Vol.l•S? 54]• Lon•z, TheCompanion in the Bird's World 245

THE COMPANION IN THE BIRD'S WORLD

BY KONRAD Z. LORENZ IntroductoryNote WnE• Dr. Lorenz'spaper on the 'Kumpan'• waspublished in 1935,it seemedto be such an original and importas•t contribution to our knowledgeof the instincts and behavior of birds, that I was as•xiousto have a summaw of its thesesand conclusions appear in the pages of 'The Auk'. Knowing that Dr. Lorenz was conversantwith English, I wrote to him, askingif he would consentto prepare suchan account of his own work, and he replied that he would be pleasedto do so if given sufficienttime. Dr. Lorenz's paper was delayed in reaching me from various causes. He found difficulty in summarizing a work which consistedin so large a measureof observa- tional detail, and accordingly has stressed those subjectsof the most theoretical importance; and he added: "I shall not refrain from using some new observations, and the resultsof experimentswhich have come to my knowledgesince the publica- tion of my paper." My function has been mainly editorial; and although it has seemedbest to condensesome parts of the author's manuscript, I trust that its value has not been impaired. The doctrine of 'releasers,' herein set forth,--or devices for the production of stimuli, which serve as the 'keys' to 'unlock' or releasethose 'innate perceptory patterns,' characteristicof the speciesand the individual, and which result in in- stinctive reactions,--seemsat last to offer a soundand satisfying explanation of that riddle so long embodiedin 'the secondarysexual charactersof birds,' and one which I believe that Darwin, when strugglingwith his 'Theory of Sexual Selection,' would have welcomedwith open arms.--FRA•c•s H. HERa•CK.

THE CONCEPTION OF THE "RELEASER" IF WE observea man performinga numberof separateactions, each of which has for its object the samething and which, taken togetherand con- sidered as a functional unit, serve an end of definite survival value to the man himself, we shall not hesitate to infer two things: first, that the man, in performingthese actions, is consciousof their goal and that it is this very goal that constitutesthe purposeof all his movements;second, that in per- forming the several actionsof the sequence,he is constantly aware of the identity of the object toward which they are directed. On seeinga similar seriesof actionsexecuted by someanimal, we are prone to cometo the same conclusions,but both would be totally erroneousin casewe shouldhave to do with a seriesof purely instinctiveacts. Much seriousharm has been doneby confoundingthe biologicalend, or survivalvalue of {nstlnct{veaction, with the purposepursued by the animal as a subject. I cannot go into the details of my argument against this wldelyspread error, but for the purposesof this paper,suffice it to say that

• 'Der Kumpan in der Urnwelt des Vogels (Der Artgenosse als aus16sendesMoment sozialer Verhaltungsweisen)' [The Companion in the Bird's World; the fellow-member of the species as a releasing factor of social behavior.] Journ. f. Ornith., vol. 83, pt. 2-3, 1935. 246 TheCompanion inthe Bird's World [July[Auk

I considerinstinctive action as 'desiredreflex action', meaningthat instinc- tive action, while in no way purposivein itself and quite as rigid as any other reflex as to the coordinationof movements,yet in its very 'goingoff' constitutesthe purposeof all truly purposiveor appetitive behaviorever met with in animals. Beingappeted, sought or desired,is the one point in which instinctive action differs from all other reflex actions. The common reflexes,like unusedmachines, may lie idle indefinitely, as if waiting for adequatestimulation, without in any way inciting the animal to searchfor suchstimulation by purposivebehavior, as it certainly doesin caseof the instinctive reactions. It is the secondof the assumptionsmentioned that concernsus more closely;namely, that the animal, in consistentlytreating one definiteobject to a seriesof separateactions, must throughoutits acting, be aware of the identity of this object. Man, in striving to conquerhis environmentand to master the problemswhich it puts before him by connectingcause and effect,is confrontedwith the absolutenecessity of conceivingevery object he encountersas a permanentunit which remainsidentical throughout the changesof spaceand time. Not so the animal, moreespecially the lower animal, that doesnot masterits environmentby insightor learningbut is innatelyadapted to it by possessinghighly differentiatedinstinctive powers of response. The lower animal very rarely 'solvesproblems' as an individ- ual, and wheneverit doesso, this is only in an abysmallysimple way. As a rule, a lowerorganism either has solved a problemas a speciesby evolving instinctiveways of reactingthat fit everysingle one of the emergenciesthat normally occur,or its fails miserablywhen confrontedwith a problemfor whichit doesnot possessany suitableinstinctive responses. Since the in- nate coordinations of movement that we call instinctive actions are not adaptedto any goalor end anyhow,it sufficesthat everyone of them be re- leased,much as in reflexes,by a simplecombination of stimuli,if only this combinationis sufficientlycharacteristic of the biologically'right' situation; that is, of the one situationin which the performanceof the innately co- ordinated movements attains its full survival value. If an instinctive action is directedtoward a particularobject, its successfulperformance is inde- pendentof the animal'sperceiving this objectas a 'thing', that is, as a per- manentunit in spaceand time as we wouldperceive it. All that is needed on the part of the animalis the dispositionto respondwith just that reaction to certain stimuli characteristicof its particular object. The sea-urchin (Sphaerechinus)possesses a very complicatedreaction of combineddefense and flight, whichis of survivalvalue exclusivelyas a meansof escapingthe sea-urchin'schief enemy, the starfish(Asterias). The whole of this highly differentiatedsequence of singleacts is releasedby a singlechemical stimu- lus emanatingfrom the mucus coveringthe ambulacral tentaclesof the Vol.1937 541J LORENZ,The Companion in the Bird's World 247 starfish. Anothergood example is that of the biting reactionof the common tick (Ixodes). The tick will bite, or try to bite, any objectpossessing the two charactersof (1) smellingof butyricacid and (2) havinga temperature of approximately37 degreescentigrade. The releasingof comparatively complicatedsets of reactionsthrough a very simplecombination of stimuli, is characteristicof the greaterpart of all purelyinstinctive reactions and by no means exclusivelyof those of lower animals. The Jackdaw (Coloeus moncdula)possesses a very interestingreaction of defendingany fellow- memberof the speciesin the grip of somebird or animal of prey. ]?or a long time I have been familiar with the fact that my tame but free-living Jackdawswould furiously attack me if I grippedone of them in my hand, but I was very much astonishedwhen I inadvertently elicitedexactly the same responseby carrying a wet, black bathing-suitin my hand. Subse- quent experimentsshowed that anything glisteningblack and dangling, carriedby any living creaturewould releasethe very samereaction in the Jackdaws. Even Jackdawsthemselves were subjectto attack from their fellowswhen they happenedto carry nestingmaterial possessingthe char- acteristicsjust mentioned. In a bird that readily recognizesand discrimi- natesfellow-members of the flock,it is rather surprisingthat the processof releasingthe socialdefending reaction so closelyresembles that which elicitsthe responsesof lowerorganisms as alreadynoticed. An instinctivereaction of survival value, when directedexclusively to a particularobject, may be releasedas if througha surprisinglysmall choice amongthe large numberof stimuli normally emanatingfrom the object. In this circumstance,the reactionsof the highestvertebrates do not differ from those of ticks, sea-urchinsand the like. The receptorypart of the animal'sreaction responds to a small but characteristiccombination of stimuli,very muchin the way in whicha highlyselective wireless receiving- station is attuned to a particularwave-length. This perceptorycorrelate to a very definitecombination of stimuli has beentermed by William 1VIc- Dougall the 'innate perceptoryinlet.' Without knowing this author's terminology,I used the term 'angeborenesAuslSse-Schema' in German which,respecting the priorityof McDougall'sterm, I shallcompromise by translating'innate perceptorypattern.' It is an old but fitting metaphorto liken the releasingset of stimuli to the key,and the innateperceptory pattern to the lockof the instinctivere- action. Even more appropriate,is the simile of a combinationlock that cannotbe openedexcept by a definiteseries of manipulationswhich, by reasonof their generalimprobability, it is practicallyimpossible to find by chance. The relationof the particularform of the lockto the key that fits it, or of any innate perceptorypattern to the set of stimulito whichit re- sponds,is ever a compromisebetween greatest possible simplicity and 248 LorENz,The Companion in the Bird's World [July[Auk greatestpossible general improbability. The improbability of the innate perceptorypattern is to guardthe instinctivereaction from beingreleased by chancethrough other than the biologically'right' influences. Sur- prisinglysimple though the innate perceptorypatterns are in the three casescited as examples,they are evidently efficientenough, when natural conditionsare taken for granted, to prevent the 'erroneous'unlocking of the reaction. It is extremelyunlikely that in its natural state the tick is ever misledby its innate perceptorypattern to bite or to try to bite any ob- ject that is of no useto it as a potential host. Nor will it often occurin wild life that a Jackdawerroneously defends a black bathing-suitinstead of a fellow Jackdaw caught in the grip of someanimal of prey. The speciesin thesecases need not reckonwith the experimentinghuman, cleverly imitat- ing the key to the instinctivereaction's lock. But certainlythere are in- stancesin whichthe very simplicityof innateperceptory patterns may offer an opportunityto entrapanimals by artificiallysupplying key stimuli that releasecertain definite actions, as the pike is caughtby meansof a pieceof tin that is drawn through the water, and suppliesthe quickly moving silveryreflections which normally release the preyingreactions in this fish. If an instinctiveaction has for its object any other thing than a fellow- memberof the speciesas, for instance,the natural prey, nestingmaterials and the llke, it is quite clear that the choiceof stimuli, effectedby means of the innate perceptorypattern, can includeonly suchas are inherently characteristicof the object. All that the evolutionof the speciescan do, is to adapt the innate perceptorypatterns as closelyas possibleto the pre- existent sets of key stimuli. The complication,and with it the highest generalimprobability of openingthe lock, that is, releasingthe reaction, cannot exceeda certain predeterminedboundary drawn by the kind and number of potential key-stimuli emanatingfrom the object. This consti- tutes a considerablelimitation to the scopeof object-treatmentby purely instinctivereactions. There is, however,one special case where this limita- tion doesnot exist and this is wheneverthe object of instinctive action is representedby a fellow-memberof the species. In this casenot only the receptorypatterns may be specializedto receivepredetermined stimuli, but the issuingof stimuli may be adaptedto the specialneeds of releasingsocial reactions. Not only the receiving,but also the issuingof stimuli is within the range of influenceof such factors as may govern the evolution of a species. The pike, to put it very crudely,is not in a positionto attach any specialcharacteristics to the minnowwhich mark it as adequateprey and thus prevent the pike from snappingat other things; but the Blue Tit- mousemay, evolutionallyspeaking, attach bright yellow marks to the cornersof the mouth of the young, or a circle of white feathersaround its cloaca,and developtwo correspondinginnate perceptorypatterns, one of Vol.1937 541J LORENZ,The Companion in the Bird's World 249 them releasingthe feedingreaction on the openingof the young bird's mouth, the other one startingthe reactionof carryingaway excretorysacs on the unfoldingof the circlet,if white, at the posteriorend of the baby bird. The innate perceptorypattern neednot, in this case,compromise between simplicity and lock-like general improbability, because the key-stimuli correspondingto it, being sent out by a contrivanceespecially evolved for that end,may be very simpleand yet of an unsurpassablegeneral improba- bility. The layman, on seeingthe glowingcolor designof a male Golden Pheasant'splumage, the striking blue and red pattern of a young Haw- finch'smouth or the curious'dance' movements executed by Mallard drakes on occasionof their communaldisplay, is often heard to remark on the 'incredibility'of thesephenomena, thereby correctly acknowledging their extremegeneral improbability. The meansevolved for the sendingout of key-stimulimay lie in a bodily character,as a specialcolor design or structure,or in an instinctiveaction, suchas posturing,'dance' movements and the like. In most casesthey are to be found in both, that is, in some instinctive acts which display color schemesor structuresthat were evolvedexclusively for this end. All such devicesfor the issuingof releasingstimuli, I havetermed releasers (Au•li•ser), regardlessof whetherthe releasingfactor be optical or acoustical,whether an act, a structure or a color. The essentialgeneral improbability of all releasersis broughtabout in a way characteristicand very similareven where they havebeen evolved in- dependentlyin very differentforms of animal life. Throughoutthe uni- verseit is order that is improbablewhile disorder,chaos, entropy or what- ever one choosesto call it, is what we shouldexpect from the laws of proba- bility. I do not think it too far-fetchedan explanationto supposethat the strikingorderliness and regularitywhich so strongly appeals to our senseof beauty in the coloring,the notes and the display movementsof so many animals,especially of birds,has its sourcein their nature as releasersand in the generaltendency of all releasersto developin the directionof the more improbable. This would alsoexplain the astonishingrhythm that we meet in very nearly everyreleasing action.. The moreor lesspure spectralcolors whichso often appearin colorpatterns functioning as releasers,very prob- ably alsofind their explanationin this way, sincethe reflexionof one wave- lengthamong the wave-mixtureof white light is in itselfrather improbable, so improbableindeed that coloralone may, in somecases, function as a re- leaser. It is for very muchthe samereasons that purecolors are appliedin all opticalsignals used by man. It may very wellbe that the sameexplana- tion holdstrue for the pure and 'beautiful' soundsoccurring in bird song, as it quite certainlydoes for the great numberof rhythmicalmovements of limb and body which play so important a part in most releasingactions 250 LORENZ,TheCompanion inthe Bird'• World œJ•lly[Auk foundin birds. The beggingmovements of manyaltriclal young birds rep- resenta goodexample of this sothat very nearly all the waving movements theoreticallypossible are realizedin differentspecies. In this they actually resembleall the differentwaving movements used in marinesignalling by man. To cite but a few instances:up-and-down movements as begging- releasershave been independently evolved by parrots,gulls, spoonbills and many others;swinging the headsideways is foundin many finches;young EuropeanOrioles (Oriolus, not the icteridebird) performa rotatingmove- ment of the head which has for its axisa line drawn throughthe upper mandible;young warblers of the genusSylvia perform an incrediblyrapid tremblingmovement of the head and neck. All thesemovements and structureswhleh functionas releasers,by their essentialquality of rhythmicregularity, have appealedstrongly from time immemorialto the senseof beauty in man. It is no wonderthat a vast' amount of theorizinghas beenspent on their behalf. Darwin, in his work on 'The Expressionof the Emotionsin Man and Animals',has indeedcome very doseto our conceptionof releasers,yet it is strangethat he attributed so little survival value to their functioningthat he couldnot believeany bodily organwas ever evolvedfor their specialuse. He statesexpressly that not a singlefacial musclewas ever differentiatedin monkeysfor the soleend of expressingsome emotion. Contrary to this, we must state that we have goodreason to believethat not only muscles,but very complicated structuresin somespecies, have beenevolved for the purposeof expressing one singleemotion or, to be more precise,of transmittingthis emotionto a fellow-memberof the species. A goodexample of this is the erectilecrest in all night herons,evolved for the purposeof indicatingpeacefulness, and comparablein its function to the tail-waggingof a dog. Darwin's well- known view about the function of bright colorsand striking structuresof male birds is so far correct,as that there are certainly a considerablenum- ber of speciesin which these releasers,in the male, function in eliciting sexualreactions in the female. Yet it would be an error to generalizethis so far as to concludethat all male coloringand all male structureshave been evolvedfor this functiononly. Noble and Bradley have shownthat in a greatnumber of reptilesthe bright coloringof the male doesnot evoke any sexualresponses in the femalebut hasits chieffunction in intimidating other males. What is true of somereptiles has also been regardedas a generalrule throughoutthe animal kingdom. It was Hingston who pro- poundedthe theorythat all bright coloringencountered in animals,as well as all peculiarstructural designs have beenevolved solely for frightening other animals,particularly, but not exclusively,those of the samespecies. In his opinioneven the axillary hair of man is there to frighten the enemy into non-resistancewhen the aggressordisplays it in raisinghis fists. There Vol.1937 54]J LORENZ,The Companion in the Bird's World 251 is one indubitablefact underlyingboth this extremetheory, and the sexual theory, namely, that structuresand colors are evolved to influence fellow- membersof the species. Yet this absolutely correct thought common to both theorieswas discardedby Wallace, in whoseopinion "the more de- veloped plumage and more vivid colo•s of the males .... may be wholly due to their greater vital energyand thosegeneral laws which lead to such superiordevelopment even in domesticbreeds; but in some casesthe need of protectionby the female, while incubating,may have suppresseda portion of the ornament which she would otherwisehave at- tained." To this we must retort that there are a considerable number of birds in which it is the female and not the male that carries 'ornamentation.' Alsothe very animalsin which the ornamentationof the malesreaches about the highestdevelopment it ever attains,--thc spidersof the familiesAttidae and Lycosidae,--cannotwell be said to have "more vital energy" in the male sex, sincethe male is much smallerthan the female. In thesespiders the Peckhamsas early as 1889, correctlyrecognized the releasingfunction of the male colorsand structureswhich, by elicitingspecific sexual responses in the female,prevent her from reactingto the male in an unadaptiveway by simply devouringhim. Every theorywhich, like that of Wallace,leaves to chancethe particular regularity of detail in all display coloringand structure, commitsa serious error againstthe laws of probability. If we refuseto resortto an altogether metaphysical,vitalistic explanation,which involvesentelechial factors, it is only by regardingthese colored and structuralpatterns as releasersthat we can hope to find somereason for the regularity and ' beauty' of their design. The very word 'ornamentation'smacks of a metaphysicalassump- tion. That this so-calledornamentation does not necessarilyhave some sexualsignificance, is exemplifiedby the many instanceswhere its releasing function is of a purely social nature. Nothing can be more ornamental than the crest of night heronsor the striking patterns of the wingsof many ducks. Yet the former functionsas the organof a peace-makingceremony without any sexualmeaning, while the latter servesas a releaserof the re- actionsof flying in a flock or of holdingsocially together. Our assumptionthat mostof the bright colorsand peculiarstructures, or what are commonlycalled secondary sexual characters, which we meet with in the bird world, actually functionas releasersof socialreactions, receives a convincingconfirmation by the fact that there are hardly any of them that do not take part in an instinctiveceremony, and are quite evidently differentiatedin sucha manneras to be strikingly unfoldedby the coordina- tion of movementscharacteristic of the latter. It is hardly an exaggeration if we state that all strikingcolor designs and bizarrestructural devices, such as elongatedplumes, inflatable pouches,turgescent naked spots and the 252 LORENZ,TheCompanion inthe Bird's World [July[Auk like, find their exclusiveuse in someinstinctive act representinga releasing ceremony. SupposingI shouldshoot, in somefar country,a bird unknown to scienceand that this bird had somewhereon its body a circumscript patch of featherslonger than thoseof the adjoiningregions, I would con- fidently assertthat this particular specieswas in possessionof somereleas- ing ceremonyin which just that patch of elongatedplumes found its use. There are indeedvery few structuraland colordesigns that functionas re- leasersautomatically, without the aid of specialinstinctive acts displaying them. The only exampleof such'automatic releasers'at presentknown to me, is representedby the deviceseliciting the flying-in-pursuitreactions in a great many socialbirds. All of them are color patterns invisiblein the sitting bird, but are suddenlyand strikingly unfoldedat the momentof taking flight. Suchpatterns are often found in the tail, as in caseof the blue-and-yellowdesign of the Shell Parakeet, the white lateral feathersin the tails of tits, shrikes,doves and hostsof other birds. Often the lower back, coveredin the sitting bird by the wings,is white or of somestriking color,as in the Bullfinch,the Bean Gooseand others. A very high differ- entiation of this type of automaticreleaser has beenattained by numerous speciesof the anatide order, of which many bear specialreleasing patterns on their wings. Heinroth has made an extensivestudy of thesereleasers and their function. Lastly, we have to speakof releasingactions, the functionof whichis not dependenton any structuresor colorsevolved for their specialuse. A great many suchactions are familiar to any studentof bird behavior,so that I may confinemyself to just oneexample. The maleRaven has a moststrik- ing threateningattitude, in which all the plumage of the blrd's head is tightly depressed,excepting a sharplydefined area just abovethe eyewhere the feathersare stiffly erectedat right anglesto the surfaceof the skin, so that two earsor hornssuddenly seem to sprouton the angryRaven's front. In a certain ceremonyof courtship,the male Raven stretchesforward his headand neckhorizontally while fluffing the plumageof the wholehead to the utmost and drawing nlctitating membranesover his eyeballs. The picturethus presentedby the Raven is so totally differentin eachof these two attitudesthat it can be readilyunderstood that two answeringreactions or, more exactly,two differentinnate perceptorypatterns, are releasedby them in the fellow-memberof the species. Releasingactions lacking structural or colordesigns are of specialinterest to us in caseswhere closely related species do possesssuch bodily differentia- tions, becausethey often tend to throw somelight upon the phylogenetic developmentof the structureused in the releasingaction. We know of a considerablenumber of instanceswhere a certain releasingceremony is commonto someclosely related species and is executedin all of them in so Vol.1937 54]J LORENZ,The Companion in the Bird' s World 253 preciselysimilar a way that even by studyingthe singleframes of a motion- picturefilm, not the slightestdifference of movementcan be found. Yet it may be that one of thesedifferent species does not possessthe least bit of structuraldifferentiation to emphasizethe optical stimulationissuing from the movementsexecuted in the ceremony,while a closelyrelated specieshas the moststriking colors and structuresthat comeinto play in the very same coordinationof movement. The Mallard (Anas platyrhynchos),as well as the doselyrelated Black Duck (Anas rubripes),and the two speciesAnas supcrciliosaand Anas (Polionetta)poecilorhyncha, have, in the male sex, a communaldisplay ceremony quite alike in all four species. I need not go into the details of movement executedin this reaction, as it sufficesto state that, while A. rubripesand A. superciliosado not possessstructures or colors taking part in this action,A. poccilorhynchahas a bright color pattern on the bill and pure white elbow-feathers,while the brilliant color designof the Mallard drake is well known. In the ceremonyin question,a certain movementof tilting up the posteriorpart of the body, togetherwith a raisingof the elbowsas strikinglydisplays the ringletin the Mallard drake's tail-eover'ts,asit doesthe whiteelbows of the poecilorhynchamale. Simi- larly, in anothermotion of the sameceremony, consisting of a quickdipping of the bill followedimmediately by a stretching-upof head and neck, two charactersdifferent in both speciesare strikingly displayed. In A. poecil- orhynchathis down-and-upmovement is made eminentlyconspicuous by the colors on the bird's bill while in the Mallard the same movement is optically emphasizedby the disappearanceand subsequentsurprising broadeningof the white stripeencircling the drake'sneck. All the conspic- uous charactersof the male of both specieshave one thing in common: they are all so placedon the bird'sbody as to be displayedby thevery same movements,which are exactly alike in both species. The night heronsof the three generaNycticorax, Nyctanassa and Cochleariushave exactly the same greeting or peace-ceremony,the function of which I have likened aboveto that of a friendlydog's tail-wagging. In this ceremonythe upper sideof the headis presentedto the fellow-memberof the specieswho is to be appeased. All of the three generapossess elongated plumes on the top of their headsthat serveto emphasizethe movementsexecuted in the cere- mony and evidently they were especiallyevolved for this particular end. Yet the ceremonyis executedwith successalso by youngbirds, thoughstill lacking the elongatedand conspicuouslycolored plumes. As a third ex- ample of the same principle, I would cite the differentiationof the head- plumagein grebes. In the smallestof Europeanspecies, the Dabchick,a certain ceremonycan be observedin which a most singularshape of the headand neckis broughtabout by depressingthe feathersof the neckup to a sharply definedline just below the lower mandible,upward from which 254 LORENZ,TheCompanion inthe Bird's World [July[Auk the plumageof the headis erected. The sharpboundary between the fluffed and the depressedparts of plumagein the Dabchickis purely functional, no differencein featherlength or colorbeing visible in the quiescentor the deadbird. Yet this line, in mostother grebes,forms the boundarybetween the short, plush-likeplumage of the neck and the area coveredwith elon- gatedand brightlycolored feathers on the bird'shead. In the grebesas well as in the ducks and night heronswe find that an instinctive action, commonto a number of closely related species,has causedsome of them to evolvecolors and structuresserving to make their movementsmore conspicuous. The evolution of structure and color has goneon independentlyin the differentspecies, but a certainsimilarity in its resultsis causedby the commonorigin of all bodily differentiationsfrom the sourceof one instinctivereleasing ceremony. I do not think it either a far-fetchedor an incautioushypothesis to assumethat in all suchcases the releasingceremony is phylogeneticallyolder than the colorsand structures usedby it. Systematiccomparative study of instinctive actionsin birds has shownvery clearly that the phyletic age of instinctivereactions is commonlyas great as that of organs. In very many casesindeed, its special form provedto be a much more conservativecharacter than that of any organ. Whitman and Heinroth were the first to use instinctiveactions as taxonomic charactersand the latter's famous paper on the ethology of Anatidae has found a very impressiveand convincingconfirmation in the work of Poll. This author made an extensivestudy of the fertility of hy- brid birds drawing conclusionstherefrom on the blood-relationshipof the parent species. His resultsconfirmed those attained by Heinroth in lit- erallyevery point in whichthe latter wereat variancewith thosegenerally heldby systematiczoologists of the time. The taxonomicvalue of releasing actions,for very simplereasons, is evengreater than that of otherinstinctive reactions. Sincereleasing actions are what we might call a private signal- ling codeof the species,and as suchpure conventions,they are largelyin- dependentof environmentalinfluences. Their specialdifferentiation is ex- plainablefrom a predominantlyhistorical point of view and not at all on the groundof theirfunction. It isbut a 'convention'that the tail-waggingof a dog expressesand transmitsa peaceablemood to the fellow-memberof the specieswhile a similarmovement in Felidaeis the expressionof quite a differentand nearly contraryemotion. As regardsthe functionof tail-wag- ging,its meaningin the differentanimals could just as well be distributed in the oppositeway. If everwe find an explanationof why the movements are executedjust so and mean just that, this explanationwill be chiefly historical. This is most relevant for the estimation of the taxonomical im- port of releasingactions, because it greatlyreduces the likelihoodof equality of releasingactions ever beingattained in two speciesby convergentevolu- Vol.1937 541J LORENZ,TheCompanion inthe Bird's World 255 tion. Equalityin instinctivereleasing actions always means hornology and it is just this that often makesit possiblefor the studentof instinctive actionsto analyzephylogenetic relations with an exaetltudehardly ever attainedby the morphologist.It wasnecessary to explainat somelength the phylogeneticage as well as the taxonomieimportance of releasing actions,so as to givesome background to my assertionthat the inherited coordinationof movementfunctioning in releasingceremonies is, for the mostpart, phylogeneticallyolder than the structuresand colorsused in them. It has beensaid alreadythat we know very little about the phyletie originof releasingactions themselves, so little indeedthat it seemsworth while to set it forth here in detail. In regardto mostreleasing actions, as in the easeof the tail-waggingof the dog,we are completelyat a losscon- cerningtheir evolutionaryorigin. The only exceptionsto this are a few instanceswhere we can trace the origin of releasingceremonies to instinc- tive actionsof an altogetherdifferent function. It may be regardedas true for all or most instinctivereactions that the very contraryof an 'All- or-nothing-law'governs their performance.Even at the lowestdegrees in intensityof instinctivereaction, the animalwill showits internalprocesses by certainmovements which, to anyonefamiliar with the reactionin ques- tion, are recognizableas its initiatingmovements. If the reactionis very low in intensity,no further movementsbut theseslight beginningsof the action will be observed. As long as the reactionis too low in intensity to carry the animal throughthe wholesequence of expectedmovements, it may breakoff at any point. Thereare all theoreticallypossible gradations betweenthe slightestmovements, hardly noticeable even to onethoroughly familiar with the reactioninitiated by them, up to the completeperform- aneeof the reaction,and attaining the full survivalvalue representingits biologicalend or goal. The frequentoccurrence of a reactionremaining incompleteand inconsequentialbecause of its beingtoo low in intensity,is of the greatesttheoretical importance. It showswith particulardearness that the biologicalend, or survivalvalue of instinctiveaction, must not be confoundedwith the purposepursued by the animalas a subject. This is most evident in easethe action, through sheerlack in intensity, is broken off at the very last moment,falling only just shortof attainingthe end for which it has beenevolved by the species. Nobody,who has ever had oc- casionto observethis phenomenon,can doubt that the animalperforms the movementspertaining to the reactionfor their own sakeand not for the sakeof any biologicaladvantages. As C. O. Whitmanputs it, the animal goesthrough the movementsof instinctivereaction because"it finds it agreeableto do so and disagreeablenot to do so." If, in early spring,we observea night heronsitting in a tree and rather suddenlybreaking into 256 LORENZ,TheCompanion inthe Bird's World •u•ky the action of tuggingat and shakinga near-by twig, only to relapseincon- tinently into its former state of quiescence,it becomesdirectly and quite indubitably evident that these initiating movementsof nest-buildingare executedfor their own sakeand not with a view to their possibleultimate result and alsothat at the presentlow intensity of the reactionsthe bird's eraring to perform them is quite satisfiedby onceor twiceshaking a twig. As a matter of course,these incomplete reactions are of no survivalvalue for the animal or its species,unless we attribute to them the functionof im- proving the action by exercise. Even when we admit the questionable value that exerelsemay or may not have for the performanceof purelyin- stlnetlve reaetlons,we must state that incompleteteaetlon is of no survival valuein regardto the primary biologicalend for whichthe completereaction was evolvedby the species. Yet there is a way by whleheven the slightest initiating movementof an instinetlyeteaetlon may attain an altogether new and biologicallyimportant function. Someinitiating movement,just hinted at in the behaviorof a bird, may tell to the humanobserver in what dlreetionactions of the birds are to be expected. Becausesuch initiating movementsare able to conveythe intentionsof the animal, they have been termed 'intention movements'(Intentionsbewegungen). It is a well- knownfact that this kind of initiating movementin socialanimals is per- feetly 'understood'by the fellow-membersof the species. A few words must be said about this so-calledunderstanding. On seeingone animal be- ginningto performsome action and therebyinciting a fellow-memberof the flock or herd to do likewise,the human observeris prone to assumean 'instinet of imitation' on the part of the secondindividual to explain its behavior. The process,however, on whleh this seemingimitation is really based,very closelyresembles that very dlreet transmittingof moodsfrom one individual to another which we can so often observe in ourselves. The 'contagious'reactions of laughing,yawning and the like, at first sight seem very different from the eomplleatedserial actions transmitted in a similar way in animals,yet, beingalso inherited possessions of the species,they are directlycomparable to them. In socialanimals the transmittingof specific excitation may clearly becomeof very considerablesurvival value. All thosefunctions of eoordinatlngthe behaviorof the individual membersof society which, in the human species,represent one of the chief tasks of speech,are performedeither by releasingceremonies or by the transmitting of specificexcitation which we have just described. There is no sharpboundary llne betweenthe simpletransmitting of ex- eltation, as we may say, "by contagion"and releasingaetlons in the strict senseof the word. Respondingto a certain teaetlonof a fellow-memberof the speciesby reactingin the very sameway is evidently wholly innate. We have, therefore,to assumethe exlsteneeof innate pereeptorypatterns Vol.1937 54]J LORENZ,The Companion in the Bird's World 257 correspondingto certain setsof stimuli emanatingfrom the individual first showingthe reaction,an assumptionwhich we have to make in all cases wherean unconditionedreaction is releasedin sospecific a manner. By the evolutionof this kind of innate perceptorypattern, almostany instinctive reactionin a socialspecies may be appointedto the secondaryand purely socialtask of releasinglike reactionsin the fellow-memberof the flock or herd. Now as a matter of course,this secondaryreleasing function is not dependenton the reactionbeing executed completely and with full inten- sity. Even slight initiating movements,wholly inconsequentialas to the primarymeaning of the reaction,may transmitits specificexcitation to the next bird, and thereby gain an undeniablesurvival value in socialspecies. The correlationof releaserand correspondinginnate perceptorypattern oncebeing established, a newdevelopment of the intention-conveyinginitia- ting movementsmay set in. By specialadaptation to their releasingfunc- tion, they may be graduallyalienated from the reaction,the beginningof whichthey originallyrepresented. This, to the bestof our knowlege,has been the origin of all thoseinstinctive actions which, with someinexacti- rude, we are accustomedto describeas 'symbolic.' It is only in very few casesthat we are able to illustrate this theory by a seriesof 'phylogenic stages'represented in living species,none of which is more convincingthan the following:a reactionwhich, in largebirds especially, always seems to re- quire a considerableamount of 'nervousenergy' or, more precisely,of re- actionalintensity, is that of taking to wing. Everyonefamiliar with bird behavior knows that none of the larger birds ever takes wing without noticeablepreparations, gathering themselves a numberof times in quick successionfor the effort of leapinginto the air and, seeminglyat the very last moment,refraining from doingso. It is only after a considerablerep- etitionof theseinitiating movements that the bird finally takesthe plunge. The importanceof taking to wing synchronouslyis evidentin all species flying in flocks,so that it is no wonderthat theseinitiating movementsof taking the air have attained a specialfunction as releasersin socialbirds. They transmitthe specificexcitation from onebird to the other until, after an interval of mutual stimulation, the thresholdof the wholereaction is so far loweredthat the final stimulationissuing from the first bird actually taking wing, is sure to causeall membersof the flock to follow. In the GreylagGoose this kind of mutual stimulationis so importantthat flocks containingmany membersactually take the air much more often than otherswhich consist of only a few birds. The individualbird sendingout specificstimulation receivesit back 'with interest' from the fellow-members of the flock and it is this 'interest' that increases with their number. An isolatedGreylag spends so little of its time on the wingthat it will become noticeablyfatter and weakerof musclethan geeseliving under the samecon- 258 LORENZ,TheCompanion in the Bird's World [july[Auk ditions but in the societyof their kind. The same phenomenoncauses largeherds of horsesor cattle to be muchmore prone to stampedethan are small numbers of the same animals. The mutual stimulationto .takewing seemsto be especiallyimportant in the anatide order, probablybecause these birds are particularly slow in their resolveto fly up. It is in this orderof birdsthat we find not only the highest differentiationof movementsreleasing this reaction, but also an exceptionallycomplete series of what we may regardas phylogeneticstages in the evolutionaryprogress of this differentiation. In the Mallard, the movementstransmitting the specificexcitation of flying up are easilyand convincinglyrecognizable as beginningsof the actual reactionitself. Any ornithologistnot acquaintedwith their specialimportance as releasers would, on observingthem, at once grasp the bird's intention to fly up. Crouchinglow and ready for the effort of leapingoff, the duck in a brief interval throwsup its head, neckand the anterior part of its body in a way extremelysimilar to, but yet a little differentfrom the actual motion of jumping upward. One might say, the action is a little over-emphasized,as symbolicactions characteristically are. In the GreylagGoose, a speciesin which all socialreactions have attained a higher degreeof specialization than in the Mallard Duck, the movement which indicates the bird's inten- tion to take flight and transmitsthe sameto the other membersof the flock is quite different,consisting in a sidewiseshaking movement of the headand beak alone,the neckbeing all the while held stiffly upright. No oneknow- ing only the movementof the Mallard just described,would ever suspect that this shakingmovement which looks exactly as if the bird wereshaking somethingoff its bill, has the samemeaning. Yet we can,with greatprob- ability of beingcorrect, assert that both theseactions of the two different anatidesare phylogeneticallyhomologous, because we find all mannerof intermediateforms of the sameceremony in otherspecies of this order. The Egyptian Goose,for instance,in its flying-upceremony, stands just mid- way betweenthe Mallard and the Greylag,for its expressionalmovement, as in the latter species,is strictlyconfined to headand beak,but, like that of the Mallard, goes'still' perpendicularlyUpward. Anybodyfamiliar with the movementexpressing the flying-emotionin the Egyptian Goosewould, without any doubt,recognize at first sightthat of the Mallard as well as that of the Greylag. Consideringthe great many intermediatestages in the developmentof movementsexpressing and transmittingflying-emotion in other anatides,I think we are fully justifiedin assuming(a) that the movementsfound in the Mallard and in someother ducksrepresent a more primitiveevolutionary stage when comparedwith thoseof other birds of the sameorder, among which thoseof geeseare to be consideredthe most differential; and (b) that all these ceremonieshave been derived phylo- ¾ol.19a7 54]I Lo•E•rZ,The Companion in the Bird's World 259 geneticallyfrom initiating movementsand representabortive performances of the reactionof taking wing, which must be regardedas one specialease of the many instancesof an instinctiveaction remaining incomplete through lack in intensity. There are a great many 'symbolic'releasing ceremonies for which a very similar originseems extremely probable, though we are not in possessionof sucha completeseries of intermediatelinks as in easeof the flying-upcere- moniesof anatides. I am referring,among some others, to the great many actionsof symbolicnest-building, which play so important a part in the mating of a great numberof birds. Especiallyinteresting is a certain sym- bolic movementof the Gannet, in which the bird sitting on the nest site 'pretends'to take buildingmaterial from the bill of an imaginarymate and to build it into a nestwhich at the time beingdoes not yet existor neednot necessarilyexist. I have seen fit to deal with these symbolicceremonies with somedetail, becausethey actually representthe only easein which we have someinkling as to the evolutionaryorigin of a releasingaction.

JAKOB VON UEXK•0LL'S CONCEPTION OF THE 'I•UMPAN' Apparently it is in birds that the parallel differentiationof releasersand innate perceptorypatterns has reachedthe highest developmentit ever attains in animals. This differentiation is one of the two alternatives which confronta specieswhenever there arisesthe biologicalneed of one indi- vidualconsistently treating another throughout a seriesof separateactions. The first of the two possibleways to securesuch consistency of object-treat- ment is representedby the animal'spossessing insight into the goalor endof its actionsand alsointo the permanentidentity of the objecttoward which its actionsare directed. The secondpossibility lies in the differentiationof a specificinnate perceptorypattern for every singleone of the actionson the part of the agent, and alsoon the part of the patient in receiptof the actions,in the differentiationof a releaserseparately eliciting each of them. The uniting factor whichensures the consistentcooperation of the several instinctiveactions thus doesnot lie in the agentat all, but in the objectof the reactionswhich, by representingthe issuingstation common to all the differentsets of releaslngstimuli, guides every single reaction to its evolu- tionallypredetermined end. The agentas a subjectneed not, evenin the vaguest way, be consciouseither of the survival value of his actions, or of the identity of their neutral object. The object in the agent'sworld need not be representedas that kind of unit in spaceand time whichwe are ac- customedto call a "thing," if only it is sendingout the specificset of stimuli and releaslngevery one of the actionswhich must be executedtoward it. Complicatedand far-fetchedthough this deviousmethod of object-treat- ment may seem to the human mind, it is certain that for anlmals on the 260 LORENZ,TheCompanion inthe Bird's World [JulyI'Auk mental level of birds this has been easierto attain throughevolution than have the mental powersnecessary to effect an object-treatmentof equal complicationand consistenceby insightand purpose. By supplyingabnormal objects which do not sendout the severalsets of releasingstimuli in a way to securethe normalcooperation of reactions,it can be provedexperimentally that in many casesit is the object of the re- actions,the patient and not the agent, who unitesthem with a functional unit of definite survival value. For example,in a femaleMallard whose ducklingsare just hatched,the reaction of defendingthem is releasedex- clusivelyby the call emitted by a ducklingin distress. Other reactionsof taking careof the young,especially that of specificallyallowing the duckling to remain near her insteadof unspecificallychasing it away, as shewould all other small living creatures,are dependentupon certain color patterns on the downof the young. Thesespecific down-patterns vary somewhatin closelyrelated speciesof ducks,while any differencein the distresssignal of the youngis hardly discerniblein the numerousspecies belonging to the anatide order. The result of these circumstances is that a mother duck's defendingreaction can be releasedby the young of very different species while other reactionsnormally directedtoward the same object cannot. A mother Mallard, on hearingthe distresspeeping of a Muscovy duckling, will respondby defendingit exactlyas shewould her ownyoung, but, hav- ing just 'heroicallysaved' it from the handsof the experimentinghuman, she will incontinentlyproceed to attack and even kill it when she seesit trying to mix with her own flockof ducklings. The possibilityof experimentallyseparating functions which ought to be united in the treatment of one particular fellow-memberof the species, certainly precludesthe assumptionthat this fellow-memberof the species appearsas an unit in the bird'sworld. In all caseswhere such a separability of the singleinstinctive acts can be proved,we have reasonto think that the fellow-memberof the speciesis perceivedas a different thing in every experimentallyseparable part-function. The mostpeculiar r61e which the fellow-memberof the speciesthus plays in the agent'sworld, being per- ceived as one thing when representingthe object of one reaction and as a differentone when being that of another,has been termed that of a "Kum- pan" by ProfessorJ. von Uexkiill. The German word, Kumpan, meansa fellowwho is our companionso far as concernsbut oneparticular kind of oc- cupation,such as hunting or drinking (Jagdkumpan,Saufkumpan). It impliesthat no deeperand noblerbonds link us to our fellowin this kind of companionship.The word certainly meetsthe caseexceptionally well, al- thoughit is hardly translatableinto English. The word 'companion'cer- tainly lacksthe detractingimplication which is so essentialfor the wonder- ful way in whichUexkiill's term describesthis lowesttype of animal com- panionship. Vol.1oa? 541I LORENZ,The Companion in the Bird's World 261

INBORN RECEPTORY •)ATTERNS AND •CoNDITIONING'

It would be a fundamental error if, on the ground of sueh observational examplesas that of the mother Mallard whieh I eited above, we shouldas- sumethat all birds in every easeare unableto pereeivea fellow-memberof the speeies,sueh as their mates, their youngor their soeialeompanions as personalitiesremaining identieal in spaeeand time. In faet, as soonas any bird attains the faculty of individually reeognizingone fellow-memberof the speeieswho representsthe mutual objeet of severalof its instinetive aetions,this personalaequaintanee may serveas an uniting faetor. It may fail, however,to do so, and it is interestingto observehow, in one of two eloselyrelated speeies,personal aequaintanee may dominate instinetire aetionsas releasedby innate pereeptorypatterns, while in the other blind instinetiveteaetlon reigns supreme. I am referringto the soeialdefending teaetlonin the Jaekdawand in the Raven. Of both thesespeeies I possessed a very tame individual who was extremelyattaehed to my person,respond- ing to me with all its soeialreaetions. When I tried to elieit the responseof defendinga eompanionby gripping a fellow-memberof the speeiesand showingit thus to eaeh bird, I got exaetly eontrary reaetionsin the two speeies. The Jaekdaw instantly attaeked me with the utmost intensity and peekedvieiously at my hand that was holdingthe other Jaekdaw, al- thoughthe latter was an individualfor whom it had a strongpersonal hate and whom, not very long before,it had so maltreated that I had to sep- arate the birds. The Raven, on the other hand, displayeda mueh lessre- flexdike response. When I graspeda young Raven, whom my tame old male thoroughlydisliked, he instantly flew on my shoulderand from this positiontook advantageof the youngbird's helplessness to peekat it, with- out givingthe eall and the movementseharaeteristie of soeialdefense. But when, sometimelater, his mate, by playingwith my mother'sknitting, got soentangled with it that sheeould not get away, the malefuriously attaeked the servant girl who was trying to extrieate her. On this oeeasionhe dis- played the full teaetlon of soeialdefense whieh, in the Raven, is similar to that of the Jaekdaw. A female Mallard, in defendingher young, aetually showsboth typesof behaviordisplayed by the Daw and the Raven in their defendingreaetions. Soonafter the hatehingof the youngshe just responds blindly to stimuli emanatingfrom them, but six or sevenweeks later, her behaviorappears on a muehhigher mental level. We find now that per- sonalreeognition of her own younghas developedto a degreethat releases her defendingreaetion. Again, in the Museovy Duek, whieh is eonsider- ably more stupid and also lesssoeial than the Mallard, the mother bird keepson defendingblindly all other dueklingsbesides her own throughout the entire summer. 262 LORENZ,TheCompanion inthe Bird's World [•u•ky

From the point of view of Professorvon Uexkiill's "Umweltforschung" this conditioningof innate reactionsto a personallyrecognized object means a radical changein the bird's world. At first, in the world of the mother Mallard, there do not exist young ducklings individually recognizedas thingsand units, but just stimuli, blindly releasingher broodingreaction in one ease,or her defendingreaction in another. Later on, the acquiredabil- ity to recognizeindividually every one of her ducklingsunites the formerly independentsets of stimuliwhich correspond to her severalinnate pereep- tory patterns. Thus by true learningon the part of the motherduck, what first may have appearedin her world as quite differentthings fuse into the unit of one thing. The 'thing to be brooded'and the 'thing to be defended' finally are united in the identity of what, followingvon Uexkiill, we may term a 'child companion'. Although I have followedyon Uexkiill's terminologyin calling such a compoundobject of severalreactions a 'Kumpan', coiningthe expressions of 'ehild-Kumpan','parent-Kumpan', and the like, it mustnot be forgotten that theseterms imply a slightbroadening of the originalconception of the 'Kumpan', as the object releasingbut one single, independantfunction. I feel inclinedto use the Englishword 'companion'as a term for this some- what differentconception of a fellow-memberof the specieswho acts as a releasingobject in regard to severalinstinctive actionsof the agent. Sucha thing as a 'parent-companion',a 'child-companion,' owes its ap- pearancein the bird's world to an acquiringprocess, that is to conditioning in a broad senseof the word. We have already said that it is true associa- tive learningwhich unites the differentsets of releasingstimuli emanating from an individualcompanion. There is, however,a very peculiaracquir- ing process,differing in somerespects from associativelearning, which unites and guidesto their properobject innate reactions directed to fellow-members of the speciesin general,but not to the objectas an individual,as is the ease with the processof personalrecognition just described.This processof ac- quiring the biologically'right' object of social reactionsby conditioning them,not to oneindividual fellow-member of the species,but to the species as such,is sovery peculiarthat Ihave thoughtit necessaryto usea particu- lar wordto describeit. I havecalled it 'Pragung'in German,which I pur- poseto translateinto Englishby the term 'imprinting'.

IMPRINTING It is a fact mostsurprising to the laymanas well as to the zoologistthat most birds do not recognizetheir own species'instinctively', but that by far the greaterpart of their reactions,whose normal object is representedby a fellow-memberof the species,must be conditionedto this object during the individuallife of everybird. If we raisea youngbird in strict isolation Vol.1937 54]J LORENZ,The Companion in the Bird'a World 263 from its kind, we find very often that someor all of its socialreactions can be releasedby other than their normal object. Young birds of most species,when reared in isolation,react to their human keeperin exactly the sameway, in which under natural conditionsthey would respondto fel- low-membersof their species. In itself this phenomenonis in no way sur- prising. We know of a great many reflexactions that can be conditionedto very differentreleasing factors without beingchanged as to their coordina- tion of movement. Also we know that a great many animals,when de- prived of the normal objectof someinstinctive reaction, will respondto a substituteobject, or, to be more precise,will react to other than the usual set of stimuli. In all theseeases the animal will preferthe normalobject as soonas it is proffered,but the bird raisedin isolationrefuses to react to its kind. In most easesexperimentally investigated, the biologicallyright ob- ject, that is, the fellow-memberof the species,was not even acceptedas a substitutefor the abnormal object, acquiredunder the conditionsof ex- periment,when the latter was withdrawn and the bird left severelyalone with other individualsof its species. Heinroth failed to breed hand-reared Great Horned .Owls, Ravens and other birds, for no other reasonthan that these tame individualsresponded sexually to their keepersinstead of to each other. In a very few casesknown, the bird whosesexual reactions were thus directed toward man, finally accepteda fellow-memberof the specieswhich, however, was always regarded as a rather poor substitute for the beloved human and was instantly abandonedwhenever the latter appeared. Portielje, of the AmsterdamZoological Gardens, raised a male of the South American Bittern (Tigrisoma) who, when mature, courted human beings. When a femalewas procured,he first refusedto have any- thing to do with her but acceptedher later when left alonewith her for a considerabletime. The birds then successfullyreared a number of broods, but even then Portielje had to refrain from visiting the birds too often, be- causethe male would, on the appearanceof the former foster-father, in- stantly rush at the female, drive her roughly away from the nest and, turn- ing to his keeper,perform the ceremonyof nest-relief,inviting Portielje to stepinto the nestand incubate! What is very remarkablein all this is that while all the bird'sinstinctive reactions pertaining to reproductionhad been repeatedlyand successfullyperformed with the female and not once had beenconsummated with a humanbeing for their object,they yet stayedir- reversiblyconditioned to the latter in preferenceto the biologicallyproper object. The performanceor better the successfulconsummation of an in- stinctivereaction is evidentlyquite irrelevantfor this peculiarway of ac- quiringits object. The object-acquiringprocess can be completedmonths beforethe actionis executedfor the first time. I oncehad a pair of Grey- lag Geesehatch a MuscovyDuck's eggs. The parent-childrelations in 264 LORENZ,TheCompanion inthe Bird'• World [JulyfAuk this artificial family dissolvedsooner than is normal for any of the two species,owing to somehitches in mutual understandingwhich occurredbe- causethe key and lock of the releasersand innate perceptorypatterns of both speciesdid not fit. From the seventhweek of their life, however,the youngMuscovies had nothingmore to do with their formerfoster-parents nor with any other Greylag Geese,but behavedsocially toward one an- other,as well as towardother members of their speciesas a perfectlynormal Muscovy Duck shoulddo. Ten monthslater the one male bird among theseyoung Muscoviesbegan to display sexualreactions and, to our sur- prise,pursued Greylag Geese instead of MuscovyDucks, striving to copu- late with them, but he madeno distinctionbetween male and femalegeese. Thesefew observationalexamples are sufficientto illustratein a general way the peculiaritiesof the acquiringprocess in question,but I wish to call the reader'sattention more especiallyto the pointsin whichthis process differsfrom what we call associativelearning. (1) The processis confmedto a very defmiteperiod of individuallife, a periodwhich in many casesis of extremelyshort duration; the periodduring which the youngpartridge gets its reactionsof followingthe parent birdsconditioned to their object,lasts literally but a few hours,beginning when the chick is drying off and end- ing beforeit is ableto stand. (2) The process,once accomplished, is totally irreversible,so that from then on, the reactionbehaves exactly like an 'un- conditioned'or purelyinstinctive response. This absoluterigidity is some- thingwe neverfind in behavioracquired by associativelearning, which can be unlearnedor changed,at least to a certainextent. In 1936,I kept a young Greylag isolatedfrom its kind for over a week, so that ! could be sure that its following-reactionwas securelyattached to human beings. I then transferredthis younggoose to the care of a Turkey hen, whomit soonlearned to useas a brooding-Kumpanfor warmth insteadof the elec- tric apparatusit had hitherto favored. The goslingthen followedthe Tur- keyhen, provided that I wasnot in sigh•,and kept this up for a fortnight; but even during that time I had only to walk near the two birds, to cause the goslingto abandonthe hen and follow me. ! did this but three times, to avoid conditioningthe goslingto my personas a leader. When, after two weeks,the goslingbegan to becomemore independent of the warming functionprovided by the Turkey hen, it left her and hungaround our front door, waiting for a human being to emergeand trying to follow it when it did so. Now this gosling,excepting the few necessarytrial runs, eachof whichdid not last more than about a minute, had never actuallyconsum- mated its following-reactionwith a human for its object. On the other hand, for more than two weeks, it had been in constant contact with the Turkey hen; yet its following-reactiondid not becomeconditioned to the Turkey in preferenceto the human. I wouldeven suspect that its instinc- Vol.1937 541.• LORENZ,The Companion in the Bird' s World 265 tive following-reactionwas neverreally releasedby the Turkey at all, and that its followingthe hen was predominantlya purposiveact, directedto the necessityof gettinga warm-upfrom time to time. It neverran directly after the Turkey hen in the intensiveway in whichit wouldfollow me and inwhich Greylaggoslings follow their normalparents, but just kept nearher in a mostcasual and deliberatesort of way, quite differentfrom the normal reaction. Most impressiveis the fact of the irreversibilityof imprinting in suchcases, in which birds becomeconditioned to an inaccessibleobject or to one with which it is physicallyimpossible to performthe reaction. (3) The processof acquiringthe objectof a reactionis in very manycases com- pletedlong before the reactionitself has becomeestablished, as seenin the observationson the Muscovydrake cited above. This offerssome diffi- cultiesto the assumptionthat the acquiringprocess in imprinting is es- sentiallythe sameas in other casesof 'conditioning',especially in associa- tivelearning. To explainthe processin questionas one of associa[ive learning, one would have to assumethat the reaction is, in somerudimen- tary stage,already present at the time whenits objectis irreversiblydeter- mined, an assumptionwhich psychoanalystswould doubtlesswelcome, but about which I have doubts. (4) In the processof imprinting, the in- dividualfrom whom the stimuliwhich influence the conditioningof the re- actionare issuing,does not necessarilyfunction as an objectof this reaction. In many casesit is the object of the youngbird's begging-reactions,or the following-reactions,in short the object of the reactionsdirected to the parent-companion,that irreversiblydetermines the conditionswhich, more than a year later, will releasethe copulatingreactions in the mature bird. This is what we might call a super-individualconditioning to the species and certainlyit is the chiefbiological task of imprintingto establisha sort of consciousnessof species in the youngbird, if we may usethe term 'con- sciousness' in so broad a sense. I wouldnot leavethe subjectof imprintingwithout drawingthe reader's attentionto somevery strikingparallels existing between imprinting and a certainimportant process in the individualdevelopment of organsknown in developmentalmechanics (Entwicklungsmechanik) as indicativedeter- mination. If, at a certain stageof development,cellular material of a frog embryobe transplantedfrom one part of its body to another,the trans- planted cellsowing to influencesemanating from their new environment, are inducedto developin a way fitting to it, andnot in the way they would have developedin their originalposition. This processof beinginfluenced by environment,termed 'induction' by Spemannand his school,is confined to very limited periodsin the ontogeneticdevelopment of the embryo. After the lapseof thisperiod, transplantation of tissueresults in the develop- ment of abnormalmonsters, because any transplantedmaterial will develop 266 LOaSNz,The Companion in the Bird's WotId [AukœJuly in a way exactlyfitting the placeof its provenience.Also, cells transplanted beforethe critical time and afterward replacedin their original situation, will developin harmonywith the part of the embryoin which they are im- planted during the critical period of inductive determination. It is cer- tainly a very suggestivefact that the two chiefcharacteristics of imprinting, in which it differs from associativelearning, namely, in being irreversible and in being strictly confinedto definite phasesof ontogenesis,coincide with those which imprinting has in commonwith induefive determination in Spemann'ssense of the word. Of course,it is a matterof personalopinion how much or howlittle impor- tance one shouldattribute to these differencesfrom associativelearning, and to analogiesto inductivedetermination. My object in drawingthe parallelsto the latter, is to showthat not only the phylogenyof instinctive reaction,but alsoits ontogenymore closely resemble those of an organthan thoseof any of the higherpsychological processes. Also, it is a purelyeon- eeptionaldispute whether imprinting is to be regardedas a specialsort of learning,or as somethingdifferent. The decisionof this questiondepends entirelyupon the contentwe seefit to assignto the conceptionof 'learning.' Imprinting is a processwith a very limited scope,representing an acquiring processoccurring only in birds and determiningbut one object of certain socialreactions. Yet rarity doesnot precludesystematic importance, but I shouldthink it rather unwiseto widen the conceptionof learningby mak- ing it includeimprinting. Suchan increaseof its contentwould bring the conceptionof associativelearning dangerously near to includinginductive determinationas well, and experiencehas shownthat this kind of stretch- ing the boundariesof a conceptionis apt to destroyits value. This is ex- actly what hasalready happened to the conceptionof the reflexand to that of instinctiveaction. Sinceit determinesthe conditionsfor the releasingof a certainreaction, imprinting certainly must be regardedas 'conditioning' in a very broadsense of the word,but I think that English-speakingscien- tists shouldbe glad to possessthis term becauseit describesa conception lessspecific than that of learning.

INTERACTION BETWEEN INNATE PERCEPTORY PATTERNS AND IMPRINTING There is a dual connectionbetween innate perceptorypatterns and the processof imprinting. On one sideit is the normal functionof innate per- ceptorypatterns to guidea reactionto their biologicallyproper object, and we have already mentionedthat setsof stimuli releasingone reactionare, strangeas it may seem,factors inducing the choice-of-objectof an entirely differentaction. For instancethe setsof stimuli which, throughinnate per- ceptorypatterns, release the following-motherreaction in many speciesof precocialbirds, in doingso determine the objectof sexualreactions not dis- Vol.1937 54]J LORENZ,The Companion in the Bird's World 267 played until a year later. On the other hand, suchcharacters of the object whichare not presentin the innate perceptorypatterns, but to which the reactionsmust becomeconditioned in individual life, serveas a uniting factor to the different reactionsonce their acquisitionis accomplished. This uniting of hithertoindependent functions has alreadybeen illustrated by the exampleof the mother Mallard learning to know her ducklingsin- dividually. A very similar uniting functionmust be attributed to the pro- cessof imprintingas well, thoughit never concentratesreactions on one in- dividual object, but only in the speciesas such. If normal imprintingis preventedexperimentally, the socialfunctions of a bird may be distributed betweena considerablenumber of species. Thus, I had a tame Jackdaw, all of whose social reactions were, once for all, directed toward Hooded Crowsas a species,while it wouldcourt humanbeings and respondwith all its caring-for-progenyreactions to a youngJackdaw. Onemight say meta- phorically,that imprintingfills out the spacesleft vacant in the pictureof the properspecies, outlined in the bird'sperceptory world by the data given by innate receptorypatterns, very much as medieval artists in drawing astronomicmaps, accommodatedthe pictures of the heraldic creaturesof the zodiacbetween the predeterminedpoints given by the positionof the stars. Just as the imaginationof suchan artist is given the more freedom the smallerthe numberof starswhich must be accommodatedin the picture, so alsois the scopeof imprintingthe greater,the fewerand the lessspecific are the charactersof the speciesrepresented in innateperceptory patterns. With very many speciesit is practicallyimpossible to directexperimentally the socialreactions of the youngto any but the normalobject, because their innateperceptory patterns are sohighly differentiated as to preventthe suc- cessful'faking' of the correspondingsets of stimuli. This is the casewith most birds of the Limicolae. Especially Curlews (Numenius arcuatus), evenwhen hatched artificially and neverhaving seen any living creaturebut their keeper,cannot be broughtto respondto him with any reactionsbut thoseof escape. Most instructiveare thosecases where it is just possible to imitate releasingstimuli normally emanatingfrom the parent bird, and thus to direct the imprintingof somereaction to a substituteobject. This is the casewith the following-motherreaction of youngMallards. It was long regardedby me as an establishedfact that Mallard ducklingswould not accepttheir humankeeper as a foster-parent,as wouldyoung Greylags, cranesand a number of other birds. I began experimentingby having Mallards hatched by a Muscovy Duck, with the result that they ran away from her as soonas they could, while she continuedincubating on the empty shells. Foster-motherand young failed completelyto respond to eachother. Heinrothhad exactlythe sameexperience when he tried to let youngWood Ducks hatch under a Mallard. On the otherhand ! knew, 268 LORENZ,TheCompanion inthe Bird's World [JIlly[Auk from formerexperience, that youngMallards without any difficultyaccept a spottedor even a white domesticduck as a foster-mother. The optical stimulationemanating from sucha domesticduck was indeedmore differ- ent from that providedby the small,brown motherMallard, than was that of the small and rather shabbyMuscovy with whom I had beenexperiment- ing. The charactersrelevant for the respondingof innate perceptorypat- ternsof the youngmust then, I concluded,be thosecommon to the Mallard and the domestic ducks. These characters evidently were represented chieflyby the call note and by the generaldemeanor, both of which have not been changedmuch in the processof domesticatingthe Mallard. ! decidedto try experimentingon the call note whichit is happilywell within the powersof the human voice to imitate. I took sevenyoung Mallards and while they were drying under the electricheater ! quackedto them my imitation of the mother Mallard's call. As soonas they were able to walk, the ducklingsfollowed me quite asclosely and with quite the samereactional intensity that they would have displayedtoward their real mother. I re- gardit as a confirmationof my preconceivedopinion about the relevanceof the call note, that I could not ceasefrom quackingfor any considerable periodwithout promptlyeliciting the 'lost peeping'note in the ducklings, the responsegiven by all younganatides on havinglost their mother. It was only very much later, probably after much conditioningto other charactersinherent to my person,that they regardedme as their mother- companioneven when ! wassilent. The distribution of function betweenthe innate perceptorypatterns and the acquisitionby imprintingis very differentin differentspecies. We find all possiblegradations between birds like the Curlew whoseinnate percep- tory patterns,corresponding to stimuli emanatingfrom their own species, are sospecialized as to leavehardly any roomfor the acquiringof characters by imprinting,up to birds like the Mallard, in which just one character, but a very 'characteristic'one, representsthe evolutionallypredetermined condition which must be fulfilled to make the object 'fit' into the general pattern of the companion. Finally we know of specieswhose innate per- ceptorypatterns are so reducedas to form but a very roughand sketchy outline of the companionwhich, under the conditionsof an experiment, may becomefilled out by a very differentobject. A goodexample of this is representedby the reactionsof the newly hatchedGreylag Goose who, on comingto the light of day for the first time, looksupward in a markedman- ner and will respondto actually any sound or movementby giving its specificgreeting reaction. If the movingand sound-emittingobject begins to move away from it, the goslingwill instantly start in pursuit and will most stubbornlytry to follow. It has been crediblyreported that boats were followedby Greylaggoslings when the parent birds had beendriven Vol.1937 54]J LORENZ,The Companion in the Bird's World 269 away, very probably at exactly the right moment to elicit the looking-up reactionabove described. I intend to experimenton the generalform and on the limits of size which the object, thus releasingthe young gosling's following-reaction,must possess.The lack of specificitywhich is so re- markablein the gosling'sfirst responsesto form and soundis, to an extent, compensatedby its specificityin time. The looking-upreaction once per- formed,it is extremelydifficult and possibleonly by very forcefulmeans to inducethe goslingto followany otherobject than the onereleasing its very first greetingreaction. In a natural environmentit is extremelyunlikely that a moving and sound-emittingobject other than the parent bird ever encountersthe goslingjust at the critical momentand, evenif someenemy of the speciesshould do so,it doesnot matterwhether the still very helpless goslingruns away from it or towardit. Another exampleof a specieswith wide and little specificinnate per- ceptorypatterns is the ShellParakeet (Melopsittaeus undulatus). I raised in isolationa youngbird of this species,which was taken out of the nestof its parentsat the ageof aboutone week. It wasreared in sucha way as to exposeit to as little stimulation from the keeper'sside as possible. When fledgedit was confinedto a cagein which a celluloidball was so attached that it wouldswing to and fro for a considerabletime if accidentlytouched by the bird. My intention to transfer the sexualand generalsocial re- actionsof this bird to the very simple contrivancementioned, succeeded beyondall expectations.Very soonthe bird kept continuouslynear the celluloidball, edgedclose up to it beforesettling down to rest and began performingthe actionsof socialpreening with the ball for an object. Not- withstandingthe fact that the celluloidball had no feathers,the bird min- utely went throughall the movementsof preeningthe short plumageof anotherbird's head. One mostinteresting item in the behaviorof this bird wasthat evidentlyhe wastreating the celluloidball as the headof a fellow- memberof the species. All actionswhich he performedin connectionwith it were suchas are normallydirected toward the head of anotherparakeet. If the ball was attachedto the bars of the cagein sucha manner that the bird was at liberty to take any positionrelative to it by holdingon to the bars,he wouldalways do so at sucha level that his own headwould be at exactlythe sameheight as the celluloidball. When I attachedit closelyto a horizontalperch, so that it was much lower than the head of the sitting bird, he wouldbe at a losswhat to do with his companionand looked'em- barrassed.' Throwing the ball looselyon the floor of the cageelicited the sameresponse as the death of the only cage-matedoes in ShellParakeets, namely, the bird fell absolutelysilent and sat still in the 'fright-attitude' with feathersdepressed close to the elongatedbody. The only instinctive reactionnot normally addressedto the head of a fellow-memberof the 270 LORENZ,TheCompanion in the Bird's World [July[Auk speciesthat I couldobserve in this isolatedbird, wasthe following:males in courtinga female excitedlyrun up and down a perch in a quick sidewise movementand finally sidlingup to her, they grip in a playful way at her lower back or at the baseof her tail, usingone foot and standing on the other. When my parakeetgrew to mature age and beganmore seriously to court the celluloidball, he would executeexactly the samemovements, but, as he was aiming them in such a way that the ball representedthe female'shead, his thrust-outclaw would grip only vacancybelow the cel- luloid spheredangling from the ceiling of his cage. All this seemsto indi- cate that someof the innate perceptorypatterns of the Shell Parakeet must, in someway, be adaptedto the receivingof formedstimuli represent- ing, at leastin roughoutline, the headand body of a companion.Portielje got analogousresults in his most interestinginvestigation of the European Bittern (Botaurusstellaris). This bird possessesan innatepereeptory pat- tern correspondingto formed stimuli emanatingfrom the enemy who re- leasesits defence-reaction.This pattern alsorepresents the head and body and alsoconsists of a very roughoutline only. The Bittern in defending itself, strikesat the head of the enemy, not at its eyesas was formerly be- lieved. Portielje could showthat the bird in this reactionwas guidedby the fact that a smallershape representing the headof the enemy,was situ- ated just abovea largerone, representing the body. Two disksof card- board were sufficientto meet the requirementsof this simple innate per- eeptory pattern. The observationsof the Budgerigar! may serve us as an example, of how, under the abnormalconditions of captivity, the roughsketch repre- sentingthe companion,outlined in the bird's pereeptoryworld by innate patterns,may be 'filledout' in an incompleteway by acceptingan object onlypartially corresponding to the innatesketch, so that someparts of the latter are left vacant, as the spacereserved for the eompanion'sbody was left vacant by the celluloidball. Instinctivereactions directed to such vacantspaces, as the oneof grippingthe female'stail in the ShellParakeet, very oftenprove their fundamentalindependence by attachingthemselves to an independentobject, to a separate'Kumpan.' The Jackdawmen- tionedabove supplies a goodexample of this. More transparentperhaps and susceptibleof a simplerexplanation is the behaviorof man-reared youngGreylag Geese. This species,as mentionedbefore, has particularly wide and sketchyinnate pereeptoryinlets which,by reasonof their very widenessenable the experimentinghuman to stepin and supplyall the needs a Greylaghas for companionship,much more completelythan is possible

• The Shell Parakeet, a corruption of Betcherrygah, native Australian for 'goodparrot,' prob- ably introduced into Great Britaiu at the time of the Crystal Palace Exhibition in Hyde Park, London, in 1851, when it soonbecame a favorite cage bird. Vol.1937 ,541J LORENZ,The Companion in the Bird's World 271 with any other speciesof bird hitherto investigated. There is, however, one reactionof the Greylagwhich constitutesan interestingexception to this rule by having a very definiteand highly differentiated'lock,' an in- nate perceptorypattern correspondingto the one and only structural re- leaserever evolvedby the species. I am referringto the reactionof flying after anothermember of the flock whichis dependentupon a strikingand beautifulgrey-and-black color pattern on the wingsof the precedingbird. This colorpattern which representsone of the prettiestexamples of an 'automaticreleaser', is quite invisiblein the sittingbird, as all the partsof the wingsthen opento view are coloredin the same'protective' gray-brown as the restof the bird. It is only the plumageof the propatagialmembrane, whichdisappears beneath the shoulderplumage when the wing is folded, and furthermorethe baseof the remigesof the hand and their primary coverts,then coveredby the dull plumageof the arm, whichdisplay a light silverygray appearingalmost white whenseen at a distance. The sudden transformationof a grayish-brownbird into one predominantlyblack and white at the momentof taking to wing, is very impressive,even for the humanobserver, and mostprobably is essentialfor the following-orflocking- togetherreaction of the fellow-memberof the species. This highly differ- entiatedway of releasingthe reactionmakes it impossiblefor the human companionto elicit it in the isolatedgoose, which results in an apparently inconsistentbehavior on the part of sucha bird. The younggoose seems to undergoa completemental transformation at the momentof taking to wing. While being completelyindifferent to any fellow-memberof the speciesand mostintensely and affectionatelyattached to its keeperas long as it stayson the groundor on the water, it will suddenlyand surprisingly ceaseto respondto the humanin any way whateverat the momentit takes to wing in pursuitof anotherGreylag. My Greylagsused to followme on my swimmingtours in the Danubeas a dogwould, and in walkingand swim- ming, the leader-companionof a Greylag Goosereleases its following-reac- tion without the useof speciallydifferentiated bodily characters;therefore, whenwalking or swimmingin front of the bird, it is able to supplythe nec- essarystimulation, but leavesunfulfilled the conditionsreleasing the fol- lowing-reactionof the bird on the wing, and leavesvacant a placewhich may be taken by any objectwhich supplies specific stimulation. The releasingof oneseparate ,social reaction independently of all others whichnormally cooperate with it in the sociallife of the species,as exempli- fied in theseobservations, is indeeda very commonoccurrence in birds rearedby man. Abnormalthough these phenomena undoubtedly are, they yet tend to throwsome light on the normalprocesses upon which social life is built up in birds. As in the caseof the reflexprocesses, our knowledgeof what is the normal sequenceof reactionsis almostentirely foundedon a carefulanalysis of abnormalreactions produced experimentally. 272 -OREz,The Companion inthe Bird's World [July[Auk

SPECIALCONSIDERATION OF THE i•ARENT-COMPANION• CHILD-COMPANION AND THE LIKE

Five chaptersof my work deal with five differentcases in whichone defi- nite fellow-memberof the speciesplays an importantpart in the bird's world,'as the parent-companion, thechild-companion, thesex-companion, the sociM-companionand finally the brother-and-sister-companion.These chaptersare madeup almostexclusively of examplesof all thesephenomena whichI havealready mentioned in the precedingsections of this abstract. Observationalevidence is difficult to condense,so perhaps! shouldcon- cludethis paperwith a list of the contentsof the 'companion'chapters. Eachof thembegins with a shortaccount of the innateperceptory patterns whichrepresent the innatesketch or outlineof the companionin the bird's world. Next comes'personal recognition of the companion',and after this I considerall the separatereactions released by the particularcompanion in question. Finally, I deal with the independenceand experimental separabilityof the differentreleasing functions performed by the companion. To enumeratea little morein detail the releasingfunctions of eachtype of companion:the releasingfunctions of the parent-companion:(a) the re- leasingof beggingreactions, (b) of food-takingreactions, (c) of following reactions,(d) of responsesto warning,and (e) of the responsesto beingde- fendedby the parent. The releasingfunctions of the child-companion:(a) the releasesof feed- ing reactions,(b) of nest-cleaningreactions, (c) of leadingreactions, (d) of defencereactions, (e) of otherlife-saving reactions, (f) of broodingreactions, and (g) the responsesto the disappearanceof the child-companion. The releasingfunctions of the sex-companion:(a) the interlockingof instinctiveactions pertaining to mating,(b) the synchronizationof mating cycles,(c) the interlockingof instinctiveactions in nest-building,(d) the releasingreactions leading to copulation,(e) the interlockingof instinctive actionspertaining to nest-relief,and (f) the responsesto the disappearance of the sex-companion. The releasingfunctions of the socialcompanion: (a) inducingof reaction by 'contagion'and so-cMledimitation, (b) the releasingof followingre- actions,(c) of responsesto warning,(d) of socialattack reactions,(e) the interlockingof instinctiveactions pertaining to the pecking-orderand to nest-protection,and (f) responsesto the disappearanceof the socialcom- panion. The releasingfunctions of the brother-and-sister-companion:(a) the releasingof followingreactions, (b) inducingof reactionsby 'contagion', (c) the releasingof responsesto warning, (d) of communaldefence re- actions,(e) the responsesto the diasppearanceof the brother-and-sister- ¾ol.1937 54]] LORENZ,The Companion in the Bird's World 273 companion;and (f) the interlockingof instinctiveactions pertaining to the pecking-order. The paper,which I have attemptedto summarize,concludes with a brief discussionof all the instancesin which my results appear to be at issue with opinionsgenerally held in comparativepsychology. Altenberg,Post Greifenstein, Nieder-Oesterreich,Austria