/LMAR=0/XLINE=3/FONT#0=BAXL30/FONT#1=BAXM30/FONT#2=BAXB30/FONT#3=BAXS30/FONT#4=METL



␈↓ ↓H␈↓␈↓ ∧Y␈↓∧Computer Science Department
␈↓ ↓H␈↓∧␈↓ ¬)Stanford University

␈↓ ↓H␈↓∧␈↓ ¬u␈↓5 March 1980




␈↓ ↓H␈↓␈↓ εεProposal to

␈↓ ↓H␈↓␈↓ ∧f␈↓∧National Science Foundation

␈↓ ↓H␈↓∧␈↓ ¬U␈↓for continuation of

␈↓ ↓H␈↓␈↓ ∧
␈↓∧Basic Research in Artificial Intelligence


␈↓ ↓H␈↓∧␈↓ ¬SJohn McCarthy
␈↓ ↓H␈↓∧␈↓ ∧VProfessor of Computer Science
␈↓ ↓H␈↓∧␈↓ ¬→Principal Investigator










␈↓ ↓H␈↓∧␈↓ ε↔␈↓αAbstract


␈↓ ↓H␈↓α␈↓This␈α⊂is␈α⊂a␈α⊂request␈α⊂for␈α⊂continuation␈α⊃of␈α⊂NSF␈α⊂Grant␈α⊂MCS78-00524␈α⊂supporting␈α⊂Basic␈α⊃Research␈α⊂in
␈↓ ↓H␈↓Artificial␈α⊂Intelligence.␈α⊂ The␈α∂renewal␈α⊂will␈α⊂continue␈α∂for␈α⊂the␈α⊂third␈α∂year␈α⊂the␈α⊂activities␈α⊂proposed␈α∂for
␈↓ ↓H␈↓three␈α∞years␈α
in␈α∞our␈α
proposal␈α∞of␈α
April␈α∞26,␈α
1978.␈α∞ The␈α
amount␈α∞requested␈α
is␈α∞$101,697,␈α∞in␈α
accordance
␈↓ ↓H␈↓with␈αthe␈αbudget␈αsubmitted␈αin␈αthe␈αprevious␈αproposal,␈αless␈αestimated␈αcarry-over␈αof␈αresidual␈αfunds␈αin
␈↓ ↓H␈↓the amount of $45,193.
␈↓ ↓H␈↓α␈↓ ∧ABasic Research in Artificial Intelligence␈↓ L1


␈↓ ↓H␈↓α␈↓ α{Proposed Research                          ␈↓ εx␈↓        2.␈αApplying␈αcircumscription␈αto␈αproofs␈αof
                                           ␈↓ εx␈↓non-knowledge seems promising.
␈↓ ↓H␈↓        For␈α)the␈α(academic␈α)year␈α(1979-80,
␈↓ ↓H␈↓Professor␈α⊃John␈α⊃McCarthy␈α⊃will␈α⊃return␈α⊃to␈α⊃the       ␈↓ εx␈↓        In␈α%1980-81␈α%McCarthy␈α&expects␈α%to
␈↓ ↓H␈↓Computer␈α_Science␈α_Department␈α_of␈α↔Stanford        ␈↓ εx␈↓continue␈α
work␈α
begun␈α
this␈α
year␈α
on␈α
concepts␈α
in
␈↓ ↓H␈↓University,␈α↔having␈α↔completed␈α↔a␈α↔year␈α↔as␈α↔a         ␈↓ εx␈↓approximate␈α
theories.␈α∞ The␈α
idea␈α
is␈α∞that␈α
many
␈↓ ↓H␈↓Fellow␈α∂at␈α∂the␈α∂Center␈α∂for␈α∂Advanced␈α∂Study␈α∂in        ␈↓ εx␈↓concepts␈αused␈αin␈α
common␈αsense␈αreasoning␈α
such
␈↓ ↓H␈↓the␈α
Behavioral␈αSciences␈α
(CASBS)␈α
at␈αStanford     ␈↓ εx␈↓as␈α∀␈↓↓causes␈↓␈α∀and␈α∪␈↓↓beliefs␈↓␈α∀cannot␈α∀be␈α∀defined␈α∪in
␈↓ ↓H␈↓where␈α
he␈α
was␈α
chairman␈α
of␈α
a␈α
study␈αon␈α
artificial     ␈↓ εx␈↓terms␈α⊃of␈α⊃the␈α⊃state␈α⊃of␈α⊃the␈α⊃world␈α⊃but␈α∩only␈α⊃in
␈↓ ↓H␈↓intelligence␈α
and␈α
philosophy␈αby␈α
a␈α
group␈α
of␈αsix     ␈↓ εx␈↓terms␈α
of␈α
an␈α
approximate␈α
theory␈α
of␈αsome␈α
aspect
␈↓ ↓H␈↓computer␈α5scientists,␈α5philosophers␈α5and         ␈↓ εx␈↓of␈α⊂the␈α⊂world.␈α⊂ They␈α⊂are␈α⊂useful␈α⊂concepts␈α∂that
␈↓ ↓H␈↓psychologists.                             ␈↓ εx␈↓seem␈α
to␈αdisappear␈α
when␈α
examined␈αclosely.␈α
 We
                                           ␈↓ εx␈↓believe␈α∃this␈α∃phenomenon␈α∃to␈α⊗be␈α∃widespread
␈↓ ↓H␈↓        McCarthy␈α
will␈α
continue␈α
his␈α∞research␈α
on  ␈↓ εx␈↓and␈α%of␈α%fundamental␈α&philosophical␈α%and
␈↓ ↓H␈↓non-monotonic␈α≠reasoning␈α≠and␈α≠the␈α~further        ␈↓ εx␈↓practical importance.
␈↓ ↓H␈↓formalization of concepts.
                                           ␈↓ εx␈↓        Three␈α⊃papers␈α⊃based␈α⊃on␈α⊃previous␈α⊂work
␈↓ ↓H␈↓        Non-monotonic␈α⊂reasoning,␈α⊂as␈α⊂described ␈↓ εx␈↓have␈α#been␈α#published:␈α$␈↓↓Ascribing␈α#Mental
␈↓ ↓H␈↓in␈α∩the␈α∩paper␈α∩␈↓↓Circumscription␈α∩Induction␈α∪-␈α∩A       ␈↓ εx␈↓↓Qualities␈α$to␈α$Machines␈↓␈α%in␈α$␈↓αPhilosophical
␈↓ ↓H␈↓↓Way␈α*of␈α*Jumping␈α*to␈α+Conclusions␈↓,␈α*is               ␈↓ εx␈↓αPerspectives␈α*in␈α*Artificial␈α*Intelligence␈↓,
␈↓ ↓H␈↓increasingly␈α⊗seen␈α⊗as␈α⊗a␈α⊗requirement␈α⊗for␈α⊗AI        ␈↓ εx␈↓Martin␈α∃Ringle␈α∃(ed.),␈α∀N.Y.,␈α∃1978,␈α∃and␈α∀"First
␈↓ ↓H␈↓systems.␈α⊃ Drew␈α∩McDermott␈α⊃of␈α⊃Yale␈α∩and␈α⊃Jon         ␈↓ εx␈↓Order␈α⊃Theories␈α⊃of␈α⊃Individual␈α∩Concepts␈α⊃and
␈↓ ↓H␈↓Doyle␈α∩of␈α∪M.I.T.␈α∩ recently␈α∪wrote␈α∩a␈α∪paper␈α∩on        ␈↓ εx␈↓Propositions",␈α in␈α ␈↓↓Machine␈α Intelligence␈α 9␈↓,
␈↓ ↓H␈↓non-monotonic␈α0reasoning␈α1proposing␈α0a           ␈↓ εx␈↓Donald␈α-Michie␈α-(ed.),␈α.Scotland,␈α-1979.
␈↓ ↓H␈↓di␈↓↓␈↓β@␈↓↓␈↓erent␈α≤approach␈α≤from␈α≤McCarthy's.␈α≤ An         ␈↓ εx␈↓␈↓↓Circumscription␈α∂-␈α∂A␈α∂Form␈α∂of␈α∞Non-Monotonic
␈↓ ↓H␈↓impromptu␈α6mini-conference␈α6on␈α5non-             ␈↓ εx␈↓↓Reasoning␈↓ in ␈↓↓Artificial Intelligence␈↓, early 1980.
␈↓ ↓H␈↓monotonic␈α∞reasoning␈α∞was␈α∞held␈α∞at␈α∂Stanford␈α∞in
␈↓ ↓H␈↓November␈α∞1978␈α∞and␈α∞attracted␈α∂50␈α∞participants     ␈↓ εx␈↓α␈↓ λπSupplementary Request 
␈↓ ↓H␈↓who␈α∂heard␈α⊂papers␈α∂by␈α∂John␈α⊂McCarthy,␈α∂Drew
␈↓ ↓H␈↓McDermott,␈α_Terry␈α_Winograd␈α→and␈α_Richard          ␈↓ εx␈↓        Recent␈α%work␈α%on␈α%formalization␈α%of
␈↓ ↓H␈↓Weyhrauch.␈α+ The␈α+proceedings␈α+of␈α*the             ␈↓ εx␈↓common␈α∞sense␈α
reasoning␈α∞has␈α
advanced␈α∞to␈α
the
␈↓ ↓H␈↓conference␈α_will␈α_be␈α→a␈α_special␈α_issue␈α→of␈α_the         ␈↓ εx␈↓point␈α
where␈α
it␈α∞is␈α
again␈α
reasonable␈α
to␈α∞plan␈α
an
␈↓ ↓H␈↓journal ␈↓↓Artificial Intelligence␈↓.           ␈↓ εx␈↓Advice␈αTaker␈α
program␈αas␈α
originally␈αproposed
                                           ␈↓ εx␈↓in␈α_McCarthy's␈α→1960␈α_paper␈α→␈↓↓Programs␈α_with
␈↓ ↓H␈↓        Recently␈α#McCarthy␈α#determined␈α#the      ␈↓ εx␈↓↓Common␈α⊃Sense␈↓.␈α⊂ Jon␈α⊃Doyle,␈α⊂who␈α⊃has␈α⊂worked
␈↓ ↓H␈↓relation␈α≡between␈α≡circumscription␈α∨and␈α≡the       ␈↓ εx␈↓on␈α$non-monotonic␈α#reasoning␈α$and␈α#truth
␈↓ ↓H␈↓McDermott-Doyle␈α⊃and␈α∩Reiter␈α⊃non-monotonic      ␈↓ εx␈↓maintenance␈αsyytems␈αat␈αM.I.T.␈αwill␈α
receive␈αhis
␈↓ ↓H␈↓formalisms.␈α& It␈α&turned␈α&out␈α'that␈α&these           ␈↓ εx␈↓PhD␈α#in␈α#June␈α#and␈α#will␈α$be␈α#available.
␈↓ ↓H␈↓formalisms␈α~di␈↓↓␈↓β@␈↓↓␈↓ered␈α~along␈α≠two␈α~dimensions.       ␈↓ εx␈↓Supplemental␈α_funds␈α→are␈α_requested␈α→for␈α_his
␈↓ ↓H␈↓McDermott,␈α∞Doyle␈α∞and␈α∞Reiter␈α∞treat␈α
possibility   ␈↓ εx␈↓support in 1980-81 as a research associate.
␈↓ ↓H␈↓modally␈α⊗by␈α↔model␈α⊗theory,␈α↔while␈α⊗McCarthy
␈↓ ↓H␈↓used␈α∂sentence␈α∞schemata.␈α∂ The␈α∂approaches␈α∞are
␈↓ ↓H␈↓substantially␈α∞but␈α∞not␈α∞precisely␈α∂equivalent,␈α∞but
␈↓ ↓H␈↓the␈α
sentence␈α
schema␈α
approach␈α
is␈α
more␈αreadily
␈↓ ↓H␈↓used␈αby␈αa␈α
computer.␈α The␈αsecond␈α
di␈↓↓␈↓β@␈↓↓␈↓erence␈αis
␈↓ ↓H␈↓that␈α∞circumscription␈α∞refers␈α∞to␈α∞minimal␈α∞models
␈↓ ↓H␈↓and␈α⊂the␈α⊂other␈α⊂methods␈α⊂use␈α⊂arbitrary␈α∂models.
␈↓ ↓H␈↓Which␈α≡is␈α∨more␈α≡useful␈α≡depends␈α∨on␈α≡the
␈↓ ↓H␈↓application.