����?presen���t�UUpurp�Gose�the�exact�date�do�es�not�matter.���6���?�3.1��]�Hilb�`ert��uT��?�Hilb�Gert��Ev���ery�w�ell�understo�Go�d��Ethat�the�same�axioms�migh�t�ha�v�e�dieren�t�mo�Gdels;����?he�(�wrote��F��;�oundations�V-of�Ge��}'ometry�(��[�35��
]�putting�non-Euclidean�and�Euclidean����?geometries��on�an�axiomatic�basis,��completing�the�w���ork�Euclid�had�b�Gegun.�O�In����?this���b�Go�ok�he�explicitly�exhibited�dieren���t�mo�dels�for�geometry�(pro�duced�earlier����?b���y�?�Beltrami�and�Klein).�j�His�famous�dictum�concerning�prop�Ger�axiomatizations����?w���as��Hthat�y�ou�had�to�b�Ge�able�to�substitute�\tables,���c�hairs,�and��Hb�Geer�m�ugs"�for����?\p�Goin���ts,�j�lines,�and���planes",�and�the�reasoning�should�still�b�Ge�correct�if�the����?axioms��Zheld.�]�These�facts�sho���w�that�Hilb�Gert�w�as�b�y�no�means�ignoran�t�of�the����?distinction��Qb�Get���w�een�syn�tax�and�seman�tics.�I�Y��*�et,��Rin�[�36��
]�Hilb�Gert�to�ok�the�p�osition����?that�0!consistency�guaran���tees�existence.�eaF��*�or�example,�7�there�are�groups�if�there�is����?no��5con���tradiction�in�group�theory��*�.�hWhat�Hilb�Gert�seemed�to�tak�e�as�a�denition����?of�fexistence�w���as�later�pro�v�ed�as�G����odel's�completeness�theorem.���As�a�denition����?of���existence,�AGho���w�ev�er,�it���w�on't�serv�e.�	tF��*�or�example,�AGjust�b�Gecause�there's�no����?con���tradiction�&'in�the�assumption�that�a�measurable�cardinal�exists,�/�that�do�Gesn't����?pro���v�e�"that�one�do�Ges�exist.���After�all,�U3there�can�certainly�b�e�dieren���t�kinds�of����?large��!cardinals�whose�existence�is�m���utually�exclusiv�e,���but�eac�h�one�b�y�itself�is����?consisten���t.��It�dseems�Hilb�Gert�just�hadn't�though�t�this�through�carefully�(this����?had�8 to�w���ait�for�G����odel�and�T��*�arksi),�p�and�it�sho�ws�that�the�relations�of�syn�tax����?and��jseman���tics�w�ere�not�en�tirely�clear�in�1904.����^��21���|�No�w�that�w�e�kno�w�ab�Gout����?the�"�completeness�theorem,�U�and�the�dierence�b�Get���w�een�"�rst-order�and�second-����?order���logic,�aUand�the�incompleteness�theorems,�w���e�can�form�ulate�the�matter����?more��dclearly��*�.�O�Hilb�Gert�had�in�mind�systems�whic���h�are�second-order�categorical,����?suc���h���as�the�P�eano�axioms.����^��22���rD�Suc�h�axiom�systems�do�in�essence�dene�their����?unique�&mo�Gdel,�#�so�if�they�are�consisten���t,�that�mo�Gdel�exists.�]
But�ev���ery��rst-or��}'der����?�axiomatization,�-#strong���enough�to�meet�the�conditions�of�the�incompleteness����?theorem,��-necessarily��"has�other�mo�Gdels�as�w���ell,�in�whic���h�non-standard�\in�tegers"����?o�Gccur.��Note�8�that�Hilb�ert's�axiomatizations�in��F��;�oundations�eof�Ge��}'ometry�8��w���ere����?second-order�1Waxiomatizations;��Xa�prop�Ger�rst-order�treatmen���t�of�geometry�w�as����?ev���en�tually��(giv�en�b�y�studen�ts�of�T��*�arski�in�mid-cen�tury�[�7��].�PBut�the�completeness����?theorem,���forging���the�link�b�Get���w�een���non-con�tradiction�and�mo�Gdel�existence,���is�a����?theorem�u:ab�Gout�rst-order�axiomatizations,��3not�second-order.��xSo�Hilb�ert�w���as����?mistak���en,�
�if���w�e�in�terpret�his�\consistency"�in�the�rst-order�sense,�
�but�not�if�w�e��?�X-�ff��v�	J=�����w���-:�21����LܻThe���history�of�the�dev�Îelopmen�t���of�the�mo�<rdern�distinction�b�et�Îw�een���syn�tax�and�seman�tics��	��of��%rst-order�logic,��cand�the�question�of�what�Hilb�<rert�knew�and�when�he�knew�it,�are�discussed���at��Xlength�in�[�56���].��	�>�����w���-:�22����Lܻ\Second-order���categorical"�means�there�is�only�one�mo�<rdel�of�the�P�Îeano�axioms�in�whic�h���the�G�second-order�v��ariables�(to�whic�Îh�mathematical�induction�applies)�range�o�v�er�all�subsets���of��Xthe�mo�<rdel.��������12����
��l�����'������?�in���terpret��2it�in�the�second-order�sense;�>and�at�the�time,��the�dierence�w�as�not�so����?clear�UUas�to�Gda���y��*�.����NHilb�Gert's�	�p�osition�liv���es�on,��ho�w�ev�er,�in�	�the�usual�in���terpretation�of�\there�ex-����?ists"��in�mathematics�when�it�is�applied�to�an�ob��8ject�rather�than�a�mathematical����?system.�:BWhen���w���e�sa�y�\there�exists�a�real�n�um�b�Ger��x��suc�h�that��P�c��(�x�)",��and�accept����?a�~�pro�Gof�b���y�con�tradiction,���w�e�are�sa�ying��x��exists,���b�Gecause�no�con�tradiction�follo�ws����?from���its�non-existence,��9i.e.,�it���is�consisten���t�to�assume�it�exists,��9therefore�it�exists.����?T��*�ec���hnically�,�-�the��rst-order�consistency�w�ould�b�Ge�that��:9�xP�c��(�x�)�is�unpro�v��q�able,����?whic���h�Cis�w�eak�er�than�pro�ving��::9�xP�c��(�x�).�:�W��*�e�w�ould�b�Ge�wrong�if�w�e�tried�to����?pro���v�e�d�the�existence�of�a�nonstandard�in���teger�from�the�indep�Gendence�of�some����?sen���tence��]�:9�xP�c��(�x�)�in�P�eano's�Arithmetic.�S But�if�w�e�in�terpret�consistency�in�the����?second-order�E�sense,�I
Hilb�Gert's�denition�of�existence�b���y�consistency�explains�the����?classical�UUin���terpretation�of�\there�exists".���6���?�3.2��]�Reductionism��uT��?�Reductionism�
�is�the�searc���h�for�fundamen�tal�\building�blo�Gc�ks".���This�scien�tic����?metho�Gd��w���as�the�engine�of�science�in�the�eigh�teen�th�and�nineteen�th�cen�turies.�N�In����?c���hemistry�_*it�led�to�the�disco�v�ery�of�atoms�and�the�explanation�of�the�structure����?of�/matter�via�the�p�Gerio�dic�/table.��It�w���as�natural�to�try�a�similar�approac�h�in����?mathematics.��I� ;am� pnot�certain�who�b�Gegan�this�eort,�S6but�Dedekind�w���as�one����?of�
the�pioneers�[�13��
].����^��23���
gβHe�sho���w�ed�
ho�w�to�construct�the�real�n�um�b�Gers�from�sets����?of��<rationals,��5and�ho���w�to�construct�rationals�from�pairs�of�in�tegers.��|F��*�rege�([�23��
],����?�x�72,��p.�ڱ85)�ͣreduced�the�concept�of�in���teger�to�that�of�class:�bcThe�in�teger�7,��ac-����?cording�2Zto�F��*�rege,�i�is�the�class�of�all�sets�with�sev���en�elemen�ts.��This�decreased����?the�!vn���um�b�Ger�of�elemen�tary�building�blo�Gc�ks�to�one:�W�sets.�`}Eviden�tly�F��*�rege�did�not����?b�Geliev���e���Kronec�k�er's�attribution�of�the�creation�of�the�in�tegers�to�Go�Gd.���During����?the�m�t���w�en�tieth�cen�tury's�op�Gening�decade,�s�Russell�and�Whitehead�carried�the�re-����?ductionist�K�program�forw���ard�in�[�51��
],�M�adopting�F��*�rege's�denition�of�in�tegers.�n�But����?as�)the�axiomatization�of�set�theory�progressed,�^and�atten���tion�w�as�fo�Gcused�on����?w���ell-founded��sets,�5�this�particular�reduction�did�not�surviv�e,�5�and�w�as�replaced����?b���y�UUthe�v�on�Neumann�in�tegers.����^��24������N�The���question�remains�whether�these�reductions�are�formal�devices�or�de-����?nitions.�?F��*�or���example,���is�the�in���teger�2��r��}'e�al���ly��¸f�0�;����f�0�gg�,�or�is�it��r��}'e�al���ly��²the�class�of����?all�ht���w�o-elemen�t�sets,�Jmor�neither,�but�just�some�indivisible,�atomic��thing��exist-����?ing�V�without�in���ternal�structure,�WTb�Geing�2�b�y�virtue�of�its�\place"�as�the�successor����?of�c�1?��Is�the�rational�n���um�b�Ger�c�2/3�\really"�an�ordered�pair�of�in���tegers,��^or�an����?equiv��q�alence���class�of�suc���h�ordered�pairs?�M`F��*�or�that�matter,���is�an�ordered�pair�of����?ob��8jects��\really"�some�kind�of�set,�qor�is�the�denition�of�sets�as�ordered�pair�just����?a���con���v�enien�t�isomorphism�that�enables�us�to�reduce�the�n�um�b�Ger�of�\primitiv�e����?concepts"?���I� �tak���e� �the�p�Gosition�that�this�question�is�not�relev��q�an�t�to�the�main��?�X-�ff��v�	J=�����w���-:�23����LܻDedekind�V�did�this�w�Îork�in�1858,�p?but�did�not�publish�it�un�til�1872,�p?according�to�the�preface��	��of��X[�14���].��	�>�����w���-:�24����LܻThe��Xv�Îon�Neumann�in�tegers�are�dened�b�y:�� 0�is�the�empt�y�set,��n�㐻+�1�\t=��n��[�f�n�g�.��������13�����?�l�����'������?�issues��Bof�truth�and�realit���y��*�.��Most�mathematicians�recognize�that�using�one�re-����?duction���or�another�is�a�matter�of�con���v�enience.�w�Opinion���is�more�divided�ab�Gout����?the��~other�cen���tral�reduction,��of�function�to�set�of�pairs.����^��25���Y)�Here�the�question�is����?dieren���t{the���question�is�whether�the�t�w�o�notions�really�are�isomorphic�or�not.����?Brou���w�er��tto�Gok�the�idea�of�sequence�of�in���tegers�as�fundamen�tal,�'|and�dev�elop�Ged����?real�URn���um�b�Gers�from�this�idea.�q�He�w�as�also�a�practitioner�of�reductionism,��Qre-����?ducing��concepts�from�the�rest�of�mathematics�to�sequences�of�in���tegers�whenev�er����?p�Gossible.�)yHo���w�ev�er,���he�|hb�eliev�ed�|hhe�w���as�really�dening�the�reduced�concepts,���since����?their�UUoriginal�denitions�did�not�mak���e�sense�to�him.����NDedekind,�Ǚand���after�him�F��*�rege�and�Russell,�attempted�the�most�am���bitious����?reduction�d�of�all:��the�reduction�of�mathematics�to�logic.���The�hard�part�here����?is��/the�reduction�of�innit���y�to�logic.�aTDedekind's�construction�of�the�in�tegers����?on���a�logical�basis�in���v�olv�ed���t�w�o�steps:�B�(1)�reication�of�adjectiv�e�\sev�en"�to����?noun�
c\sev���en",�7�and�(2)�reication�of�adjectiv�e�\is-in�teger"�to�noun�\In�tegers",����?the�ڧcompleted�totalit���y�of�natural�n�um�b�Gers.�	�Brou�w�er�criticized�the�latter�in����?his���1907�thesis�[�9��],�j#p.�	p�78.�on���the�grounds�that�the�totalit���y�of�Dedekind's����?though���ts,�)Ahis�ˬ�Ge��}'dankenwelt�,�\cannot�b�Ge�view���ed�mathematically��*�,�so�it�is�not����?certain��cthat�with�resp�Gect�to�suc���h�a�thing�the�ordinary�axioms�of�whole�and�part����?will��remain�consisten���t."���This�step�w�as�basic�to�Dedekind's�attempt�to�reduce����?mathematics�A�to�logic.��The�fundamen���tal�m�ystery�is�innit�y��*�,�x�and�Brou�w�er�though�t����?it�S0to�b�Ge�an�irreducible�m���ystery��*�.��Logic�for�Brou�w�er�w�as�merely�a�linguistic�trace�of����?mathematics;�Ԓmathematics��0consists�in�men���tal�constructions;�these�constructions����?exhibit�Dregularities�whic���h�are�describ�Ged�b�y�the�la�ws�of�logic�in�the�same�w�a�y�as����?the�UUla���w�of�gra�vit�y�describ�Ges�regularities�in�the�motions�of�the�planets.���6���?�3.3��]�Implications��of�Kleene's�w��ork�for�Brou�w�er's�ideas��uT��?�Brou���w�er��had�sev���eral�notions�of�sequence,��and�it�is�v�ery�instructiv�e�to�review�the����?reasons��wh���y��*�.�E�First,��qif�existence�means�\w�e�can�nd",��qas�it�did�for�Brou�w�er,��qthen����?in��order�that�a�sequence��x����	0e�rcmmi7�n���W�satisfy��8�n�9�m�(�m���=��x����n��q~�),�#Ythe��sequence�m���ust�b�Ge�giv�en����?b���y��a��law�,��or�as�w�e�w�ould�no�w�sa�y��*�,��there�m�ust�b�Ge�an�algorithm�for�computing��x����n��q~�,����?giv���en�s��n�.�̎On�the�other�hand,�{(Brou�w�er�felt�that�w�e�ha�v�e�a�geometrical�in�tuition����?ab�Gout���the�real-n���um�b�er���line:��Lto�eac���h�p�oin���t�of�the�visible�(or�visualizable)�line,����?there�	�corresp�Gonds�a�n���um�b�er.���Based�	�on�this�in���tuition,�6�Brou�w�er�	�felt�that�com-����?putable��afunctions�on�the�reals�m���ust�b�Ge�con�tin�uous.�O�That�is,���an�appro�ximation����?to�9the�output�should�b�Ge�computable�from�an�appro���ximation�to�the�input.��sIn����?other��w���ords,��/an�y�function�guaran�teed�to�w�ork�on�all�real�inputs�m�ust�w�ork�b�y����?computing���with�appro���ximations�to�the�input.�n�A���step�function,���for�instance,�is����?not�єdened�on�all�real�n���um�b�Gers,��but�єonly�on�those�real�n���um�b�ers�єfor�whic���h�w�e����?can�UUdecide�on�whic���h�side�of�the�step�they�lie.��?�X-�ff��v�	J=�����w���-:�25����LܻThe�Treduction�of�the�concept�of�function�to�that�of�set�(of�pairs)�w�Îas�not�made�b�y�Dedekind��	��or��Can�Îtor.���During�their�time�the�concept�of�function�itself�w�as�still�ev�olving.���See�the�discussion���in��X[�2��@],�pp.�� 418.��������14����̻�l�����'������N�Brou���w�er�BZcouldn't�justify�this�con���tin�uit�y�BZtheorem�(called����^��26���
�@�the�\fan�theorem")����?using�;only�la���wlik�e�;sequences.�"�There�app�Geared�to�b�e�a�gap�b�et���w�een�;existence����?justied��Kb���y�\w�e�can�nd",�#�and�existence�as�justied�b�y�geometrical�in�tuition.����?He��bturned�to�the�second�criterion�for�existence�w���e�discussed�ab�Go�v�e:�Nwhen�can�w�e����?sa���y�yw�e�are�giv�en�a�real�n�um�b�Ger?�V)He�though�t�w�e�could�b�Ge�\giv�en"�a�real�n�um�b�Ger����?b���y��:a�sequence�of�c�hoices,��sfor�example�c�hoices�made�b�y�tossing�a�die�or�making����?an��"arbitrary�decision�to�determine�the�next�decimal�place.��.Suc���h�sequences�he����?called��%\c���hoice�sequences".�2cOf�course,��/y�ou�could�mak�e�y�our�c�hoices�in�accordance����?with�O�a�la���w,�P�so�the�la�wless�sequences�are�a�sub�Gclass�of�the�c�hoice�sequences;�Qubut����?the��bcon���v�erse�w�as�not�ob�vious.�2wBrou�w�er�claimed�to�pro�v�e�the�con�tin�uit�y�theorem,����?but�+his�pro�Gof�in���v�olv�ed�+reasoning�ab�out�the�nature�of�p�ossible�though���ts,�4and�few����?w���ere�UUcon�vinced.����NThese��^issues�w���ere�greatly�claried�in�the�thirties�and�forties,��when�the�deni-����?tion��and�theory�of�computable�functions�w���ere�dev�elop�Ged,���and�applied�b�y�Kleene����?and���others�to�the�sub��8ject�of�\recursiv���e�analysis".��RKleene�sho�w�ed�that�there�is����?a�u�binary�tree�in�whic���h�all�recursiv�e�paths�run�out�of�the�tree,��but�y�et�there����?are�i_arbitrarily�long�paths�in�the�tree.���That�is,�nbK����onig's�lemma�fails�in�recursiv���e����?analysis.���This�^implies�the�failure�of�the�con���tin�uit�y�^theorems�and�sho���ws�that����?there�!�really�are�more�c���hoice�sequences�than�la�wless�sequences.�`�F��*�or�more�details����?see�UU[�2��],�Chapter�IV.����NOur�1in���terest�in�the�matter�here�is�in�the�meaning�of�existence.�VfIt�seems�that����?the���w���ork�of�Kleene�sho�ws�that�y�ou�can't�ha�v�e�it�b�Goth�w�a�ys:��if�y�ou�can�b�Ge�\giv�en"����?a��(real�n���um�b�Ger��(b�y�a�sequence�of�c�hoices,��as�our�geometrical�in�tuition�suggests,����?then�D	\for�eac���h��n��there�exists�an��m�"�cannot�mean�\w�e�can�nd��m��from��n��b�y�an����?algorithm".�;By��haccepting�the�geometric�criterion�for�the�existence�of�reals,��,w���e����?are��:forced�to�accept�a�wider,���p�Gossibly�non-algorithmic�meaning�for�\for�eac���h��n����?�there���exists�an��m�".��Brou���w�er���w�as�w�ell�a�w�are�of�that,���although�he�did�not�ha�v�e����?Kleene's�lfprecise�denitions�and�theorems.���What�Brou���w�er�lfdid�ab�Gout�it,�r*w���e�will����?discuss�UUb�Gelo���w�under��Existenc��}'e���and�the�L�aws�of�L�o�gic.����N�One���can�use�Kleene's�tree�to�construct�a�recursiv���e�co�v�ering�of�the�set�of����?recursiv���e�D�reals�in�[0,1],���whic�h�has�arbitrarily�small�measure.����^��27������Hence�the�set����?of��Wrecursiv���e�reals�has�measure�zero,��Wev�en�in�constructiv�e�mathematics,��Wat�least����?with�Jathe�most�ob���vious�denition�of�measure.�P�There�is�a�direct�con�tradiction����?b�Get���w�een���the�viewp�oin���t�that�all�reals�are�recursiv�e�and�our�geometric�in�tuition,����?whic���h�:9sa�ys�the�unit�in�terv��q�al�should�ha�v�e�measure�1.�h�This�is�a�particularly�clear����?example�|�of�a�case�in�whic���h�mathematical�and�logical�w�ork�has�yielded�a�result����?of�UUphilosophical�signicance.�q�W��*�e�can't�ha���v�e�UUall�three�of�the�follo���wing:����N��UU�ev���ery�(computable)�metho�Gd�is�recursiv�e����N��UU8�x�9�y�[�A�(�x;���y��)�means�there�is�a�(computable)�metho�Gd�to�get��y��.�from��x��?��f�ff��v�	J=�����w���-:�26����LܻT��J�ec�Îhnically�,�;�the�'Ffan�theorem�is�ab�<rout�trees;�P=the�con�tin�uit�y�theorem�is�deduced�from�the��	��fan��Xtheorem�in�a�1924�pap�<rer�[�9��@],�p.�� 286.��	�>�����w���-:�27����LܻThis�7theorem�is�due�to�Lacom�Îb�<re,�@rand�w�as�pro�v�ed�indep�<renden�tly�sligh�tly�later�b�y�Zasla�vskii���and��XTseitin.�� A�pro�<rof�and�original�references�can�b�e�found�in�[�2��@],�page�69.��������15������l�����'������N��UU�The�unit�in���terv��q�al�has�measure�1����NBrou���w�er�[�w�asn't�faced�with�this�dilemma�directly����^��28���x�,��b�Gecause�to�see�this����?dilemma�Dclearly��*�,��w���e�need�the�precise�denition�of�computabilit�y��*�,��and�the�de-����?v���elopmen�t�Mof�recursiv���e�analysis.�oOur�geometric�in�tuition�w�on't�allo�w�us�to�giv�e����?the��kunit�in���terv��q�al�measure�0.��
W��*�e�ha�v�e�to�giv�e�up�either�Ch�urc�h's�thesis�(ev�ery����?computable���metho�Gd�is�recursiv���e),���or�the�fundamen�tal�constructiv�e�explanation����?of�UUexistence�as�\w���e�can�nd".����NThese�D�facts�b�Gecame�kno���wn�in�the�1950's�as�the�sub��8ject�of�recursiv�e�analysis����?w���as�S�dev�elop�Ged�(in�parallel�in�Russia�and�the�W��*�est,��Mat�that�time�on�opp�osite����?sides�A�of�the�Iron�Curtain).�k>It�seems�that�the�Mark���o�v�A�sc�ho�Gol�to�ok�the�incredible����?step��wof�accepting�that�the�unit�in���terv��q�al�has�measure�zero.�P-In�the�W��*�est,���p�Geople����?simply���con���tin�ued�to�study�formal�systems�whic�h�did�not�include�Ch�urc�h's�thesis����?as�UUan�axiom,�sa���ying�that�Ch�urc�h's�thesis�is�\problematic".����NIf��Pexistence�means�computabilit���y��*�,���there�aren't�enough�reals.�߹Brou�w�er�real-����?ized���that,���and�in���v�en�ted���c�hoice�sequences�to�pro�vide�more.�2�But�these�same�c�hoice����?sequences�Lcan�o�Gccur�in�the�un���winding�of�quan�tiers,�2�e.g.����8�n�9�mm���=���	z�(�n�),�so����?existence���m���ust�go�b�Gey�ond�computabilit�y��*�.�?�A�t�least��r��}'elative��c�omputability�,��using����?the���parameters�of�the�form���ula�as�oracles,��m�ust�b�Ge�in�v�olv�ed.�HBKleene�claried�this����?situation���in�his�b�Go�ok���[�37��
],���in�whic���h�he�rst�dened�formal�systems�using�func-����?tion��/v��q�ariables�for�c���hoice�sequences,��&and�then�ga�v�e�a�realizabilit�y�in�terpretation����?for��these�systems�using�what�w���ould�no�w�b�Ge�called�\Kleene's�second�mo�del�of����?the�Əlam���b�Gda-calculus".��vThis�notion�of�realizabilit�y�sho�w�ed�that,���in�spite�of�the����?presence��of�axioms�ab�Gout�c���hoice�sequences�that�con�tradict�Ch�urc�h's�thesis,�J�it����?is�N�still�true�that�when��8�x�9�y�[�A�(�x;���y��)�N�is�pro���v��q�able,�Pthen��y����can�b�Ge�found�recursiv�ely����?from�UU�x�.���6���?�3.4��]�Set��Theory��uT��?�Ideas�}�ab�Gout�the�existence�of�sets�precede�and�guide�the�form���ulation�of�axioms.����?In���turn,��Qthe�axioms�and�theorems�pro���v�ed���from�them�and�ab�Gout�them,�help�to����?clarify�I6our�notions�ab�Gout�the�existence�of�sets.�m�A���t�the�b�eginning�of�the�cen���tury����?w���e�T1had�only�informal�w�ork�on�set�theory��*�,�Tkb�y�Dedekind�and�Can�tor,�Tkand�formal����?but��vinconsisten���t�w�ork�b�y�F��*�rege.�L((Russell�used�Russell's�parado�x�to�sho�w�F��*�rege's����?system�UUinconsisten���t�in�1903.)����NRussell�3
and�Whitehead's�axiomatization�in��Principia��w���as�based�on�an�anal-����?ysis��of�existence.��
Classes�exist;���to�b�Ge�giv���en�a�class��A�~�=��f�x��:���(�x�)�g�,�އw�e��need�a����?prop�Gositional���function���,�� whic���h�pro�duces�a�prop�osition���(�x�)�when�giv���en�an��x�.����?But���functions�ha���v�e���to�b�Ge�dened�on�classes,�зso�w���e�need�to�kno�w�that�the�domain����?of�$���exists,�W�b�Gefore�the�denition�of��A��mak���es�sense.���Therefore,�w���e�need�some����?classes���to�get�started�with,��[and�for�this�Russell�used�the��typ��}'es�;��Zessen���tially�one�it-����?erates�the�p�Go���w�er�set�to�get�the�t���yp�es.�U�Ev�en�tually��*�,��Russell�and�Whitehead�found��?�X-�ff��v�	J=�����w���-:�28����LܻHo�Îw�ev�er,�1in��an�article�written�in�1952�(p.��509�of�[�9��@]),�he�clearly�sho�Îws�a�w�areness�of�the��	��dicult�Îy��Xthat�a�system�of�la�wlik�e�n�um�b�<rers�will�ha�v�e�measure�zero.��������16�����l�����'������?�themselv���es�\�forced�to�assume�another�axiom,���the�axiom�of�reducibilit�y��*�,���whic�h����?they���could�not�justify��*�,���so�the�exercise�w���as�a�failure�in�explaining�the�existence����?of�UUsets,�though�it�ma���y�ha�v�e�b�Geen�successful�on�other�fron�ts.����NIt��5w���as�against�this�bac�kground�that�Brou�w�er�form�ulated�his�theory�of�sets,����?whic���h��/he�called�\sp�Gecies".�aWThese�sets����^��29����w�ere�at�rst�essen�tially�determined�b�y����?functions��
dened�on��N����^��N���̲,��but�later�sets�of�sets�of�sets�w���ere�also�allo�w�ed.���Y��*�ou����?could��b�Ge�giv���en�a�sp�ecies�essen���tially�b�y�the�separation�axiom,���as�the�set�of�p�Goin�ts����?satisfying�3oa�giv���en�prop�Gert�y��*�,�:7but�there�w�asn't�m�uc�h�y�ou�could��do��with�a�sp�Gecies,����?except�(�use�it�to�dene�another�sp�Gecies.�b�In�particular,�1�there�w���as�no�go�o�d�w���a�y�(�to����?dene��a�n���um�b�Ger��set-theoretically��*�,��4since�y���ou�couldn't�get�a�n�um�b�Ger�b�y�dening����?(for���instance)�a�Dedekind�cut�as�a�sp�Gecies.����^��30���ǁ�Brou���w�er's���notion�of�set�is�coheren���t����?but�UUimp�Goten���t.����NZermelo���and�F��*�raenk���el�ga�v�e�an�axiomatization�of�set�theory�that�ev�en�tually����?b�Gecame���widely�used.�~�In�essence�their�axioms�sa���y�\what�y�ou�can�do�with�a����?set"�O�(mak���e�more�sets�according�to�these�rules.)�o�The�axioms�do�not�address�the����?question�T�of�ho���w�w�e�can�b�Ge�giv�en�a�set;�T�but�no�w�ada�ys�the�usual�justication�for����?the���axioms,��0and�the�stated�\in���tended�mo�Gdel"�of�the�axioms,�is�the��cumulative����?hier��}'ar�chy�S�of�sets�,��dened��b���y�iterating�the�p�Go�w�er�set.�Z�This�construction�dep�Gends����?on�c�a�prior�understanding�of��or��}'dinal�,��Gso�w���e�ha�v�e�to�ask�ho�w�w�e�are�giv�en�an����?ordinal.��vSince�y:an�ordinal�is�a�transitiv���e�set�linearly�ordered�b�y��2�,��3according�to����?v���on��Neumann,�(�w�e�ha�v�e�to�kno�w�ho�w�w�e�are�giv�en�a�set�b�Gefore�w�e�kno�w�ho�w�w�e����?are���giv���en�an�ordinal;��and�w�e�ha�v�e�to�kno�w�ho�w�w�e�are�giv�en�an�ordinal�b�Gefore�w�e����?can���construct�the�cum���ulativ�e���hierarc�h�y;��Jand�w�e�ha�v�e�to�do�that�b�Gefore�w�e�can����?kno���w��ho�w�w�e�are�giv�en�a�set.� �So�the�notion�of�existence�underlying�Zermelo-����?F��*�raenk���el��+set�theory�is�circular.�NdBut�as�I��said,�gw�e�don't�discuss�this�at�the�dinner����?table.�X�Presumably��Jthe�circle�is�not�vicious:��it�just�means�that�there�is�some��a����?priori�UU�notion�of�set�or�ordinal�whic���h�is�required�rst.����NThat�Ěb�Geing�the�case,��lQuine�set�out�to�giv���e�a�new�foundation�to�set�theory����?with�N:his�axiomatization�NF�M�(for�New�F��*�oundations).�\vThis�theory�is�b�Geautiful����?in���its�simplicit���y:��
it�has�only�t�w�o�axioms,��extensionalit�y�and�the�axiom�that����?�f�x�w��:���(�x�)�g��i�exists�whenev���er����is�a��str��}'atie�d��i�form�ula.�	IStratied�means�that����?in���tegers�%R(called�\lev�els")�can�b�Ge�assigned�to�all�the�set�v��q�ariables�in����in�suc�h����?a��jw���a�y�that�when��a���2��b��o�Gccurs�in���,�c�then��a��is�assigned�a�lo���w�er�lev�el�than����?�b�.�5The��pRussell�form���ula��x��C�62��x��p�ob�viously�is�not�stratied,��so�the�Russell�set����?can't�ޞb�Ge�directly�dened�in�NF.�Whatev���er�the�b�eauties�of�NF,�there�seems�to����?b�Ge�e�no�underlying�\in���tended�mo�del",���other�than�p�erhaps�the�true�univ���erse�of����?sets,�|and�Anob�Go�dy�has�ev���er�b�een�able�to�dene�a�mo�del,�|so�it�is�still�an�op�en����?problem�T�whether�NF�is�consisten���t.�q�The�axioms�are�uninformativ�e�for�problems����?of�v�existence:��!they�don't�tell�us�an���ything�ab�Gout�ho�w�w�e�can�b�Ge�giv�en�a�set.��OSee��?�X-�ff��v�	J=�����w���-:�29����LܻBrou�Îw�er�PEused�the�w�Îord�\sp�<recies"�in�his�review�of�F��J�raenk�el's�b�<ro�ok�PEon�set�theory�([�9��@],�op.��	��441),��so��pclearly�not�to�<ro�m�Îuc�h��psignicance�should�b�e�placed�on�the�dierence,��if�an�Îy��J�,�b�<ret�w�een���a��X\set"�and�a�\sp�<recies".��	�>�����w���-:�30����LܻOf��ecourse,��0y�Îou�can��dene��Dedekind�cuts,�but�y�Îou�can't�pro�v�e�they�ha�v�e�the�prop�<rerties�of���the��Xreal�n�Îum�b�<rers.��������17�������l�����'������?�[�22��
]�UUfor�more�information�ab�Gout�NF.����NAnother,�޿and��m���uc�h�more�recen�t,�޿fora�y�in�axiomatic�set�theory�is�Aczel's����?non-w���ell-founded�Z>sets�[�1��].���As�usual�when�a�new�mathematical�concept�is�in�tro-����?duced,�\�most�[wp�Geople�regard�non-w���ell-founded�sets�as�con�v�enien�t�men�tal�ctions,����?as���opp�Gosed�to�the�\real"�sets,�Qwhic���h�are�mem�b�Gers�of�the�cum�ulativ�e�hierar-����?c���h�y��*�.�fRemem�b�Gering�8�that�complex�n���um�b�ers,�q�the�8�gra�vitational�eld,�q�and�atoms����?w���ere�_once�so�regarded,�Jap�Gerhaps�w�e�should�pause.���What�do�w�e�ha�v�e�to�do�to����?construct��Ha�non-w���ell-founded�set?�:�Ho�w�can�w�e�b�Ge�giv�en�a�non-w�ell-founded����?set?���If���these�questions�can�b�Ge�answ���ered,��Mthen�they�exist,�in�the�same�sense�as����?w���ell-founded�UUsets.���6���?�3.5��]�Sets,��Classes,�and�Prop�`erties��uT��?�A���t�e/the�b�Geginning�of�the�cen�tury�the�Russell�parado�x�had�just�sho�wn�the�necessit�y����?of��distinguishing�carefully�b�Get���w�een��a�prop�ert���y�(suc�h�as��x���62��x�)��and�its�extension����?(the�
�set�or�class�consisting�of�things�ha���ving�the�prop�Gert�y).�X�Clearly�it�w�as�partly����?b�Gecause���of�these�diculties�that�Brou���w�er���w�as�led�to�dev�elop�his�in�tuitionistic����?philosoph���y��*�,��with�m	its�leading�ideas�of�men�talism�and�constructiv�e�logic.�$YThe�exact����?relation�b�b�Get���w�een�sets,�fRprop�Gerties,�and�the�extension�of�prop�Gerties�w���as�not�clear����?at�-that�time.�xQIf�mem���b�Gership�in�a�set�is�considered�a�prop�ert���y��*�,�-cthen�ev�ery�set����?is�"9the�extension�of�a�prop�Gert���y;���but�the�Russell�parado�x�sho�ws�that�not�ev�ery����?prop�Gert���y��has�an�extension,�O�or�p�erhaps�it�sho���ws�that�not�ev�ery�extension�of�a����?prop�Gert���y�y�is�a�set.�(�F��*�or�instance,���the�prop�ert���y�of�b�eing�a�set�has�the�class�of�all�sets����?for�Ӈits�extension,��}but�this�class�is�not�a�set,�or�p�Gerhaps�it�has�no�extension�at�all.����?By�c�mid-cen���tury�these�matters�had�b�Geen�claried�considerably��*�,�g|b�y�means�of�the����?denition�w�and�study�of�the�cum���ulativ�e�w�hierarc�h�y�of�sets,���the�dev�elopmen�t�of�b�Goth����?ZF�~�set�~�theory�and�GB�class/set�theory��*�,��4and�the�understanding�of�the�relations����?b�Get���w�een��them.�"In�particular,���the�understanding�that�the�cum���ulativ�e��hierarc�h�y����?can�T�b�Ge�cut�o�at�an���y�inaccessible�cardinal,�T�pro�ducing�a�mo�del�of�ZF�whic���h�can����?b�Ge�%6extended�to�a�mo�del�of�GB�%*b���y�considering�its�p�o���w�er�%6set�to�dene�the�classes����?of�qthe�mo�Gdel,�xw���en�t�qfar�to���w�ards�qdetermining�the�presen���t-da�y�qphilosophical�view����?of��the�mathematical�comm���unit�y��*�,�whic�h��amoun�ts�to�this:��rw�e'll�use�as�m�uc�h�of����?the��qcum���ulativ�e�hierarc�h�y�as�w�e�need,���and�w�e�usually�don't�need�m�uc�h�of�it,���so����?w���e��w�on't�w�orry�either�ab�Gout�just�ho�w�m�uc�h�more�of�the�cum�ulativ�e�hierarc�h�y����?there��sis,�)�or�ab�Gout�whether�there�are�an���y�sets�not�in�the�cum�ulativ�e�hierarc�h�y����?at��Nall.�.�This�comfortable�conclusion�seems�to�ha���v�e��Nsettled�the�matter�in�the����?minds��zof�most�mathematicians,���lea���ving�only�a�few�philosophers,�logicians,�and����?set�UUtheorists�to�p�Gok���e�around�the�edges.����NIn��fthe�sixties,�ΫP���aul�Cohen�in�tro�Gduced�the�tec�hnique�of�\forcing",�Ϋand�used����?it�;�to�sho���w�that�the�axiom�of�c�hoice�and�the�con�tin�uum�h�yp�Gothesis�are�indep�en-����?den���t��#of�the�Zermelo-F��*�raenk�el�axioms�[�12��
].�H�Other�set�theorists�follo�w�ed�this�lead,����?and���h���undreds�of�long-standing�questions�w�ere�\settled"�in�the�same�fashion.����?The�Mtec���hnique�extends�nicely�to�all�kno�wn�plausible�axiom�systems.���Some�of����?these��dresults�impro���v�ed��dour�grasp�of�the�facts�relev��q�an���t�to�questions�of�realit�y�and�������18�����l�����'������?�truth,���for��example,�the�question�of�whether�eac���h�set�is�denable.�.Mo�Gdels�w�ere����?constructed���in�whic���h�ev�ery�set�is�denable�(from�ordinal�parameters{of�course����?parameters�
jare�necessary�since�there�are�only�coun���tably�man�y�parameter-free����?denitions).���Other�-�mo�Gdels�w���ere�constructed�in�whic�h�there�are�real�n�um�b�Gers����?whic���h�
are�not�so�denable.�X�The�conclusion�is�that�the�kno�wn�axioms�of�set�the-����?ory�|do�not�settle�that�matter.��FThe�geometric�con���tin�uum,��Iaccording�|to�mo�Gdern����?set�¬theory��*�,�con���tains�a�\sea"�of�reals�whose�simplest�denitions�are�arbitrar-����?ily���complicated�(e.g.�S�constructed�as�generic�o���v�er���the�cum���ulativ�e���hierarc�h�y�up����?to��larger�and�larger�inaccessibles).��]It�is�tempting�to�iden���tify�these�reals�with����?Brou���w�er's��5c�hoice�sequences:�Èfrom�within�a�giv�en�lev�el�of�the�cum�ulativ�e�hier-����?arc���h�y��*�,�R�they��app�Gear�to�b�e�generated�b���y�random�c�hoices.�цThe�non-randomness����?in�ptheir�denition�is�only�apparen���t�on�a�higher�lev�el.��Indeed,�I7formal�theories����?of�`�la���wless�sequences�turn�out�to�ha�v�e�mo�Gdels�constructed�b�y�forcing,�c�so�precise����?logical�UUresearc���h�b�Gears�out�this�in�tuition.����NOne���result�deserv���es�particular�men�tion�here:�6FF��*�eferman�[�17��
]�pro�v�ed�that�no����?set-theoretically�ϡdenable�w���ell-ordering�of�the�con�tin�uum�can�b�Ge�pro�v�ed�to�exist����?from��the�Zermelo-F��*�raenk���el�axioms,���ev�en�with�the�aid�of�the�axiom�of�c�hoice�and����?the���generalized�con���tin�uum���h�yp�Gothesis.�9�This�conrms�our�in�tuition�that�w�e�can't����?nd�UUsuc���h�a�thing,�in�spite�of�Zermelo's�theorem.���6���?�3.6��]�Predicativit��y��uT��?�Russell�@@and�Whitehead�in���tro�Gduced�a�\vicious�circle�principle"�[�51��
],�z�p.�2�37,�to����?prev���en�t���the�parado���xes.���They�sa�y�\the�vicious�circles�in�question�arise�from����?supp�Gosing���that�a�collection�of�ob��8jects�ma���y�con�tain�mem�b�Gers�whic�h�can�only����?b�Ge�mdened�b���y�means�of�the�collection�as�a�whole."��The�\principle"�is�simply����?that��suc���h�collections�\ha�v�e�no�total",�COthat�is,�do�not�exist.���On�the�Platonist����?view,�Gthis�C�principle�is�nonsensical{wh���y�should�the�p�Gossible�denitions�of�an�ob-����?ject��&aect�its�existence?�;On�the�other�hand,��Zif�w���e�tak�e�it�seriously��*�,��Zw�e�ha�v�e�to����?outla���w��denitions�that��might��violate�the�principle,�H"for�example�all�denitions����?of�Q�a�set�of�in���tegers�that�in�v�olv�e�quan�tication�o�v�er�all�sets�of�in�tegers�in�the����?denition.�aTSuc���h�#�denitions�are�called��impr��}'e�dic�ative�.�Questions�#�ab�Gout�the�v��q�alid-����?it���y�5of�impredicativ�e�denitions�w�ere�raised�b�y�W��*�eyl�and�P�oincar���Ge�long�b�Gefore����?the�\Qparado���xes�and��Principia�.���In�the�t�w�en�tieth�cen�tury�a�thorough�metamathe-����?matical�1,study�of�the�implications�of�the�predicativ���e�p�Gosition�w�as�carried�out�b�y����?F��*�eferman�_�[�16��
],�b�[�18��],�[�19��].��gThe�_�upshot�of�these�studies�is�a�clear�iden���tication�of����?the�@�p�Gortion�of�classical�mathematics�that�can�b�e�pro���v�ed�@�using�predicativ���e�prin-����?ciples,�=�and�7�a�precise�iden���tication�of�the�pro�Gof-theoretic�strength�of�predicativ�e����?formal�@>systems.�j�Space�do�Ges�not�p�ermit�a�thorough�discussion�of�the�issues,�Dvbut����?see�UUthe�pap�Gers�of�F��*�eferman�just�cited.�������19����4�l�����'�������?�4��WL�Realit���y���f,�ffT�ruth,�and�Set�Theory�����?�G����odel's�[�w���ork�sho�w�ed�us�that�existence�at�higher�t�yp�Ges�is�connected�to�truth����?at�~>lo���w�er�t�yp�Ges,��xsomething�w�e�migh�t�otherwise�nev�er�ha�v�e�susp�Gected.��But�if�w�e����?use�r�higher�t���yp�Ges,�z1w�e�r�can�dene�truth�predicates�for�lo���w�er�r�t�yp�es,�z1and�r�pro�v�e�the����?consistency�0of�theories�men���tioning�only�lo�w�er�t�yp�Ges,�f�so�that�w�e�obtain�corre-����?sp�Gonding�1�to�a�sequence��T����n����of�theories�of�higher�and�higher�t���yp�es,�8�a�sequence�of����?purely�$�arithmetical�facts��A����n��q~�,�.teac���h�one�pro�v��q�able�in��T����n���:�but�not�in��T����m��
�W�for��m��<�n�.����?This��Ais�a�philosophically�imp�Gortan���t�result,��since�it�sho�ws,��or�at�least�suggests,����?the���futilit���y�of�the�non-reist�p�Gosition.��kIf�w�e�giv�e�up�sets�of�sets�of�in�tegers,����?w���e're�9lgiving�up�hop�Ge�of�settling�certain�questions�ab�out�in���tegers�alone.�This����?refutation���of�the�non-reist�p�Gosition�is�not�airtigh���t,��ho�w�ev�er,�since���w�e�ha�v�en't����?sho���wn�� that�higher�t�yp�Ges�are�the��only��w�a�y�to�pro�v�e�these�arithmetical�facts;��only����?that�E�they��do��pro���v�e�E�more�arithmetical�facts.�l�W��*�e�will�return�to�this�p�Goin���t�b�elo���w.����NEv���en���ZF,�or�an�y�true�axiomatization,�Üwill�still�lea�v�e�y�ou�only�a�recursiv�ely����?en���umerable�
/set�of�theorems,�7fwhile�after�T��*�arski�w�e�kno�w�that�the�truth�set�of����?arithmetic�/is�m���uc�h�/more�complex.�]�T��*�uring�[�60��
],�%7and�later�F�eferman�[�15��
],�%7studied����?transnite��[progressions�of�axiomatizations,���eac���h�obtained�in�a�natural�w�a�y�b�y����?\re
ection���principles"�asserting�the�soundness�of�the�previous�theories.�hIn�this����?w���a�y�}�w�e�can�get�w�ell�b�Gey�ond�a�recursiv�ely�en�umerable�set�of�theorems;��(so�more����?realit���y��can�buy�more�truth�in�this�sense�also:��kno�wledge�of�the�existence�(of�w�ell-����?orderings)��3enables�the�construction�of�longer�progressions�of�axiomatizations����?based�UUon�the�re
ection�principle,�and�hence�to�more�arithmetic�truths.����NSince�yG����odel�emphasized�the�p�Goin���t,���the�b�elief�that�realit���y�at�higher�t�yp�Ges����?is��directly�related�to�truth�ab�Gout�the�in���tegers�has�b�een�widespread�among�lo-����?gicians.��Let�5�me�therefore�presen���t�some�evidence�to�the�con�trary��*�.��A�5Ytheorem����?of�ܷKreisel,���Sho�Geneld,�and�W��*�ang�[�40��
]�states�that�ev���ery�true�sen�tence�of�arith-����?metic�E�is�deriv��q�able�in�P��*�A�E�b���y�transnite�induction�on�some�primitiv�e�recursiv�e����?w���ell-ordering��nof�the�in�tegers����^��31���x�.��So,��tif�w�e�w�an�t�to�kno�w�the�whole�truth�ab�Gout����?arithmetic,��*it�j�w���ould�b�Ge�enough�to�p�ossess�the�abilit���y�to�recognize�of�sp�ecic����?primitiv���e��recursiv�e�linear�orderings�whether�they�are,��or�are�not,�w���ell-orderings.����?It�?'follo���ws�from�a�\basis�theorem"�due�to�Kreisel�[�53��
],�v�page�187,�that�an���y�primitiv�e����?recursiv���e���ordering�whic�h�is�not�a�w�ell-ordering�has�a����^���0��l�2���N6�descending�sequence.����?That�}Fis,��Bthe�descending�sequence�is�itself�arithmetically�denable,�and�using����?only���t���w�o�quan�tiers.�tTherefore,���esoteric�sets�lik�e�inaccessible�cardinals�are��not����?�after�Qall�necessary�for�arithmetic�truth.���W��*�ell-orderings�seem�to�b�Ge�imp�ortan���t����?but�UUit�w���ould�b�Ge�enough�to�kno�w�all�ab�Gout��c��}'ountable��w�ell-orderings.����^��32����?�X-�ff��v�	J=�����w���-:�31����LܻF��J�or���those�who�need�a�clear�statemen�Ît�of�the�theorem:���If��R��Q�is�a�primitiv�e�recursiv�e�binary��	��relation,�	�pro�Îv��able��Ain�P��J�A��to�b�<re�a�linear�ordering�on�the�in�tegers,�	�the�sc�hema��T�.:I��һ(�R���)�is�the���collection��Xof�all�form�Îulae�of�the�form�����nb��8�y�I{�(�8�xR��y�A�(�x�)�\t�!��A�(�y�I{�))��!�8�z�V�A�(�z��)���	�>�����w���-:�32����LܻThis�7is�recursion-theoretically�sensible�as�the�truth�set�for�arithmetic�is����-:��1��wq�1���*��,�7�while�the�set�of��������20����-��l�����'������N�A�	philosophically�imp�Gortan���t�p�oin���t�ab�out�the�theorem�of�Kreisel,�#�Sho�eneld,����?and���W��*�ang�is�that�the�reduction�of�arithmetic�truth�to�transnite�induction�on����?primitiv���e�6recursiv�e�w�ell-orderings�is�not�sub��8ject�to�the�argumen�t�giv�en�ab�Go�v�e����?for���higher-t���yp�Ge�sets,�that�p�erhaps�there�are�other�w���a�ys���to�establish�arithmetical����?truth.��DIndeed,��3the���statemen���ts�of�transnite�induction�on�primitiv�e�recursiv�e����?w���ell-orderings��are�themselv�es�arithmetical,���so�they�are�among�the�truths�to�b�Ge����?established.�8�Therefore,���there���can't�b�Ge�an���y�easier�w�a�y�to�nd�out�the�truth�ab�Gout����?arithmetic:�,�ho���w�ev�er���w�e�pro�Gceed,�
$w�e�will�still�ha�v�e�to�settle�whic�h�primitiv�e����?recursiv���e�UUorderings�are�w�ell-orderings.����NThere��Jis�a�tec���hnical�problem�of�in�terest�here:�Y�What�if�w�e�use�in�tuition-����?istic���arithmetic�instead�of�classical?���W��*�e�can�form���ulate�the�conjectures�that����?ev���ery��constructiv�ely�true�sen�tence�of�arithmetic�is�deriv��q�able�b�y�transnite�in-����?duction�on�some�w���ell-founded�primitiv�e�recursiv�e�ordering.��This�statemen�t�is����?expressible�rin�v��q�arious�in���tuitionistic�theories,�N9but�I�?b�Geliev�e�not�m�uc�h�is�kno�wn����?ab�Gout���its�metamathematical�status,�"�except�of�course�that�(restricted�to�nega-����?tiv���e�]arithmetical�statemen�ts)�it�is�consisten�t�with�theories�that�ha�v�e�a�classical����?in���terpretation.�T�But��for�example,��is�it�consisten�t�with�in�tuitionistic�analysis�plus����?Ch���urc�h's�UUthesis?����^��33����!�?�5��WL�Existence�ffand�Ph���ysics�����?�Although�x�the�primary�purp�Gose�of�this�article�is�to�discuss�realit���y�(existence)����?and��_truth�in�mathematics,���I��Pb�Geliev���e�that�the�fundamen�tal�issues�of�realit�y�and����?truth���naturally�extend�b�Gey���ond�the�b�ounds�of�mathematics,�
and�that�funda-����?men���tal���insigh�ts�ma�y�b�Ge�obtained�b�y�considering�truth�and�realit�y�in�ph�ysics����?as��jw���ell,��if�for�no�other�reason�than�to�see�what�the�dierences�in�the�analysis����?ma���y��$b�Ge.�4Moreo�v�er,���some�concepts,�suc���h�as�space,�b�Gelong�to�b�oth�mathematics����?and���ph���ysics.�J�Finally��*�,���I���will�put�forw�ard�the�view�that�quan�tum�mec�hanics�has����?essen���tial�UUimplications�for�the�foundations�of�mathematics.���6���?�5.1��]�The��existence�of�space��uT��?�Kan���t��though�t�the�nature�of�space�inheren�t�in�the�nature�of�mind,�but�he�w�as����?pro���v�ed�qVwrong.���In�fact,��Vhe�w���as�pro�v�ed�wrong�sev�eral�times.���First,��Vin�regard����?to��mathematical�space,��the�dev���elopmen�t��of�non-Euclidean�geometry�shattered����?Kan���t's��p�Gosition.�	4�But�a�Kan�tian�could�still�retrenc�h,�Q'b�y�claiming�that�non-����?Euclidean���spaces�w���ere�gmen�ts�of�the�imagination,�ȧwhile�real�(ph�ysical)�space����?w���as�`�Euclidean,�c�and�its�nature�w�as�inheren�t�in�the�mind.��-This�holdout�p�Gosition��?�X-�ff��v�	J=��primitiv�Îe��xrecursiv�e�w�ell-orderings�is����-:��1��wq�1���*��,�ȥi.e.���denable�using�one�univ�ersal�function�quan�tier.��	��A��Xset�is����-:��1��wq�1������if�b�<roth�it�and�its�complemen�Ît�are����-:��1��wq�1���*��.��	�>�����w���-:�33����LܻOne��Acan�also�ask�ab�<rout�the�consistency�with�Ch�Îurc�h's��Athesis�and�in�Îtuitionistic�second-���order��`arithmetic�of�the�statemen�Ît�that�the�true�sen�tences�of�arithmetic�are�those�deriv��able���using�l}the�recursiv�Îe��!�I{�-rule.���The�classical�truth�of�this�statemen�t�is�a�theorem�of�Sho�<reneld���[�54���].��������21����>��l�����'������?�w���as�
seriously�set�bac�k�b�y�Einstein's�sp�Gecial�theory�of�relativit�y��*�,�0�whic�h�unied����?space��Oand�time,���and�dealt�the�death�blo���w�b�y�Einstein's�theory�of�general�rela-����?tivit���y�UUin�1915.����^��34������N�Can�a]w���e�sa�y�that�space�exists?� uGeneral�relativit�y�sa�ys�that�mass-energy�struc-����?tures�ȯspace.���The�mathematical�description�in���v�olv�es�ȯassigning�15�n���um�b�Gers�ȯ�T���������?�to���ev���ery�\p�Goin�t"����^��35���x�.��This�assignmen�t�of�n�um�b�Gers�is�called�the�\energy�tensor."����?It���seems�that�in�general�relativit���y��*�,���space�is��c��}'ompletely��describ�Ged�b�y�the�energy����?tensor,�+
so� {w���e�ha�v�e�to�ask�whether�p�Gerhaps�space��is��just�the�energy�tensor.�`)Cer-����?tainly�_when�w���orking�with�relativit�y��*�,�2�w�e�are�\giv�en"�a�space�b�y�b�Geing�\giv�en"����?an�p^energy�tensor.���But�the�answ���er�to�this�question�is�certainly�\no,�� space�is����?not�R�just�the�energy�tensor."�p�The�reason�is�that�w���e��know��that�general�relativit�y����?is���not�a�complete�description�of�real,��Tph���ysical�space.���The�energy�tensor�de-����?scrib�Ges�j�energy�b���y�assigning�n�um�b�Gers�to�ev�ery�p�Goin�t.���In�realit�y�energy�comes�in����?quan���tum-mec�hanical�@�swirls�rather�than�precise�v��q�alues�at�ev���ery�p�Goin�t.�3�So�the����?question��of�the�nature,�%wand�p�Gerhaps�ev���en�the�existence,�of�ph���ysical�space�is�still����?an��op�Gen�question.��!This�seems�to�b�e,���ho���w�ev�er,�a���scientic��rather�than�a��philo-����?sophic��}'al�XȲquestion;�Z�or�at�least,�Y�w���e�can�exp�Gect�that�the�philosophical�asp�ects�will����?b�Gecome��1m���uc�h�clearer�once�a�suitable�scien�tic�theory�is�found�that�encompasses����?b�Goth�UUrelativit���y�and�quan�tum�theory��*�.����NCurren���t��theories�of�quan�tum�gra�vit�y�require�radical�revisions�in�our�concep-����?tualization�K:of�space,���requiring�the�use�of�spaces�of�e.g.�Sx23�dimensions,�all�of����?whic���h�[except�the�usual�four�time�and�space�dimensions�are�\rolled�up"�to�a�cir-����?cle��Aof�v���ery�small�diameter,���so�w�e�don't�p�Gerceiv�e�them.�C�The�consensus�is�that�the����?\nal"�ʥtheory�will�in���v�olv�e�ʥquan�tum�
uctuations�in�the�top�Gology�of�space-time����?itself.��Clearly��philosophical�ideas�ab�Gout�the�nature�of�space�will�b�e�struggling����?to�UUcatc���h�up�to�science�for�some�time�to�come.���6���?�5.2��]�Quan��tum��mec�hanics�and�virtual�existence��uT��?�In��@an�exp�Gerimen���t�often�used�to�illustrate�quan�tum�mec�hanics,�S;electrons�are����?\red"��at�a�target�through�a�diraction�grating.�s`The�diraction�grating�can,����?at�C�least�in�principle,�@b�Ge�reduced�to�t���w�o�C�small,�closely�spaced�holes�in�a�solid����?plate.�WThe���electron�m���ust�(at�least�it�seems�it�m�ust)�pass�through�one�of�these����?holes��to�reac���h�the�target.�'W��*�e�re�a�lot�of�electrons�and�the�target�measuring����?device��rrecords�where�they�land.�. Instead�of�t���w�o��rpiles�of�electrons,�zin���terference����?patterns��
(suc���h�as�y�ou�w�ould�exp�Gect�to�see�when�using�ligh�t�instead�of�electrons)����?are�\observ���ed,��unless�w�e�measure�carefully�whic�h�hole�the�electron�go�Ges�through.����?If��w���e�do�lo�Gok�carefully�at�whic�h�hole�is�used,�Sthe�electrons�b�Geha�v�e�lik�e�particles.��?�X-�ff��v�	J=�����w���-:�34����LܻA�Vmore�V5generous�view�of�Kan�Ît�migh�t�hold�that�he�only�though�t�that�the�nature�of�our��	��ev�Îeryda�y�*limmediate�exp�<rerience�of�space�(and�medium-sized�ph�Îysical�ob�x�jects)�is�inheren�t�in�the���nature��of�mind.�II��think�mo�<rdern�kno�Îwledge�of�the�eects�of�neurological�damage�and�disease���dispro�Îv�es��Xev�en�this�claim,�but�the�details�are�not�relev��an�t�to�the�topic�of�this�pap�<rer.��	�>�����w���-:�35����LܻThe�Vop�<roin�Îts�in�question�are�p�oin�Îts�in�a�mathematical�space,�o�not�in�the�ph�ysical�space�itself,���so��Xthere�is�no�philosophical�circularit�Îy�here.��������22����M��l�����'������?�The��qupshot�is�that�(when�w���e�see�the�in�terference�pattern)�w�e�can't�kno�w�the����?electron's��0p�Gosition�w���ell�enough�to�sa�y�whic�h�hole�it�w�en�t�through,��gand�indeed����?ma���yb�Ge�Q�it�didn't�really�go�through�either�hole,�Rnbut�someho�w�w�en�t�through�b�Goth����?holes�?sim���ultaneously��*�.���It's�ev�en�p�Gossible�that�it�didn't�go�through��either��hole,����?but��Tinstead�\tunneled"�through.�j�(This�happ�Gens�in�ev���ery�transistor,��it's�not����?an��>esoteric�o�Gccurrence.�B�The�fact�that�y���our�radio�pla�ys�m�usic�dep�Gends�on�this����?quan���tum���b�Geha�vior�of�electrons.)���In�other�w�ords,�װit's�not�merely�that�w�e�can't����?�know��ݲits�p�Gosition,��@it�simply�is�false�that�the�electron�at�all�times��has��a�unique����?p�Gosition���and�v���elo�cit�y��*�.�?�The���uncertain�t�y�principle�is�not�a�matter�of�epistemology��*�,����?but�UUof�on���tology��*�.����NDo�Ges� �the�electron�exist�while�it's�tunneling?��4or�do�es�it�
ash�momen���tarily����?out�UUof�existence�and�then�bac���k�in�to�existence�again?����NThese�pWquestions�turn�up�again�in�quan���tum�electro�Gdynamics�and�in�subatomic����?ph���ysics,�U�where�"jin�teractions�b�Get�w�een�photons�and�electrons,�U�or�more�generally����?b�Get���w�een��an�y�t�w�o�particles,��are�calculated�b�y�summing�o�v�er�man�y�(usually�in-����?nitely�man���y)�p�Gossible�in�termediate�states.���These�in�termediate�states�cannot����?b�Ge��observ���ed,��but�in�those�states�neither�the�original�nor�the�nal�particles�exist,����?and�psome�other�particles�ma���y�exist�that�didn't�exist�b�Gefore�and�w�on't�exist�af-����?ter�vthe�in���teraction.���These�are�called�\virtual"�particles.�They�ha���v�e�va�nebulous����?status:���they��don't�\really�exist"�in�the�sense�that�they�are�unobserv��q�able{y���ou����?are��nev���er�\giv�en"�a�virtual�particle.��Moreo�v�er,���all�innitely�man�y�of�these�in-����?termediate�ڰstates�\happ�Gen"�at�the�same�time,��and�mak���e�their�con�tribution�to����?the�-2nal�result.�dgSo�there�is�an�observ��q�able�result;�:�it�is�therefore�hard�to�sa���y�that����?the�virtual�particles�\don't�exist"{�ho���w�could�a�non-existen�t�particle�pro�Gduce����?an�UUobserv��q�able�result?����NIf�j�some�mathematicians�are�tempted�to�sw���eep�these�diculties�under�the�car-����?p�Get�Maas�irrelev��q�an���t�to�mathematics,�N�the�new�eld�of��quantum���c��}'omputation��should����?put�	xa�stop�to�that.��0Algorithms�ha���v�e�	xb�Geen�dev���elop�ed�to�sp�eed�up�v��q�arious�dif-����?cult�˞computations�b���y�dividing�the�computations�in�to�parts�that�can�b�Ge�done����?sim���ultaneously��*�.�f(This��4is�called�\parallel�computation".)�In�quan���tum�comput-����?ing,�Authe�<}idea�is�to�get�parallel�computations�p�Gerformed�b���y�the�virtual�states,�so����?that�D�a�single�piece�of�hardw���are�can�p�Gerform�a�large�n�um�b�Ger�of�computations�at����?the�Isame�time.�YA�7quan���tum�algorithm�has�b�Geen�dev�elop�Ged�to�sp�eed�up�factoring����?in���tegers,�	�and���p�Geople�are�no�w�w�orking�to�build�mac�hines�whic�h�can�actually�exe-����?cute�
Asuc���h�algorithms,�Fusing�v�ery�thin�optical�b�Gers�capable�of�trapping�photons����?in�UUa�single�quan���tum�state.����^��36������N�The�G�philosophical�moral�is�that�the�concepts�of�existence�that�w���e�ha�v�e����?formed,��based�܌on�our�exp�Gerience�with�the�w���orld�of�tables,�c���hairs�and�b�Geer�m�ugs,����?and�5�ev���en�on�our�exp�Gerience�with�n�um�b�Gers�and�simple�sets,�n"are�clearly�inade-����?quate�[to�deal�with�virtual�particles.��Realit���y�is�just�more�complicated.�There�is����?another��kind�of�existence,�	virtual�existence,�whic���h�is�dieren�t�from�ordinary�ex-��?��Éff��v�	J=�����w���-:�36������LܻSee�a���C�scmtt8�http://p23.lanl.gov/Quantum.html��for�an�in�Îtro�<rduction�to�quan�tum�computation,�x�as��	��w�Îell��as�v�e�tutorials�and�links�to�rep�<rorts�of�curren�t�researc�h�at�Los�Alamos�National�Lab�<roratory���and��Xelsewhere.��������23����^(�l�����'������?�istence.�'�Virtual�w
ob��8jects�cannot�b�Ge�observ���ed�but�they�can�ha�v�e�observ��q�able�eects.����?One��"real�ob��8ject�(enjo���ying�the�ordinary�kind�of�existence)�giv�es�rise�to�sev�eral,����?ev���en�6�innitely�man�y��*�,�o7virtual�ob��8jects,�whic���h�coalesce�in�to�one�ordinary�ob��8ject����?again.��MAccording��to�quan���tum�electro�Gdynamics,�P�this�is�happ�ening�constan���tly{����?ev���en�zZthe�v��q�acuum�is�seething,�Ûit�is�a�sea�of�virtual�particles�constan�tly�b�Geing����?created��}and�destro���y�ed.�@@Ph�ysicists��}b�Geliev�e,�as�men�tioned�ab�Go�v�e,�that�a�correct����?theory�Mof�quan���tum�gra�vit�y�will�require�the�existence�of�virtual�space,�Dthat�is,����?in���termediate��5states�will�ha�v�e�to�b�Ge�summed�o�v�er�all�p�Gossible�higher-dimensional����?top�Gologies.���The���roadblo�c���k�to�a�successful�theory�of�quan�tum�gra�vit�y�seems�to����?b�Ge�UUthat�nob�o�dy�kno���ws�ho�w�to�form�ulate�suc�h�a�sum.����NSc���hr����odinger��sho�w�ed,� �in�a�famous�though�t-exp�Gerimen�t,� �that�virtual�existence����?cannot��+b�Ge�conned�to�subatomic�ob��8jects.�UKHe�describ�ed�an�apparatus�designed����?to�h�place�a�cat�in�a�state�of�virtual�existence.���More�precisely��*�,�m�the�initial�state�is����?a�x�liv���e�cat,��uand�there�are�t�w�o�in�termediate�virtual�states,��uliv�e�cat�and�dead�cat,����?corresp�Gonding���to�the�t���w�o���holes�in�the�t���w�o-slit���exp�erimen�t.�0=The���nal�state�will�b�e����?liv���e�g�cat�or�dead�cat,�l�but�the�apparatus�can�b�Ge�arranged�so�that�a�b�o���x�con�tains����?t���w�o�c�virtual�cats,�gone�aliv���e�and�one�dead.��kWhen�the�b�Go�x�is�op�Gened,�gthe�cat�will����?b�Ge�w�in�a�normal�state,��Teither�aliv���e�or�dead,�but�the�result�is�as�unpredictable�as����?whic���h�+�hole�the�electron�will�pass�through.�c�Mathematically��*�,�43the�t�w�o�virtual�cats����?are��describ�Ged�b���y�a�w�a�v�e�function�whic�h�is�linear�com�bination�of�the�liv�e-cat�and����?dead-cat�UUw���a�v�e�functions.����^��37����|����?�5.3��]�Quan��tum��mec�hanics�and�the�con�tin�uum��uT��?�W��*�e��ha���v�e�already�discussed�the�diculties�in�accoun�ting�for�the�\fullness"�of�the����?geometrical��fcon���tin�uum�b�y�a�\n�um�b�Ger�line"�comp�osed�of�computable�n���um�b�Gers.����?Brou���w�er�w�an�ted�to�allo�w�\c�hoice�sequences"�to�ll�up�the�con�tin�uum.��These����?w���ere��mean�t�to�allo�w�for�n�um�b�Gers�whose�appro�ximating�sequences�(decimal�ex-����?pansions,��roughly)�U�w���ere�generated�b�y�random�c�hoices�or�b�y�\free�c�hoices"�of����?a�W�conscious�b�Geing.�x�No���w,��8randomness�in�ph�ysical�pro�Gcesses�arises�either�from����?uncertain���ties��Jin�our�kno�wledge�of�the�initial�conditions,��or�from�quan�tum�uncer-����?tain���t�y��*�.���The��former�(kno���wn�as�\c�haos")�is�not�relev��q�an�t.���Only�quan�tum�uncer-����?tain���t�y�P�can�lead�to�a�truly�random�ph���ysical�pro�Gcess.����^��38�����Also�the�free�c�hoices�of�a����?conscious��Rb�Geing�m���ust,���so�far�as�mo�dern�ph���ysics�is�concerned,���in�v�olv�e�quan�tum����?mec���hanical��Kuncertain�ties.�FoSince�the�la�ws�of�classical�ph�ysics�are�deterministic,��Nif����?h���uman�g�b�Geha�viour�is��not��deterministic,�l;then�there�m�ust�b�Ge�quan�tum�systems�in����?the���brain�whic���h�propagate�quan�tum�uncertain�ties�from�the�microscopic�to�the����?macroscopic���lev���el.��See�[�46��
]�for�sp�Geculations�as�to�what�these�quan�tum�systems��?����ff��v�	J=�����w���-:�37����LܻF��J�or��\a�discussion�of�Sc�Îhr����odinger's�cat�and�other�philosophical�problems�asso�<rciated�with��	��the��Ximplications�of�quan�Îtum�mec�hanics�for�macroscopic�ob�x�jects,�see�[�25���],�pp.�� 11-15.��	�>�����w���-:�38����LܻT��J�ec�Îhnically��Vsp�<reaking,���this�is�an�unpro�v�en�claim.�1P�erhaps�there�are�classical�situations���in��whic�Îh�the�results�are�not�computable.��\In�this�connection�Kreisel�long�ago�prop�<rosed�the���problem��qwhether,��in�the�three-b�<ro�dy��qproblem�of�Newtonian�mec�Îhanics,�it�is�or�is�not�a�recur-���siv�Îely��*solv��able�problem�whether�three�spheres�with�giv�en�masses,��initial�p�<rositions,�and�initial���v�Îelo�<rcities��Xwill�ev�en�tually�collide.�� So�far�as�I�kno�w�this�problem�is�still�op�<ren.��������24����n��l�����'������?�migh���t�b�Ge.�xIn�either�case,�-Hthen,�quan�tum�mec�hanics�oers�the�p�Gossibilit�y�that����?ph���ysical��pro�Gcesses�ma�y�exist�that�migh�t�go�b�Gey�ond�the�computable.�U�It's�easy�to����?giv���e�R�an�example�of�suc�h�a�sequence:�pba�Geiger�coun�ter�measures�the�radioactiv�e����?deca���y�q�of�cesium-137�atoms;���the�next�mem�b�Ger�of�the�sequence�is�the�n�um�b�Ger����?of���seconds�un���til�the�next�deca�y��*�.�a�After�ev�ery��N�òdeca�ys�(for�some�xed��N��)�w�e����?replenish�pthe�supply�of�cesium-137,�wJso�that�the�pro�Gcess�is�theoretically�innite.����?It��Qshould�not�b�Ge�tec���hnically�dicult�to�arrange�a�mapping�of�these�sequences����?on���to��the�unit�in�terv��q�al�in�suc�h�a�w�a�y�that�an�y�real�n�um�b�Ger�in�the�unit�in�terv��q�al�is����?equally�	Plik���ely�to�b�Ge�generated.�XqThen,��since�the�measure�of�the�computable�reals����?is���zero,�*�the�probabilit���y�that�a�quan�tum�pro�Gcess�w�ould�generate�a�computable����?real��ois�zero.�wAll�those�classical�reals�that�the�set�theorists�lo���v�e��oto�classify�ma���y����?ha���v�e�UUph�ysical�realit�y�after�all.�������?�5.4��]�Quan��tum��Set�Theory?��uT��?�Those��who�ha���v�e��though�t�ab�Gout�the�matter�enough�to�realize�that�there�is�no����?essen���tial�:,dierence�in�kind�b�Get�w�een�the�existence�of�electrons�and�the�existence����?of�j�in���tegers,����^��39����D�ma�y�w�onder�then�whether�the�naiv�e�notions�of�existence,�p^deriv�ed����?from�Y�our�exp�Gerience�with�tables,�Z�c���hairs,�and�Y�b�eer�m���ugs,�Z�but�inadequate�for�the����?foundations���of�ph���ysics,���migh�t���also�b�Ge�inadequate�in�the�foundations�of�mathe-����?matics.�X:A���t��the�presen�t�time,�titling�a�pap�Ger�\Quan�tum�set�theory"�w�ould�prob-����?ably�\nbrand�the�author�as�a�crank.��But�who�kno���ws�what�the�future�ma�y�bring?����?Thirt���y�UUy�ears�ago�quan�tum�computing�w�ould�ha�v�e�sounded�equally�ridiculous.�������?�5.5��]�Information��Structures�Energy��uT��?�I�Dprop�Gose�Dthat�existence�in���v�olv�es�Denergy�structured�b���y�information.�l
In�the�case����?of��Zparticles,�Ȍthe�information�is�carried�(or�describ�Ged)�b���y�the�quan�tum�w�a�v�e�func-����?tion.�h�Mathematical���ob��8jects�are�the�extreme�case�in�whic���h�there�is�no�energy��*�,����?only��Qpure�information.�TqThe�n���um�b�Ger��Q7�consists�of�information�only��*�.�Add�a�small����?quan���tit�y��Eof�energy�(p�Gerhaps�in�the�form�of�ink�and�pap�er)�and�y���ou�can�create����?an���instance�of�the�n���umeral�`7'.�FDAccording�to�the�uncertain�t�y�principle,���an�ything����?in���v�olving�Kxenergy��E���m���ust�ha�v�e�a�nite�lifetime,�Mqat�least�����h��	x=E����.�n~Mathematical�ob-����?jects��can�b�Ge�eternal�b�ecause�their�energy�is�zero.�]GP���erhaps�the�quan�tum�v��q�acuum����?is�UUan�example�of�the�other�extreme,�energy�without�information.�� �"���?�6��WL�Existence�ffand�the�La���ws�of�Logic�����?�In��mathematics,��it�is�through�logical�reasoning�that�w���e�attempt�to�arriv�e�at�the����?truth�|ab�Gout�mathematical�ob��8jects.��<W��*�e�w���ould,�O�it�seems,�lik���e�to�separate�that��?�w��ff��v�	J=�����w���-:�39����LܻT��J�o���a�Îv�oid�confusion:���the�sen�tence�do�<res�not�sa�y�there�is�no�essen�tial�dierence�b�<ret�w�een��	��electrons�{and�in�Îtegers.���W��J�e�use�dieren�t�metho�<rds�to�in�v�estigate�the�t�w�o,�+�of�course,�and�they���ha�Îv�e��|dieren�t�prop�<rerties.���But�w�e�cannot�sa�y�that�electrons�are�\more�real"�than�in�tegers,���or���vice-v�Îersa.��������25�����>�l�����'������?�pro�Gcess���from�the�more�fundamen���tal�pro�cess�of�constructing�or�obtaining�ob��8jects����?themselv���es.��T��*�o�Y�dra�w�an�analogy��*�,��/in�mathematics�w�e�tend�to�lo�Gok�at�the�existence����?of�monk���eys�as�one�kind�of�question,�}and�the�existence�of�monk�eys�that�can�pla�y����?�Til���l�1ther��}'e�was�you�)�on�the�accordion�as�another�kind�of�question.�rCThat�is,�*�w���e����?rst�
address�the�question�of�what�the�basic�ob��8jects�allo���w�ed�
in�mathematics�are,����?and���then�the�question�of�pro���ving�the�existence�of�ob��8jects�with�certain�prop�Gerties����?is���considered.�T�But�I���main���tain�that,�din�mathematics�as�in�life,�this�is�an�articial����?distinction.�&4F��*�or��yexample,���the�question�as�to�what�sets�exist�(in�general)�is�not����?philosophically���far�remo���v�ed���from�the�question�whether�a�measurable�cardinal�(a����?certain�UUkind�of�set)�exists.����NThe�Еplan�of�rst�circumscribing�a�class�of�ob��8jects�whose�existence�is�assured,����?and�1�then�regarding�the�phrase�\there�exists"�as�a�kind�of�selection�from�this����?class,���go�Ges��bac���k�at�least�to�Russell�and�Whitehead.���When�they�in�tro�Gduce�the����?prop�Gositional�>tfunction��9��(page�17�of��Principia�,�x�v���olume�1),�they�sa���y�it�means����?\some���prop�Gositions�of�the�range�[of�prop�ositions��P�c��(�x�),�Uwhere�the�allo���w�ed���v��q�alues����?of��D�x��are�those�for�whic���h��P�c��(�x�)�is�meaningful]�are�true".�K�This�implies�that�the����?range��of�allo���w�ed��v��q�alues�is�already�determined.�[KIndeed,�^the�driving�idea�of��Prin-����?cipia����is�to�circumscrib�Ge�the�\univ���erse"�of�t�yp�Ges�once�and�for�all,��hat�the�b�eginning����?of�UUmathematics.���6���?�6.1��]�Brou��w�er��and�the�unreliabilit��y�of�logic��uT��?�Brou���w�er's��starting�p�Goin���t�w�as�that�\there�exists"�m�ust�mean�\w�e�can�construct".����?But��some�v���ery�basic�la�ws�of�logic�suce�to�pro�v�e�that��A��b�_��B�k
�is��equiv��q�alen�t�to����?asserting�)qthe�existence�of�an�in���teger��n��suc�h�that�if��n���=�0�)qthen��A�,�28and�otherwise����?�B��q�.��It�2`follo���ws�that�for�Brou�w�er,�i�to�pro�v�e��8�x�(�A�(�x�)��<�_��B��q�(�x�))�2`is�to�sho�w�ho�w�to����?determine,�o/giv���en�6еx�,�whic�h�6�alternativ�e��A�(�x�)�or��B��q�(�x�)�holds.�:No�w�in�case��B��A�is����?the���negation�of��A�,���w���e�nd�that�the�\la�w�of�the�excluded�middle",��˵A���_�:�A�,����?has���a�non-trivial�con���ten�t,��and���indeed�there�are�man���y�examples�in�whic�h�w�e����?cannot�� at�presen���t�pro�v�e��8�x�(�A�(�x�)����_�:�A�(�x�))�� in�a�st�yle�Brou�w�er�w�ould�appro�v�e.����?F��*�or�x�example,��[w���e�do�not�kno�w�whether��e���^��e���زis�rational�or�not,��[and�w�e�don't�ev�en����?ha���v�e�aan�algorithm�for�deciding�of�a�giv���en�rational��x��whether�it�equals��e���^��e��M��or�not,����?so�UUw���e�can't�pro�v�e��8�x�(�x���=��e���^��e���+�_�8�x��6�=��e���^��e��KK�),�UUwhere��x��ranges�o�v�er�rational�n�um�b�Gers.����NBrou���w�er���seized�the�bull�b���y�the�horns:�l�he�declared�the�la�w�of�the�excluded����?middle�.�to�b�Ge�\unreliable",�6�and�set�o�to�dev���elop�mathematics�without�it.�eBoth����?Brou���w�er��and�Hilb�Gert�though���t�that�this�w�ould�result�in�w�eak�er�mathematics.����?They�^�w���ould�surely�b�Goth�ha�v�e�b�Geen�surprised�b�y�the�tec�hnical�results�whic�h�G����odel����?obtained�Z|using�the�double-negation�translation.��=This�simple�translation�had�to����?a���w�ait�UUthe�dev���elopmen�t�UUb�y�Heyting�of�formal�rules�for�in�tuitionistic�logic����^��40���x�.��?�X-�ff��v�	J=�����w���-:�40����LܻBrou�Îw�er�y�w�as�an�an�ti-formalist;��he�felt�that�sym�b�<rols�w�ere�only�a�means�of�comm�unication,��	��whic�Îh��mcarry�or�suggest�meaning,��and�he�felt�that�no�system�of�formal�rules�could�encompass���mathematics.�q&G����odel's�cZincompleteness�theorem�should,���therefore,�not�cZha�Îv�e�surprised�him�at���all.�wA�Îttempting��to�v�erify�this�prediction,��I��searc�hed�Brou�w�er's��Col�p[le���cte�d�H�Works���for�his�reac-���tion�kto�the�incompleteness�theorems,��Jand�found�that�there�is�not�one�direct�reference�to�G����odel��������26�����٠l�����'������N�Brou���w�er's�pxrejection�of�the�la���w�of�the�excluded�middle�b�Gecame�far�b�etter-����?kno���wn��ethan�his�p�Gosition�of�men�talism�and�the�resulting�restrictions�on�his�math-����?ematical�x9on���tology��*�.��uAlthough�Brou�w�er's�logic�w�as�stim�ulated�b�y�the�parado�xes,����?c���hanging��Gto�in�tuitionistic�logic�b�y�itself�do�Ges�nothing�to�stop�the�parado�xes.�9The����?Russell���parado���x,��dfor�instance,�can�b�Ge�deriv���ed�without�an�y�app�Geal�to�argumen�t����?b���y���cases.���After�dening��R�u1�=�aj�f�x��:��x��62��x�g�,��w�e���then�pro�v�e��R�u1�62�aj�R�Ŵ�directly��*�,��from����?whic���h�\3w�e�conclude��R����2�}2�R�Dz.��aW��*�e�don't�need�to�divide�in�to�cases�according�as����?�R��߸2���R�i�or�UUnot.���6���?�6.2��]�The��negativ��e�translation��uT��?�A�|��ne��}'gative�|��form���ula�is�one�whic�h�do�Ges�not�con�tain��9��or�disjunction.��Classically��*�,����?�9�x�can�b�Ge�replaced�with��:8:��and��A�P�_��B��}�can�xb�e�dened�in�terms�of�conjunction����?and��negation,��so�ev���ery�form�ula�has�a�classically�equiv��q�alen�t�negativ�e�form.�	The����?negativ���e���translation�of�a�form�ula�is�this�form,��where�in�addition�atomic�form�ulae����?are�UUto�b�Ge�double-negated.����NOnce���formal�rules�for�in���tuitionistic�and�classical�logic�w�ere�b�Goth�a�v��q�ailable,����?it�#�w���as�easy�to�pro�v�e�that�this�translation�preserv�es�logical�pro�v��q�abilit�y��*�.����^��41����(�Hence,����?ev���ery�Wrclassical�pro�Gof�can�b�e�translated�in���to�a�corresp�onding�constructiv���e�pro�of.����?In���the�case�of�certain�axiom�systems,���notably�P���eano's�Arithmetic,�the�transla-����?tions��of�the�axioms�are�pro���v��q�able�in�the�constructiv�e�system,�wso�ev�ery�classical����?theorem�–of�P��*�A��phas�its�negativ���e�translation�pro�v��q�able�in�Heyting's�arithmetic�HA.����?Since��Vdouble�negations�on�prime�form���ulae�can�b�Ge�dropp�ed�in�arithmetic�(they����?are�kfdecidable),���ev���ery�negativ�e�theorem�of�P��*�A�kis�a�theorem�of�HA.�In�tuition-����?istic�M�arithmetic�is�not�w���eak�er�M�than�classical�arithmetic,���a�result�whic���h�surely����?surprised�UUsome�p�Geople.���6���?�6.3��]�Set��theories�with�in��tuitionistic�logic��uT��?�Brou���w�er��dev�elop�Ged�in�tuitionistic�mathematics�in�a�w�a�y�that�used�not�only�a�re-����?stricted���logic,���but�a�dieren���t�notion�of�the�fundamen�tal�mathematical�ob��8jects,����?since�h
he�put�the�con���tin�uum,�l�lled�h
with�somewhat�m���ysterious�c�hoice�sequences,����?at�k!the�cen���ter.��+This�basic�approac�h�did�not�c�hange�un�til�the�w�ork�of�Bishop����?[�4��].��Bishop���k���ept�the�restricted�logic,��but�eliminated�the�c�hoice�sequences�in��?��f�ff��v�	@��an�Îywhere�U$in�the��Col�p[le���cte�d��EWorks�U$�(Brou�w�er�rarely�refers�to��anyone�).�F�There�is�the�follo�wing��	��indirect��reference�in�a�surv�Îey�pap�<rer�published�in�1952([�9��@],���p.�3V508):�d�\[T]he�hop�e�originally���fostered�[�b�Îy�the�Old�F��J�ormalists�that�mathematical�science�erected�according�to�their�princi-���ples�T�w�Îould�b�<re�cro�wned�one�da�y�with�a�pro�<rof�of�noncon�tradictorit�y��J�,�t�w�as�nev�er�fullled,�t�and,���no�Îw�ada�ys,��zin��Cview�of�the�results�of�certain�in�Îv�estigations��Cof�the�last�few�decades,�has,�I��8think,���b�<reen��Xrelinquished."�� This�is�tak�Îen�almost�v�erbatim�from�[�10���],�p.�� 2,�written�four�y�ears�earlier.��	�>�����w���-:�41����LܻCredit�^ufor�the�negativ�Îe�translation�b�<relongs�to�Kolmogoro�v�[�39���],�v<who�published�in�Russian.���He��Nalso�constructed�the�rst�in�Îtuitionistic�formal�logical�systems.�G����odel�[�29���]�extended�the���translation�JOfrom�predicate�logic�to�arithmetic.�&See�[�26���]�for�an�extension�of�the�result�to�ZF���set��Xtheory��J�.��������27������l�����'������?�fa���v�or�wof�a�dieren���t�mathematical�dev�elopmen�t.��When�Heyting�and�Kleene�for-����?m���ulated��$their�in�tuitionistic�systems,�Θthey�simply��p��}'ostulate�d��$�the�con�tin�uit�y�of����?real-v��q�alued��dfunctions�(or�some�equiv�alen���t�prop�Gert�y).�?�This�they�no�doubt�con-����?sidered���an�impro���v�emen�t���o�v�er�Brou�w�er,���who�attempted�to�justify�con�tin�uit�y�b�y����?argumen���ts���ab�Gout�the�p�ossible�men���tal�pro�cesses.�KTBrou���w�er's���an�ti-formalist�p�Gosi-����?tion��prev���en�ted�him�from�p�Gostulating�an�ything�at�all.�͡Bishop�to�Gok�a�dieren�t����?course:�٩he�	Fsimply�restricted�his�atten���tion�to�the�con�tin�uous�functions����^��42����,�,�6Band����?said�}�that�if�there�are�an���y�other�functions,���nev�er�mind�them,���w�e're�only�in�terested����?in��the�con���tin�uous��ones.��This�p�Gosition�stim���ulated�the�consideration�of�v�ersions����?of���set�theory�with�in���tuitionistic�logic.�U�Suc�h�theories�w�ere�rst�constructed�b�y����?Myhill�>and�b���y�F��*�eferman.���T�ec�hnical�>results�on�these�theories�are�describ�Ged�in����?[�2��].�e�The�0�p�Goin���t�to�b�e�made�here�is�that�most�of�these�theories�p�ermit�some�form����?of���the�G����odel�negativ���e�translation,�jso�that�the�phenomenon�p�Gersists�when�higher����?t���yp�Ges���are�added:�?�using�in�tuitionistic�logic�alone�do�Ges�not�mak�e�a�theory�w�eak�er.����NIf��eone�tak���es�the�idea�of�rule�or�op�Geration�as�primitiv�e,�/as�w�ell�as�sets,�/so�that����?for�E�example�the�union�op�Geration�exists�ev���en�though�its�graph�is�not�a�set�but�a����?prop�Ger���class,�Ȯone�can�form���ulate�coheren�t�formal�systems�based�on�a�com�bination����?of��9the�lam���b�Gda-calculus�for�rules�and�some�axioms�of�set�theory��*�.�uThis�w�as�the����?approac���h�8=pioneered�b�y�F��*�eferman�in�[�20��
],�p�[�2��].��F�eferman's�set-theoretic�axioms����?are�=�v���ery�dieren�t�from�those�of�ZF.�It�is�p�Gossible�to�create�a�coheren�t�theory�of����?rules���and�sets�b���y�simply�taking�the�usual�ZF���axioms,�Badding�a�new�primitiv�e����?for��1rules,��(and�the�axioms�of�the�lam���b�Gda-calculus.��]The�replacemen�t�axiom�can����?then��b�Ge�form���ulated�as,���the�range�of�an�op�eration�on�a�set�is�a�set.�HXSuc���h�a�theory����?has��b�Geen�studied�in�[�3��],���where�v��q�arious�results�standard�for�in���tuitionistic�theories����?ha���v�e��b�Geen�pro���v�ed��ab�out�it.�W�The�p�oin���t�is�that�the�mo�dications�of�the�axioms�of����?set���theory�and�the�inclusion�of�rules�as�a�primitiv���e�are�more�or�less�indep�Genden�t����?c���hanges.�q�Once�URy�ou�ha�v�e�decided�whether�or�not�rules�are�fundamen�tal�ob��8jects,����?y���ou��still�ha�v�e�to�mak�e�an�indep�Genden�t�c�hoice�ab�Gout�what�sets�y�ou�think�exist,����?at�4�least�as�far�as�kno���wn�formal�systems�indicate.��There�is�no�direct�relation����?b�Get���w�een��y�our�on�tological�commitmen�t�(represen�ted�b�y�set-theoretic�axioms)�and����?y���our�W�epistemological�commitmen�t�(represen�ted�b�y�in�tuitionistic�logic�and�the�use����?of�UUrules�instead�of�functions).���6���?�6.4��]�Recursiv��e��realizabilit�y��uT��?�W��*�e���usually�tak���e�it�for�gran�ted�that�prop�Gositions�can�b�e�com���bined�b�y�means����?of�H�the�logical�op�Gerations�of�conjunction,�K(negation,�disjunction,�and�H�implication.����?Classically��#these�op�Gerations�are�dened�b���y�their�truth�tables,��abut�constructiv�ely����?this���w���on't�w�ork.�-�In�in�tuitionism,���w�e�can�only�accept�a�prop�Gosition�as�meaningful����?if���w���e�kno�w�what�it�means�to�construct�a�pro�Gof�of�it.�9IHeyting,�Ϳfollo�wing�Brou�w�er,����?ga���v�e�?Lexplanations�of�the�logical�connectiv���es�and�quan�tiers�in�these�terms.�These����?explanations��:are�straigh���tforw�ard��:for�all�the�op�Gerations�except��8��and�implication.��?�X-�ff��v�	J=�����w���-:�42����LܻT��J�ec�Îhnically�,��Xthe�functions�uniformly�con�tin�uous�on�compact�sets.��������28�������l�����'������?�A��jpro�Gof���of��8�xA�(�x�)�amoun���ts�to�a�metho�d�for�transforming�an���y��x��(in�the�range����?of��the�v��q�ariable)�to�a�pro�Gof�of��A�(�x�).�dSimilarly��*�,��Fa�pro�of�of��A�M��!��B�&��is��a�means�of����?transforming��'a�pro�Gof�of��A��to�a�pro�of�of��B��q�.��>When�Heyting�ga���v�e��'this�denition,����?recursion�jJtheory�had�not�y���et�b�Geen�dev�elop�Ged,��Mand�\metho�d"�w���as�still�an�informal����?term.�L�But��Nin�1945,��Kleene�replaced�\pro�Gof��"�b���y�\in�teger"�and�\metho�Gd"�b�y����?\recursiv���e�K)function",�M2and�the�result�w�as�the�denition�of�recursiv�e�realizabilit�y��*�.����?This�UUpro���vided�a�new�seman�tics�for�in�tuitionistic�arithmetic.����NThe��Arealizabilit���y�seman�tics�has�b�Geen�extended�to�almost�ev�ery�constructiv�e����?formal��Hsystem;�Bindeed,��Eit�is�hard�to�imagine�calling�a�system�constructiv���e�if����?it�E'do�Ges�not�p�ermit�a�realizabilit���y�in�terpretation.�A?The�realizabilit�y�in�terpreta-����?tion�connected�the�constructiv���e�unpro�v��q�abilit�y�of�the�la�w�of�the�excluded�middle����?to�[�the�recursiv���e�unsolv��q�abilit�y�of�the�halting�problem,��a�thoroughly�satisfying����?connection.����NThere�'�has�b�Geen�a�lot�of�w���ork�in�the�last�t�w�en�t�y�y�ears�that�spun�o�from�the����?notion��8of�realizabilit���y;���I��+refer�to�category-theoretic�in�terpretations�of�construc-����?tiv���e���systems,��.including�the�realizabilit�y�top�Gos,��.and�m�uc�h�w�ork�on�the�extraction����?of�A�algorithms�from�formal�pro�Gofs.�6�Although�this�w���ork�is�in�teresting�and�sat-����?isfying,��kit�p�do�Ges�not�address�the�philosophical�issues�with�whic���h�this�pap�er�is����?concerned.���6���?�6.5��]�The��meaning�of�implication��uT��?�Realizabilit���y���giv�es�us�a�m�uc�h�b�Getter�picture�of�the�meaning�of�in�tuitionistic�logic����?than�^�w���e�had�b�Gefore�it�w�as�in�v�en�ted,�aebut�the�question�arises�whether�the�realiz-����?abilit���y��in�terpretation�is�the�\in�tended�in�terpretation"�of�in�tuitionism�(at�least����?of�͏in���tuitionistic�arithmetic).��uMark�o�v�and�his�sc�ho�Gol�of�Russian�constructivists����?to�Gok�Gthis�in���terpretation�at�face�v��q�alue,�JCbut�in�the�W��*�est,�in�the�fties,�only�a�few����?logicians�w���ere�in�terested�in�constructiv�e�mathematics�an�yw�a�y��*�,��and�they�did�not����?accept���the�realizabilit���y�in�terpretation,���b�Goth�b�ecause�of�doubts�ab�out�Ch���urc�h's����?thesis�I!(is�ev���ery�\metho�Gd"�really�recursiv�e?)�m�and�b�Gecause�the�realizabilit�y�in�ter-����?pretation��seems,��lik���e�the�T��*�arksi�truth�denition,�to�presume�that�w���e�kno�w�the����?meanings�=fof�implication�and�quan���tication�as�applied�on�the�righ�t-hand�side�of����?the��rdenition.�p F��*�or�this�reason,�)�b�Goth�the�truth�denition�and�the�realizabilit���y����?denition��5can�only�b�Ge�used�to�dene�truth�(or�constructiv���e�truth)�in�a�formal����?language,�UUnot��a���priori�.����NKreisel�Iattempted�to�revise�the�Heyting�explanation,�~�adding�\second�clauses"����?to�Vthe�denitions�of�what�it�means�to�pro���v�e�Van�implication�and�a�univ���ersally����?quan���tied��1statemen�t.�7]A���pro�Gof�of��A�߅�!��B���w���ould�b�Ge,��according�to�Kreisel,�a����?metho�Gd�H$of�transforming�a�pro�of�of��A��in���to�a�pro�of�of��B��q�,�J��to��}'gether���with�a�pr�o�of�of����?this���fact�.���F��*�or�pTthis�to�b�Ge�an���y�reduction,�ww�e�m�ust�assume�that�statemen�ts�of�the����?form���\�p��pro���v�es��еB��q�(�x�)"�are�simpler�than�arbitrary�statemen���ts;��Kreisel�supp�Gosed����?they�!w���ere�decidable.�^The�plan�w�as,�%�to�turn�this�in�to�a�precise�in�terpretation�b�y����?using�recursion�theory��*�.���There�w���ere�man�y�tec�hnical�diculties,�G|but�Go�Go�dman����?ev���en�tually�A+pro�Gduced�a�\theory�of�constructions"�based�on�this�idea.�5JMo�dern�������29������l�����'������?�w���ork���has�connected�the�theory�of�constructions�with�the�F��*�rege�structures�of����?Aczel.����NAlthough�J�this�\second�clause"�in���terpretation�has�sometimes�b�Geen�claimed����?to�3�represen���t�Brou�w�er's�idea�of�the�meaning�of�the�logical�connectiv�es,�:Tin�realit�y����?Brou���w�er's��%view�w���as�rather�dieren�t.�IbBrou�w�er�explained�the�quan�tier�com�bina-����?tion�x��8�x�9�y��|�in�terms�of�functions.�۲If��x��and��y��are�in���tegers,��wthen�this�means�there����?is�!/a�function���*��that�gets��y�}�from��x�.�`fIf��x��is�a�function�itself,�+�then�it�means�that��y����?�can�b�Ge�gotten�from��x��b���y�a�con�tin�uous�function.�_VIf��x��ranges�not�o�v�er�all�in�tegers����?or�jfunctions,�50but�o���v�er�ja�sp�Gecies�of�in���tegers�or�functions,�since�sp�Gecies�ha���v�e�jto����?b�Ge�sdened�as�the�ranges�of�functions,�3�it�is�p�ossible�to�reduce�this�case�to�the����?former�ccase.���Brou���w�er�did�not�b�Geliev�e�that�linguistic�constructions�necessarily����?corresp�Gonded���to�underlying�mathematical�constructions,��so��a��!priori��it�migh���t����?not�˞b�Ge�meaningful�to�put�an�implication�arro���w�b�et���w�een�˞an�y�t�w�o�prop�Gositions.����?After�DGin���v�estigation,�G�though,�ev���erything�is�reduced�to�functions�from�in�tegers�to����?in���tegers,��whic�h���in�turn�are�giv���en�b�y�c�hoice�sequences.�2USo�Brou�w�er�really�reduced����?logic�Ƣto�on���tology:�Taall�w�e�ha�v�e�to�do�is�explain�what�c�hoice�sequences�are,��and����?w���e�UUwill�understand�in�tuitionism,�logic,�sp�Gecies,�and�all.����NA��6more���mo�Gdern�attempt�to�explain�the�logical�connectiv���es�w�as�made�b�y����?Martin-L����of��in�his�theory�of�t���yp�Ges�[�42��
].��The�leading�idea�of�this�theory�is�that����?w���e�l2construct�\t�yp�Ges",�q�whic�h�are�domains�o�v�er�whic�h�quan�tiers�ma�y�range,�q�b�y����?certain��primitiv���e�op�Gerations.�P�This�idea�w�as�presen�t�already�in��Principia�8#Mathe-����?matic��}'a�;�Oebut�LmMartin-L����of�retains�the�idea�of�\op�Geration"�as�a�primitiv���e�notion�as����?w���ell.�UUThis���theory�has�b�Geen�of�in�terest�b�Goth�to�philosophers�and�computer�scien-����?tists.�Y7A��n���um�b�Ger��of�in���teresting�tec�hnical�results�ha�v�e�b�Geen�obtained,�bwhic�h�sho�w����?the��?connections�of�this�viewp�Goin���t�with�those�of�category�theory�and�recursion����?theory��*�.���6���?�6.6��]�The��Dialectica�in��terpretation��uT��?�When���Heyting�in���tro�Gduced�his�formal�systems�for�in�tuitionism,��Iof�course�he�al-����?lo���w�ed��the�construction�of��A��\�!��B�Y�from��an���y�form�ulas��A��and��B��q�.��EThe�fact�that����?Brou���w�er��gdid�not�do�so�w���as�hardly�noticed.�Z�One�is�tempted�to�sp�Geculate�that����?G����odel�	&ma���y�ha�v�e�studied�Brou�w�er�carefully��*�,�6b�Gecause�in�1941�he�in�tro�Gduced�an����?formal�YCin���terpretation�of�in�tuitionistic�n�um�b�Ger�theory�whic�h�has�a�remark��q�able����?parallel��ito�Brou���w�er's��iviews.����^��43�����This�in���terpretation,�nwhic�h��iis�called�the�Dialec-����?tica�SXin���terpretation�(after�the�journal�in�whic�h�it�w�as�published),�S�presumes�that����?quan���tier-free��statemen�ts�are�meaningful�(ev�en�those�in�v�olving�implication);����?and�ndstatemen���ts�of�the�form��9�x�8�y�[�A�(�x;���y��)�ndare�meaningful,�t�where��A��is�quan�tier-����?free.���Let��us�call�this�EA�tform.�Here��x��and��y�q~�range�o���v�er��functionals�of�nite����?t���yp�Ge.�G����odel's�6idea�w�as�to�explain�the�meaning�of�an�y�comp�Gound�form�ula�b�y����?reducing���it�to�an�equiv��q�alen���t�one�of�this�form.�!�He�ga�v�e�a�recursiv�e�denition�of��?�X-�ff��v�	J=�����w���-:�43����LܻI���could���nd�no�evidence�for�or�against�this�sp�<reculation.���Ho�Îw�ev�er,��there���is�some�evidence,��	��in��a�lecture�G����odel�ga�Îv�e��in�1938,�ׄthat�he�w�Îas�in
uenced�b�y�Hilb�<rert�and�Ac�k�ermann's�w�ork�on���functionals��Xof�higher�t�Îyp�<re.�� See�[�52���]�for�details�and�references.��������30������l�����'������?�an��in���terpretation�that�do�Ges�just�that.�^�When�he�presen�ted�the�in�terpretation�for����?publication�!C(o���v�er�a�decade�after�he�in�v�en�ted�it),�T>it�w�as�presen�ted�as�a�purely����?formal�&�reduction�of�an�in���tuitionistic�theory�with�all�the�usual�connectiv�es,�Z�to����?a���quan���tier-free�target�theory��*�.��2Eac�h�form�ula��A��has�a�translation�in�EA�ӭform,����?�9�x�8�y�[�A���^��O!�cmsy7�����(�x;���y��).�SThe�9�theorem�is�that�if��A��is�pro���v��q�able�in�nite-t�yp�Ge�arithmetic,����?then�UUfor�some�term��t�,��A���^������(�t;���y�[ٲ)�is�pro���v��q�able�in�G����odel's�quan�tier-free�theory�T.����NT��*�o��pcarry�out�the�reduction�of�form���ulas�to�EA��form,��vG����odel�had�to�use�a����?few�Slogical�principles�that�are�not�constructiv���ely�deriv��q�able,�S�so�it�is�not�the�case����?that�R!�A��is�pro���v��q�ably�equiv�alen���t�to�its�translation��9�x�8�y�[�A���^������(�x;���y��).�h+This�R!do�Ges�not����?in���terfere��]with�the�translation's�soundness�as�stated�ab�Go�v�e,���but�it�probably�made����?G����odel�reluctan���t�to�adv��q�ance�the�in�terpretation�as�an�explanation�of�constructiv�e����?meaning,�26and�)op�Gerhaps�accoun���ts�for�the�dela�y�of�o�v�er�a�decade�in�publishing�the����?w���ork,�x@although�qDof�course�this�is�only�sp�Geculation.�ŔNo�suc�h�reserv��q�ations�did�Er-����?rett�qBishop�ha���v�e�qin�the�late�sixties,�"8when�(in�[�5��])�he�indep�Genden���tly�redisco�v�ered����?the�S�Dialectica�in���terpretation�and�adv��q�anced�it�as�an�explanation�of�constructiv�e����?meaning.�LQBishop's��-pap�Ger�w���as�not�widely�read,��cand�few�p�eople�realize�that�the����?Dialectica���in���terpretation�is�a�candidate�for�a�fundamen�tal�philosophical�deni-����?tion�U�of�the�meaning�of�the�constructiv���e�logical�connectiv�es.�s�This�in�terpretation����?is,��in��km���y�opinion,�quite�close�to�the�original�in���ten�tions��kof�Brou���w�er.�@zIt��kis�in���terest-����?ing��that�similar�notions�w���ere�outlined�b�y�Brou�w�er,��-made�metamathematically����?precise�UUb���y�G����odel,�and�redisco�v�ered�b�y�Bishop.��!č��?�7��WL�Conclusions�����?�In�|this�section,�,�w���e�will�tak�e�the�basic�issues�listed�in�the�in�tro�Gduction�one�b�y����?one,�UUand�consider�them�in�the�ligh���t�of�the�scien�tic�progress�discussed�ab�Go�v�e.����N�What�:�is�r��}'e�al,�L�and�:�how�do�we�know�it?�Qy�Brou���w�er��jansw�ered,��that�whic�h�can�b�Ge����?constructed,�adand�^�w���e�kno�w�it�b�y�constructing�it.���The�dev�elopmen�ts�of�recursiv�e����?analysis��rha���v�e�giv�en�the�lie�to�this:��w�e�kno�w�that�lots�of�non-constructiv�e�real����?n���um�b�Gers�+3m�ust�exist�to�ll�up�the�geometric�con�tin�uum.��aBishop�has�skillfully����?pap�Gered�}o���v�er�this�dicult�y�so�that�w�e�can,�J�if�w�e�lik�e,�J�just�not�men�tion�these����?non-constructiv���e�C6real�n�um�b�Gers,�~�but�b�y�considering�only�con�tin�uous�functions,����?w���e�eQmak�e�sure�that�our�functions�are�dened�on�the�non-constructiv�e�reals�that����?w���e���don't�men�tion.�t�By�dening�measures�in�terms�of�functions,��Sand�restricting����?functions�oto�b�Ge�con���tin�uous,��(the�oevil�conclusion�that�the�unit�in���terv��q�al�has�measure����?0�A�is�a���v�oided�A�([�6��]�con���tains�the�denitiv�e�treatmen�t,�E{impro�ving�on�[�4��]).�k-But�what����?do�Ges���Bishop�think�is�lling�up�the�unit�in���terv��q�al?�QLThis�question�will�b�e�answ���ered����?b�Gelo���w.����N�What���do��}'es�it�me�an�to�say�a�thing�exists?��6�Brou���w�er���answ�ered,��.it�means�that����?w���e�ϯcan�construct�it.���His�c�hoice�sequences�w�ere�only�\p�Goten�tial�innities",��Fnot����?ob��8jects���that�could�b�Ge�said�to�exist;��but�existence�in�the�con���text��8�n�9�m�	z�(�n�)��=��m����?�means��that�\c���hoices"�are�allo�w�ed�as�\constructions".�	A�The�man�y�results�on����?realizabilit���y��ksho�w�us�that�it�is�at�least�consisten�t�to�think�that�the�real�n�um�b�Gers�������31���� �l�����'������?�are�UUthe�only�source�of�non-constructiv���e�existence.����N�Can�b�things�exist�that�we�c��}'an���P't�know�ab�out?�_�Brou���w�er��said�no,�*�b�Gecause�things����?exist���only�if�w���e�can�construct�them�in�our�minds.�;But�the�real�n�um�b�Gers�con-����?structed���b���y�forcing�tec�hniques�m�ust�giv�e�us�pause.��These�seem�to�b�Ge�p�erfectly����?acceptable�}�c���hoice�sequences;�Dindeed�forcing�can�b�Ge�used�to�construct�mo�dels����?of�ctheories�of�c���hoice�sequences.���Fixing�a�mo�Gdel�of�set�theory��*�,�G�then,�there�care����?plen���t�y�Tof�reals�that�w���e�can't�kno�w�an�ything�ab�Gout�\within�that�mo�del".�WrThis�is����?somewhat�o�incoheren���t,�v=as�w�e�kno�w�in�our�minds,�v=rather�than�within�a�mo�Gdel�of����?set���theory��*�.�)�But�it�sho���ws�that�there�is�a�transnite�sequence�of�more�and�more����?recondite�UUreal�n���um�b�Gers,�UUmore�and�more�\generic"�if�y���ou�lik�e.����NAnother���w���a�y�of�expressing�the�matter�is�this:�¶View�ed�from�within�a�xed����?axiomatic���theory��*�,���most�reals�are�\generic".�0�Strengthening�the�theory�(b���y�adding����?more�"true�axioms)�will�\precipitate"�some�more�reals�from�the�\sea�of�gener-����?icit���y",�Qso��y�ou�will�b�Ge�able�to�distinguish�them�b�y�their�prop�Gerties.��But�most����?(\almost�UUall")�will�remain�generic.����NT��*�aking�̆up�the�defense�of�the�constructivist�p�Goin���t�of�view,���w�e�migh�t�main�tain����?that�7%these�n���um�b�Gers�7%don't�exist,�=/b�ecause�they�are�nev���er�completed.�g�They�repre-����?sen���t�&not�a�single�construction,�/�but�an�innite�series�of�constructions,�and�hence����?they��}don't�exist.�mAW��*�e�ma���y�then�answ�er�the�question�ask�ed�ab�Go�v�e,�(�\what�do�Ges����?Bishop�l�think�is�lling�up�the�unit�in���terv��q�al"?��LThe�constructivist�m�ust�main�tain����?that���the�unit�in���terv��q�al�is�lled�up�with�non-existen�ten�t�p�Goten�tial�innities.��The����?v��q�ast�j7ma��8jorit���y�of�mem�b�Gers�of�[0,1]�do�not�exist,��pb�ecause�w���e�cannot�construct����?them,���but�|�b�Gecause�w���e�can�construct�arbitrarily�long�partial�Cauc�h�y�sequences,����?they�/\migh���t�exist".�e�This�p�p��}'osition�is�not�absur�d.�e�Indeed,�6�it�/sounds�a�lot�lik���e�the����?p�Gosition�>Fof�mainstream�ph���ysics,�B�that�the�v��q�acuum�is�seething�with�virtual�parti-����?cles�Z�that�don't�really�exist�either.��dBut�it�do�Ges�require�conceiving�of�sets�ha���ving����?mem���b�Gers��that�don't�exist.��KCertainly�no�presen�t-da�y�formal�theories�allo�w�for����?that.����N�Can��things�exist�that�we�don���P't�know�how�to�nd?�	E6�F��*�or��$example,�Xa�w���ell-����?ordering��Cof�the�reals,��~or�a�uncoun���table�subset�of�the�reals�whic�h�is�not�in�one-����?one��corresp�Gondence�with�all�the�reals?�V�While�w���e�ma�y�b�Ge�no�closer�to�unanimit�y����?on�Υthis�question�than�w���e�w�ere�in�1905,��w�e�understand�the�ramications�m�uc�h����?b�Getter.�q7Indeed,�S�w���e�S�seem�to�ha�v�e�a�c�hoice�in�the�matter:�p�w�e�can�tak�e�Brou�w�er's����?meaning��sof�existence,��in�whic���h�case�his�answ�er�is�v��q�alid,��or�w�e�can�tak�e�Hilb�Gert's,����?whic���h��the�G����odel�negativ�e�translation�has�sho�wn�can�b�Ge�dened�in�terms�of����?Brou���w�er's.�M�W��*�e�IAjust�ha���v�e�IAto�b�Ge�careful�not�to�mix,��<in�the�same�con���text,�t�w�o����?dieren���t�UUmeanings�of�\existence".����NAs�tlcomputer�assistance�in�mathematical�problem-solving�is�b�Gecoming�more����?commonplace,���the�u�in���terest�of�the�mathematical�comm�unit�y�in�pro�Gofs�that�supply����?algorithms��Ais�increasing.�=�F��*�or�example,��yin���v��q�arian�t��Atheory�,��ywhic�h��Aw�as�a�\hot�topic"����?in�m�the�nineteen���th�cen�tury��*�,�s�languished�for�most�of�the�t�w�en�tieth,�s�but�recen�tly�it����?has��Gb�Geen�reviv���ed,��and�a�b�o�ok�has�b�een�published�with�the�title��A���lgorithms�,in����?Invariant���The��}'ory����[�57��
].�Z�This�is�not�an�isolated�inciden���t,��ias�a�p�Gerusal�of�an�y�������32����!��l�����'������?�mathematics�8Mpublisher's�catalog�will�sho���w.�hThe�comm�unit�y�presen�tly�tak�es�the����?view��suggested�ab�Go���v�e:���w�e��ha�v�e�existence,�Zand�w�e�ha�v�e�constructiv�e�existence.����?If�#�y���ou�can�pro�v�e�constructiv�e�existence,�W�that's�b�Getter,�b�ecause�#�algorithms�are����?useful,���but��6v���ery�few�p�Geople�regard�nonconstructiv�e�existence�as�meaningless.����?Since�x�the�negativ���e�translation�(and�the�formal�to�Gols�to�mak�e�its�form�ulation����?p�Gossible),�3Vit��really�isn't�p�ossible�to�main���tain�that�p�osition:��classical�existence����?can���b�Ge�in���terpreted�as�non-con�tradictory�existence.�9,So�Brou�w�er's�conclusion�that����?classical�UUmathematics�is�meaningless�w���as�o�the�mark.����^��44������N�What�@Pdo��}'es�it�me�an�to�say�something�is�true?����Brou���w�er��said�that�it�mean���t����?w���e�uScan�pro�v�e�it.��F��*�or�simplicit�y�let�us�restrict�the�discussion�to�the�truth�of����?sen���tences���in�the�language�of�P�eano�Arithmetic.��XNo�w,��QT��*�arski's�truth�denition����?can�(
b�Ge�constructiv���ely�understo�o�d,�\�lik���e�an�y�other�inductiv�e�denition.���And�it����?has��a�con���vincing�c�haracter:�M{this�do�Ges�seem�to�b�e�what�w���e�mean�b�y�truth.�Y�Then����?Brou���w�er's��claim�b�Gecomes�a�conjecture:�2&ev���ery�true�arithmetical�sen�tence�is�pro�v-����?able.�H�No���w,���b�y���G����odel's�incompleteness�theorem�w���e�kno�w�that�an�y�notion�of����?\pro���v��q�abilit�y"�u�for�whic���h�this�w�ould�b�Ge�true�cannot�b�e�captured�in�a�formal�sys-����?tem.���But�`that�is�what�Brou���w�er�`b�Geliev�ed�all�along,���so�one�could�imagine�his����?reaction����^��45���x�:�q�\Of�UUcourse."����NT��*�o�c�main���tain�Brou�w�er's�conjecture�requires,���in�view�of�the�incompleteness����?theorems,�ia���strong�faith�in�the�non-mec���hanical�nature�of�the�h�uman�mind.����?There�,�do�Gesn't�seem�to�b�e�the�sligh���test�evidence�in�fa�v�or�of�the�conjecture;��zit����?m���ust�UUb�Ge�tak�en�as�an�article�of�faith.����NAfter��4a�cen���tury�of�tec�hnical�progress,���this�is�the�conclusion:��what�Brou�w�er����?to�Gok�UUto�b�e�the�denition�of�truth�is�an�unpro���v��q�able�conjecture.����N�How��c��}'an�we�know�whether�something�is�true?��b�Brou���w�er��3said,���the�only�w���a�y����?is��pto�pro���v�e��pit.�bThis�seems�to�b�Ge�the�common�viewp�oin���t�of�all�mathematicians,����?constructiv���e���or�not.�;�It�is�what�distinguishes�a�mathematician�from�a�ph�ysicist����?(who��will�accept�empirical�evidence),�!�or�from�a�p�Gerson�of�faith�(who�will�accept����?a�UUrev���elation).����NThis��Whallo���w�ed�principle�of�mathematics�is�undergoing�a�c�hallenge�these�da�ys,����?from�%Qsome�who�b�Geliev���e�there�is�ro�om�for�the�exp�erimen���tal�metho�d�in�mathe-����?matics.���F��*�or�i�example,���computers�ha���v�e�i�b�Geen�used�to�pro���vide�evidence�for�the����?Riemann�g�h���yp�Gothesis�����^��46���H0�that�w�ould�b�Ge�con�vincing�enough�for�a�ph�ysicist,�l:sho�w-����?ing��Flots�and�lots�of�zero�Ges�on�the�critical�line�and�none�o�the�critical�line,���y���et����?mathematicians�UUstill�aren't�satised.����N�Can�ththings�b��}'e�true�that�we�c�an���P't�ever�know�to�b�e�true?�M۲Brou���w�er�I[said�no.��?�E�ff��v�	J=�����w���-:�44����LܻSp�<recically��J�,�Y�on�:�page�510�of�[�9��@],�Brou�Îw�er�:�asks�whether�a�\linguistic�application"�of�classical��	��logic��can�alw�Îa�ys��b�<re�paralleled�b�Îy�\languageless�mathematical�constructions",��Aand�answ�ers�in���the�ZDnegativ�Îe�if�the�principle�of�the�excluded�third�is�in�v�olv�ed.�U�But�the�negativ�e�translation���sho�Îws��Xthat��some��construction�do�<res�accompan�y�the�use�of�classical�logic.��	�>�����w���-:�45����LܻIndeed,��Xone��must��imagine�his�reaction,�since�he�did�not�record�it.�������w���-:�46����LܻThe�O�Riemann�h�Îyp�<rothesis�sa�ys�all�the�complex�zero�<res�of�a�certain�function,�n?the�analytic��	��extension�*|of�1�R+�1�=�2���-:�x���Q�+���:�j�:�:��
q��1�=n���-:�z��lP�+���:�j�:�:���,�?�lie�*|on�the�\critical�line"��R��e�(�z�V��)��Z=�1�=�2.�ƌIt�*|is�a�famous���unsolv�Îed��Xproblem.��������33����"p�l�����'������?�But���the�thrust�of�the�cen���tury's�progress�is�that,���ho�w�ev�er�little�w�e�lik�e�it,���there����?are��Blots�of�true�things�that�w���e�aren't�ev�er�going�to�kno�w.�?�It�seems�that�w�ell-����?orderings�۹are�the�k���ey�to�truth�in�arithmetic.�I>But�ho�w�are�w�e�supp�Gosed�to�recog-����?nize���w���ell-orderings,�#b�Ge�they�primitiv�e�recursiv�e�w�ell-orderings�of�the�in�tegers,�#or����?measurable�n�cardinals?��rIn�b�Goth�cases,�t�w���e�ha�v�e�run�out�of�ideas,�t�and�are�relying����?on�29mere�sp�Geculations.�fIt�seems�that�a�practical�limit�to�arithmetical�kno���wledge����?has��rb�Geen�reac���hed.�J Ev�en��rif�future�generations�in���v�en�t��rsev�eral�new�kinds�of�large����?cardinals,�Fthey�
Bwill�just�ha���v�e�
Badv��q�anced�a�few�steps�do���wn�a�transnite�road,�and����?will�hb�Ge�stuc���k�again.�Z#The�inscription�on�Hilb�ert's�gra���v�e�h(\w�e�m�ust�kno�w,��w�e�will����?kno���w")�UUlo�Goks�false�to�us�to�da���y��*�.����^��47������N�Chaitin�c�[�11��
]�has�giv���en�a�form�ulation�of�the�incompleteness�theorems�that����?mak���es���the�limits�of�our�kno�wledge�esp�Gecially�clear.����^��48���
|i�Chaitin's�form�ulations�of����?the�h�incompleteness�theorems�mak���e�it�clear�that�the�true�arithmetical�theorems����?of�Qthe�formal�theories�w���e�kno�w�and�lo�v�e�are�just�a�drop�in�the�o�Gcean�compared����?to�Giall�the�arithmetical�truths,�J2and�w���e�don't�ha�v�e�an�y�clear�ideas�ho�w�to�extend����?our���theories�m���uc�h���further,��Dand�ev���en�if�w�e�did,��Dthey�w�ould�still�b�Ge�formal�theories����?and��Btheir�theorems�w���ould�still�b�Ge�just�a�drop�in�the�o�cean�of�arithmetical�truths.����?No���w�w�y�ou�migh�t�think�that�\a�drop�in�the�o�Gcean"�is�a�gross�exaggeration,��[since����?the�$�incompleteness�theorem�sa���ys�only�that�there�is��some��true�arithmetical�non-����?theorem.�a*Ma���yb�Ge�#}so,�-uand�there�is�certainly�disagreemen�t�o�v�er�whether�Chaitin's����?w���ork�}$has�added�to�the�philosophical�signicance�of�the�incompleteness�theorems.����?But�1Jev���en�without�Chaitin,�hGthere�is�clear�evidence�that�there�are�arithmetical����?truths���that�w���e�can't,��=in�an�y�practical�sense,��=ev�er�pro�v�e.�Z[Ho�w�ev�er�little�w�e�lik�e����?it,�UUthat's�the�w���a�y�UUthings�are.����^��49����?�X-�ff��v�	J=�����w���-:�47����L��Historic���al�dLR�emarks��D�:�w+Hilb�<rert�-]b�eliev�Îed�that��every�dLwel�p[l-formulate���d�pr�oblem�has�a�solution��	��and�Hwil�p[l�eventual�ly�b���e�solve�d�.�ĊThis�͕is�not�exactly�the�same�as�the�question�whether�whatev�Îer���is���true�can�b�<re�pro�Îv�ed,��wbut���it�certainly�implies�it.��F��J�or�Hilb�ert�the�question�had�a�dieren�Ît���meaning�ιthan�for�Brou�Îw�er,�
since�ιBrou�w�er's�approac�h�in�v�olv�ed�questioning�the�meaning�of���\w�Îell-form�ulated�problem".��RIt�is�therefore�not�surprising�that�Brou�Îw�er�explicitly�repudiated���Hilb�<rert's���con�Îviction�just�men�tioned.��OF��J�or�example,�”Brou�w�er�w�ould�ha�v�e�denied�that�the�prob-���lem�
of�determining�whether�t�Îw�o�
giv�en�real�n�um�b�<rers�are�equal�is�solv��able.�e]But�a�sp�ecic���statemen�Ît,���suc�h���as�\�e���-:�e���f�is�irrational",�w�Îould�not�b�<re�a�truth�for�Brou�w�er�unless�w�e�had�a�pro�<rof���of�][it.�_*It�w�Îas,�\and�still�is,�the�custom�in�Dutc�Îh�univ�ersities�that�a�thesis�is�accompanied�b�y���some�c�short�statemen�Îts�to�b�<re�defended�b�y�the�candidate.��SThe�last�suc�h�statemen�t�in�Brou�w�er's���thesis��;([�9��@],���p.���101)�is�that�Hilb�<rert's�con�Îviction�is�\unfounded."�A��exact�quotation�of,���and���reference��Xto,�Hilb�<rert's�original�statemen�Ît�can�also�b�e�found�there.��	�>�����w���-:�48����LܻGiv�Îen��@a�p�<rolynomial��f���n���w�of�sev�eral�v��ariables��x�,��dep�<rending�on�a�parameter��n�,�let��A���n���w�b�<re���the��eansw�Îer�(y�es�or�no)�to�the�question�whether��f���n��7�(�x�)���=�0��ehas�innitely�man�y�solutions�in���the���in�Îtegers.�1�Chaitin�pro�v�es�that,�A�giv�en�a�xed�formal�theory��J�,�A�suc�h�as�P�eano�Arithmetic���P��J�A�>�or�?JZermelo-F�raenk�Îel�set�theory�ZF,�there�is�a�p�<rolynomial��f���n��	]��suc�h�that�the�sequence����A���n���i�lo�<roks�d2random�from�within�the�xed�formal�theory��J�.�s�That�is,���using�the�resources�of��Z���F�.:�,���w�Îe�8�could�not�distinguish�the�sequence��A���n��	V��from�the�tosses�of�a�fair�coin.��Morev�er,��Kin�a���precise�Psense,�this�is�true�of�an�Îy�randomly�selected�sequence��f���n��7�,�not�just�of�this�sp�<recial�one.���Although���the�form�Îulation�in�this�fo�<rotnote�is�not�precise,�9Chaitin's�theorems�use�precisely���dened��Xmathematical�concepts.��	�>�����w���-:�49����LܻThis�b�line�is�deliv�Îered�with�appropriate�in
ection�in�the�mo�vie��Pig�,��when�the�y�oung�pig���learns�?wh�Îy�certain�farm�animals�are�k�ept�ev�en�though�they�do�no�useful�w�ork.�k�The�order�of���the��Xuniv�Îerse�w�as�not�necessarily�established�for�our�con�v�enience.��������34����#w�l�����'������?�References��
u����D�[1]���S�<Aczel,���P��*�.,��Non-wel���l-founde��}'d���sets�,�Cen���ter�jQfor�the�Study�of�Language�and����S�<Information,�UUStanford,�Calif.�(1988).��&Z����D[2]���S�<Beeson,��7M.,��F��;�oundations�mof�Constructive�Mathematics�,��7Springer-V��*�erlag����S�<(1985).������D[3]���S�<Beeson,��rM.��8T��*�o���w�ards�a�computation�system�based�on�set�theory��*�,��r�The��}'or�etic�al����S�<Computer���Scienc��}'e�UU�/�"V

cmbx10�60�:�q�297{340�(1988).������D[4]���S�<Bishop,�I�E.�F��F��;�oundations��Qof�Constructive�A���nalysis�,�McGra���w-Hill,�New�F�Y��*�ork����S�<(1967).��sThe�original�edition�is�out�of�prin���t.�There�is�a�revised�and�extended����S�<second��edition,���E.�Bishop�and�D.�Bridges,��Constructive�05A���nalysis�,�Springer-����S�<V��*�erlag�UU(1985).������D[5]���S�<Bishop,�#(E.,�Mathematics��as�a�n���umerical�language,�#(in:�RkKino,�A.,�Myhill,�J.,����S�<and�toV��*�esley�,�|6R.E.�(eds),��Intuitionism���and�Pr��}'o�of���The�ory:���Pr�o�c�e�e�dings�of�the����S�<Summer��Confer��}'enc�e�at�Bualo,�5New�Y��;�ork,�1968�,��9pp.��r53-71,�North-Holland,����S�<Amsterdam�UU(1970).������D[6]���S�<Bishop,��E.,�and���Cheng,�H.,��Constructive��cMe��}'asur�e�The�ory�,��Memoirs���of�the����S�<A.M.S.�UU�116�,�Amer.�Math.So�Gc.,�Pro���vidence,�R.I.�(1972).������D[7]���S�<Borsuk,���K.�|�and�Szmielew,�W.,��F��;�oundations���of�ge��}'ometry:��|Euclide�an���and����S�<Bolyai-L��}'ob�achevskian�7ge�ometry�,��Rev.��iEnglish�translation.�[T��*�ranslated�from����S�<P���olish���b�y�Erwin�Marquit].�North-Holland,���Amsterdam�and�In�terscience,����S�<New�UUY��*�ork�(1960).������D[8]���S�<Brou���w�er,�UUL.�E.�J.,�V��*�olition,�Kno���wledge,�Language,�in�[�9��],�pp.�443-446.������D[9]���S�<Brou���w�er,�JgL.��1E.�J.,��Col���le��}'cte�d��Works�,�V��*�ol.��11,�ed.�b���y�A.�Heyting,�North-����S�<Holland,�UUAmsterdam�(1975).������?[10]���S�<Brou���w�er,��cL.��-E.�J.,��Br��}'ouwer's�Cambridge�L�e�ctur�es�on�Intuitionism�,��ced.��-b���y����S�<D.�UUv��q�an�Dalen,�Cam���bridge�Univ�ersit�y�Press,�Cam�bridge�(1981).������?[11]���S�<Chaitin,��.G.,��A���lgorithmic���information�the��}'ory�,��.Cam���bridge��iUniv�ersit�y�Press,����S�<Cam���bridge/New�UUY��*�ork�(1987).������?[12]���S�<Cohen,���P��*�.,��Set�ǣThe��}'ory�and�the�Continuum�Hyp�othesis�,���Benjamin,�New�wLY��*�ork����S�<(1966).������?[13]���S�<Dedekind,�LPR.�J�Continuity���and�Irr��}'ational�Numb�ers�J�(rst�published�in�1872),����S�<in�UU�Essays���on�the�The��}'ory�of�Numb�ers�,�UUDo���v�er,�New�Y��*�ork�(1963).������?[14]���S�<Dedekind,�W'R.��c�The�Natur��}'e�and�Me�aning�of�Numb�ers�,�W'(rst��cpublished�in����S�<1887),���in�O��Essays��
on�the�The��}'ory�of�Numb�ers�,���Do���v�er,�New�O�Y��*�ork�(1963).�T�rans-����S�<lation�UUof��Was���Sind�und�was�Sol���len�die�Zahlen?�.�������35����$&��l�����'��������?�[15]���S�<F��*�eferman,���S.,�T�ransnite�r�recursiv���e�progressions�of�axiomatic�theories,���J.����S�<Sym���b�Golic�UULogic��27��(1962),�259-316.������?[16]���S�<F��*�eferman,���S.,�Systems�Fjof�predicativ���e�analysis,����Journal�q�of�Symb��}'olic�L�o�gic����S�<�29�UU�(1964)�1-30.������?[17]���S�<F��*�eferman,�g�S.,�Some�d(applications�of�the�notions�of�forcing�and�generic�sets,����S�<�F��;�und.���Math.�,�UUv���ol.�56,�pp.�325{345,�1965.������?[18]���S�<F��*�eferman,��LS.,�Systems��Jof�predicativ���e�analysis�I�GI:�represen�tations�of�ordinals,����S�<�Journal���of�Symb��}'olic�L�o�gic�UU�33��(1968)�193-220.������?[19]���S�<F��*�eferman,�+�S.,�Predicativ���ely��reducible�systems�of�set�theory�,�+�in��Axiomatic����S�<Set���The��}'ory,�L�Pro�Gc.���Symp�os.�in�Pure�Math.�v���ol.�XI�I�I,�P���art�2,���pp.�11{32,����S�<Amer.�UUMath.�So�Gc.,�Pro���vidence,�1974.������?[20]���S�<F��*�eferman,�]�S.,�Constructiv���e�(�theories�of�functions�and�classes,�]�pp.�159-224����S�<in:�_�M.�1QBoa,�8�D.�v��q�an�Dalen,�and�K.�McAlo�Gon�(eds.),��L��}'o�gic�r�Col���lo�quium�'78:����S�<Pr��}'o�c�e�e�dings��&of�the�L��}'o�gic��&Col���lo�quium�at�Mons�,��sNorth-Holland,�Amsterdam����S�<(1979).������?[21]���S�<Field,��]H.��[H.��Scienc��}'e���without�numb�ers:�ea�defenc�e�of�nominalism�,Princeton����S�<Univ���ersit�y�UUPress,�Princeton,�N.J.�(1980).������?[22]���S�<F��*�orster,�~�T.�I[E.��Set��_The��}'ory�with�a�Universal�Set�,�Oxford�I[Science�Publications,����S�<Clarendon�UUPress,�Oxford�(1995).������?[23]���S�<F��*�rege,� G.,��The�F��;�oundations�of�A���rithmetic�,� Blac���kw�ell,�Oxford���(1980).�En-����S�<glish�UUtranslation�of��Grund���lagen���der�A���rithmetik�,�UUoriginal�date�1884.������?[24]���S�<F��*�raenk���el��A.,��!and�Bar-Hillel,�Y.,��F��;�oundations�xof�Set�The��}'ory�,�North-Holland,����S�<Amsterdam�UU(1958).������?[25]���S�<F��*�eynman,�ՌR.,��F��;�eynman��L��}'e�ctur�es�on�Gr�avitation�,�ՌAddison-W��*�esley�,�Reading,����S�<Mass.�UU(1995).������?[26]���S�<F��*�riedman,���H.,�The��iconsistency�of�classical�set�theory�relativ���e�to�a�set�theory����S�<with�UUin���tuitionistic�logic,�J.�Sym�b�Golic�Logic��38��(1973)�315-319.������?[27]���S�<G����odel,�M�K.,��Col���le��}'cte�d��Works�,�M�V��*�ol�K�I,�edited�b���y�F�eferman��et.��al.�,�M�Oxford�Uni-����S�<v���ersit�y�UUPress,�New�Y��*�ork�(1986).������?[28]���S�<G����odel,��K.,��Col���le��}'cte�d���Works�,��V��*�ol���I�GI�I,�edited�b���y�F��*�eferman��et.���al.�,��Oxford����S�<Univ���ersit�y�UUPress,�New�Y��*�ork�(1995).������?[29]���S�<G����odel,�x�K.,�Zur��in���tuitionistisc�hen�Arithmetik�und�Zahlen�theorie,�x�1933,����S�<reprin���ted�UUwith�English�translation�in�[�27��
],�pp.�282-295�������36����%2v�l�����'��������?�[30]���S�<v��q�an�EHeijeno�Gort,���J.�(ed),��F��;�r��}'om�ptF�r�e�ge�ptto�G���odel,���A�p;Sour�c�e�Bo�ok�in�Mathe-����S�<matic��}'al���L�o�gic,�(1879-1931�,���Harv��q�ard��)Univ���ersit�y�Press,���Cam�bridge,�Mass./����S�<London�UU(1967).������?[31]���S�<Hellman,���Georey��*�,��Mathematics���without�numb��}'ers:�
˞towar�ds���a�mo��}'dal-����S�<structur��}'al���interpr�etation�,�{[Clarendon�s�Press,�Oxford,�and�Oxford�Univ���ersit�y����S�<Press,�UUNew�Y��*�ork�(1989).������?[32]���S�<Herbrand,�B�J.,�Rec���herc�hes�+sur�la�th�����Georie�de�la�d���Gemonstration,�B�T��*�ra�v��q�aux�de����S�<la�~uSo�Gci�����Get��e�~udes�Sciences�et�des�Lettres�de�V��*�arso���vie,��Classe�I�GI�I�}�sciences����S�<math�����Gematiques��2et�ph�ysiques,���33�,�128��2pp.�(1930).�Reprin�ted�in�[�33��
]�and����S�<translated�UUin�[�34��
].������?[33]���S�<Herbrand,�JPJ.,��E'crits�H6L��}'o�giques�Q�(J.�v��q�an�Heijeno�Gort,�JPed.),�Presses�QUniv���ersi-����S�<taires�UUde�F��*�rance,�P���aris�(1968).������?[34]���S�<Herbrand,�`_J.,��L��}'o�gic�al�XrWritings�*��(English�translation�of�Herbrand�1968�b���y����S�<W.�UUGoldfarb),�Reidel,�Dordrec���h�t/�UUBoston/�Lancester/�T��*�oky���o�(1971).������?[35]���S�<Hilb�Gert,�4D.,��F��;�oundations��of�Ge��}'ometry�,�4English��translation�of��Grund���lagen����S�<der���Ge��}'ometrie�,�UUOp�Gen�Court,�La�Salle,�Illinois�(1971).������?[36]���S�<Hilb�Gert,��D.,�On�
|the�foundations�of�logic�and�arithmetic,��in:�M�v��q�an�Heijeno�ort����S�<[1967],�UUpp.�129-138.������?[37]���S�<Kleene,�N>S.�LyC.��The���F��;�oundations�of�Intuitionistic�Mathematics,��besp��}'e�cial���ly���in����S�<r��}'elation���to�R�e�cursive�F��;�unctions�,�UUNorth-Holland,�Amsterdam�(1965).������?[38]���S�<Klong-c���hen��[rab-b�y�ams-pa,�؍Ho�w�the�samsara�is�fabricated�from�the�ground����S�<of��@b�Geing,�n�translated�b���y�K.�Lipman,��Crystal��Mirr��}'or��5�,�1977,�344-364.����S�<Dharma�UUPublishing,�Emeryville,�Calif.������?[39]���S�<On��8the�principle�of�tertium�non�datur,��Mat.�Sb.�32�(1925)�646-667�(Rus-����S�<sian).�UUEnglish�translation�in:�q�[�30��
],�pp.�414-437.������?[40]���S�<Kreisel,��(G.,�Sho�Geneld,�J.,�and���W��*�ang,�H.,�Num���b�er-theoretic���concepts�and����S�<recursiv���e�&�w�ell-orderings,�[,Arc�h.�Math.�Logik�u.�Grundlagen�5�(1960)�371-����S�<391.������?[41]���S�<Liss�vT.,�~3and�Tipton,�P��*�.,�The�disco���v�ery�vof�the�top�quark,��Scientic���A���meri-����S�<c��}'an�,�UUSept.�1997,�pp.�54-67.������?[42]���S�<,�UUMartin-L����of,�P��*�.,��Intuitionistic���T��;�yp��}'e�The�ory�,�UUBibliop�Golis,�Naples�(1984).������?[43]���S�<Myhill,�UUJ.�Constructiv���e�set�theory��*�,��J.���Symb��}'olic�L�o�gic�UU�40��(1975)�347-382.������?[44]���S�<Nelson,�ocE.�j-In���ternal�Set�Theory:��xA�j(new�approac�h�to�nonstandard�analysis,����S�<�Bul���l.���A.�M.�S.�UU�83��(1977)�1165-1198.�������37����&=٠l�����'��������?�[45]���S�<Nelson,���E.,�The�Q�syn���tax�of�nonstandard�analysis,����A���nnals��of�Pur��}'e�and�Applie�d����S�<L��}'o�gic�UU�38��(1988)�123-134.������?[46]���S�<P���enrose,���R.,��Shadows���of�the�mind�:���a�se��}'ar�ch���for�the�missing�scienc��}'e�of����S�<c��}'onsciousness�,�UUOxford�Univ���ersit�y�UUPress,�Oxford/New�Y��*�ork�(1994).������?[47]���S�<Reid,��C.,��Hilb��}'ert�,�Cop�Gernicus,�an��Fimprin���t�of�Springer-V��*�erlag,��New�Y�ork����S�<(1996).������?[48]���S�<Robinson,�ijA.,��Non-standar��}'d��A���nalysis�,�revised���edition,�Princeton�Univ���ersit�y����S�<Press,�UUPrinceton�(1996).�First�published�in�1966.������?[49]���S�<Ruc���k�er,�2�R.,��Mind�6�to��}'ols:��*the�ve�levels�of�mathematic�al�r�e�ality�,�2�Hough���ton����S�<Miin,�UUBoston�(1987).������?[50]���S�<Russell,��B.,��The��Pr��}'oblems�of�Philosophy�,��Oxford�xPUniv���ersit�y�Press,��Oxford����S�<(1912).������?[51]���S�<Russell��B.,�#SandWhitehead,�A.�N.,��Principia�ZdMathematic��}'a�,�Cam���bridge�Uni-����S�<v���ersit�y�UUPress,�Cam���bridge�(1963).�First�published�in�1910.������?[52]���S�<Sieg,�UUW.,�and�P���arsons,�C.,�In�tro�Gductory�Note�to�1938a,�in�[�28��
],�pp.�62-85.������?[53]���S�<Sho�Geneld,��J.,��Mathematic��}'al�L�o�gic�,��Addison-W��*�esley�,�Reading,�Mass.����S�<(1967).������?[54]���S�<Sho�Geneld,��bJ.,�On��%a�restricted�[omega]-rule,��bBull.�Acad.�P���olon.Sci.�7�(1959)����S�<405-407.������?[55]���S�<Shree�$3Purohit�Sw���ami�and�Y��*�eats,�W�W.�B.�(translators),��The�R9T��;�en�Princip��}'al����S�<Up��}'anishads�,�UUF��*�ab�Ger�and�F�ab�Ger�Limited,�London�(1937).������?[56]���S�<Sieg,�UUW.,�Hilb�Gert's�Programs:�q�1917-1922,�to�app�ear.������?[57]���S�<Sturmfels,��B.,��A���lgorithms��in�Invariant�The��}'ory�,��Springer-V��*�erlag,�Vi-����S�<enna/New�UUY��*�ork�(1993).������?[58]���S�<T��*�ro�Gelstra,��A.��S.��Metamathematic��}'al�5�Investigation�of�Intuitionistic�A���rith-����S�<metic���and�A���nalysis�,�UUSpringer�Lecture�Notes�in�Mathematics��754��(1973).������?[59]���S�<T��*�ro�Gelstra�0�A.�S.,�g�and�v��q�an�Dalen,�D.,��Constructivism�]�in�Mathematics:�-AA���n����S�<Intr��}'o�duction�,�UUt���w�o�v�olumes,�North-Holland,�Amsterdam�(1988).������?[60]���S�<T��*�uring,�*A.,�Systems��|of�logic�based�on�ordinals,�*�Pr��}'o�c.�0rL�ondon�Math.�So�c.����S�<�45�UU�(1939),�161-228.������?[61]���S�<T��*�ro�Gelstra,��/A.��S.,�and�Sc���h�wic�h�ten�b�Gerg,��/H.,��Basic��sPr��}'o�of�The�ory�,��/Cam���bridge����S�<Univ���ersit�y�UUPress,�Cam���bridge�(1996).�������38����I����;�l�&�8�':

cmti10�/�"V

cmbx10�.��N�ffcmbx12�"#�f�cmti8�!���@cmti12��C�scmtt8���N�cmbx12��K�cmsy8�;�cmmi6��2cmmi8��Aa�cmr6�|{Ycmr8�X�Qcmr12�D��tG�G�cmr17�
!",�

cmsy10�O!�cmsy7�
�b>

cmmi10�	0e�rcmmi7�K�`y

cmr10�ٓ�Rcmr7�Tt�����

Sindbad File Manager Version 1.0, Coded By Sindbad EG ~ The Terrorists