ATheoryofPLAYandFantasy*

Thisresearchwasplannedandstartedwithanhypothesistoguideour
investigations, the task of the investigators being to collect relevant
observational data and, in the process, to amplify and modify the
hypothesis.Thehypothesiswillherebedescribedasithasgrowninour
thinking. Earlier fundamental work of Whitehead, Russell, 10
Wittgenstein,11Carnap,12Whorf,13etc.,aswellasmyownattempt14
to use this earlier thinking as an epistemological base for psychiatric
theory,ledtoaseriesofgeneralizations:
(1) That human verbal communication can operate and always does
operate at many contrasting levels of abstraction. These range in two
directionsfromtheseeminglysimpledenotativelevel("Thecatisonthe
mat"). One range or set of these more abstract levels includes those
explicit or implicit messages where the subject of discourse is the
language. We will call these metalinguistic (for example, "The verbal
sound`cat'standsforanymemberofsuchandsuchclassofobjects,"or
"Theword,`cat,'hasnofurandcannotscratch").Theothersetoflevels
of abstraction we will call metacommunicative (e.g., "My telling you
wheretofindthecatwasfriendly,"or"ThisisPLAY").Inthese,the
subjectofdiscourseistherelationshipbetweenthespeakers.*Thisessay
wasread(byJayHaley)attheA.P.A.RegionalResearchConferencein
MexicoCity,March11,1954.ItisherereprintedfromA.P.A.Psychiatric
ResearchReports,II,1955,bypermissionoftheAmericanPsychiatric
Association.10A.N.WhiteheadandB.Russell,PrincipiaMathematica,
3vols.,2nded.,Cambridge,CambridgeUniversityPress,191013.11L.

Wittgenstein,TractatusLogicoPhilosophicus,London,HarcourtBrace,
1922. 12 R. Carnap, The Logical Syntax of Language, New York,
Harcourt Brace, 1937. 13 B. L. Whorf, "Science and Linguistics,"
TechnologyReview,1940,44:22948.14J.RueschandG.Bateson,
Communication: TheSocial Matrixof Psychiatry,New York,Norton,
1951.183Itwillbenotedthatthevastmajorityofbothmetalinguistic
and metacommunicative messages remain implicit; and also that,
especially in the psychiatric interview, there occurs a further class of
implicitmessagesabouthowmetacommunicativemessagesoffriendship
andhostilityaretobeinterpreted.
(2)Ifwespeculateabouttheevolutionofcommunication,itisevident
thataveryimportantstageinthisevolutionoccurswhentheorganism
graduallyceasestorespondquite"automatically"tothemoodsignsof
anotherandbecomesabletorecognizethesignasasignal:thatis,to
recognizethattheotherindividual'sanditsownsignalsareonlysignals,
whichcanbetrusted,distrusted,falsified,denied,amplified,corrected,
and so forth. Clearly this realization that signals are signals is by no
meanscompleteevenamongthehumanspecies.Wealltoooftenrespond
automaticallytonewspaperheadlinesasthoughthesestimuliweredirect
objectindications of events in our environment instead of signals
concocted and transmitted by creatures as complexly motivated as
ourselves.Thenonhumanmammalisautomaticallyexcitedbythesexual
odorofanother;andrightlyso,inasmuchasthesecretionofthatsignis
an"involuntary"moodsign;i.e.,anoutwardlyperceptibleeventwhichis
apartofthephysiologicalprocesswhichwehavecalledamood.Inthe
human species a more complex state of affairs begins to be the rule.

Deodorantsmasktheinvoluntaryolfactorysigns,andintheirplacethe
cosmeticindustryprovidestheindividualwithperfumeswhicharenot
involuntarysignsbut voluntarysignals, recognizable as such.Many a
manhasbeenthrownoffbalancebyawhiffofperfume,andifweareto
believetheadvertisers,itseemsthatthesesignals,voluntarilyworn,have
sometimes an automatic and autosuggestive effect even upon the
voluntarywearer.Bethatasitmay,thisbriefdigressionwillserveto
illustrateastageofevolutionthedramaprecipitatedwhenorganisms,
havingeatenofthefruitoftheTreeofKnowledge,discoverthattheir
signalsaresignals.Notonlythecharacteristicallyhumaninventionof
language can then follow, but also all the complexities of empathy,
identification,projection,andsoon.Andwiththesecomesthepossibility
ofcommunicatingatthemultiplicityoflevelsofabstractionmentioned
above.184
(3)Thefirstdefinitestepintheformulationofthehypothesisguidingthis
researchoccurredinJanuary,1952,whenIwenttotheFleishhackerZoo
in San Francisco to look for behavioral criteria which would indicate
whetheranygivenorganismisorisnotabletorecognizethatthesigns
emittedbyitselfandothermembersofthespeciesaresignals.Intheory,I
hadthoughtoutwhatsuchcriteriamightlooklikethattheoccurrence
of metacommunicative signs (or signals) in the stream of interaction
betweentheanimalswouldindicatethattheanimalshaveatleastsome
awareness(consciousorunconscious)thatthesignsaboutwhichthey
metacommunicate are signals. I knew, of course, that there was no
likelihoodoffindingdenotativemessagesamongnonhumanmammals,
butIwasstillnotawarethattheanimaldatawouldrequireanalmosttotal

revision of my thinking. What I encountered at the zoo was a

phenomenon well known to everybody: I saw two young monkeys
PLAYing, i.e., engaged in an interactive sequence of which the unit
actionsorsignalsweresimilartobutnotthesameasthoseofcombat.It
wasevident,eventothehumanobserver,thatthesequenceasawhole
wasnotcombat,andevidenttothehumanobserverthattotheparticipant
monkeysthiswas"notcombat." Now,thisphenomenon,PLAY,could
onlyoccuriftheparticipantorganismswerecapableofsomedegreeof
metacommunication,i.e.,ofexchangingsignalswhichwouldcarrythe
message"thisisPLAY."
(4)Thenextstepwastheexaminationofthemessage"ThisisPLAY,"
and the realization that this message contains those elements which
necessarilygenerateaparadoxoftheRussellianorEpimenidestypea
negative statement containing an implicit negative metastatement.
Expanded, the statement "This is PLAY " looks something like this:
"Theseactionsinwhichwenowengagedonotdenotewhatthoseactions
forwhichtheystandwoulddenote."Wenowaskabouttheitalicized
words,"forwhichtheystand."Wesaytheword"cat"standsforany
member of a certain class. That is, the phrase "stands for" is a near
synonymof"denotes."Ifwenowsubstitute"whichtheydenote"forthe
words"forwhichtheystand"intheexpandeddefinitionofPLAY,the
resultis:"Theseactions,in185whichwenowengage,donotdenote
whatwouldbedenotedbythoseactionswhichtheseactionsdenote."
ThePLAYfulnipdenotesthebite,butitdoesnotdenotewhatwouldbe
denotedbythebite.AccordingtotheTheoryofLogicalTypessucha
messageisofcourseinadmissable,becausetheword"denote"isbeing

used in two degrees of abstraction, and these two uses are treated as
synonymous.Butallthatwelearnfromsuchacriticismisthatitwould
bebadnaturalhistorytoexpectthementalprocessesandcommunicative
habitsofmammalstoconformtothelogician'sideal.Indeed,ifhuman
thought and communication always conformed to the ideal, Russell
wouldnotinfactcouldnothaveformulatedtheideal.
(5)Arelatedproblemintheevolutionofcommunicationconcernsthe
originofwhatKorzybski15hascalledthemapterritoryrelation:thefact
thatamessage,ofwhateverkind,doesnotconsistofthoseobjectswhich
itdenotes("Theword`cat'cannotscratchus").Rather,languagebearsto
theobjectswhichitdenotesarelationshipcomparabletothatwhicha
mapbearstoaterritory.Denotativecommunicationasitoccursatthe
human level is only possible after the evolution of a complex set of
metalinguistic(butnotverbalized)16ruleswhichgovernhowwordsand
sentencesshallberelatedtoobjectsandevents.Itisthereforeappropriate
to look for the evolution of such metalinguistic and/or meta
communicativerulesataprehumanandpreverballevel.Itappearsfrom
whatissaidabovethatPLAYisaphenomenoninwhichtheactionsof
"PLAY" are related to, or denote, other actions of "not PLAY." We
thereforemeetinPLAYwithaninstanceofsignalsstandingforother
events,anditappears,therefore,thattheevolutionofPLAYmayhave
beenanimportantstepintheevolutionofcommunication.
(6)ThreatisanotherphenomenonwhichresemblesPLAYinthatactions
denote,butaredifferentfrom,otheractions.Theclenchedfistofthreatis
differentfrom thepunch,butitreferstoapossible15A.Korzybski,

ScienceandSanity,NewYork,SciencePress,1941.16Theverbalization
ofthesemetalinguisticrulesisamuchlaterachievementwhichcanonly
occur after theevolutionof anonverbalizedmetametalinguistics. 186
future(butatpresentnonexistent)punch.Andthreatalsoiscommonly
recognizable among nonhuman mammals. Indeed it has lately been
arguedthatagreatpartofwhatappearstobecombatamongmembersof
asinglespeciesisrathertoberegardedasthreat(Tinbergen,17Lorenz18
).
(7)Histrionicbehavioranddeceitareotherexamplesoftheprimitive
occurrence of mapterritory differentiation. And there is evidence that
dramatizationoccursamongbirds:ajackdawmayimitateherownmood
signs(Lorenz19),anddeceithasbeenobservedamonghowlermonkeys
(Carpenter20).
(8) We might expect threat, PLAY, and histrionics to be three
independent phenomena all contributing to the evolution of the
discriminationbetweenmapandterritory.Butitseemsthatthiswouldbe
wrong,atleastsofarasmammaliancommunicationisconcerned.Very
brief analysis of childhood behavior shows that such combinations as
histrionicPLAY,bluff, PLAYfulthreat, teasing PLAY inresponse to
threat,histrionicthreat,andsoonformtogetherasingletotalcomplexof
phenomena.AndsuchadultphenomenaasgamblingandPLAYingwith
riskhavetheirrootsinthecombinationofthreatandPLAY.Itisevident
alsothatnotonlythreatbutthereciprocalofthreatthebehaviorofthe
threatenedindividualareapartofthiscomplex.Itisprobablethatnot

only histrionics but also spectatorship should be included within this

field.Itisalsoappropriatetomentionselfpity.
(9)Afurtherextensionofthisthinkingleadsustoincluderitualwithin
this general field in which the discrimination is drawn, but not
completely,betweendenotativeactionandthatwhichistobedenoted.
Anthropological studies of peacemaking ceremonies, to cite only one
example, support this conclusion. In the Andaman Islands, peace is
concludedaftereachsidehasbeengivenceremonialfreedomtostrikethe
other. This example, however, also illustrates the labile nature of the
frame"ThisisPLAY,"17N.Tinbergen,SocialBehaviorinAnimals
withSpecialReferencetoVertebrates,London,Methuen,1953.18K.Z.
Lorenz,KingSolomon'sRing,NewYork,Crowell,1952.19Ibid.20C.
R.Carpenter,"AFieldStudyoftheBehaviorandSocialRelationsof
HowlingMonkeys,"Comp.Psychol.Monogr.,1934,10:1168.187or
"Thisisritual."Thediscriminationbetweenmapandterritoryisalways
liabletobreakdown,andtheritualblowsofpeacemakingarealways
liabletobemistakenforthe"real"blowsofcombat.Inthisevent,the
peacemakingceremonybecomesabattle(RadcliffeBrown21).
(10)ButthisleadsustorecognitionofamorecomplexformofPLAY;
thegamewhichisconstructednotuponthepremise"ThisisPLAY"but
ratheraroundthequestion"IsthisPLAY?"Andthistypeofinteraction
alsohasitsritualforms,e.g.,inthehazingofinitiation.
(11)Paradoxisdoublypresentinthesignalswhichareexchangedwithin
thecontextofPLAY,fantasy,threat,etc.NotonlydoesthePLAYfulnip

notdenotewhatwouldbedenotedbythebiteforwhichitstands,but,in
addition,thebiteitselfisfictional.NotonlydothePLAYinganimalsnot
quite mean what they are saying but, also, they are usually
communicating about something which does not exist. At the human
level,thisleadstoavastvarietyofcomplicationsandinversionsinthe
fieldsofPLAY,fantasy,andart.Conjurersandpaintersofthetrompe
l'oeilschoolconcentrateuponacquiringavirtuositywhoseonlyrewardis
reachedaftertheviewerdetectsthathehasbeendeceivedandisforcedto
smileormarvelattheskillofthedeceiver.Hollywoodfilmmakersspend
millions of dollars to increase the realism of a shadow. Other artists,
perhaps more realistically, insist that art be nonrepresentational; and
pokerPLAYersachieveastrangeaddictiverealismbyequatingthechips
forwhichtheyPLAYwithdollars.Theystillinsist,however,thatthe
loseraccepthislossaspartofthegame.Finally,inthedimregionwhere
art,magic,andreligionmeetandoverlap,humanbeingshaveevolvedthe
"metaphorthatismeant,"theflagwhichmenwilldietosave,andthe
sacramentthatisfelttobemorethan"anoutwardandvisiblesign,given
unto us." Here we can recognize an attempt to deny the difference
betweenmapandterritory,andtogetbacktotheabsoluteinnocenceof
communicationbymeansofpuremoodsigns.
(12)WefacethentwopeculiaritiesofPLAY:(a)thatthemessagesor
signalsexchangedinPLAYareinacertainsenseuntrueornot21A.R.
RadcliffeBrown, The Andaman Islanders, Cambridge, Cambridge
UniversityPress,1922.188meant;and(b)thatthatwhichisdenotedby
thesesignalsisnonexistent.Thesetwopeculiaritiessometimescombine
strangelytoareverseaconclusionreachedabove.Itwasstated(4)that

thePLAYfulnipdenotesthebite,butdoesnotdenotethatwhichwould
bedenotedbythebite.Butthereareotherinstanceswhereanopposite
phenomenonoccurs.Amanexperiencesthefullintensityofsubjective
terrorwhenaspearisflungathimoutofthe3Dscreenorwhenhefalls
headlongfromsomepeakcreatedinhisownmindintheintensityof
nightmare.Atthemomentofterrortherewasnoquestioningof"reality,"
but still there was no spear in the movie house and no cliff in the
bedroom.Theimagesdidnotdenotethatwhichtheyseemedtodenote,
butthesesameimagesdidreallyevokethatterrorwhichwouldhavebeen
evokedbyareal spear or arealprecipice.Byasimilartrickofself
contradiction, the filmmakers of Hollywood are free to offer to a
puritanicalpublicavastrangeofpseudosexualfantasywhichotherwise
would not be tolerated. In David and Bathsheba, Bathsheba can be a
Troilistic link between David and Uriah. And in Hans Christian
Andersen,theherostartsoutaccompaniedbyaboy.Hetriestogeta
woman,butwhenheisdefeatedinthisattempt,hereturnstotheboy.In
allofthis,thereis,ofcourse,nohomosexuality,butthechoiceofthese
symbolisms is associated in these fantasies with certain characteristic
ideas,e.g.,aboutthehopelessnessoftheheterosexualmasculineposition
whenfacedwithcertainsortsofwomenorwithcertainsortsofmale
authority.Insum,thepseudohomosexualityofthefantasydoesnotstand
foranyrealhomosexuality,butdoesstandforandexpressattitudeswhich
mightaccompanyarealhomosexualityorfeeditsetiologicalroots.The
symbols do not denote homosexuality, but do denote ideas for which
homosexualityisanappropriatesymbol.Evidentlyitisnecessarytore
examine the precise semantic validity of the interpretations which the
psychiatristofferstoapatient,and,aspreliminarytothisanalysis,itwill

be necessary to examine the nature of the frame in which these

interpretationsareoffered.
(13) What has previously been said about PLAY can be used as an
introductoryexampleforthediscussionofframesandcontexts.Insum,it
is our hypothesis that the message "This is PLAY" establishes 189 a
paradoxicalframecomparabletoEpimenides'paradox.Thisframemay
be diagrammed thus: The first statement within this frame is a self
contradictorypropositionaboutitself.Ifthisfirststatementistrue,thenit
mustbefalse.Ifitbefalse,thenitmustbetrue.Butthisfirststatement
carries with it all the other statements in the frame. So, if the first
statementbetrue,thenalltheothersmustbefalse;andviceversa,ifthe
firststatementbeuntruethenalltheothersmustbetrue.
(14)Thelogicallymindedwillnoticeanonsequitur.Itcouldbeurged
thatevenifthefirststaementisfalse,thereremainsalogicalpossibility
thatsomeoftheotherstatementsintheframeareuntrue.Itis,however,a
characteristic of unconscious or "primaryprocess" thinking that the
thinkerisunabletodiscriminatebetween"some"and"all,"andunableto
discriminatebetween"notall"and"none."Itseemsthattheachievement
ofthesediscriminationsisperformedbyhigherormoreconsciousmental
processeswhichserveinthenonpsychoticindividualtocorrecttheblack
andwhitethinkingofthelowerlevels.Weassume,andthisseemstobe
anorthodoxassumption,thatprimaryprocessiscontinuallyoperating,
and that the psychological validity of the paradoxical PLAY frame
dependsuponthispartofthemind.

(15)But,conversely,whileitisnecessarytoinvoketheprimaryprocess
asanexplanatoryprincipleinordertodeletethenotionof"some"from
between"all"and "none,"thisdoes notmeanthatPLAYissimplya
primaryprocessphenomenon.Thediscriminationbetween"PLAY"and"
nonPLAY,"likethediscriminationbetweenfantasyandnonfantasy,is
certainlyafunctionofsecondaryprocess,or"ego."Withinthedreamthe
dreamerisusuallyunawarethatheisdreaming,andwithin"PLAY"he
mustoftenberemindedthat"ThisisPLAY."190Similarly,withindream
orfantasythedreamerdoesnotoperatewiththeconcept"untrue."He
operateswithallsortsofstatementsbutwithacuriousinabilitytoachieve
metastatements.Hecannot,unlessclosetowaking,dreamastatement
referringto(i.e.,framing)hisdream.ItthereforefollowsthatthePLAY
frame as here used as an explanatory principle implies a special
combination of primary and secondary processes. This, however, is
relatedtowhatwassaidearlier,whenitwasarguedthatPLAYmarksa
stepforwardintheevolutionofcommunicationthecrucialstepinthe
discoveryofmapterritoryrelations.Inprimaryprocess,mapandterritory
areequated;insecondaryprocess,theycanbediscriminated.InPLAY,
theyarebothequatedanddiscriminated.(16)Anotherlogicalanomalyin
this system must be mentioned: that the relationship between two
propositionswhichiscommonlydescribedbytheword"premise"has
become intransitive. In general, all asymmetrical relationships are
transitive.Therelationship"greaterthan"istypicalinthisrespect;itis
conventionaltoarguethatifAisgreaterthanB,andBisgreaterthanC,
thenAisgreaterthanC.Butinpsychologicalprocessesthetransitivityof
asymmetrical relations is not observed. The proposition P may be a
premiseforQ;QmaybeapremiseforR;andRmaybeapremiseforP.

Specifically,inthesystemwhichweareconsidering,thecircleisstill
more contracted. The message, "All statements within this frame are
untrue"isitselftobetakenasapremiseinevaluatingitsowntruthor
untruth.(Cf.theintransitivityofpsychologicalpreferencediscussedby
McCulloch.22Theparadigmforallparadoxesofthisgeneraltypeis
Russell's23 "class of classes which are not members of themselves."
Here Russell demonstrates that paradox is generated by treating the
relationship,"isamemberof,"asanintransitive.)Withthiscaveat,that
the"premise"relationinpsychologyislikelytobeintransitive,weshall
usetheword"premise"todenoteadependencyofoneideaormessage
uponanothercomparabletothedependencyofonepropositionupon22
W. S. McCulloch, "A Heterarchy of Values, etc.," Bulletin of Math.
Biophys.,1945,7:8993.23WhiteheadandRussell,op.cit.191another
whichisreferredtoinlogicbysayingthatthepropositionPisapremise
forQ.
(17)Allthis,however,leavesunclearwhatismeantby"frame"andthe
relatednotionof"context."Toclarifythese,itisnecessarytoinsistfirst
thatthesearepsychologicalconcepts.Weusetwosortsofanalogyto
discussthesenotions:thephysicalanalogyofthepictureframeandthe
moreabstract,butstillnotpsychological,analogyofthemathematical
set. In set theory the mathematicians have developed axioms and
theoremstodiscusswithrigorthelogicalimplicationsofmembershipin
overlapping categories or "sets." The relationships between sets are
commonlyillustratedbydiagramsinwhichtheitemsormembersofa
largeruniversearerepresentedbydots,andthesmallersetsaredelimited
byimaginarylinesenclosingthemembersofeachset.Suchdiagrams

thenillustrateatopologicalapproachtothelogicofclassification.The
firststepindefiningapsychologicalframemightbetosaythatitis(or
delimits)aclassorsetofmessages(ormeaningfulactions).ThePLAYof
twoindividualsonacertainoccasionwouldthenbedefinedasthesetof
all messages exchanged by them within a limited period of time and
modifiedbytheparadoxicalpremisesystemwhichwehavedescribed.In
asettheoreticaldiagramthesemessagesmightberepresentedbydots,
andthe"set"enclosedbyalinewhichwouldseparatethesefromother
dotsrepresentingnonPLAYmessages.Themathematicalanalogybreaks
down, however, because the psychological frame is not satisfactorily
represented by an imaginary line. We assume that the psychological
framehassomedegreeofrealexistence.Inmanyinstances,theframeis
consciously recognized and even represented in vocabulary ("PLAY,"
"movie,""interview,""job,""language,"etc.).Inothercases,theremay
benoexplicitverbalreferencetotheframe,andthesubjectmayhaveno
consciousnessofit.Theanalyst,however,findsthathisownthinkingis
simplifiedifheusesthenotionofanunconsciousframeasanexplanatory
principle;usuallyhegoesfurtherthanthisandinfersitsexistenceinthe
subject'sunconscious.Butwhiletheanalogyofthemathematicalsetis
perhaps over abstract, the analogy of the picture frame is excessively
concrete. The psychological concept which we are trying to define is
neither physical nor logical. Rather, the actual physical frame is, we
believe,192addedbyhumanbeingstophysicalpicturesbecausethese
humanbeingsoperatemoreeasilyinauniverseinwhichsomeoftheir
psychologicalcharacteristicsareexternalized.Itisthesecharacteristics
which.wearetryingtodiscuss,usingtheexternalizationasanillustrative
device.

(18)Thecommonfunctionsandusesofpsychologicalframesmaynow
belistedandillustratedbyreferencetotheanalogieswhoselimitations
havebeenindicatedinthepreviousparagraph:(a)Psychologicalframes
areexclusive,i.e.,byincludingcertainmessages(ormeaningfulactions)
within a frame, certain other messages are excluded. (b)Psychological
framesareinclusive,i.e.,byexcludingcertainmessagescertainothersare
included.Fromthepointofviewofsettheorythesetwofunctionsare
synonymous,butfromthepointofviewofpsychologyitisnecessaryto
listthemseparately.Theframearoundapicture,ifweconsiderthisframe
asamessageintendedtoorderororganizetheperceptionoftheviewer,
says,"Attendtowhatiswithinanddonotattendtowhatisoutside."
Figureandground,asthesetermsareusedbygestaltpsychologists,are
not symmetrically related as are the set and nonset of set theory.
Perceptionofthegroundmustbepositivelyinhibitedandperceptionof
the figure (in this case the picture) must be positively enhanced.
(c)Psychologicalframesarerelatedtowhatwehavecalled"premises."
Thepictureframetellstheviewerthatheisnottousethesamesortof
thinkingininterpretingthepicturethathemightuseininterpretingthe
wallpaperoutsidetheframe.Or,intermsoftheanalogyfromsettheory,
themessagesenclosedwithintheimaginarylinearedefinedasmembers
of a class by virtue of their sharing common premises or mutual
relevance.Theframeitselfthusbecomesapartofthepremisesystem.
Either,asinthecaseofthePLAYframe,theframeisinvolvedinthe
evaluationofthemessageswhichitcontains,ortheframemerelyassists
the mind in understanding the contained messages by reminding the
thinker that these messages are mutually relevant and the messages

outside the frame may be ignored. (d)In the sense of the previous
paragraph,aframeismetacommunicative.Anymessage,whicheither
explicitly or implicitly defines a frame, ipso facto gives the receiver
instructions 193 or aids in his attempt to understand the messages
included within the frame. (e)The converse of (d) is also true. Every
metacommunicativeormetalinguisticmessagedefines,eitherexplicitly
orimplicitly,thesetofmessagesaboutwhichitcommunicates,i.e.,every
metacommunicativemessageisordefinesapsychologicalframe.This,
forexample,isveryevidentinregardtosuchsmallmetacommunicative
signalsaspunctuationmarksinaprintedmessage,butappliesequallyto
such complex metacommunicative messages as the psychiatrist's
definitionofhisowncurativeroleintermsofwhichhiscontributionsto
the whole mass of messages in psychotherapy are to be understood.
(f)Therelationbetweenpsychologicalframeandperceptualgestaltneeds
tobeconsidered,andheretheanalogyofthepictureframeisuseful.Ina
painting by Roualt or Blake, the human figures and other objects
representedareoutlined."Wisemenseeoutlinesandthereforetheydraw
them."But outsidetheselines,whichdelimittheperceptualgestaltor
"figure,"thereisabackgroundor"ground"whichinturnislimitedbythe
pictureframe.Similarly,insettheoreticaldiagrams,thelargeruniverse
withinwhichthesmallersetsaredrawnisitselfenclosedinaframe.This
doubleframingis, we believe, not merelya matter of "frames within
frames"butanindicationthatmentalprocessesresemblelogicinneeding
anouterframetodelimitthegroundagainstwhichthefiguresaretobe
perceived. This need is often unsatisfied, as when we see a piece of
sculptureinajunkshopwindow,butthisisuncomfortable.Wesuggest
thattheneedforthisouterlimittothegroundisrelatedtoapreference

foravoidingtheparadoxesofabstraction.Whenalogicalclassorsetof
itemsisdefinedforexample,theclassofmatchboxesitisnecessary
todelimitthesetofitemswhicharetobeexcluded,inthiscase,allthose
thingswhicharenotmatchboxes.Buttheitemstobeincludedinthe
backgroundsetmustbeofthesamedegreeofabstraction,i.e.,ofthe
same"logicaltype"asthosewithinthesetitself.Specifically,ifparadox
is to be avoided, the "class of matchboxes" and the "class of
nonmatchboxes" (even though both these items are clearly not
matchboxes) must not be regarded as members of the class of
nonmatchboxes.Noclasscanbeamember,ofitself.Thepicture194
frame then, because it delimits a background, is here regarded as an
externalrepresentationofaveryspecialandimportanttypeofpsycho
logical framenamely a frame whose function is to delimit a logical
type.This,infact,iswhatwasindicatedabovewhenitwassaidthatthe
pictureframeisaninstructiontotheviewerthatheshouldnotextendthe
premiseswhichobtainbetweenthefigureswithinthepicturetothe.,wall
paperbehindit.But,itispreciselythissortofframethatprecipitates
paradox.Theruleforavoidingparadoxesinsiststhattheitemsoutside
anyenclosinglinebeofthesamelogicaltypeasthosewithin,butthe
pictureframe,asanalyzedabove,isalinedividingitemsofonelogical
type from those of another. In passing, it is interesting to note that
Russell'srulecannotbestatedwithoutbreakingtherule.Russellinsists
that all items of inappropriate logical type be exluded (i.e., by an
imaginaryline)fromthebackgroundofanyclass,i.e.,heinsistsuponthe
drawingofanimaginarylineofpreciselythesortwhichheprohibits.(19)
Thiswholematterofframesandparadoxesmaybeillustratedintermsof
animal behavior, wherethree types of message maybe recognized or

deduced:(a)Messagesofthesortwhichweherecallmoodsigns;(b)
messageswhichsimulatemoodsigns(inPLAY,threat,histrionics,etc.);
and (c) messages which enable the receiver to discriminate between
moodsigns andthose other signs which resemble them.Themessage
"ThisisPLAY"isofthisthirdtype.Ittellsthereceiverthatcertainnips
and other meaningful actions are not messages of the first type. The
message"ThisisPLAY"thussetsaframeofthesortwhichislikelyto
precipitateparadox:itisanattempttodiscriminatebetween,ortodrawa
linebetween,categoriesofdifferentlogicaltypes.
(20) This discussion of PLAY and psychological frames establishes a
type of triadic constellation (or system of relationships) between
messages.Oneinstanceofthisconstellationisanalyzedinparagraph19,
but it is evident that constellations of this sort occur not only at the
nonhumanlevelbutalsointhemuchmorecomplexcommunicationof
humanbeings.Afantasyormythmaysimulateadenotativenarrative,
and, to discriminate between these types of discourse, people use
messagesoftheframesettingtype,andsoon.195
(21) In conclusion, we arrive at the complex task of applying this
theoreticalapproachtotheparticularphenomenaofpsychotherapy.Here
thelinesofourthinkingmaymostbrieflybesummarizedbypresenting
andpartiallyansweringthesequestions:(a)Isthereanyindicationthat
certain forms of psychopathology are specifically characterized by
abnormalities inthepatient's handlingof framesandparadoxes?(b)Is
there any indication that the techniques of psychotherapy necessarily
dependuponthemanipulationofframesandparadoxes?(c)Isitpossible

to describe the process of a given psychotherapy in terms of the

interaction between the patient's abnormal use of frames and the
therapist'smanipulationofthem?
(22) In reply to the first question, it seems that the "word salad" of
schizophrenia can be described in terms of ,the patient's failure to
recognizethemetaphoricnatureofhisfantasies.Inwhatshouldbetriadic
constellationsofmessages.,theframesettingmessage(e.g.,thephrase
"asif")isomitted,andthemetaphororfantasyisnarratedandactedupon
inamannerwhichwouldbeappropriateifthefantasywereamessageof
themoredirectkind.Theabsenceofmetacommunicativeframingwhich
was noted in the case of dreams (15) is characteristic of the waking
communicationsoftheschizophrenic.Withthelossoftheabilitytoset
metacommunicativeframes,thereisalsoalossofabilitytoachievethe
moreprimaryorprimitivemessage.Themetaphoristreateddirectlyasa
messageofthemoreprimarytype.(Thismatterisdiscussedatgreater
lengthinthepapergivenbyJayHaleyatthisConference.)
(23)Thedependenceofpsychotherapyuponthemanipulationofframes
followsfromthefactthattherapyisanattempttochangethepatient's
metacommunicative habits. Before therapy, the patient thinks and
operates in terms of a certain set of rules for the making and
understandingofmessages.Aftersuccessfultherapy;heoperatesinterms
of a different set of such rules. (Rules of this sort are in general,
unverbalized,andunconsciousbothbeforeandafter.)Itfollowsthat,in
theprocessoftherapy,theremusthavebeencommunicationatalevel
metatotheserules.Theremusthavebeencommunicationaboutachange

inrules.Butsuchacommunicationaboutchangecouldnotconceivably
occur in messages of the type permitted by the patient's 196
metacommunicativerulesastheyexistedeitherbeforeoraftertherapy.It
wassuggestedabovethattheparadoxesofPLAYarecharacteristicofan
evolutionarystep.Herewesuggestthatsimilarparadoxesareanecessary
ingredientinthatprocessofchangewhichwecallpsychotherapy.The
resemblance between the process of therapy and the phenomenon of
PLAYis,infact,profound.Bothoccurwithinadelimitedpsychological
frame,aspatialandtemporalboundingofasetofinteractivemessages.
In bothPLAYand therapy,the messages have a special and peculiar
relationshiptoamoreconcreteorbasicreality.Justasthepseudocombat
ofPLAYisnotrealcombat,soalsothepseudoloveandpseudohateof
therapyarenotrealloveandhate.The"transfer"isdiscriminatedfrom
real love and hate by signals invoking the psychological frame; and
indeeditisthisframewhichpermitsthetransfertoreachitsfullintensity
and to be discussed between patient and therapist. The formal
characteristicsofthetherapeuticprocessmaybeillustratedbybuilding
upamodelinstages.ImaginefirsttwoPLAYerswhoengageinagame
ofcanastaaccordingtoastandardsetofrules.Solongastheserules
governandareunquestionedbybothPLAYers,thegameisunchanging,
i.e:, no therapeutic change will occur. (Indeed many attempts at
psychotherapyfailforthisreason.)Wemayimagine,however,thatata
certainmomentthetwocanastaPLAYersceasetoPLAYcanastaand
startadiscussionoftherules.Theirdiscourseisnowofadifferentlogical
type from that of their PLAY. At the end of this discussion, we can
imagine that they return to PLAYing but with modified rules. This
sequenceofeventsis,however,stillanimperfectmodeloftherapeutic

interaction,thoughitillustratesourcontentionthattherapynecessarily
involves a combination of discrepant logical types of discourse. Our
imaginaryPLAYersavoidedparadoxbyseparatingtheirdiscussionofthe
rules from their PLAY, and it is precisely this separation that is
impossibleinpsychotherapy.Asweseeit,theprocessofpsychotherapy
is a framed interaction between two persons, in which the rules are
implicit but subject to change. Suchchange can only beproposed by
experimental action, but every such experimental action, in which a
proposaltochange197therulesisimplicit,isitselfapartoftheongoing
game.Itisthiscombinationoflogicaltypeswithinthesinglemeaningful
actthatgivestotherapythecharacternotofarigidgamelikecanastabut,
instead,thatofanevolvingsystemofinteraction.ThePLAYofkittensor
ottershasthischaracter.
(24)Inregardtothespecificrelationshipbetweenthewayinwhichthe
patienthandlesframesandthewayinwhichthetherapistmanipulates
them, very little can at present be said. It is, however, suggestive to
observethatthepsychologicalframeoftherapyisananalogueofthe
framesettingmessagewhichtheschizophrenicisunabletoachieve.To
talkin"wordsalad"withinthepsychologicalframeoftherapyis,ina
sense,notpathological.Indeedtheneuroticisspecificallyencouragedto
dopreciselythis,narratinghisdreamsandfreeassociationssothatpatient
and therapist may achieve an understanding of this material. By the
processofinterpretation,theneuroticisdriventoinsertan"asif"clause
intotheproductionsofhisprimaryprocessthinking,whichproductions
hehadpreviouslydeprecatedorrepressed.Hemustlearnthatfantasy
containstruth.Fortheschizophrenictheproblemissomewhatdifferent.

Hiserrorisintreatingthemetaphorsofprimaryprocesswiththefull
intensityofliteraltruth.Throughthediscoveryofwhatthesemetaphors
standforhemustdiscoverthattheyareonlymetaphors.
(25) From the point of view of the project, however, psychotherapy
constitutes only one of the many fields which we are attempting to
investigate.Ourcentralthesismaybesummedupasastatementofthe
necessityoftheparadoxesofabstraction.Itisnotmerelybadnatural
historytosuggestthatpeoplemightorshouldobeytheTheoryofLogical
Typesintheircommunications;theirfailuretodothisisnotduetomere
carelessness or ignorance. Rather, we believe that the paradoxes of
abstraction must make their appearance in all communication more
complexthanthatofmoodsignals,andthatwithouttheseparadoxesthe
evolutionofcommunicationwouldbeatanend.Lifewouldthenbean
endlessinterchangeofstylizedmessag